Turquoise Energy News #134
covering July 2019 (Posted August 5th 2019)
Lawnhill BC Canada
by Craig Carmichael

www.TurquoiseEnergy.com = www.ElectricCaik.com = www.ElectricHubcap.com

Batteries: "Everlasting" Zinc Electrodes! √
  (See Month in Brief, Electricity Storage)

Month In "Brief" (Project Summaries etc.)
 - All the Answers at Once! - Picking Up Steam - CNC Table - More Things - The Covered Wagon (solar E-Bike) - Donation

In Passing (Miscellaneous topics, editorial comments & opinionated rants)
  - Stop Colds Fast with OJ - Homegrown, Homemade Quinoa Flour Bread: Mmm! - Another Hair Loss Preventative - Small Thots - ESD

- Detailed Project Reports -
Electric Transport - Electric Hubcap Motor Systems
* Ground Effect Vehicle ("GEV") - only slight progress as usual - wing shape and stability
   - Testing Ducted Fans, Radio and Motor Controllers - Revised Ducted Fan positions: Outboard, in line with wing
* More Unipolar BLDC Motor and Motor Controller Ideas - ? Turns per coil - Sizing
   - Permanent Magnet Assist, Magnet Fitting - Tuned Circuit Coil Drive - Rotor Design - Figuring Out Activation Sequences
* More Variable Torque Converter Ideas

Other "Green" Electric Equipment Projects
* 36 Volt DC Wiring & Infrastructure
* Some Handheld Bandsaw Sawmill Notes - Update

Electricity Generation
* My Solar Power System: - "Floating" Panel Voltage Grid Tie - Combining More Panels on One Grid Tie or Charge Controller? - east-west facing panels - Monthly Solar Production log et cetera

Electricity Storage - * Dried Out(?) NiMH Dry Cells
Turquoise Battery Project (Mn-Zn or Ni-Zn in Potassium Hydroxide electrolyte ?)

* Non Dendrite-forming Nickel-Zinc Cells:
 - Cost comparison and estimate - Preparing a Zinc Alkaline Electrode - Electroplating - Egg Albumen Coating - Assembly and testing - Antimony Sulfide or Zirconium Silicate? - 14th - Duck Eggs - Electroplating Rack - A Positive Electrode - Parchment Paper Separator - 3D Printed Positive Electrode Shells - Ethaline Deep Eutectic Solvent (DES) - Ethaline Electrode Jell for Zinc? (nope) - Agar Jell (seems to work, including in KOH, but low currents) - Getting Current - Osmium Doped Film Again + Agar? - Apparently Successful and Practical 'Fuzzy' Electroplated, Osmium Filmed and Agar Jelled Zinc Negative Electrode! - Microcontroller for Cycle Tests and Cycling Cells? - How Many Ni-Zn Cells for 12 Volt Battery? (Seven.) - Positive Electrodes and 3D Printers

July in Brief

All the Answers at Once!

   Omitting the previously started original wave power project (from 2006-2007) which I soon ceased to pursue, the inspirations of June and July bring the original goals of Turquoise Energy in January 2008 right back and into clear focus:

1. A highly efficient electric car motor, mounted on the outside of a car wheel.
2. An efficient, friendly, repairable and reliable motor controller for said motors.
3. Better, cheaper batteries for electric cars.

   And when I finally decided a motor small enough to fit directly on a car wheel at 1-to-1 ratio (at least any in line with how I was making them) couldn't have the torque to drive it:

4. An efficient variable torque converter to couple the motor to a wheel - spring 2009.

   I made great axial flux BLDC motors by 2012 or 2013. My BLDC motor controllers work but usually end up with blown transistors when stressed too hard. (A 300 amp Kellycontroller BLDC controller finally put the Electric Hubcap motor to work and showed its power.) A lasting battery cell was elusive until this June, and I stumbled around in the dark trying various things mechanical and magnetic to make a torque converter. (I finally had all the right parts for a mechanical one, but I wasn't putting them together in a way that worked. The "Magnetic Impulse" torque converter idea should work, but it needs too many magnets and rotors with too much copper to be practical in a car - and it would probably make too much vibration.)

   In 2016 I came up with or ran across concepts for "unipolar" motors where each coil is activated in only one magnetic direction instead of opposite directions at different points in the rotation. From my point of view, the biggest benefits were simplicity and ruggedness of the high power electronic circuits of the motor controller. A unipolar controller couldn't short straight from B+ to B- if a transistor turned on for an instant at the wrong time, and so the chance it would blow transistors in heavy use was greatly reduced.
   For reluctance motors a 3-phase unipolar controller would work because the rotor iron is attracted to either polarity. But (limiting my imagination to 3-phase motor controllers) I couldn't figure out a "unipolar BLDC" configuration that would work right and well. Reluctance motors were however a new type for me, very different from the ultra-efficient BLDC motors I already knew how to build well. I did get a sample reluctance motor (the "ARM" axial reluctance motor) running, but it needed a lot of changes to be effective, and a reluctance motor would probably never have the "ultra-efficiency" of my axial flux BLDC motors (~95%).
   An additional potential advantage of the unipolar motor is that it is amenable to a type of permanent magnet assist that adds the flux of a permanent [super]magnet, seemingly for free "two for the [energy] price of one", to that of the coil electromagnet. I would very much like to try this out and see what actually happens.

   Suddenly in July 2019 I have answers for what seem to be excellent batteries, for rather simple and effective variable torque converters, and for unipolar BLDC motors that will allow me to make reliable unipolar motor controllers that won't blow up. Since I already build great BLDC motors and simply need to create a (4/3 larger) variant of the Electric Hubcap motor, this means that I quite suddenly have answers to all the thorny design questions I've beat my head against for so many years, in all my main projects.

   If the time I have lavished on these projects over the years had been spent building them in the right or optimal way from the start, I'd have had them all done long ago. But with the inspirations finally coming now and having further valuable projects on the go as well, I wonder when and even if I'll ever be able to build these marvelous concepts and put together the fabulous, ultra-efficient, longer range electric car that I now, quite abruptly, see just how to create. They will each take a lot of work.
   Of course things progress from a beginning, and for new technology ideas - inventions - from something of a blank slate, as puzzles with pieces missing. If it wasn't so, the invention would already exist. I learned as I went. The designs and principles I now present seem pretty simple (that is to say, easily grasped by those in the relevant fields), but if it seems like I was really slow off the mark to take so many years stumbling around to get to such simple places, it must also be considered that no one else has come up with any of these things, from any past time to the present day. If they had, the thing would be out there and I could have just copied the work -- or perhaps have just gone into a store and bought finished products. Instead here are some new concepts and designs that never existed before, for anyone to use to create working models. (I acknowledge of course all those things developed by previous inventors that I can just go into a store and buy, or order on line, and which my own new work is built on top of!)

Picking up Steam

   In June I was lamenting that I didn't have time for projects. In July, with so many answers to imponderables suddenly there in front of me, I made time. The garden was mostly planted and I did the minimum watering and almost no weeding. I cut way down on milling wood. I skipped naps. I skipped trips to town, where it seemed sometimes the only purpose to the trip was to check the mail anyway (but I usually buy a couple of pieces of fresh fruit). Seeing real progress I did battery experiments and developments one after another instead of weeks apart.

CNC Table

   From a far corner in the shop I brought in the new Geckodrive, the old CNC table controller box and the CNC table. If I was to build a new design of motor, this had to work - at least the router to make the molds. I put them in my 3D printer alcove (clean, uncluttered... was) and started putting things together for the CNC router and CNC plasma cutter. (15th, 16th) I disassembled the stepper motor/limit switch assemblies to trace out the connections to the DE-9 connectors. (I tried months ago to get the DE-9 pinouts from "Techno", the company that made it, to no avail. Too old, I guess. But they were easily traced once opened.) The CNC table was one "hi-tech" item that wasn't obsolete after 30 years. Then I made up cables to connect them to the "Gecko", which also used DE-9 connectors but with a completely different pinout. Having got that far, on the 17th I ripped the old electronics out of the box, keeping just the power supply, and installed the Geckodrive on the back panel. I started in on the instructions, testing that the unit worked properly - the right lights came on, and the CNC table stepper motors had holding torque.

The Geckodrive mounted on the back of the original controller box for the CNC table.
In 1990 the whole box was choc full of electronics to do the same job. I wasn't confident
of getting it to work as I couldn't get info on it, and anyway it only had two stepper
motor drives and I would need at least three. (The Gecko has four.) Now I'm just
using the box's power supply with its on-off switch and emergency stop button.
(and the fan... that loud fan has got to go!)

Front View

   Then I reached the part about "using your software" to step the motors and found that LinuxCNC was a 1.1 gigabyte download. Oops. With my internet, that put an end to that for the time being. On the 23rd I went to town to the library to use the WIFI and try to download some short version, but I spent much of that day and the next in frustration unable to get anything I'd downloaded to work.
   Later I phoned a friend in Victoria, Jim Lawrence. He downloaded the 1.1 gigabyte disk image (in a minute and a half!) and sent it to me on a DVD, which arrived August 3rd. I look forward to getting the software, the CNC table and (at least) the CNC router to work. Getting the plasma cutter going (to do steel motor rotors and make equipment for recycling plastic) will need additional work - a separate project. (Oh boy, another project!)

More Things

   In the meantime I did battery experiments. Doing the batteries is a given. The osmium doped film and agar seems able to make 'everlasting' zinc electrodes, so in principle I seem to have everything needed for nickel-zinc or manganese-zinc "prismatic cell" batteries, with zinc sheet and plastic shelled Ni/Mn electrodes simply stacked in a row filling the cell end to end. These would be "liquid filled" but would require only a small amount of liquid.

L: Sheet Zinc Electrode with "fluffy" zinc electroplating giving it high surface area per volume for high amp-hours. ("full size")
R: 3D printed Porous Plastic Electrode Shell to encase positive oxide electrodes. ("test size" shell/electrode bits)

Test cell in use. It didn't look much different
from test "A" to "Z", but the things inside
definitely got better!

   By the end of the month I had the makings of what should be an "everlasting" zinc electrode with good performance, and "porous" plastic shells for positive electrodes. The detailed writeup on all this under Electricity Storage is about 1/2 of this newsletter, which may be the longest one yet.

   Of special note in this connection, Jim Harrington sent me a link to a Canadian battery design challenge. It looks much more promising than NRC's IRAP offerings. Instead of asking you to supply over half of the money yourself in order to qualify for funding with many strings attached (why bother?), it was sponsored by multiple organizations who seemed more interested in getting better batteries developed and made in Canada than in delving into every petty detail about exactly what the future is expected to hold. (The future is always changing as it unfolds, as even the last couple of months of battery development in these newsletters abundantly discloses.) There is more than one "prize" for promising sounding projects. If this project doesn't sound promising at this point, I don't know what would.
   The "up to" amount of each prize, apparently intended for research, should be enough to start some small Everlasting Ni-Zn battery factory -- and up the pace of more of my other research as well. The grand prize of a million dollars to be awarded later would be enough to expand on that factory and perhaps move it to a more central location than "the edge of the world".

   And I worked out some of the basic details for a unipolar BLDC motor for an EV (etc): configuration, size, PM assist, coil sequencing and rotor magnet sensing, and a new "resonant drive" system to replace PWM for driving the coils. (More is under Electric Transport.)

   And I thought I might manage to do the variable torque converter in less than geologic time if I could contrive to use the same basic "experimental transmission box" for the Chevy Sprint that has housed all such experiments, and fit everything in using all the existing parts including the 'forklift' drive motor. I cut a new steel plate to mount the motor on. I took out the motor and transmission to see what could be fitted. The motor wouldn't fit the way I had hoped, but I figured out another configuration (the second of two possibilities that looked good to me) that should work equally or almost as well. I'll do my very best to make the whole thing sturdy enough not to bust or slip instead of moving the car. (In fact, for once I think I'll do a drawing first, to make sure everything actually fits exactly where I vaguely expect it to, and to know exactly how to make each part.) (more under Electric Transport)

[pitcher]   I didn't need any special new insights to continue (at glacial pace) the radio controlled model for the ground effect vehicle. But it did occur to me that if the ducted fans were in-line with the wing, that they would have no tendency to push the nose down or need to be angled to where they would be less efficient.    So I re-made a couple of longer, 'streamlined' spars so I could mount them solidly at the sides. They should still be above the water there. (more under Electric Transport)

   And I continued to monitor and record the performance of my electric solar panel system. It has been cutting 1/3 or more off my electricity bill (which is pretty good considering it includes the cost of driving my electric car all over the place). Even with a month of nothing but clouds and overcast it put out about 55% of what it would have in full sunlight, which makes such installations worthwhile at today's solar equipment prices. I look forward sometime to being able to store a couple of days worth of electricity in the new batteries.

   By the end of the month I had some lengthy project reports and a ton of pictures on my cellphone (camera) all to be edited and put together, and I could see this newsletter wasn't going to be out in a day or two. And there seemed to be one thing or another taking up much of each of the first few days in August, too.

The Covered Wagon: E-Bike

   My friend Tom Sawyer sent me pictures of a three wheeled solar electric cycle he ran across. He said it was made by a homeless person in Vancouver. If so the guy has a lot of talent, imagination, drive and for sure someplace to work. It would be long enough to sleep in (it seems to be insulated), and assuming a reasonable battery capacity the 600 watts of solar panels that unfold would allow considerable daily travel distance with (two?) typical e-bike motors of 500+ watts, especially as it would continue to charge at 300 watts while riding. I imagine as long as one can afford to eat, one could get to most anywhere, slowly but almost for free, with this unit. (Tom's pinstriped electric Nissan Leaf with "Tesla" logos is in the background.)



   Someone sent me a very generous donation in July. I've been meaning to ask if he wanted his name and or the amount in the newsletter, but somehow I didn't get to it. He said he has been reading TE News for years with interest, and has made use of the ideas of ilmenite for magnetic circuits and molded polypropylene-epoxy composite material. Thank you again, it is much appreciated!

In Passing
(Miscellaneous topics, editorial comments & opinionated rants)

Stop Colds Fast

   Ever wake up and feel that unpleasant sensation at the back of your throat that says you're getting a cold? I used to feel helpless, knowing that it was taking hold and that in a few hours it would spread and I'd be stuck with it, perhaps for weeks, until it decided to leave.
   Then I found a secret weapon: orange juice. by itself, with no other food or drink. I woke up and had that nasty "cold starting" feeling in the early morning hours of the 27th. I thought, "Uh-oh! Surely this has progressed too far to be stopped?" But I got up and rummaged through the freezer and finally found a small container of orange juice I had frozen, probably 3 years ago. (At the bottom, at the back. I knew it was there somewhere! I split big containers of OJ into small ones so I can use some and save the rest. Somehow I had stopped drinking it as a regular drink, so the last couple had been just sitting in the freezer. BTW this freezer is around -25° C -- that keeps things much better for much longer than say -15°.)
   I put the bottle into a larger container of hot tap water so some would thaw quickly. I took that into my bedroom and managed to get a couple of swigs before I lay down, and another couple before I got to sleep again. Drinking too much too quickly isn't good anyway. And I took the juice bottle out of the warm water. In the morning there was no sign of a cold. I poured the thawed juice into a glass for breakfast, but most of it was still frozen and I put it back in the freezer for next time. When there's no more, I'll buy another bottle and freeze it in smaller containers again, just for this purpose.

   Something related that many aren't aware of (but may be subconsciously aware of through experience of colds and flus): your digestive system wants to do somewhat different things with alkali reacting foods than with acid reacting foods. Citrus fruits don't go well with starches, and they especially don't go well with milk products. The digestive system can't optimally handle both at once, and the environment is more conducive to viruses. Drinking fruit juice, especially orange juice, with milk (eg, with cereal) is a ticket to getting colds and flus. They're all good, but not together. Juice and milk should be spaced hours apart.

[The source for this information is (if I remember the title right) "The Edgar Cayce Handbook for Health Through Drugless Therapy" by Dr. Harold J. Reilly (or was it Riley?) who Cayce referred a lot a patients to. I read it decades ago. I can vouch that most any of the things I've tried in it work. I seem to have lost it in my house move.]

Quinoa Bread... Mmm!

   On the one hand I had been baking my own whole wheat bread in a bread machine. On the other, in spite of finding a better way it was a bit heavy and dense. ("knead dough" cycle and then "bake only" cycle works better than "whole wheat bread cycle", which makes it extra heavy and low rising.) Then on the other hand, I had a big sack of quinoa I bought at a health food store in Victoria over two years ago when I lived there. And again on the other hand, I also had a couple of liters of quinoa that I grew myself last year. (Mentioned in TE News #...123-125?)

   I had used quinoa in bread. But it was grainy and not much of an addition. Why wasn't there quinoa flour?

   There's a lot of bitter stuff, little bits of chaff, in with the home-grown quinoa. That's how it protects itself from being eaten by bugs. (The store bought is all nicely cleaned.) It was said to just rinse it in a collander under the tap, but that just didn't work. I put some in a small mixing bowl, got a hair dryer, took it out on the porch, and blew the hair dryer into the bowl. Shake the bowl as you blow. When you have it right, most of the bitter and lighter chuff comes out. Bingo!
   Some is in clumps that still have some seeds in them. To get those, rub a bunch of it between your palms, dropping it into the mixing bowl. That breaks it up and then the chaff can be blown out without the seeds.

   So I took 1/2 cup of that and ground it in the coffee grinder. Then I did the same with 1/2 cup of the store-bought quinoa. The store bought ground finer, easier. The home-grown was somewhat of "quinoa meal" while the store bought was definitely "quinoa flour". (They may have been different varieties?)

   Then I made bread as usual except using 3 cups of whole wheat flour and 1 cup of quinoa flour instead of 4 of the wheat flour. It rose to the lid of the breadmaker. It was delicious! The texture was ideal and the flavor was great! 100% whole wheat flour is now "out". 25% quinoa flour is "in". Now I'll seriously be using up the quinoa instead of just a bit now and then. I may actually finish it all some day. ...I bet it would be great in pie crust! - a bit of a "graham wafer"-like effect? (Yup, it's quite nice. Mmm, delicious pie from thimbleberries picked along the edge of the highway! You wouldn't want to do that near a very busy highway, but this is Haida Gwaii.)

   It's great that 1/8 of the makings (potentially 1/4) of the best bread came from a small patch in my own garden. It seems pretty impractical to grow wheat, although I did grow a little once along with a few oats, and this summer I have a small patch of barley. As well as corn, which just might produce a few palatable ears. Some "volunteer" quinoas in the barley patch (last year's quinoa patch) are growing well. (They're not in a greenhouse, but the main garden is along the south wall of the house, so it's a bit warmer and calmer than out in the open. Growing a garden here this cloudy, cool July reminded me I'm not so far south of Alaska.)

Another Hair Loss Preventative

   Using a brush on the scalp instead of a comb didn't seem to be the whole answer. At the top at the back, it gradually started getting thinner again anyway, and then I realized I had lost some all along the top, too. Hardly noticeable without looking carefully - I'm sure baldness creeps up on people that way. I saw some "Rogaine" in the drug store and on impulse I bought some. There was some patented ingredient, but I looked and it seemed the main constituents were ethyl alcohol and propylene glycol. I thought, "Wow! They've sure found a good way to sell alcohol and antifreeze for a really high price!"
   So I started looking on the web, and some clues seemed to indicate that putting ethyl alcohol on your head is a very good way to help prevent hair loss. Not just a little bit on the thinning patch as Rogaine recommends, but over the whole scalp. In my enthusiasm and experimentative nature I started spraying ethyl rubbing alcohol on daily. (The squirter from the Rogaine fit on the alcohol bottle - the one real benefit to the Rogaine besides triggering the search for information.) But that must have been seriously overdosing - it's how I gave myself shingles.
   Maybe once or twice a month may be about right. The ostensible reason for it is that it seems a primary cause of hair loss is the hair follicles going "dormant". The reason for that is lack of blood flow in the scalp. (That's why the stimulation of a brush helps.) The reason for the poor blood flow is more or less continual vaso-constriction because of cold - the sort of cool conditions many of us in northern climes get used to all winter. (Hmm, how prevalent is baldness in the tropics?)
   Most people don't wear hats any more. Is your house warm? Do you go around in the cool to cold all day? Is your bedroom cold at night too? What part of your head is against the warm pillow? What part goes bald?
   Alcohol is a vaso-dilator when drunk. Apparently it dilates the blood vessels as well or better if sprayed directly on the whole scalp.

   But I have a suspicion that there may be more to it. Perhaps some bacteria or something that get growing in the cold scalp that are part of why the hair follicles quit working. Not the shingles (which bacteria never go away entirely once you've had chicken pox), but something that persists along those lines and spreads over the poor circulation areas of the scalp. If that is the case, killing them off once a month with alcohol may go much farther than expected to prevent hair loss in the first place whether or not it restores lost hair. But that's just my present theory. It may simply be periodic vaso-dilation.

   I can see one can't expect miracles once hair has been lost. Hair isn't lost in day, either. But maybe, eventually, it will get thicker again? (<= wishful thinking alert!)

Small Thots

* The whole world is fully populated (and then some). Very large families have had their purpose, but may now be considered immoral. Even 3 child families will continue to grow the population, albeit at a much slower rate than with so many couples, so often in poor circumstances, having 5 or more children.
   The best and most effective aid more developed nations could send to the "third world" is free birth control products, freely proffered but with the notice "We can't take any more of your ever-growing crowds of desperate refugees. However many kids you have, you'll have to house them and feed them yourselves." There is now, too little and much too late to stave off the coming disaster, the beginnings of a backlash against large families and in favor of birth control in the world's most crowded countries such as Bangladesh.

* How the financial system is ruining everybody. Charles Hugh Smith explained it all on Zerohedge.com one day in July. Central bank prints money galore and lends it almost 'free' to other financial institutions. They use it to speculate in investment assets. One of these is real estate.
   So say a mall cost 1 million dollars. The stores in it pay 1000 $/month rent. (just for nice round numbers.) Now an entity comes along and buys it for 3 million - the free money for some is pushing all real estate prices up. Now they have to charge the stores 3000 $/m because of their large mortgage. The stores have very little room to raise prices in the declining market where the consumer isn't getting proportionate increases in salary/income. Stores go out of business and become empty. If the mall can't charge the stores 3000, then the whole mall goes under. Here we have the makings of the retail apocalypse which has been underway for a decade.

   Smith explains it much better:

Main Street Melancholy - Small Business On The Precipice   --  Authored by Charles Hugh Smith via OfTwoMinds blog  --  Small businesses on the precipice need only one small shove to go over the edge, and there won't be replacements filling the fast-multiplying empty storefronts.

* In the early to mid 1970s one could buy "Duralex" ("Made in France") unbreakable glass drinking glasses. (I think they were associated with the new "Corelle" by Corning dishware that is so much better than the old thick, heavy, easily broken ceramic dishes. At least, they became available at about the same time.) As with the Corelle dishes they were very popular and everyone assumed they were here to stay.
   I had (have) an 8 oz. Duralex glass that I somehow got from my Mom. When I bought my house in 1977 I looked for some more in the stores. I have a vague recollection of at least one person saying then that they had looked all over but hadn't been able to find any. I wanted the 8 oz glasses, but all I could find was a set of four 6 oz. ones. I bought those "for now"... but I have never again seen any Duralex glasses in any store. They were the last ones.
   You already know I'm the suspicious type... I think someone of means deliberately got rid of them (one way or another) so they could continue selling their regular easily broken glasses. But that's only the same suspicion as other people voiced at the time. Once you bought glasses that didn't break, you never needed any more, so "they" got rid of them. I have bought several sets of regular breakable glasses since then, since 6 oz is pretty small. Almost all of them are gone - bought, used and then broken over the years. I still have three of the four Duralex I bought in 1977 (I don't know where the other one went), and the 8 oz one from my Mom (possibly bought 1971-1972?).

* Floating ice sheets extend variously around Antarctica. Variable warmer sea currents are causing some of these to break up unexpectedly fast. "Average" temperature doesn't count; it's the warmest points causing the break-ups. New info based on observing the collapsing Jakobshavn glacier in Greenland would suggest that the ongoing breakup of the sea ice shelves destroys the outer edge support and would cause the mile to two miles thick on-land glaciers to start to collapse from the shorelines. Since they are so thick they can't support their own weight if the edge isn't supported, and so they will break up and flow faster into the sea. As the Antarctic ice melt, thus far minor in global terms, increases, it would reach epic proportions. If it happens to the area in the news some time back, this would cause a very rapid global 11 foot sea level rise. It is probably now just a matter of "when", and it didn't sound like it would be many decades off. Say g'bye to most of the port facilities and coastal cities around the globe. Say hello to hundreds of millions of displaced and destitute refugees seeking higher ground. (I'm glad I found this place, 25 feet above the present high tide mark! I could as easily have taken one just above sea level and might then have wound up in the refugee category in my lifetime.)

Eagles feasting on a rotting sea lion carcass on the beach out front.
Where it came from I have no idea. It was huge. Evidently somebody thought it was a whale.
The first day I saw it someone had skinned the accessible part, but it was too heavy to flip over.

Bits of crystal(?) sponge have washed up all over the beach everywhere. Why?
I've never seen more than the odd lone piece before.
I have only been here 2 years, so I don't know if this might be normal every few years.
I hope it isn't yet another mass marine die-off - the end of the world's only sponge reef!

(Eccentric Silliness Department)

* Conponent: That piece you got at such a great discount price, that doesn't actually work when installed.

* Ah! I get it now! The glass labware item is called "beakers" because they have "beaks" to pour out of. How mundane, ha ha! (How many decades did that take me?)

* Az sumwun faymus wunss sed, "Hee iz uv smoll maend, hoo kan onlee thingk ov wun wey tu spel u wurd."

* Rumor has it one should stay away from the bar in Epstein. You're likely to catch something there. (...At first I thought it was just a virus.)

* Do Linux computers have to be run in "sudo" mode to get permission to play sudoku?

* Parachuting, like fine wine, is good 'til the last drop!

   "in depth reports" for each project are below. I hope they may be useful to anyone who wants to get into a similar project, to glean ideas for how something might be done, as well as things that might have been tried or thought of... and even of how not to do something - why it didn't work or proved impractical. Sometimes they set out inventive thoughts almost as they occur - and are the actual organization and elaboration in writing of those thoughts. They are thus partly a diary and are not extensively proof-read for literary perfection, consistency and completeness before publication. I hope they add to the body of wisdom for other researchers and developers to help them find more productive paths and avoid potential pitfalls and dead ends.

Electric Transport

Ground Effect Vehicle (first the R/C Model)

   On the 9th I put the spars and ribs together to form the wing, and did a bit of sanding. It started to take on more form. Then I got onto other things.

   On the 23rd the motor controllers ("ESC"s) ordered in May finally arrived. Well, they did say up to 60 days delivery. There was two days to spare.
   The motors had come with banana jacks on them. The controllers came with bare wire and some bits of heatshrink to cover the solder joins. I found a brass tube of the right diameter for the jacks. I cut six 1 inch lengths and soldered three on the wires of one controller, with some bigger, longer pieces of heatshrink to match.
   Then I got out the radio controls and a 6 volt NiMH battery tube to try one out. Everything worked fine. It blew a lot of air. Then I tried 12 volts (10 NiMH AA cells tube). It blew a hurricane and I could feel the fan want to move forward. Nothing ventured and all that I put them both in series for 18 volts. Up to 22 volts is allowed. This time it blew a hurricane at lowest throttle, and then sucked in and ripped a nearby piece of paper. I think the craft will take off with two of those on it.
   (Ack, 1AM! How do I end up always staying up so late?)


   Someone thought that the craft as configured would be unstable, that it would "hobby horse" [in waves?]. Well, some ground effect craft have done that. He mentioned that wings have their center of lift at about 1/4 of the way from the front to the back. But that was the point to to Ryland's special ground effect wing profile, that the center of suction lift from the top would be much further back. The wing is actually thicker toward the back than near the front. It's a bit more like a wing "with its flaps down". Ryland says that the compression lift from underneath, the ground effect lift, is more toward the front. I'm not sure myself how pronounced that effect will be.
   Anyway we shall see! It may be that the batteries (the heaviest part) should be placed more toward the rear than the front, which would also be unusual for an aircraft. But these questions are just more good reasons for making a model first. If there is anything like "hobby horsing" instability that is hard to get rid of by physical design, my backup plan is to take the "stealth fighter" technology route: a microcontroller connected to inertial sensors and servo motors makes instant microadjustments to the elevator to keep it on an even keel. If that is done, it may be that the driver?/pilot?/captain? doesn't even need to have an elevator control. After all, it isn't intended that the craft fly up into the sky, only that it ride just above the wave crests.

   Near the end of the month I decided the ducted fans should go on the outside of the hulls, in line with the wing. That way they would only be pulling/pushing forward, and not trying to push the nose down as propellers above the wing do. That meant a couple of the spars had to be made longer to mount them, as well as made streamlined. Working on batteries as I was, I didn't get that done until the 30th.

GEV with the ducted fans placed outside, in line with the wing

   Finally to visualize what I have in mind, on the 31st I pinned on a scrap of the paper-thin polypropylene cloth. (non-woven "landscaping fabric") My intention is to epoxy this, as thinly as possible so as to add very little weight. I expect this strong, featherweight, cloth composite to be a modern version of the doped canvas that early aircraft skins were made of.

Three views with the motors set in place with sticks and a scrap of fabric pinned on for effect.
(The elevator that goes behind the wing at the very back isn't made yet.)


   In related news... I was thinking how the unipolar BLDC motors (next article down) would probably make great motors for the ducted fans on the full size ground effect craft, except for their large "pancake" diameter taking up so much room in the middle of the duct.
    Then on the 19th I happened across a short video on youtube: Rolls-Royce | Permanent Magnet Technology  by Rolls-Royce. They were making PM / BLDC motors for marine use, as underwater ducted fans.
   In their configuration the stator coils were on the inside of a drum around the outside: in the outer duct. The magnets were on a rotor rim inside from that. Of course that configuration would be quite applicable to the unipolar motor. The stator was the outside of the duct, and the rotor with propeller was almost the whole inner diameter, with no lump bigger than the bearings on the center shaft, and of course a few stationary spokes. It seemed ideal.

Rolls-Royce electric ducted fan motor for marine use.
Outside: ring of stator coils. Inside: magnet rotor with fan.

Stator ring and magnet motor rim form parts of the duct,
thus only a small center spindle is required to mount the rotor.

   Somehow designing another whole new motor configuration (and rather different than my others) wasn't quite what I had in mind of when I thought of applying the (potential) new motor to the ground effect vehicle.

Unipolar BLDC Motor & Motor Controller Design

[drawing]   On the 5th I decided to cut circles out of paper and draw magnets and coils on them. Then I could actually rotate the "rotor" and see which sets and sequences of coils (if any) would rotate it consistently. I drew 8 magnets on the rotor and 12 coils on the stator. I wasn't sure that 12 and 8 would work. Rotating the rotor around the stator 1/3 of a coil at a time disclosed that it was the same as 6 coils and 4 magnets, doubled, and so of course it would work the same. (Duh!) What was different was that being doubled, whatever coils were activated next to each other was duplicated on the other side of the rotor, so the forces were all symmetrical. Perfect!
   It would seem that ideal conditions are: both the rotor and stator have to have an even number of elements, and that there are 3 coils per 2 magnet poles. So, a multiple of 6 coils and 4 magnet poles, but at least two sets so forces on the rotor are balanced against the axle. Hence, 12 coils and 8 magnets.
   So the next size up that would work so perfectly would be 18 coils and 12 magnets, which would be a triple set of 6 coils and 4 magnets, with three sets of coils activated symmetrically 1/3 of the way around the rotor from each other. Then 24 and 16 with four sets. I had tried that, 24 coils with the "Electric Weel" and it was just too big for me to make solid enough with my "Electric Hubcap" techniques. The slightest off-angle on the axle bearings and the huge rotor would be touching the coils at one point and too far away at the other side. I would presume that the 18/12 size would likewise more than tax my capabilities. It looked like the 12/8 size was thus IT! If other sizes of motor were desired, they would have to also be made with 12 coils and 8 magnet poles, with different size coils and magnets than the sizes I already have lots of.

   I started with rotor magnet "A" (north) approaching coil "1" (north). This meant that also the opposite north magnet ("E") was approaching north coil 7. The other two north magnets ("C", "G") were approaching south coils 4, 10. The coils that would drive the rotor clockwise from this position were: 2 & 8 coming on, 3 & 9 already on, and 4 & 10 just turning off.
   Next I moved "A" just past "1". Now it was 1 & 7 coming on, 2 & 8 already on, and 3 & 9 turning off.

   1/4 of a rotation thus needed this sequence of coils driven:

  2,8,   3,9
  1,7,   2,8
12,6,   1,7
11,5, 12,6
10,4, 11,5
  9,3, 10,4

   1/4 of a rotation also brought the next set of north magnets to coils 1,4,7 & 10, and so completed a magnetic cycle, with four magnetic cycles to one physical rotation.


   I then checked out the 3-4 (or "4-3") unipolar configuration I had conceived of in 2016, with the 8 magnet rotor and just 6 coils. Sure enough, that could be made to run too, with the coil activation sequence being the opposite direction to the 4 magnet, 6 coil arrangement:

3,4 (and repeat)

   Again with 12 coils (and 16 magnets) instead of 6 coils, it would be bi-symmetrical with even forces around the rotor. But there would be twice as many coil switchings for the same rotation. Since I would be using 16 magnets anyway for the 8 magnet pole version, it would be just a matter of mounting the same magnets one way or the other:

   (Gosh, I actually have the unipolar Electric Caik with 6 coils and 8 magnet poles made in 2016 - asymmetrical forces notwithstanding. That should work to test the required 6-phase motor controller that would be the same as for the 12 coils 16 magnets size. Then I can put on the regular rotor with 4 magnet poles NNSSNNSS and try out both options. They would be sure to perform a little differently. Might one have more torque than the other?)

   Of course another new motor and controller is a very considerable project. However, since I have already made very good axial flux BLDC motors I am sure the unipolar one would be likewise an excellent runner, and since the motor controller will doubtless be more reliable, I am sure they would make a good set for an electric car. If it wasn't for the large diameter - or maybe in spite of it - they could also be a good set for a propeller drive - the ducted fans for the ground effect vehicle or for a boat propeller. [See Ground Effect Vehicle, above, for a new idea.]
   And it would be great to be able to try out the "permanent magnet assisted" coils idea, which just might bring a whole extra level of performance per power used, perhaps somewhat akin to a heat pump making a building warmer by an amount of energy greater than the amount of energy supplied to pump it.

? Turns Per Coil

Now... How many winds per coil? In the Electric Hubcap there were 20 or 21 winds per coil with the three coils of the phase in series. Then there were always 2 phases activated in series between ground and 36 volts (via the "Y" point). That's 120 (to 126) winds total. In this unit just two coils are driven by each phase, which would suggest 60 turns per coil. But the currents will need to be higher because ony 1/3 of the coils are on at a time instead of 2/3. That might suggest using 30 turns per coil for a total of 60, which would mean #12 or #13 magnet wire. Which I don't have any of. Or perhaps double winding a #15 and a #16.
   Then again, coils are inductors... how many winds for double the current again? I have a feeling it's not a straight linear relationship. I should probably be able to look it up on the web and figure it out but my brain is rebelling. 30 turns sounds good.

Motor Sizing

coils   I laid out 12 coils/cores on a table. I left a bit more room between each two than in the original Electric Hubcap for fatter wires and better cooling - 78 mm instead of 73. When I had what seemed like a good size circle, it appeared the magnet rotor should be about 350 mm diameter, and the stator inner diameter then 370. (Make it 390 and have enough room for the wiring for a change!) That seemed very substantially larger than the 250/280 mm of the original Electric Hubcap. I guess saying "huge" would be an overstatement. It would probably weigh over 40 pounds. I don't suppose such a large rotor could take much over 2000 or 2200 RPM safely.
   I would probably make the stator plate almost an inch thick to be sure it didn't warp. Polypropylene-epoxy is quite light, so thick ends don't add much weight. Later it occurred to me that "spoke" ribs on the outside might be the best way to go, with a thick bearing mount place at the center. (The longer the short shaft is between end bearings, the straighter it will be. and it will need to be very straight for such a big platter.)

   To review the figures: a 13 inch car wheel turns about 10 RPM for every kilometer per hour it's traveling. So to do 100 KmPH (63 MPH) the wheels are turning around 1000 RPM. I was counting on 3000 RPM motor speed to be able to have two such motors, one on each front wheel, to provide sufficient torque to get a car moving at 3 to 1 reduction yet do 100 KmPH on the highway, in case the variable transmission didn't pan out. But at just 2200 RPM, the vehicle wouldn't go much over 70 KmPH. I'd call that a success as a town vehicle, but I'm not sure other traffic on the highway would appreciate it. A reduction of only 2.2 to 1 to the wheels is asking for a lot of motor torque. (...at least 35 foot-pounds each or better. At 3 to 1 it's 25 - still a lot, but just a minimal 150 foot-pounds at the wheels.)

Now... about that 6-phase unipolar motor controller!

Permanent Magnet Assist - Magnet Fitting

   I had found in 2016(?) that I could machine the iron powder toroid cores on the lathe - I had done one. A typical washer for a 1/2 inch bolt could fit perfectly, flush with the top end of the core. The purpose of the washer was as a "keeper" for a permanent magnet assist cylinder magnet placed in the center of the toroid. But I had never tried to fit one.

[imidj]   On the 11th I dug out some 3/4 inch and 1 inch PVC water pipe. The 1" one was too big for the core center and the magnet rattled around loose in it. The 3/4" one telescoped inside the 1". It was too small for the magnet and was very loose inside the core. I thought I could probably shrink the 1" in the oven. Then I thought of cutting a slit in the 3/4". It could open up a bit to fit the magnet. It was too stiff. I heated it to over 200°F in the oven and it was much more compliant. I put it over the magnet. Then it looked a little too big for the core, but I tried it anyway. It was a perfect, fairly tight friction fit. No looseness, no jamming anywhere! There was a gap where the slit was, but that didn't matter.

[imidj]   Then I put the washer on the end and everything was just right. (The washer stuck out just a bit but the depth of the machined space could easily be made a little deeper.) The other end was still open, and predictably a second core was far more attracted to the open end than the closed end.
   All was going just as according to theory! Now, if the coil on the core was energized so as to have the same strength as the magnet, the magnet and the coil should project their united magnetic field at twice the strength the electricity being applied to the coil alone would make. How could it not make more torque than just the coil alone, yet without using more electrical energy?
   My prime concern was that the magnet and the washer would make for substantially higher iron losses than the miniscule losses of the iron powder core and would thus warm the cores. The next day it occurred to me I could machine the washer and magnet in at least a little deeper, recessed in instead of flush with the top of the core. That would help a little.

   There was another takeaway from this little trial: in order to allow for later addition of permanent magnets into the cores, the 12 cores should be machined to take the washers before winding them, and the stator plate should be designed to lock down the open end - i.e. the core center mounting "buttons" shouldn't be too tall to accommodate the magnets. Easy to do and shouldn't make any substantial performance difference if the magnets weren't put in, or if they were put in and then removed. Simply adding the magnets to the same cores in the same motor after initial tests, should allow some very fair performance comparisons!

Tuned Circuit Coil Drive

(20th) I came up with an idea for a new sort of coil driver, especially appropriate to 'permanent magnet assisted' technology. The permanent magnet assist, according to sources I've run across, reduces currents at higher currents, but doesn't do much at lower levels of magnetic flux. The assistance is probably maximum when the permanent magnet and the electromagnetic coil are magnetized to the same strength. So even if lower power is needed from the motor, it would pay to have the flux strength at a high level. So instead of applying high frequency PWM to get less overall flux in the coil, short bursts of full strength coil activation would be used to gain high flux, with rests between since full torque isn't needed. In other words, a very slow PWM, perhaps measured in tens of hertz or even ones, instead of 16 KHz. (This would also eliminate what surely must be an [ultra]sonic noise irritant.)

   First I thought to put a capacitor in series with each coil. That seemed ideal for turning on the coil and holding it on for the resonant period. But then I couldn't see how to discharge the capacitor when the coil was turned off.

   So the tuned circuit would be created by putting a capacitor in parallel with the coil. Three B+ power line chokes (per 2016 unipolar designs) for the six phases would be placed in series with the coils so that the capacitor doesn't present an initial short circuit to the driver. (I expect we can get away with 3 instead of 6 because an oppositely timed coil is off when the active one is being switched, and for a period before and after, so the one motor coil isn't affected by the switching of the other one.) In a coil, current rises with pulse on time. In a capacitor it drops. Thus the current, for the given resonant frequency period, stays somewhat even instead of simply rising with time, and a magnetically high level of flux is maintained at a relatively even strength while the mosfet remains on.
 When the mosfet switches off, the coil voltage continues to be supplied by the capacitor. Decreasing flux continues to be supplied until the capacitor (and coil) voltage is exhausted.
   No flyback diodes are needed in this configuration, because the coil loses voltage "gradually" as the capacitor discharges instead of being suddenly switched off. Likewise, the main power line choke suddenly has a voltage across it when any coil switches on, but as it is a lower inductance this recovers rapidly during the coil's on period and there is no significant flyback effect. (I should check this out to make sure.) The saving of energy usually wasted in flyback diodes through their forward voltage drop is not insignificant, without recourse to "active rectification" that means turning on mosfets that just might short across the power supply. (improved reliability.)

   Based on the component values, the microcontroller (ie the programmer writing the code) must determine how long each resonant pulse should be, and the time between pulses, and based on the "throttle" control and RPM, how many pulses to apply and when, as a magnet passes between two coils.

Rotor Design

   I decided that wrapping each magnet with PP webbing/strapping is probably a waste of time. Maybe in a dirty environment it protects the magnets better. But if the epoxy came loose, the magnet might not fly out, but the rotor would still need repair. I saw one youtuber (making a generator, was it?) who cast his magnet rotor in epoxy, but he simply didn't fill it full, in order that each magnet protruded and provided some "centrifugal fan" action for cooling. That would surely be simpler and cleaner than the strapping.
   Could the whole cast epoxy disk come loose from the steel rotor? I had the thought that putting some smaller holes through the rotor for the epoxy to put "roots" through would be a good improvement to strength. Especially if they were drilled out on the back side with a "V", so the plugs of epoxy would look like flat-head bolts. With that, might I even dare push the max. up a couple of hundred RPM? 2400 or 2500 RPM sounds so much better than 2000 or 2200. OTOH 350 mm, almost 14", is a big disk to spin! Also I'm thinking 1/4" steel rotor might be just as good as 5/16", and 20% lighter. (If only I had that plasma cutter working... OTOH I could just have Victoria Waterjet do the rotor and mail it to me. OTOH I have to get the CNC table working to make the stator molds anyway.)

Activation Sequence

[drawing](16th) In 3-phase motors, one uses 3 hall effect sensors each midway between two coils to determine magnet positions - the rotation of the rotor. Would the 6-phase controller need 6? Obviously one will be using a microcontroller as the brain. There are no 6-phase motor controller chips (much less unipolar ones). Three hall sensors do in fact provide six different polarity combinations, and there are 6 phases. But with unipolar coils it seemed lopsided - two sensors would be between N-S coils and only one between a S-N pair. I put three hall sensor positions between three coils on my paper 'motor' to figure it out. Somehow it all worked out. Each of the six combos of three "norths" and "souths" corresponded to one of the six coil activations. The microcontroller could easily be programmed to pick the right ones. Only three hall sensors are required.

Variable Torque Converter

   Having groped around in the dark for a way to do this from 2009 to 2016, I gave it a rest after I moved to Haida Gwaii. I thought maybe making reluctance motors that could freely do very high RPMs and so not need a variable converter was a better answer - at least for me. But during June I had thought of a couple of fixed 3-speed gear designs, and somehow started to realize that for a variable torque converter, just two main components had to work together:

1. A three element gear set such as a planetary with one of the elements being tied to the driving motor and the second one to the output - the car wheels or the differential. If the motor turned but the car didn't move, the third gear had to spin.

2. The third gear's turning had to be controlled; obviously it couldn't just free-spin. Controlling its turning would control the coupling ratio between the other two. If it spun near its free-spinning speed, the RPM reduction between the other two would be very great. If all three elements spun together, the drive ratio (regardless of the number of teeth on each) would be 1 to 1.


3. If the control element ran around the same speed as the other two, then the torque needed to adjust the control element would be as high as that needed to move the car. The faster the control element spun, the less the torque that would be needed to modify its speed. This was driven home to me by the large control forces needed for the experiments of 2016, driving one wheel and using the car's differential itself as a 3-element variable torque converter gear set. A control gear spinning 3 times as fast as the motor would need 1/3 the torque.

4. If the control element ran opposite to the motor, then the force required to slow it would also try and slow the motor down. If the control element ran the same direction as the motor but free-spun faster, then the force needed to slow it down would add to the motor's force instead. The losses incurred that way would be [I presume] lower. I was trying to accomplish that in the 2016 designs. But now I realize they were needlessly complicated and less able to supply good results than just (1) a planetary gear with ratios that would lend themselves to the job, and (2) a centrifugal clutch (or a flat belt & pulleys "clutch") to control the spin of the control element - both correctly configured of course. (The way I was doing it before 2015, a clutch that tried to halt the control gear entirely, was simply wasting energy.)

   It also seemed to me that the large centrifugal clutch I made in 2016 (2015?) was an ideal means to couple the motor to the control element. As the motor started the shoes would be retracted, and the drum on the control gear would spin up freely. The car wouldn't move. When the motor started imparting force to the drum via the shoes, pressure would be put on the output gear and the car would want to move.

   I found a planetary that looked about right - one I had used before, early on. If the motor drove the planet gears assembly, and the ring gear (connected to the car wheels) wasn't moving, the sun gear (centrifugal drum) would turn 2.8 times as fast and in the same direction. If the sun gear was slowed toward the motor speed, the ring gear would have to start turning and speed up toward the motor speed - the car would move. If I have it right, the torque to move the car would be (at most!) 1 / 2.8 times the torque needed to slow the drum, or just over 1/3 as much. This would be similar to having a 2.8:1 gear reduction. Multiply that by 3:1 reduction to the differential gear and it's the equivalent of 8.4:1 total reduction, which should be sufficient to start the car rolling with a fairly small motor. The difference is that as the torque requirement drops, so does the ratio. When all three gears are spinning together with the shoes locked to the drum, the 1:1 ratio from the variable section has dropped the overall ratio to 3:1, which will prevent the motor from over-revving on the highway. If that loaded a small motor too heavily, a 3.5:1 at the differential would give 9.8:1 overall reduction at low speed and still allow ~87.5 KmPH (instead of ~100).

   Or if that wasn't good enough, I should look for a suitable planetary gear with a higher ratio from the planets to the sun gear, such as 4 or 5 to 1.

[desired config]   After working all that out on the night of the 5th, on the 6th I showed the gear and explained the concepts to a friend, and as I did so, I realized it still wasn't the best arrangement. I was putting the centrifugal shoes on the motor because that was the only part that was surely driven at a known speed. But until the car started moving, the sun gear would actually spin 2.8 times as fast as the motor. There was the part to fling the shoes outward with maximum force! The drum could be on the motor (planets assembly) and the shoes on the sun gear instead.

 And it appeared this configuration would simplify construction. When laid out, it appeared that the ring gear that had to drive the chain to the differential would be off to the left of the 'experimental transmission box' entirely.

[config 2]   Still later I realized that the drum could be either on the motor or on the output (ring) gear with quite similar effect. Which was better? At the start, if the drum was on the motor, it would be turning 1/3 of the speed of the centrifugal shoes, and in the same direction. If it was on the output gear, it wouldn't be turning at all. Which way made for higher torque to the output? Probably the latter? Either way, as the shoes took more hold and the car accelerated, everything would end up rotating at the same speed, 1 to 1 ratio.

   It looked like this might work well, but only if the motor was on the port side of the box instead of the starbord. It looked close, so at this point I wanted to fit the box into the car to see.

   Just how many ways were there to configure one planetary gear, anyway? It seemed there were 6 possible ways to connect the 3-element gear to the 3 external components:


   For each of those, one part of the centrifugal clutch (either one) had to go to the ratio control, and the other part could go to either the motor or to the output. So by picking which gear (x 3), and picking which element of the clutch went to which gear (x 2 again), that made 36 possible ways, each with its own advantages or disadvantages. But putting the centrifugal shoes on the output obviously won't work. And any configurations that cause reverse motion of any of the elements is probably counterproductive. Without checking, that's likely to be over half of them. Of the remainder, only one can be optimum so the rest are relatively counterproductive by comparison.

   I wonder how many counterproductive ways I've already tried? The two options mentioned above the table seem likely to be the best.


   It looked like I could contrive to cludj everything together mostly with parts and pieces I already had, perhaps fabricating and welding up a couple more things. The next question was, what to use for a motor? I had the Electric Hubcap, but the Kelly motor controller was blown, and I had dismantled the surrounding parts and used them for the forklift motor controller. So the simple thing would be to use the forklift motor, already wired in to the car.

   And... amid all the other fabulous projects, when on Earth was I going to find time to put this together? But of course it was easy to dig out the transmission box and various parts just to see how they all fit together. When I did so on the 9th there seemed to be one very big fly in the ointment: There seemed to be no way to arrange it to fit the ring gear anywhere where a chain or belt or whatever from it could turn the differential. The ring gear would be right outside the box.
   I turned the shaft around. It looked like it just might work with the motor on the right instead fo the left. That seemed preposterous, but measurements showed that it just might work. Maybe! Just! To check it out I would have to disassemble the present (fixed ratio) transmission, install the box, make careful measurements, and if it could work, some new pieces for mounting things - the motor and a steady bearing at the far end. There would sure be a lot of empty space under the hood on the other side. One could easily place more batteries there.

[imidj]   On the 22nd I took out the forklift motor and the transmission, and put in the 'transmission experiments box'. It didn't seem the motor could be fitted on the narrow side of the box. I would have to rebuild the box as a sort of mirror image of itself. Or would I? There were after all two similar configurations. My plan was to place the centrifugal drum on the motor shaft. But if instead I placed it on the output shaft, and reversed the car's differential (chain sprocket on the other side), and made a couple of new mounting plates with spacers, it looked like the motor could go in the larger space of the starboard side after all. It might even be easier to mount things, and maybe even get a bearing on to help stabilize the drum/ring/drive sprocket.

   Finally on August 4th, editing this newsletter, I looked at the first photo again. If the drum was shifted way over to the right by making an offset mounting for the steady bearing and maybe cutting a bit of a dish into the top of the right side, the sprockets could line up and everything should fit that way. I went out to the car where I had already set 'the box' in place and checked it out. It looked like the drum could just clear everything if it was all done just right. That should actually be the most practical. (I couldn't place it there to take a photo without making changes.)


   In any event, (way premature of course) for something besides a prototype I might look at something with larger plastic gears that don't need to be immersed in oil, since it would be at least difficult if not impossible to seal oil into a regular planetary with the external clutch components attached to it.

   The "all spur gears" differential or planetary gear might be a good choice for that. I had come up with that (derived from a sketch in Wikipedia) in October 2016 - the last month before I got involved in buying a house on Haida Gwaii and moving. I think the sizes of all the gears could be adjusted to provide any desired ratios.
   Then one could probably drive gear "D" from the shaft going out the back and attach the centrifugal shoes to it on the motor side of the drum. The outer drum (planets) could extend behind to engage the shoes, all on the motor's side of the planetary. The output would then come out the front on a shaft from "C" (attached to "C" and not with a bearing there as shown, of course).
    And then (after all that) since only a small center shaft comes out each face, it could easily be sealed to keep oil in for metal gears.
   (I also note that gears "A" and "B" could be fixed on one shaft, reducing three gear meshings to two and reversing the direction of "C" -- albeit with more limitations on their relative sizes and ratios. Gear "B" as it is would be too small to reach "C" if it was on the same shaft as "A".)

Other "Green" Electric Equipment Projects

36 V DC "Off Grid" Infrastructure

HAT36V-50A Plugs & Ceramic Sockets

   On the 12th I finally got around to making a HAT 36V, 50 amp plug for the kitchen water tank. (If only so I could turn it off at night without going out to the garage and flipping off the circuit breaker.)

   Before I wired the socket I plugged them in together and everything was fine. But when I had wired the ceramic socket somehow there was enough room in the 'minus' side that the 'hairpins' had twisted around when I connected the wire, so the plug wouldn't plug in. I twisted the socket around so the wire turned the 'hairpins' straight again, and the plug went in. However, when it did a chunk broke off the corner on the 'plus' side. The pressure on the 'hairpin' there from the plug pin was too much for it.

The ceramic socket cracked from the pressure against the side when the 3D printed plastic plug was inserted.
I taped it up and am using it. But it's not good advertising!

socits   Apparently I either need to make them thicker or else out of porcelain, which is considerably harder than ceramic. 'Cone 6' porcelain is the top limit of what my poor little mini-kiln will do, and it takes it four hours to get up to that temperature. But perhaps a worse problem is that in this very hot firing porcelain shrinks more, and tends to warp. That means (as I recall from making Supercorder beaks) making a still more oversize mold and unreliable results.
   I thought I'd try making them from thicker ceramic first. That still means designing and 3D printing a new mold. I tried sticking some fresh clay on the unfired sockets at the weakest points, but it didn't work. Predictably the new clay shrank as it dried and fell off.

   At the Tlell fall fair on August 4th I saw someone selling homemade pottery (nice selection!) and thought he might have advice. As I explained about my broken ceramic, he said "They're usually made out of porcelain." I guess that's my answer. He insisted it didn't shrink so much and he usually expected 16%. My mold is 18% oversize.
   I said it took 4 hours for my mini-kiln to get up to cone 6. He said "That's fast!" After a quick double-take I thought, "of course it is, compared to heating up a larger kiln." A large kiln firing can take 2 or 3 days. He offered to sell me a few pounds of 'cone 10' porcelain clay and fire them for me when he was doing a firing.

   First I'll get out my big block of 'cone 6' Dove Porcelain and try that in my mini-kiln. (It's probably dried out and hard as a brick, so I should find it and get onto re-hydrating it well in advance.) If it warps or shrinks too much, I might take him up on his offer and see if his clay and kiln work any better.

Handheld Bandsaw Sawmill Notes - Update

gummy   They say spruce is a tough wood to mill. It has "interwoven" grain that makes for tougher cutting, plus it is pitchy and has hollow pockets full of sticky liquid pitch that gum everything up. On the 7th I stopped cutting because of slow cutting and worse, a lot of vibration - which turned out to be lumps of sawdust glued even to the slippery UHMW wheels and to the stainless steel band by pitch. I had to scrape them off.

Cutting this piece was especially nasty. I went through
about 3 band sharpenings before I realized there must
be sand or something in the gummy voids. I finished it
up with the chainsaw mill.

I 'set' the offset of the teeth with two crescent wrenches.

   On the 9th I sharpened a band which I had already "set" the teeth on on the 7th. The "set" made for straighter cutting, but I still wasn't happy with the performance. Sharpened or not, all the cutting seemed to take twice as long as I expected. I was cutting 1"x8"s, 10 feet long. and taking 7, 8, 9 minutes per board. At the bottom, the last piece, the wood was only wide enough for a 1"x6", and I trimmed 2" off one side with a skill saw (and chainsaw at the fat end where it was too thick for the skillsaw). The 6" wide cut, literally, went twice as fast as the 8" ones. ("3 minutes" according to my cellphone, which is just another stupid "timer" that won't break time down any finer than a minute. So, somewhere between 2 and 4 minutes.) Doubtless a brand new band, or even a newer one, would have done better on all cuts. But but 3/4 wide in 1/2 the time shows how gummy spruce bogs everything down. I was happily cutting 10" and even 12" wide boards in alder last September.

   After setting the teeth one one band with two crescent wrenches, I decided to make a pair of tools for the job. It was just two pieces of small bar steel with slots in one end. No doubt I could have gotten fancier -- How about a single item with a hinge? or a pair of "crimping" pliers with the right bend built-in? I found that roughly setting the teeth got the band cutting again, but then it made rough cuts. Something that would set them all the same, evenly, would surely be better.

I made a couple of little stands with 6"x6"es to put the remainder of cants up on when they get thin.
If you're pushing the saw forward with your legs (that turns out to be the easiest way),
you don't want the cutting band down by your toes.

Here are a couple of pictures of the saw mostly opened up on the bench to change the band.
A couple of people have written me about building one, but I haven't had time to draw up any instructions.

The key pieces are two 36" Alaska Mill rails, pillow block or similar bearings, axles 10" UHMW wheels
(they shed sawdust etc), the cover pieces, and especially the double "L" angle pieces that hold the self-
adjusting guide wheels et al, and the other pieces to hold and adjust the rests that set the cutting height.
And the skillsaw, pulleys and V-belt (I used a link belt), of course.
The prospective builder will have to look in previous issues of TE News for details on each component.
If the bearings were mounted more offset from the rails with the wheels lower down, eg, with one inch
spacers, the top covers would be shorter, and there would be more cutting height adjustment.

Latest change: 1 inch pulley replacing (larger) adjustable pulley.

This pulley had a 1/2" center. I drilled it out on the lathe to 5/8" for the saw spindle,
but not far enough so it could go on far enough to hit the stationary bearing holder.
A longer "saw blade" bolt holds it on tightly.
I think the band slow-down is better but I wouldn't swear to it.
The "optimum" size might even depend on what wood is being cut.

So now it's a 9 to 1 speed reduction.

   One last thing... That tube with a sponge at the band for cooling water... it's pretty primitive. I just saw someone on youtube make a cooling water squirter by punching a small hole in the side of a plastic pop bottle. Once he filled it, No water came out until he loosened the lid of the bottle, then there was a thin stream out the hole. When he tightened the lid again, it stopped. Wow, simple, ingenious! I think I'll adopt that system for the saw!

Bathtub Insert Notes

   While in the bath (AM of 10th) I figured that if I was going to use up a spray foam can to glue styrene foam and wood together and fill a few gaps for the ground effect vehicle, and since you can't save the can very long once you've started using it, and since also I had lots of scraps of styrene foam, this would be the time if there ever was one to try out making a bathtub insert. I dripped water to the kitchen and got a magnetic tape measure off the fridge, a pencil and a handy chocolate bar wrapper (half foil - more wet strength than paper) and came out with dimensions on paper.
   I also measured the water in my present bath. Sitting in it, it was 5" deep - a pretty shallow bath. When I got out it was 3-1/2". (If I had lain down it would have been about 5-1/2" to 6" deep.) I emptied the water into a pitcher and poured it down the sink. I measured a little over 46 liters.

I came out with the inside dimensions. A person heavier than 160 pounds might want to go larger, but I tried to leave extra everywhere. I didn't want a tight fit at any point. I didn't make any compromises for showers - pull the insert out and set it aside for a shower.

   The tub had at least 53". I would shorten that just by the thickness of the end walls. Those would be 2" foam, but I could cut an inch away at the top-back since it sloped, so measure 54" max. outside length. The back would slope inward at the bottom by 4", so (with the top trim), 51" (and cut foot end at 45° to allow for the bathtub curve up).
   I figured 17" inside width for the first 22" from the back end, then tapering to 14" at the 34" mark, and no less than 13" at the foot end. I might have to curve or cut the side walls a bit. And I might go just a bit narrower but slope the walls outward toward the top.
   To cover my thighs (which I was never able to do in the regular tub - it would have used the whole tank of hot water) needed a 9" depth. Since it would be inside the regular tub, I wasn't worried about splashes. So 10" seemed like a good depth.

   During the next bath I decided to make it all 2 inches wider. If it was too narrow for comfort, there wouldn't be much I could do about it. But if it seemed needlessly wide (wasting hot water), I could glue an inch of foam on the inside on one or both sides to narrow it.

   Is it relevant to "green energy"? Well, it should save energy (hot water) while providing a deeper bath. OTOH, I got rid of the red spots on my legs with witch hazel, so my prime motivation for wanting a deeper bath covering my thighs is gone. But the idea and the plan is now in my head. I fervently hope this project will be as simple and short as it looks, and not grow into some monster like so many of them do!

   By the end of the month I hadn't started using a spray foam can, and I might not at all.

Electricity Generation

My Solar Power System

Charge Controller Troubles

   On the 8th I got up wondering which project I would work on: batteries, ground effect craft, or maybe start putting together the CNC table with the "Gecko drive" controller toward being able to do all the things that would enable - especially including body section molds for the new design of unipolar BLDC motor. And I had a cant of wood outside waiting to be milled into 1"x8"s. I decided to start with the house solar instead.

   I had had a house guest for a few days (4th-7th) and we used more hot water in the kitchen. The MPT7210A charge controller continued to fail to meet the demand. It would sit there putting out maybe 1.5 amps while the water heater was drawing 6.5, and of course the batteries would discharge to supply the load. It didn't seem to care that the voltage was dropping way below the setpoint. There was lots of power coming from the panels which it didn't try to use.
   Notwithstanding the countless bells and whistles, one can only question the thinking or lack of thought that went into vital operational parts of the programming. (It rather smacks of having been done by zealous and and academically inclined programmer without much real-world experience in electrical/solar/battery systems.) With more hot water use, 1.5 amps (or whatever) didn't even recover the batteries in the shorter slack times when the water heater was off. So in the evening after suppertime dishes the batteries would already be low, and then there was no more solar. I had to turn the hot water off on all four nights to prevent the batteries from going too low. (That doesn't say much for the old NiMH "D" cell batteries, either, which should have held plenty of charge to compensate.)
   If I turned it off and on again the charge controller would wake up, rise to the occasion and start putting out 5, 6 or even 7.5 amps (the highest I'd seen on this supposedly 10 amp unit). But I couldn't always be there to babysit it! And I shouldn't have to.

   So by the 8th I was pretty fed up. I removed the MPT7210A (boost) and installed another PowMr (buck) charge controller. I put the panels 2x2 in series to get the over 50 volts required by a buck controller instead of having all 4 in parallel - which had also allowed connecting to to the grid tie, in parallel with the charge controller, if the DC system didn't need the power.
   I saw from the last PowMr that if the hot water heater drew 6 amps, within seconds it was putting 6 amps back in from the solar panels to keep the battery voltage from dropping. That was the way it should be!
   The four panels were now at two different voltage levels (~0-30 V and ~30-60 V). It would seem that the Y-Solar grid tie 'micro'inverters were isolated. (The case was grounded to the AC plug ground pin, but no other terminal connected to any other on the ohmmeter, and the manual said the output was transformer isolated.) So if I got another one, I should be able to put it on the two panels that were floating at ~~30-60 volts without issue and still have all four going to the grid tie, but on two separate inverters. Unless I got another one to do that (or brought one back from the trailer and put all four panels there on the other one - easier!), I would lose two panels from the grid and much of their power would go to waste.

   With the batteries very low, when I turned on the solar panel breaker, the PowMr immediately started putting out 10 amps, and didn't drop below 8 (it was cloudy) until... actually, it seemed determined to fry the batteries! One thing I don't like about the PowMr charge controller is that it "auto-detects" whether the battery system is 12, 24, 36 or 48 volts. Up to 40 volts it's supposed to detect "36" - but 36 volt batteries of any type can charge to over 40 volts. So if powered off and on again, it could decide that the 36 volt batteries it itself had charged, were 48 volts. It seemed to have decided my batteries were 48 volts regardless of them being a little under 40 on its own meter. At almost 41 volts it still said "undervoltage" and wouldn't turn the load circuit on, and it kept pumping in all the watts it could get from the solar panels. I changed some settings and fiddled around with it, turned it off and on a couple of times, and finally it decided my system was 36 volts after all and then seemed to behave itself. Disquieting!
   I set the battery float charge voltage to 39.6 volts to be easier on them than the 40 or 40.5 I had had earlier (max float for 30 NiMH in series would be 42.0) -- and to be under 40 volts at all times. Or rather, I set it to 13.2, which it multiplies by 3 if the batteries are "36 volts". Later I changed it to 13.1 (39.3 V) as it seemed to read a little low and so charged them a little higher.
   But setting it to the same value is multiplied by 4 if it decides your system is "48 volts". I do NOT like the charge controller "automatically" deciding how many batteries you have. Might it do it again? If it was a manual setting, you would only need to set it once anyway. You would never want it to change by itself! There are few electrical system things I'm more afraid of, especially having almost burned the house down with the disastrous Suzuki Swift EV experiment in June 2017, than stupid charge controllers obliviously (or is it deviously) trying, hour after hour, to pump many times more juice into batteries than they can possibly hold - or withstand. But I left it connected, and it did continue to behave.

   Once everything was running well, I made up another cord with MC4 connectors to connect the panel on the lawn to the grid tie with the other two 305 W panels. I cut the ends off an old 100 foot extension cord I got at the waste disposal site. (The cable I had used before I had made to use for the tugboat panels, and - the owner still not having called me to finish the installation - I finally took the cable, PWM/Lead-acid charge controllers et al, pretty much ready to install, and dropped them off at his house when I was going up that way. There's not much to it and he may finish it up himself.) Having lots of length I moved the panel toward the east side of the house and aimed it to the west to catch more afternoon and evening sun.

   By the time everything was done, the day was pretty much done too. So much for the little job to do before deciding what to do for the day!

"Floating" Panel Voltage Grid Tie

   (12th) I took one grid tie inverter from the trailer and connected it to the other two of the series panels, whose outputs were sitting at +30 and +60 volts (or +25/+50 or +35/+70 depending on load). Sure enough, it worked. Nothing blew up. The inverters are isolated, both input and output. Only the case is grounded, to the AC plug ground. Now all 11 panels (with the one on the lawn) are grid tied again. The 4 in the trailer are all back on one inverter. On all those cloudy days there's not much difference. It will limit maximum production on sunny days - which have been rare this summer.

Combining More Panels on One Grid Tie or Charge Controller?

   The Y-solar grid tie inverters are "1000 watts" maximum. In practice they seem to put out up to about 850. So it would seem there's not much point connecting more than 3 or at most 4 panels to one. But what happens if some panels are oriented east to get more the morning sun, and others west for the afternoon/evening sun? The morning panels will be putting out well below rated capacity in the evening and vice versa. And neither ones will be delivering full power at noon. So one might connect more panels - say 3 morning and 3 evening panels to the same inverter or charge controller. Depending on facings and angles - and the season - the input to the unit would never be higher than with just 4 panels all facing the same direction. But it would run more hours of the day. This could perhaps accommodate a couple of or a few extra panels without adding to infrastructure cost.

   One could also potentially install extra panels just to account for the dull days without buying more of the connecting equipment. If one hooked up (say) five 305 watt panels to a 1000 watt Y-Solar grid tie inverter instead of three, at 50% those 1530 rated watts of panels would put out 765 watts, which is almost as much as the inverter puts out at any time (~830W). So it would be more fully utilized more of the time. It just wouldn't put out more on those less common sunny days when the panels could give it more. (Actually, the four panels I have may be a good compromise. ...or maybe six if three face eastward for early morning sun and three face westward for the late afternoon.)

   (Somehow this seemed like a better idea in cloudy July than after the first 5 fabulous sunny days of August! And after all the Y-Solar grid ties are only 100-150$, and the PowMr charge controllers equally cost effective.)

"MC4" connector combiners for combining 2, 3 or 4
 solar panels onto one grid tie or charge controller

Westward facing solar panels on lawn - 2nd panel added July 31st.

Once the tree shadows are off, they'll be making extra power later in the day.
Hey, wait a minute... there were no tree shadows there a couple of weeks ago!
In another week or so they'll be shading the garden by the house, too.

   On the 31st I thought, "I have have solar panel combiner plugs. A palette of panels sitting around. Why don't I add another panel on the lawn on the same cord, also facing somewhat westward?" So in the afternoon the house system collection went up to 8 panels. (The #16 cord - an old #16 AWG 100' extension cord from the dump - got a bit warm, but nothing like hot.) So one grid tie now has four 305 W Hanhwa panels - same as the trailer roof. The other two ties each have just two 250 watt panels, so they aren't being fully utilized, but with the panels being 2+2 in series/parallel I can't combine them onto one and have a spare tie left over to do more.

Solar Panel Post

Post   When I bought the house there was an old 1980s Satellite TV dish on a post at the edge of the trees. I soon took it off. This month I checked out how the post was mounted and with just a little digging I found it had a heavy square concrete pad hidden in the grass. I think I want to move that into the open somewhere and mount 2 or 4 solar panels on it. The top fixture is at latitude angle, the optimum fixed solar panel angle. Ideally I would have two panels facing eastward for morning sun and at least two facing westward for the evening (which is when I use the most electricity).

   (Now I just need for my neighbor to come back from Lac La Hache and move it for me with his big forklift.)

Month of July Log of Solar Power Generated [and grid power consumed]

   The whole of July had only a couple of mornings or afternoons of sunshine. Virtually the whole month was cloudy without a single whole sunny day after the 6th. The afternoon of the 31st was sunny, and then the first five days of August. It didn't make for good garden growth or solar collection, but it did provide data about how much collection one might expect in overcast weather, which proved to be around 55% of full sun production for the month.
   Notwithstanding, my power bills for March through July 2019 were only about 3/5ths of what they were the previous year. If 161$ for 2 months from mid March to mid May, and 117$ for mid May to mid July sounds high to some, it must be remembered that that cost includes the energy for my driving in the electric Leaf. Gasoline alone for a gas car would have been more, and it wouldn't have cut the power bill even in half.

(All times are in PST: clock 48 minutes ahead of sun, not PDT which is an hour and 48 minutes ahead.)

Date  House solar KWH(Grid+DC), +Trailer Roof solar KWH - day total KWH made [power co. meter readings] weather, usage...

June 30th 11.82+.68,508.53- 14.78 [laundry; 45Km, finish chj;  66860@22:00] Cloudy AM, Sunny PM
July1st 15.90+.63,512.76 - 8.94 [66866@21:30] Clouds, dim sun all day. Oops, grid tie on 1000 W panels was turned off until afternoon. (shoodabin over 10 KWH!)
2nd 22.47+.78, 517.66 - 13.25 [66872@20:30] cloudy, dim sun. again. (not chemtrails... I don't think?)
3rd 25.74+.66, 520.24  -  6.61 [65 Km, chj.car 3.8KW; 66886@21:00] cloudy
4th 32.14+.73, 525.44  - 12.32 [66897@21:30] Cloudy earlier AM, Sunny PM
5th 39.79+.92, 532.47  - 15.60 [85 Km, car chj.@3.8KW; 66915@21:30] Sunny. Panel on lawn disconnected, 1KW panels feeding DC only part of the day.
6th 42.90+1.12,538.36 - 10.12 [55 Km, chj@3.8KW; 66930@21:00] mostly sunny. 1 KW panels went to DC only, Lawn panel disconnected. (This doesn't make the readings more consistent!)
7th 45.10+1.04,541.42 -  6.30 [66938@20:00; 55 Km,car chj.] Overcast.
8th 47.65+.49, 544.24  -  5.86 [66949@12:30; 66950@20:30; bath] Cast over again. Replaced DC charge controller MPT7210A with PowMr. - now just 2 of the 4  250 W panels are grid-tied. Reconnected 305 W panel on lawn.
9th 50.50+.51, 547.95 -   7.07 [66956@9:00; Milling with Electric bandmill; 66958@20:30] Overcast.
10th 54.74+.57, 552.31 - 9.17 [66961@10:00; bath; laundry; 55Km,slow chj; 66970@21:30] Overcast with a few less overcast spots
11th 57.67+.63, 554.96 - 6.21 [fini.car chj; 66976@20:30] Heavy overcast. Where is summer?
12th 61.45+.55, 558.42 - 7.79 [85Km drv,fast chj.3.8KW; 66992@20:00] Lighter Overcast. Got all panels grid tied again.
13th 68.49+.48, 563.71- 12.81 [bath; 66996@14:00; 55Km,slow chj] Overcast, PM with sunny periods. WOW! 1570 watts from the house panels alone when the sun came out. (while the panels were still cool.)
14th 73.72+.53, 567.69 - 9.74 [67003@14:00; 50Km,Fast chj car from 37%; 67014@19:30] Light overcast
15th 78.41+.48, 571.04 - 8.52 [67019@20:00] Overcast. Some rain.
16th 79.77+.60, 572.56 - 3.48 [60 Km drv, chj 3800W; 67036@19:00] Clouds and rain. WOW, what a dull day! The PowMr couldn't get enough power to keep the batteries up (the solar hot water heater used 600 WH) with the grid tie inverters hogging it, and I had to turn one of them off! (After going to all that work wiring it in - but partly just to see if it would work with non-zero floating voltage level panels! But then it stayed on on other days.)
17th 83.72+.66, 575.98 - 8.03 [BR heat(in July!?!); 67044@10:00; 67046@20:30]
18th 88.91+.49, 580.50- 10.20 [BR heat; 67052@20:00] Mixed sun (yay!) and clouds (with chemtrails above)
19th 92.45+.41, 583.79 - 7.24 [67057@15:00; 85 Km,slow chj.car; 67062@20:00] Overcast
20th 96.03+.03, 588.70 - 8.52 [BRHeat; 67068@10:00; bath, some car charging; 67075@21:00] Overcast.
21st 03.54+.76, 592.31 - 7.91 [Car full charge 3.8 KW; laundry; 67094@22:00] mostly overcast. The energy monitor in the garage 'froze up' and I had to unplug/replug it to get it working again. That was in the morning, but by no means first thing.
22nd  8.37+.60, 596.10 - 9.22 [oops, left BR heat on all day. But it probably didn't come on; 67102@21:30] AM: Sunny with clouds and jettrails (for a while I thought it just might be summer). PM: Overcast and some rain.
23rd 14.01+.49, 600.46- 10.49 [55Km,part chj.@1500W; 67112@19:20] clouds with sunny breaks
24th 19.28+.47, 604.62 -  9.90 [67122@21:30] AM: a bit of sun. PM: Overcast, cold, wind and rain.
25th 21.16+.54, 606.23 -  4.03 [67128@21:00] overcast & cold. I had to shut the 250W panels' grid ties off so the PowMr could get enough to run the water heater & keep the batteries up.
26th 23.21+.62, 608.37 -  4.81 [85Km drv. 3.8KW chj.car; 67150@21:00] Clouds & rain.
27th 27.34+.45, 611.94 -  8.15 [55Km,start.chj.car@1500W; 67167@20:30] Mostly overcast.
28th 31.46+.54, 615.58 -  8.20 [finish Chj.car; 67173@20:30] Clouds, a bit of sun.
29th 37.88+.35, 620.17- 12.36 [65Km,stt.car chj.@1500W; 67181@19:30] Clouds with occasional sun.
30th 41.59+.62, 623.72 -  7.98 [finish car chj.; 67193@21:00]
31st 49.85+.53, 629.39 - 14.46 [67202@ 20:30; Br.Heat] Early: rain, Most of day: Sun w. scattered clouds, very light haze, no jet trails.


August 1st 57.62+.53, 634.98 - 13.89 [67207@11:00; bath; 67208@21:00] Mostly Sunny, no jet trails. Forgot I had turned 2 inverters off previous evening, until late morning.
2nd 66.90+.44, 641.21 - 15.95 [67213@20:30] Sunny
3rd 75.19+.43, 646.98  - 14.49 [55Km,chj.car slow(oops, into evening); 67222@20:30] Mostly sunny
4th 84.59+.52, 653.39  - 16.33 [55Km,car chj.@1500W; bath; 67233@20:00] Sunny, warm day for the Tlell Fall Fair!
5th 91.79+.45, 659.80  - 13.76 [finish car chj; 67240@21:30] Sunny. Oops, left 2 grid ties turned off again. (Wouldabin ~16 KWH)

KWH-  # of Days (July)
3.xx  - 1
4.xx  - 2
5.xx  - 1
6.xx  - 3
7.xx  - 5
8.xx  - 6
9.xx  - 4
10.xx- 3
11.xx- 0
12.xx- 3
13.xx- 1
14.xx- 1
15.xx- 1

   As there wasn't a single full sunny day in all of July, the average value of "full-sunlight collection" in July is rather speculative. The few sunny days in June were around 18 KWH. The best sunny day in May, the 27th, was 16.96 (17 KWH by any other name.) In early August (finally sunny) it only hit about 16 KWH, and that was with the 12th panel connected. So perhaps it would be fair to say that the average maximum that might have been expected in July (with 11 panels), had the days been sunny, would have been about 16.5 KWH.
   Total power generated in July was 273.98 KWH, divided by 31 days was 9.13 KWH/Day. 9.13/16.5 = .5535. So the overall collection in July, through all the clouds and overcast weather of the whole month, was nevertheless about 55% of full-sunlight collection. It doesn't get much worse than "not one sunny day all month". It's said that the cloudy west coast is the worst place to install solar, but given the low prices of solar equipment these days, it's worth while anyway. And if our electricity wasn't being subsidized, it would be much more worth while and there would be solar panels springing up everywhere. (Much of the power for this island is made by diesel generators and the cost for generating that is said to be 50¢/KWH, but we only pay around 12¢ like the rest of BC.)
   And This July was nothing like sunny July 2018. Collection would have been much higher last year. The 31st mercifully had a lot of sun and some warmth, and it was good to see figures of over 1400 watts coming from the six house roof collectors plus two on the lawn (the second one placed just that day), instead of 400 or 800, or even 200 on the dullest days. (And August has started out sunny. If the sunshine keeps up, I might just get some corn by this fall!)

Monthly Tallies: Generated KWH [Power used from grid KWH] ("April 0" = March 31)

March 1-31: 116.19 + ------ + 105.93 = 222.12 KWH [786 KWH]
April - 0-30: 136.87 + ------ + 121.97 = 258.84 KWH [608 KWH]
May  - 0-31: 156.23 + ------ + 147.47 = 303.70 KWH [543 KWH] (11th solar panel installed on 26th)
June - 0-30: 146.63 + 15.65 + 115.26 = 277.54 KWH [374 KWH] (36V, 250W Hot Water Heater installed on 7th)
July  - 0-31: 134.06 + 19.06 + 120.86 = 273.98 KWH [342 KWH]

5 month total: 1336.18 KWH made; [2653 KWH consumed from grid]

   Power generated that went to the grid from my ad-hoc arrangements was not subtracted from the utility meter readings or otherwise accounted for. By the end of July I was looking at getting & installing approved equipment to 'legitimize' my system. (A 1200 watt approved grid tie inverter I'm considering is over 500$. I paid a bit less than that for all four 1000 watt Y-Solar grid tie inverters. And the Y-Solars just plug in. That was actually their biggest attraction and the reason I got them actually working. The 'proper' one has to be 'installed' and fed to its own double circuit breaker in the electrical panel.)

Electricity Storage (Batteries)

Dried Out Dry Cells

   I had got the theory about 3 years ago that the NiMH dry cells quit working because they had become too dried out inside. And I thought that I should try throwing them in water for a while before discarding them entirely. Having opened one "dead" NiMH "D" dry cell and determined that it was indeed quite dried out inside except right near the middle, on the 6th I took a bunch of the others from the garbage pail and threw them in a barrel of water, about 3 feet deep.
    A few hours in water didn't help. A few days didn't help. 2-1/2 weeks didn't help. Maybe a long vertical pipe full of water, so that there was some pressure pushing the water into the cells? Or maybe once they wouldn't charge, it was too late.

   Anyway I'm very much souring on NiMH "D" dry cells for high capacity storage. The ones I'm using for my solar storage work, but they hardly seem to have 10 or 20% of their original rated capacity. Admittedly the majority were somewhat abused in the Mazda RX7-EV and they're five years old, and admittedly I'm not charging them to maximum "float" voltage and capacity any more (trying to get them to last): 39.5 volts rather than 42.0. (1.35 V/cell rather than 1.40) But dry cells in charging and discharging I think just gradually lose what little water they start with. I don't think my drills with NiMH sub-C cells run as long as they did a couple of years ago either. Neither does my 'netbook' with a lithium battery run even half as long as when it was new 2-1/2 years ago.

   For serious energy storage, better to make my own batteries - and not dry cells!

Turquoise Battery Project: Ni-Zn or Mn-Zn in KOH ? or in KCl ?

[imidj]    I wondered what sort of cost projections I would be facing to do Ni-Zn or Mn-Zn cells. Without wanting to do anything like a detailed analysis, I found this chart of "cost goals" on line, from the US DOE.
   The "dissing" of Ni-MH big flooded cells has apparently worked so well that they aren't even mentioned, notwithstanding what a great success they were in the EV-1 et al. (Ovshinsky - and maybe Jungner and Edison - must be groaning from Mansonia!)

   Mn-Zn however has more energy per weight than Ni-MH and the second lowest potential cost listed -- but it's only useful if the zinc dendrite growth problem is solved. With my developments of June and this month I believe it has been.
   Ni-Zn has a higher voltage potential than Mn-Zn (1.7+ vs 1.4 or 1.5 depending on your criteria) and shouldn't cost much more the way I'll be constructing cells, so the price per capacity may be about the same, or perhaps even a little less with fewer cells needed to attain a voltage. (Using nickel manganates will increase the current and capacity per weight and size and cost, and I'll be using copper and conductive carbon black in urethane conductive paint rather than more costly nickel for current collector components.)
   So my estimate would be that my batteries, if "manufactured at scale", should come out in the cost ballpark indicated, 50 $/KWH. And if they are also "everlasting", who would want any of the other ones listed for most bulk storage uses?

Preparing a Zinc Alkaline Electrode

   Here are microscope pictures of the zinc in the stages of preparation as I did it on the 11th.

   The first interesting change of procedure was to etch the surface in hydrochloric acid instead of ferric chloride. With copper one must add hydrogen peroxide to get a gradual reaction to copper chloride to take place. The zinc boiled vigorously from the instant it was dipped into the acid. So I only left it in for a second. (I knew there was a reason I put safety glasses and a plastic glove on! Now, about that fume hood that any self-respecting chemist would have...) Then I washed it off with varsol. It was much harder getting the black color off than it was after ferric chloride, and I didn't get it very clean.
   I tried a second piece and left it in longer - 3 seconds. I could feel it was getting warm fast. Did it look any different? It was harder cleaning off the black off, so it probably had some deeper recesses.
   Surely the black color must be zinc chloride, ZnCl2? What else could it be? But then why didn't it dissolve in water? Another mystery! Could there be such a thing as Cl-Zn-Zn-Cl , Zn2Cl2 ? Oh... It's most probably zinc carbonate. First zinc oxide on the surface from the air, then it gets a carbon dioxide from the air. 2 Zn + O2 => 2 ZnO ; ZnO + CO2 => ZnCO3 . Why is there so much after etching it? Is it already there, just in unseen concentration until some of the pure Zn is etched away? (Looked it up. Nope. ZnCO3 is translucent white crystals, a zinc ore called "smithsonite". Still a mystery!)

Microscope images of zinc surface

1. As-made, straight off the roll of "moss killer" zinc for roofs.

2. Scratched up with nylon scouring pad ("Scotchbrite") to clean and make some surface convolutions.

3. Etched with hydrochloric acid (~1 second) and cleaned with varsol.

4. Etched 3 seconds & cleaned. (I think it was a little thinner - ?)

   The "zincate" ion, sodium-zinc hydroxide, had made a pretty good fuzzy plating on the zinc sheet. The video had mentioned doing it with acid to get a smooth zinc sheet plating. I decided to try in between - zinc as a chloride salt. And if I was next going to try plating the zinc in a bath of zinc chloride, why worry about getting the chloride off now? It was just more zinc to plate on - or in this case, back on.

[imidj]   First thing was to make the zinc chloride bath. That was a no-brainer: throw pieces of zinc into the hydrochloric acid until it stopped devouring them. I poured 105 cc of acid into a 250 cc beaker and dipped in a piece of zinc. It was immediately a roiling boil and the piece got hot. Hmm, an open window wasn't good enough. It took it outside. I dipped it in a few seconds at a time, and soon the bottom of the piece was eaten away. After a couple of those, I started cutting the zinc into very small pieces and tossing them in, one or two at a time. Unquestionably zinc has a lot of energy in it when it reacts. as does acid of course. When it was headed for 20 grams of zinc dissolved, the reactions slowed visibly. The 24th gram hardly bubbled and the bubbles were so tiny the liquid turned 'milky'.

[imidj]   I had put a big pail over it to dissuade any curious crows or ravens. Every time I tipped it away I could see a cloud of hydrogen gas come out. As things slowed down I put bigger chunks of zinc in, and I went inside for a while. A while later gram 26-27 was floating on top and still bubbling away slowly. It became apparent that there would be no abrupt end to the reaction - it would just get slower and slower before it petered out. I threw in a bit more, 3 little pieces making a little over 30 grams of zinc - 25 theoretical amp-hours worth. Thin slivers of it remained, and another chunk I threw in hardly seemed touched. Finally pH rose to 2, and the reaction seemed to have stopped. 95 cc remained, so liquid lost in all the bubbling was maybe 10 cc, or 10% give or take a few.

[imidj]   The black stuff stayed as solid flakes floating on or drifting in the liquid. Obviously it wasn't chloride at all - it was probaby zinc oxide or zinc carbonate formed by air on the surface of the sheets. (But then why wasn't it black before the chloride? Zinc oxide powder is very pale yellow.) Or (horrors!) the zinc wasn't pure and it was some foreign impurity. Is there an insoluble chloride? Anyway it was guck I could filter out. I just hoped no foreign substance that would have odd effects was dissolved into the liquid. (I'd have said "negative" effects, but that's what we want. What we don't want is any positive effects in a negative electrode! Or I could have said "deleterious" effects, but that's so long! Maybe somebody should rename the polarities the "yin" and the "yang" charges or something.)


   I took a small funnel and a piece of coffee filter and poured the 100 cc of liquid into a 100 cc beaker, which was then deep enough to stick the little zinc electrodes into vertically. If I put a flat plate plus electrode vertically right next to the flat plate minus electrode, might it give an even plating instead of doing mainly around the edges? For that matter, would the plating have the desired porous consistency anyway?

[pitcher]   I arranged for two zinc battery electrodes, one on each side of a sacrificial zinc electrode. I cut two more pieces of the plastic grille to keep them from shorting. Then I went back to the shop and the varsol to better clean the electrodes. The black obviously wasn't zinc chloride, so I really did want to get it all cleaned off. I took an old toothbrush to scrub with this time. Iso alcohol? Toothpaste? Baking soda? Scotchbrite? Nothing seemed to work and I finally gave up.
   I put the 'trodes in the liquid and turned on 1.5 volts. By the time all was arranged and running for a while, it was doing over 4 amps. That seemed like a lot of current, but thinking about it... if I wanted to electroplate amp-hours of zinc onto the zinc plates, it only made sense that it was going to take amps for hours to do so. ...or would it just grow dendrites and short out? Sure enough, it soon went up to the power supply's limit of 5 amps. If I pulled the center (sacrificial) plate up and down, it would drop back to 4.3 amps. But I had to do that repeatedly.

[foto]After 15? minutes the plating made it very apparent where the holes in the grille were. At least the black color was gone.

The sacrificial zinc electrode.
It turned black, and black flakes fell off of it.

Microscope 5. The uniform plating on the main surface an electrode after 15? minutes.

6. The fuzzy dendrite plating near and at the bottom edge. Dendrites growing up
and those growing down over the edge couldn't all be brought into focus at once.

7. Plating where the mesh was and in the holes in between.

8. Fuzziness squashed down some in press.

9. Fuzzy bottom edge squashed. (Is this the right slide?)
The electrode picked up bits of rust (or paint) from the steel
blocks I squished it between - oops. But rust should just sit
or else turn into metallic iron - it has a lower voltage than zinc.

   These images were all at 40x magnification. The relative coarseness shows that zinc structures build and dissolve electrochemically at "micro" scales - meaning sub-millimeter - but nothing like a "nano" scale that would be hard or impossible to see in the microscope even at 800x. Those ultra-fine zinc powders I bought in years past were much finer than what I was now getting by electroplating processes.

I decided to put just one battery electrode back in with two sheets of grille between it and the other. Current started at 2.54 amps. Perhaps 5 amps for two wasn't a short circuit after all? Maybe it just was that high doing the plating? I turned it around to do the other side. Current rose to 3 amps... of course! as it plated, the electrode gained surface area, so the current went up. Jumping up to 5 amps and hissing as it was still doing, however, was surely dendrites shorting the electrodes.
   Then again, it started becoming quite plain that much of the plating was happening right at the edges. They were very fuzzy and it was obvious the fuzz was getting fat enough to short to the other electrode. It would seem some sort of strip around the edge was needed. The zinc at the edges couldn't grow larger than the slot it was in, and it would limit electrolyte circulation in the edge areas. Perhaps the electrodes should also be farther apart to slow it all down? Would that make it more even? Perhaps a box, where all the electrodes lid down into slots along both edges, each spaced at a distance so as not to touch each other. This seemed to be moving from the area of experiments into production issues. I suppose it already was when I decided to try another liquid for electroplating even tho the first one worked.
   Enough for that one! I could scrape off the edges, and it had some plating across both faces. It was only about 1/2 hour's worth, but I pulled it out and put the other one back in, with the previously back face toward the other 'trode. 2 amps rose to 3.5... then limited at 5.0 again until disturbed.
   The alternative to taming the edges was if the entire electrode could be so fuzzy. Then I would press it to a uniform thickness and chop the edges to a set size. Worth considering!

   In the meantime, even if not plated the sheet metal zinctrodes worked quite well, and now they had some sort of plating. On to the next and perhaps most vital experiment.

Egg Albumen Coating

   After all these preliminaries I was back to the idea from last month. I noticed that the ester coating seemed to deteriorate, probably through electrolytic action, and so did the whole electrode surface, as evidenced by the break of the first one at the water line. Yet no dendrites had formed. Did the ester leave the osmium embedded in the zinc surface and that somehow prevented dendrites?
   But something I had thought about many times was that egg shells, and the outer and inner membranes, allow oxygen to enter the egg, carbon dioxide to exit, and yet prevent entry of bacteria and escape of the egg's liquid. Here again we might gain an ion selective membrane that would keep the zinc out of contact with the electrolyte and so prevent it from dissolving into it. What I didn't see was a way to harvest these layers from an egg and use them inside a battery.

   The one layer that is accessible is the eggwhite, the albumen. Perhaps, if the alkali didn't dissolve it, egg albumen once dried could form a membrane to keep the zinc out of the electrolyte and yet allow the ions in to react? It sounded like it might work. Or perhaps the albumen in combination with the osmium doping would prevent deterioration as well as dendrites? Mañana!
   (12th) I got the powdered egg albumen out of the fridge. I went to get some distilled water to mix it in. Was that what I should use? Then I thought that if I used an actual egg, it was already liquid without adding anything.

[eggy]   What luck - I had eggs! I cracked one into a bowl. The yolk broke, but I managed to dig out a couple of teaspoons from the other side of the bowl, and put them on a dessert plate. Then I put the electrode on the plate in the egg white, and used a small paintbrush to coat the top and edges. I got out a small vise and turned it sideways to hold the electrode level by its terminal tab. There was some left and I got the other electrode and coated it too. (So much for doing the osmium doping first on one of them! Not very methodical, am I?) Should I cook it, or just let it air dry? I decided on the latter, again for both. It seemed to me cooking it would wreck it.

   The rest of the egg I put a bit of cheese on and poached in the microwave, and had on a toasted bagel with chard and green onion from the garden, and (store) tomato. Nice!

   There they were, two zinc electrodes, coated in drying eggwhite. Could that be all that was needed, after 11-1/2 years of trying to figure something out? After over 2 centuries of battery making and dozens if not hundreds of patents by as many researchers for how to sort-of solve the zinc dendrite problem? It remained to be seen.

   It occurred to me that if the electrode already had an insulating layer of albumen, there was - in theory - no need for a separator sheet between electrodes. That might be leaving a lot to chance, but perhaps I would use a thinner separator sheet than I would otherwise have dared to use - some thin white polypropylene non-woven cloth (thin "landscaping fabric".)

   When I looked at the electrode surface under the microscope, the egg layer which had seemed so sticky when wet looked to be vanishingly thin except where there was some fuzzy zinc holding a glob, or where there had been a bubble, which there had been several of. It might be a good, thin membrane, or it might be that several coats are required for good isolation of the zinc. (I'd better buy some more eggs!)

10. Virtually transparent egg Layer

Assembly and Testing

   I made up a batch of electrolyte I'll call the "20-20" mix. It was 100 cc of distilled water to which I added 20 cc of methyl hydrate and 20 grams of potassium hydroxide flakes. The methyl hydroxide, contrary to my original expectations, doesn't work as electrolyte. It doesn't carry current itself, seems to cause rather severe self discharge, and it even seems to inhibit the potassium hydroxide from conducting current.
   But the purpose here was to condition the eggwhite surface.

   In the evening by which time I thought the eggwhite must be well dried, I put one of the electrodes into the same little cell with the same little pieces of nickel hydroxide electrode from the dry cell. I folded a piece of the polypropylene cloth around it, but to protect the eggwhite. I inserted the same grille as before between the electrodes.

 It started about 1.63 volts and started dropping bit by bit. I conceived that perhaps the egg layer needed some charging and I hooked up a 1.9 volt charge. It started at 50 mA but went down quickly to 20 mA or so, and after a bit it was below 10.
   I shorted the current meter across the cell. 150 mA, dropping to 90 in 10 seconds. Well, let's see what a little charging would do. Of course, it wasn't charging in any hurry. I left it off, and in the morning (13th) it was sitting a little over 1.2 volts.

   I drained the liquid. It was even more of a purple color than the zinc chloride, and nothing like the yellowish sodium zincate. I'm confused. I refilled the cell with distilled water, and did that 3 times to dilute out whatever might be in it. Then I refilled it with the 20% potassium hydroxide (20 g KOH in 100 cc H2O) electrolyte.

   After a bit of charging, still at very low currents, it seemed it would steadily put out 200 mA. It recovered voltage quite slowly. The electrode was 33 x 55 mm, so 18.15 sq.cm, so that was just 11 mA/Sq.cm. At that rate it was going to take a lot of electrode surface to get good amps!
   Then I remembered the "+" side had acted up before and I wiggled the "+" terminal with pliers. Charging current went up and the short circuit drive became 290 mA. I was going to have to do something about that side before coming to too many conclusions. And I was going to have to look a long way back through TE News issues to find a good positrode formula!

Antimony Sulfide or Zirconium Silicate?

   After a couple of hours, the short circuit current hadn't gone up much, but it would drive a 15 ohm load at 150 mA/1.5 V. Definitely the zinc had started out fully charged and it was the nickel hydroxide side that had to charge. So it occurred to me that the zinc side must be bubbling hydrogen, and that hydrogen could surely get through the eggwhite layer. That meant the zinc would be turning to zinc hydride. Ugh!
   If the idea of isolating the zinc from the liquid was working, it didn't prevent all problems. I had once raised the hydrogen overvoltage of manganese to the point where it would hold a charge. With zinc, even in flooded cells it would probably be be advantageous to raise the overvoltage. Was there a way to incorporate a trace component into a plated plate of zinc instead of a powder electrode like the manganese ones were? One can perhaps plate a metal, not a compound. Perhaps I could plate zirconium? Maybe I could just paint powdered zirconium silicate on/into the fuzzy plated electrode surface before the egg? Mixed into the egg?

   (Meanwhile, the zinc chloride I had made with zinc and hydrochloric acid and which was still quite acidic last checked, appeared to be growing a white mold on the surface. In less than one day? In a beaker with a rubber stopper on it? Egads! I spooned it off, whatever it was, and it didn't reappear.)

   I hooked up a 60 ohm load... and forgot about it. Some hours later, the cell was very discharged.


   I reconnected the charge and it started charging with quite high current, almost 1/2 an amp and over 1/4 amp for a while. The voltage slowly came up from .3 volts to 1.9. By then the current dropped to 30 mA. I stopped charging and shorted the cell - over .3 A for 10 seconds.
   Soon there was some weirdness. I took the cell apart and found dendrites - nearly all right at the fuzzy bottom of the electrode. They had grown through the PP cloth, and one in one corner had grown across the separator grille

Microscope 11. dendrites at the bottom

Separator cloth with dendrites

Microscope 12: Patchy surface

   The surface of the zinc sheet was patchy. This seemed to follow the pattern of the PP cloth and of the separator grille. Was the cloth absorbing the egg and leaving the zinc bare? protecting the egg, and the zinc was bare where there was no cloth? Should I try painting the cloth with the eggwhite?

   I wasn't ready to give up the theme yet. Noting how vanishingly thin the eggwhite seemed in places, I suspected it would take more than one coat to fully cover the zinc. It was certainly easy to imagine the eggwhite shrinking back as it dried, and leaving some or even all of those protruding points exposed. Maybe even several coats would be necessary. A good, thick layer. Then too, chickens weren't the only eggs. Most bird eggs are pretty similar in color and taste. An exception is duck eggs.

Duck Eggs

   Some friends had mentioned they would have duck eggs for sale. They have darker, richer yolks that some like better and others don't. At first I thought, nah!... but on looking them up, it seems duck eggs are more oily and more waterproof than most bird eggs. Might some difference apply to their whites too? So I bought a half dozen.

Electroplating Rack

   I also gave some thought to electroplating the zinc electrodes. They had once again plated mainly around the edges and especially at the bottom where they sat against the glass. How might I make them more uniform? The first thought was to make something so the sheets had slots around the sides and bottom, limiting the extent to which the edges could grow. If I was to do that, I would probably want to make something that would do several electrodes at once, of my intended production size. The second thought was that the sacrificial electrodes could be smaller, and not extend out to the edges of the sheets being plated. Then the currents would go more to the centers. So the plastic cube would have slots, and other slots for the sacrificial electrodes between each of the electrodes being plated as well as on the outer two faces. I decided to make one that would fit in a particular jar - Adams Peanut Butter 1L. They have the widest lid of any glass jar I've seen. Hoping I wasn't wasting my time, here would be the first little piece of equipment to make production go better and faster.

[ajar]   I made the rack the next day (15th) from some scraps of plastic. It just fit in the jar - after I sanded the corners off. And I cut some bits off the top so the lid could go on the jar with the rack still in it. My intent was that it should hold up to 3 battery electrodes between 5 smaller sacrificial electrodes. But having got 4 slots, it seemed easier just to put in one battery electrode with equal size sacrificial ones on each side of it. Especially just to try it out.
   Some zinc chloride had settled on the bottom of the beaker, so obviously it was too concentrated. But 100 cc became pretty diluted, I'm sure, in a 1 liter jar filled almost full with distilled water. Nevertheless, when I connected power, it started out at 3-3/4 amps and went up from there. Well, if it's to have that many amp-hours of zinc plated onto it, it needs that many amps for an hour. Soon it started bubbling. Let's see...

Positive: Zn + 2 Cl-  =>  ZnCl(aq) + 2 e-
Negative: ZnCl(aq) + 2 e-  => Zn + 2 Cl-

   What part of that is bubbles? It must be hydrogen. Lets see... Zn in pH 14 alkali is -1.28 volts. Zn in neutral salt is ~ -1.0V. I turned the power supply down from 1.5 volts to 1.0. The current dropped to 2-3/4 amps. Hmm, the sacrificial zinc could be bubbling oxygen and making zinc hydroxide, which would be balanced by the negative side bubbling hydrogen. Of course. And then it would all turn more and more alkaline with hydroxide as water was converted to H2 + Zn++:(OH-)2. Probably to around pH 12. So what was the chloride for, again? Well, that would be in there too! Maybe it would plate with a more favorable porosity?

[foto]   I pulled two of the three electrodes out. The sacrificial one had some black on it. The one being plated seemed to be getting a nice uniform coat except around the edges that were in the slots, and various places that obviously weren't clean. Apparently the roll of zinc had got wet and had a discoloration all along one side, and I hadn't etched these electrodes, only buffed them with scotchbrite. It obviously wasn't a good cleaning.

   Now I tried scouring with baking soda but nothing would clean off that line. I dipped it in the old, weak ferric chloride. Then it was black. But if the black was zinc chloride, and I was using zinc chloride electrolyte for the plating, did I need to clean it off? I put it in the tank and hooked it up. I turned the voltage up to 1.1 and it was about 2-1/2 amps.
 After a few minutes it didn't really look very good, so I dipped it again and then cleaned it with varsol. It still didn't look very clean. But I put it in the plating solution again. The plating didn't look perfect, but it was more even. The pH was 5. All 3 electrodes had lots of very tiny bubbles coming off them. I turned it down to .9 volts, and then .8, but it seemed to be working much the same. I left it for maybe a couple of hours(?) and worked on somthing else.
   When I came back it seemed to have a nice, even plating - not a bunch of dendrites. The side edges in the slots had almost nothing, but it seemed the bottom edge hadn't gone into the slot and it was coated - but this time pretty evenly compared with the rest - even a bit lighter. The the original sheet was .3 mm thick. The plated electrode was 1.2 mm. Before coating the piece would have been about 18 grams. After coating it was 25.55 grams (wet) so say 25 grams. That would be a theoretical 7 g * .820 AH/g = 5.74 amp-hours added by the plating - in practice hopefully a couple of amp-hours on each side. And it could easily be made thicker by putting it in longer. It was an excellent result.

   Instead of making this one thicker, I dipped the first one with the splotchy plating in the ferric chloride, cleaned it off in varsol (still didn't look clean), and put it in to plate. Starting current was just under 2 amps, which rose to a little over 2 amps, still at .8 volts.

   The rack is great and I'll use it for now. But having used it now I have a "next version" in mind.

* There seems to be no need to have the sacrificial electrodes smaller in area - the coating is pretty uniform.
* So then, the 3/8 to 1/2" spacing of the slots can be reduced to 1/4" or so - no smaller electrodes between the ones being plated, just an evenly spaced stack with every second one being sacrificial.
* Second, the edge slots will be narrower so the plating comes closer to the edges of the sheet. (Caution: the other roll of zinc is 1 mm narrower than the first one, so don't make it too close to the edge! How variable is the width of the rolls?)
* Third, the bottom slots will have tapered entries so the sheets don't get stuck just before they enter.
* Tapered entries to the top of the edge slots would be better, too.

   It still looked pretty splotchy, so I made two new electrodes, dipped and cleaned them as best I could, and put one in. I left it on a couple of hours at 2 amps, .8 volts. The coating was thick, but soft and flaky. It rubbed off easily. It was a bit like fluffy trees growing up from the surface, and only one height could be brought into focus at a time under the microscope. One side was pretty uniform, but the other had some missing patches.

Microscope slide 13. Heavy zinc electroplating

3 faces of the 2 plated electrodes

   If the fragile plating could be coerced to stay where it was, it certainly had a lot of surface area for a relatively small volume. Obviously the way to get it to stay, if it would work, would be a thick coating of eggwhite encasing it all.
   The thickness went from .3 mm to 2.3 mm, yet the weight only (almost exactly) doubled. That meant that .3 of 2.0 was zinc, 15%; the other 85% was air spaces. That certainly must mean lots of surface area for electrochemical reactions.

   The weight increase asked a question: how did it go from 17 grams to 34, adding 17 g * .820 AH/g = 14 amp-hours with around 2 hours plating at 2 amps - 4 amp-hours? That doesn't make sense to me.

[foto]   The sacrificial electrodes were now almost "aluminum foil" thickness and had gone from 17 or 18 grams to 9 and 11, losing about 17 grams of zinc. (Quite even un-plating - no holes in them yet! Oops, there are a few pinholes in one patch on one. That whole patch will surely be gone next time!) The other few grams for the first one must have come from the zinc chloride already in solution. (The excess becoming exhausted probably explains why the currents went down from 4 amps to 2.)

   Examination of the previous electrode showed that it was really much like the latest except thinner and stiffer. The start of the dendrite growth with air spaces between was there. It would make a good electrode.

   I got out a duck egg. It looked suspiciously like a chicken egg, but the shell was a lot harder and thicker. The membrane inside it was thick and rubbery. I poured the 'white' into a small ointment jar. I spooned a little zirconium silicate (AKA zircon) into a small plastic cup and added some eggwhite. I put the lid on the jar and put it in the freezer. I mixed the eggwhite and the zircon in the cup and painted it onto both the small electrodes (which already had been painted with chicken egg), including the one that had shorted with dendrites. I considered doing one of the big ones just plated, but decided to stick to practicing on the small ones.

A Positive Electrode

   The performance of the cells, and especially the last one, seemed to be limited by the pieces of NiMH dry cell I was using for the positive electrode. What with take-aparts and reassemblies, the conductive nickel 'tape' had come loose and off and it just got worse.
   I got a piece of nickel foam out of the cupboard. With pH 14 alkali, at least I knew nickel was one metal that would remain intact in the positive electrode, so here was a good current collector. IIRC it was 97% air. What should I fill that 97% with? I picked the first 'positrode' mix from TE News #73, calculated to make a good nickel-manganate electrode compound. But I would use the nickel foam instead of carbon foam. When I actually mixed it I used less carbon black (5 g vs 20 g) because it was so light and fluffy that a little weight was a lot of volume. It was the only thing that filled the 50cc beaker to overflowing, with just 5 grams. Also a little less samarium oxide (5 g vs 7 g) because it just seemed like so much of what is just a trace ingredient, compared to the rest.

Monel Powder    - 16 g
Ni(OH)2 - 17 g
KMnO4  - 40 g
Graphite Powder - 5 g
Sm2O3  -  5 g

total 83 grams instead of 100. The proportion of nickel to manganese for forming nickel manganates was the main consideration.

   I then tried to make a porous 3D printed plastic shell for the electrode, but without success. More on that is below.

   Finally I decided to just use another piece of the NiMH dry cell nickel hydroxide electrode for the time being.
   I put it together with parchment paper wrapped around the nickel parts and no grille between electrodes. The short circuit current was disappointing, around 200 mA, but it kept up that level. Probably it can be improved. But better low current and a battery that lasts than high current and a short cycle life. (I looked up last month and I see it was only after some cycles they started getting the higher currents I was remembering.)
   I charged it some hours, then turned it off. Some hours later (20th) it read 1.593 volts. But it still put out similar current when shorted, so it wasn't much discharged. Since they charged to over 1.7 volts before, something seemed to have changed. I suspect that with the zinc no longer in direct contact with the electrolyte, the pH level at the zinc is reduced. The reaction voltage of zinc drops with the pH. A 1.6 volt level resting might suggest pH of 8 to 11(?).
   I tried a discharge test later with a 60 ohm load. The voltage dropped over the seconds instead of the minutes, and within 5 minutes it was down to 1/2 a volt. Over a couple of minutes it recovered to 1.6 volts and more gradually to 1.65. That doesn't say much for the current capacity! Next question: was the problem in the nickel electrode chunks, the parchment paper or the duck eggwhite? Removing the paper and putting the grille back in made no notable difference. Neither did replacing the coated zinc electrode with a plain piece of zinc. (Of course, it was now just a flat smooth surface.) I put a piece of nickel foam beween the nickel hydroxide and the cupro-nickel collector plate hoping for better conductivity. There was no evident difference in the short circuit current, but the charging current went up substantially and it took many minutes to get up to 1.9 volts instead of many seconds. So the problem was probably the nickel electrode.

Parchment Paper Separator

   I note that the parchment paper seems to be a very good separator paper. It still seemed good after a couple of uses and it held the little bits coming off the nickel oxides electrode contained. (Now... does it gradually deteriorate, or last "forever"? Looks like the latter so far.) But with the porous plastic positrode shells to hold the "trodes" apart I decided to use the also paper thin non-woven PP cloth as being likely to conduct ions more freely.

3D Printed Positrode Shells

   But whatever the mix, how was this positrode to be held together? The nickel foam was only .5mm thick. The dry cell electrodes had been together packed into the cell so long that they tended to hold together by themselves even in the open. My experience with my own powder electrodes has been that the compacted "briquettes" tend to expand into whatever space they can find once wetted in the cell. And yet, I didn't want them packed tightly against the zinc electrodes. But all the extra metal in pocket electrodes seemed so wasteful.
   The "plastic pocket" electrode idea came back to me on the night of the 18th. What had I been trying to achieve? porous plastic sheets for pocket faces. What had I been missing all this time? All I needed to do was thin out the extrusion with a one-time (per session) manual command to the printer until the plastic was thin and porous instead of "pretty solid": M92 E750 (or whatever), where M92 E1000 was the "neutral point" for the amount of plastic filament to extrude and E750 would only print 3/4 as much, which should leave tiny gaps in the "weave". When I checked I had actually done that before so I was revisiting old territory, but now I had a couple of different ideas.

[picture]   When I had the thought I immediately went and turned on the laptop computer I use with the 3D printer and started on a better design. Instead of making it so the electrode "innards" are slipped in through a gap at the top as I had done before, they would be placed in the bottom half shell, and then the top half would be glued on with methylene chloride. That should make for a much tighter fit with no leaks.
   Another idea is to put some paper inside to line the plastic. That way no little gaps in the plastic can leak material. And between the paper and the plastic lining the active electrode, the electrodes in the cell could actually touch each other - stack against each other, in principle without risk of anything shorting.
   But the shells wouldn't have the "porous" gaps that I was hoping for. The second layer was fine, but the first layer gets smeared against the glass printing bed. If it doesn't the part comes loose. The smear effect ensured that there were few gaps in the first layer.

box   I made a second, improved box. The first had been two halves. This one was one piece being a box with sides, and the other a flat cover. The full height (electrode thickness) sides would contain the electrode better during assembly.

[foto]   But the box was only porous as intended in one area. In the rest the plastic was all merged.

   Then I clicked on a command to heat up the extruder while the "skeinforge" was turning the 3D .STL file into .GCODE printer commands. I've never made any interruption before during the skeinforge "slicing" process. Apparently it didn't like it. Nothing worked after that. The temperatures and temperature setpoints just said 0/0, 0/0 . When I tried to tell it to move it said "Printer halted due to error. (Temperature reset. Set temperature before issuing M999 to restart printer.)" I didn't even know if it was the computer software or the printer. (Later I found that string of text in the source code for the printer itself, so it was the printer returning that message.) Anyway, setting the temperature of the bed or the extruder did nothing. Perhaps the temperature setting commands too were ignored because the printer was "halted", so it couldn't fix itself. (Incomplete testing of the printer firmware? It was an old 2011 version.) The temperature setting return from the printer continued to say "0" along with the other numbers. (And didn't they used to say "-273" initially before they were set? Why now "0"?) I worked long into the night, and I was unable to get anything going on the 19th, either. There was almost nothing on line about the problem. What there was said to fix the extruder thermistor. But the thermistor had been fine. It read 2.5 volts, which seemed reasonable.
   I was now getting into the "Marlin" source code of the firmware program inside the microcontroller of printer itself. No doubt as a programmer I could eventually get to the bottom of this. But it wasn't anything I wanted to get involved with. What frustration!

   Regardless, by the 21st I had come up with a modified plan for the shells: Make the bottom layer, which can't seem to be made porous, into a grille, just thin lines going across. That would definitely have holes in it, albeit too big to stop particle leakage. Above that (inside of that in the electrode 'box', is optional. (It might take a coon's age to print that bottom layer. So what?)

Option 1: Add two more layers, which would come out with crosshatch pores if printed thinly, on top of the grille. Thickness would be 1.2 mm.

Option 2: The grille could be continued for a second layer, line going crossways to the first - .8 mm thickness, and a sheet of parchment paper could be layed down flat on the inside. (3 or 4 layers, making the grille 1.2 or 1.6 mm thickness, would make it stiffer.) A solid strip (no grille) around the edges would prevent leakage at the edges of the paper. The parchment paper is only .04 mm thick, so even several layers of it wouldn't add much thickness. The plastic would be extruded at normal thickness.

   The thickness is important. If the electrode is say 2 mm thick, then 1.2 mm for each face, 2.4 mm, would mean there's more plastic than electrode. OTOH a short circuit through a sheet of parchment paper plus 1.2 mm of plastic grille will be pretty unlikely in the absence of dendrites forming bridges.
   I think of stiffness as being important to hold the electrode powders compacted together. But if the cell was wholly filled with alternating plastic encased oxide "+" and zinc "-" plates stacked against each other, the stiffness of the plastic only has to hold until and during cell assembly, and before it is wetted. That shouldn't take much. (And it shouldn't take much electrolyte, either.)

Option 3: Get the "Anycubic I3 Mega" printer software working and see what could be done with that printer. It prints finer beads than the old RepRap.

   I set it aside for a while.

Ethaline Deep Eutectic Solvent (DES)

   I found the following text reading a write-up on zinc oxides that has been on my drive for some time.

(ZnO_Zn(OH)2-info.txt - source uncertain)

"The solubilities of zinc oxide and zinc hydroxide are at a minimum at pH 9.3. At this pH, the solubility of zinc hydroxide is 0.0822 mg/L for the amorphous form and 0.0041 mg/L for the <greek-e>-Zn(OH)<INF>2 form. Zinc oxide and hydroxide are insoluble in organic solvents, including alcohols and acetone (Refs. 3, 5-9)."

 At first glance the mention of solubilities at lower alkaline pH would seem to be in conflict with Pourbaix pH diagrams showingZnO and Zn(OH)2 to be insoluble from about pH 7 to 13. However, 82 and 4.1 micrograms per liter could be said to be pretty much "insoluble". OTOH, my unprotected zinc electrodes seemed to form dendrites at any pH, which I hadn't expected.
   But the really interesting part is that it says it's not soluble in organic solvents... perhaps like ethaline DES? Perhaps all one needs to do to prevent zinc electrodes from dissolving is to use that electrolyte? (Wouldn't someone have found that out by now? But I didn't see much of relevance on the web. And how soluble is "insoluble"? Do they really mean zero PPM?)

   Until now I had thought that the key advantage to ethaline was that it had a wider voltage range than water. A chart shows ethaline as being stable from about -1 volt to +1.25 volts, but gives no indication of change with pH. (Water is only -.8 to +.45 at pH 14.)

   But if zinc was really insoluble in ethaline it would solve the dendrite problem. That was a different tack than I had been taking, but it seemed well worth exploring. The eggwhite might work fine, but it wouldn't need much of a gap in the coating to annul the effect, and it wouldn't be exactly a hard thing to scratch a slot in accidentally during assembly. This should be more reliable.

   On the 21st I decided to try it. I took the cell apart and rinsed everything off, then let it dry. I got out the ethaline, and found there was only 15 cc of it. I added 3 grams of potassium hydroxide flakes - 20%, as before. For a while I wasn't sure it would dissolve. If not, what would I do? But very gradually the flakes started turning to mush and I thought after an hour there was less substance on the bottom. (Certainly no worries about the reaction generating too much heat like it does in water!) In several hours the lumps were all gone but it still hadn't clarified. I used it anyway. It poured like molasses.

   In the meantime, I wanted a new, electroplated, non-coated zinc electrode in the test cell size, so I set about making one of those. Before electroplating it would have been around 4.4 grams. After it weighed 6.1, so 1.7 g of high surface area plating, a theoretical 1.4 amp-hours - .7 AH on each face.

   By evening I finally put the cell together. Currents were very disappointing - 5 mA charge current. 20 mA initial short circuit and dropping by the second from there. It might work fine and grow no dendrites, but I wasn't sure it was much good for anything with such low current capacity. And the only thing between the electrodes is two thicknesses of parchment paper. Apparently the high viscosity severely limited ion flow - but that has long been noted by others. Wiggling the positive lead seemed to make little or no difference, and it didn't feel loose. A while and some charging later it would start at 35 mA shorted. It was still a long way from even 150 mA, let alone an amp or two. Was there any way to improve it? Could some thin solvent be added and mixed in to reduce the viscosity?

   Longer charging disclosed the certainty that it was self discharging, and pretty rapidly. Not good! It would discharge down to around 1.2 volts and then stay there. Shorting it said there was still some energy in it. What was happening? The voltages might be a clue: NiOOH +49V; Zn -1.24V. Might it be that it was discharging the nickel electrode, but not the zinc? Ethaline is supposed to handle up to +1.25V, but maybe there was some chemical incompatibility? What would happen if I tried an MnO2 trode with its lower +.15V reaction voltage?
   But when I went to pour the electrolyte out of the cell, nothing came out. Was there not enough to pour? Or had it turned to a jell? Turning it sideways and shining a flashlight in seemed to indicate the latter. Although the cell was no longer full**, a little fluid was visible and it didn't start oozing toward the lower side. The cell had had a reasonably tight fitting lid on it. The ethaline with KOH still in the beaker, with an even less perfect cap (the rubber stopper has a gap at the beak) was still viscous liquid.
   Here was an interesting and unexpected overnight development. The electrolyte itself seemed to be changing as the cell was being charged. What would happen if I just kept charging it? or even if I just let it sit some hours or days? I decided to let it sit a few hours before trying to charge it again. But nothing changed, and when I opened the cell on the 25th, I found my perception was wrong: the DES was still liquid, however viscous.

   I noticed something odd about the ethaline still in the beaker. It had still not clarified since adding the KOH. Of more note, there was a clarified layer 2-3mm deep on top (yellowish), then 7-8mm of unclear liquid (colorless) under it. As if the KOH had dispersed as an emulsion but not dissolved, or not entirely dissolved, in the lower part, but had dissolved or was being expelled from the upper part. This was supposed to be one liquid: why would it separate into two layers? If it did have two components, was only one of them the DES for the cell? Could the presence of the other be what was causing problems?
   The jelled(?) ethaline in the cell couldn't be modified. It surely contained both components. I could probably separate the clearly defined layers in the beaker, but there wasn't enough in the beaker to do much with. (Maybe I could fill one small cell with the cloudy but colorless part.) If I wanted to try more things, I would have to make more ethaline and new cells.

** Being so viscous, I suspect the ethaline only gradually seeped into small cracks and wetted more electrode surfaces in the cell, so the level dropped some, gradually, after I had filled it and closed the cover.

[imidj]   Also on the 25th I made a cell case for "full size" electrodes but only thick enough for two (or maybe three with two thin zincs around a single oxide one). Basically it's for testing "full size" 'trodes. But I didn't get as far as to use it. I look forward to getting amps instead of 100 milliamps.

Ethaline Electrode Jell for Zinc?

   On the 23rd a new idea occurred to me: If the ethaline would jell or solidify, perhaps it could be painted on the zinc similar to the eggwhite, and 'ordinary' water with KOH could be the electrolyte for the positive. If the ethaline didn't touch the positive electrode, it couldn't cause it to discharge. And maybe the current capacity would go up too. And jelled electrodes last 'forever'. And ethaline has a higher hydrogen reaction voltage than water, presumably reducing gassing versus eggwhite & water.
   I had already started thinking, "What might one synthesize that would do the same job as eggwhite, but probably better?"

... maybe ethaline was it? It seemed worth trying.

   I painted the best looking of my electroplated "full size" zinc electrodes with the ethaline + KOH still in the test tube (simply mixing the layers), both sides and up the terminal tab a ways. (I had been about to paint it with duck eggwhite.) I left it to dry a while before a second coat, but I don't suppose it was very dry. In the second coat I added zircon powder as an extra buffer against hydrogen generation. But after a couple of days I had the impression that the DES had simply evaporated rather than dried. After all, it is a solvent, of sorts.

   Having already mixed some duck eggwhite and zircon, I was painting the other large plated Zn electrode with that for possible comparison. But only coating one side at a time and letting it dry, then flipping it and doing the other side. After two coats and as many days, it was starting to look coated. I went for 3 coats and was still painting on the 25th. On the 24th I made a skinny battery box for just 2 or 3 of the large size electrodes - one that wouldn't take a lot of electrolyte to fill.

   But when I opened the cell I found I was mistaken and that it hadn't actually jelled. Well, then... could it be induced to jell somehow? How about agar? I got some and painted it onto the small DES soaked electrode. It seemed to go on well, but the fuzzy zinc plating seemed to come off into the goo. Not to change the topic here, It seemed to make a nice thick layer, dark colored with the zinc "powder" bits in it. But did I need to heat it as one normally does to jell agar? If I did, would it jell in DES anyway? And what would happen to the DES? Would the zinc bits poke through the jell and start making dendrites? Would the whole thing work anyway?
   It was like I was painting zinc powder back onto the electrode. I think my zinc electroplating is too fuzzy - the stems of the leafy 'trees' of zinc are too thin and so the tops break off easily. There must be some way to adjust the plating process. Maybe have the plus and minus electrodes somewhat farther apart? Adjust the voltage?

Agar Alone - Zinc Electroplating Adjustments?

   Okay, what about going back to something I tried years ago but didn't try with zinc... just coat the whole electrode with agar? (instead of "egger"... and no DES.) Could it really be that simple? Again, it seemed like something worth trying.

   So I went and read up on agar. "Ideal pH 5.4 to 5.7". Although that was under the "horticulture" section, a first question then was: what happens to the agar in KOH solution? If it doesn't like such strong alkali, it may once again be necessary to try lower pH electrolytes. Agar also has temperature state-change hysteresis: It melts at 85°C and freezes (jells) at 32 to 40°. That means once it's solid the battery could get up to 40 or 50 without the agar liquifying.

   I was making (yet another) new electrode to try with no coating except agar. (I would have just used plain zinc sheet, but I figured the agar would adhere better to a fuzzy surface.) There was another thing I could try. I had been trying plating at .8 volts. I went to go lower, but it turned out the power supply wouldn't go any lower. What about higher then? I went back to my original 1.5 volts. That should plate it fast, anyway.
   I cut the zinc sheet a bit too thick and too tall. I dipped it in hydrochloric acid for a moment. (wow that works fast!) Then I trimmed it by hand and ended up at 29mm x 66mm - 19.14 sq.cm. It would fit in the test cell easily even with a thick coating of jell.
   I plated it for a while, not very thickly. It drew 1.1 to 2 amps, increasing as the sheet plated and gained surface area.

   I put some distilled water in a petri dish and heated it to boiling in the microwave. I sprinkled in some agar powder and stirred it with a fork. In the fridge it turned to a fairly soft jell. (Later it was substantially harder.) I heated it again and part of it melted. I dipped in the electrode. Of course it solidified quickly as it cooled, and it was thick enough to just need one coat. (After days I was still waiting to paint the next coat of egg on the other electrode!)

   ...I forgot to add the zircon. Oh well.

   First test: simply dip it in KOH electrolyte and see if the agar dissolved. After a couple of hours it looked fine and felt no different! It was actually hard to see much under the microscope as the jell was almost transparent.

   Next: put together a cell with it. More chunks of the NiMH nickel oxides electrode again. It started off at 1.45 volts but dropping. Soon it was charging at 1.9 V, 30 mA, and would deliver hundreds of milliamps, increasing as it charged. The voltage didn't dive off a cliff as with the DES. After a while I saw "2.2 A" flash by just as I shorted it - over 100 mA/sq.cm. (Later I never saw more than 1.5 A.) After some charging, it would continue to put out 290 or 300 mA when shorted.

   So... a good nickel-zinc cell. The agar conducted ions, apparently, at usable but not high current. The next question was how it would hold out inside the cell with charging and discharging. Unless it disintegrated, it looked like this was a simple, high energy, long life battery. If there was a gap and dendrites formed, yet most of the coating was good, it should be just a matter of improving the coating technique - perhaps painting it on as I did in 2010.
   The rest would just be working out manufacturing with established or new techniques, optimum formulas and proportions, and any further improvements that might be found.

Electrodes jelled with agar, January 2010.      
...painted quite thickly on quite thick electrodes.   
If the negative had been zinc, perhaps I'd have   
got much farther years sooner, with fewer detours 
along the way? But that Mn negative wouldn't have 
held a charge in 2010 as I hadn't found the trace  
additives Mn requires at that time.          

   I dug back to see when I had first thought of using agar to jell electrodes, thinking 2012 or 2014. It was worse than that: I first mentioned the idea of using agar in electrodes in TE News #8, covering September 2008. At that time I didn't understand some key points of what I was doing and I didn't realize how to use it. It would seem I used it, but in electrodes and cells that weren't successful for one reason or another. Have I been sitting on a key factor for 11 years? The new working agar coated zinc plate electrode seemed to say so.

   It's odd no one else has come up with this idea in a couple of centuries, or even recently. It's hinted at on Wikipedia (uses of agar: electrochemistry -- read between the lines). And this picture was in TE News #24 and I guess it didn't fire anyone's imagination... or was it? Egads, No! Somehow TE News #24 had been replaced by a very incomplete draft that must have been done early in the month. I looked on my computer and found
"TENews24-real.html", which looked like the finished version with a file date of months later. Something funny must have happened way back then. I'll upload the "-real" one when I upload this one. Perhaps I meant to then and forgot. (I think that was probably about when I was transferring the site from shaw.ca .)

Salted Jell

   I wondered if maybe adding KOH to the agar would make it more conduct current better. I put a gram into a shallow petri dish of agar. The pH went from 6 to 12. (I was surprised it didn't jump to 14.) It got thick but it didn't properly jell, even in the fridge. What did that mean for its long-term prospects in KOH electrolyte?

[trode]   So what else? Next I made and plated another electrode (with about 1g Zn plating). I tried a gram of potassium chloride salt in a similar batch of agar. It jelled. I painted some onto the electrode. Normally a zinc electrode wouldn't last in salt, as it doesn't in 'standard' dry cells. I hoped the agar would change that picture.

   Next was a test: I would open the cell and inspect the plain agar coated electrode. If its agar looked okay, I would exchange it for the salty agar electrode. If it didn't, I would drain and rinse out the potassium hydroxide and replace it with potassium chloride. Now I feel like I'm going back to 2010. Except with zinc (known to work) and salted agar. I sort of remember that my positive electrodes would slowly self-discharge in KCl. Was it the graphite powder? voltage of nickel oxide too high at a more neutral pH? Or was my assessment of what was happening wrong or is my memory faulty? Well, let's check it all out again. If nickel has trouble, manganese oxides in graphite or conductive carbon black definitely work in neutral salt. Billions of 'standard' dry cells say so.

   Well, maybe no need to go there... the agared electrode I took out looked fine. If there was any problem with agar in potassium hydroxide solution, it would apparently take weeks to show. The jell seemed to cover it all and it was still solid. There was no sign of any dendrites forming. (Not that I had run it that long.)
   So I just put the new electrode with the salted agar into the cell. It seemed to perform almost identically to the other one. That seemed disappointing - I was hoping for substantially higher currents. Well, I just heated up some of the previous jell and mixed a gram of KCl into it. Next time I would add the salt to the water, heat it and then add the agar.

   So it appeared I had a cell method - agared zinc - that wouldn't soon form dendrites and and short out. But it was anything but high current. One might conceivably make quite large batteries for bulk solar storage this way, where they would charge slowly during the day and put out a few amps for a long time overnight. But running electric transport would be pretty much out of the question.

Getting Current - Osmium Doped Film Again

   On the 28th I looked back at June's newsletter, #133. Currents had definitely been higher before the albumen and agar coatings - and with the osmium film. What would happen if I electroplated a zinc trode, painted it with the film, and then dipped it in agar?

FWIW: Osmium Pourbaix Diagram.
Any osmium in contact with the zinc at
-1.x volts will stay in solid metallic form.

[imidj]   So... another new electrode! It weighed 4.1 g before plating. After it was 5.55. There were 'bald' patches - not clean enough? I rubbed off the crumbly bits around the edges and it was down to 5.3. Theoretically an amp-hour of plating - half an amp-hour per side. I only painted the osmium film on one side, since there was only one other electrode in the test cell and I don't have much.


Electroplating at the edge:
Crumbly bits go around edge, a smoother line,
more uniform interior coating.

   For the agar dip I sprinkled in maybe a couple of teaspoons of samarium oxide. (Raises oxygen overvoltage. How absent minded I am -- it's supposed to be zirconium silicate (zircon) to raise hydrogen overvoltage! I only realized the mistake late the next evening. I guess I can toss that batch! Still, it's not a critical point unless the zinc corrodes from hydrogen bubbles, and otherwise wouldn't have.) Then I poured in 250 cc of distilled water. Then I added 10.0 grams (~3 teaspoons) of agar. Then I heated it to boiling on a stove burner. Then I dipped the electrode in it. After it cooled I found it was a pretty stiff jell. ...another recipe to work out the best proportions for.

(Why does it look pink and rough in the picture? It was very light yellowish, almost white (samarium oxide color), and the surface was quite smooth.)

   I replaced the previous (salted agar, no osmium) electrode. Its agar looked just a bit the worse for wear. But the salt seemed to degrade the agar even in the petri dish without the alkali, so that may account for it and I ignored it for now.

I put the cell together without even refilling the electrolyte (a bit dry-cellish) and connected it. It started charging at 100 mA. instead of 30. When shorted I immediately saw 2.49 amps. This dropped way off quickly, but it was the highest figure I'd seen for a while. After a few minutes charging it was over 3 amps instant and it dropped off more slowly. That may be the highest I've seen from this size electrode (30.5mm x 67mm = 20.435 sq.cm). 3 A / 20 sq.cm = 150 mA/sq.cm. That's good current. A short load test with a 27 ohm resistor showed it to be performing similarly to the first agared electrode with 120 ohms - with 4-1/2 times heavier load, putting out 63 mA instead of 14. Admittedly that's just 3 mA/sq.cm , but a large cell could at least handle solar charging and loads. And it hadn't had a long charge yet.
   Later I added some more electrolyte. Short circuit currents continued to start off at over 3 amps but dropped below an amp in a few seconds. (The next day they stayed above an amp for 10 seconds.) A 5 ohm load (300 mA) seemed to tax it pretty heavily, driving it below 1.7 volts, but it could run 15 ohms (110 mA) at over 1.7 for a while. After driving a load it would recharge (with 1.9 volts) at up to 150 mA, and continue on at 50 mA for quite a while, and it would hold over 1.8 volts for some hours, staying near 1.79V overnight.

   It was in fact very similar to the previous electrode with no jell in June.

1st Load Test

(29th) I did a first load test with 15 ohms. It put out over 1.65 volts for almost 15 minutes, dropping sharply after 17. (~28 mA-H with a heavy load.)

2nd Load Test

   Next test was with a 20 ohm load - about 80 milliamps. It ran for about 44 minutes before dropping below 1.65 volts, then it fell to 1.600 in 51 minutes, and dropped like a rock after that. That made it about 68 mA-H rather than 28. Now we could use some more improvement with each cycle. But which electrode is the limiting factor? One thinks of the zinc, but after all the NiMH cell I used bits of the nickel oxide electrode from was dead. And the live NiMH cells are showing very poor storage performance compared to their stated ratings.
   I must get the 3D printer going again to make my own "plus" electrodes!

   We have jelled zinc that should last for ages, and decent currents. Liftoff: We appear to have the makings of a working, long life battery!

   For negative electrodes then, the rest is optimization. The first question that comes to my mind is: How "optimum" is the osmium film? I made the acetal ester, but then I just sprinkled in a little osmium powder. I had and have no idea what a good proportion is. Could the currents be still better with a different mix, eg, with more concentration of osmium? or less?

3rd Load Test

   After somewhat over 3 hours recharge I ran a third load test, with a 15 ohm load. With the resistance of the connections and the milliamp meter itself, it must have been almost 17 ohms, because the currents started at 105 mA instead of about 115. Anyway it ran for 53 minutes mostly at about 100 mA, total 88 mA-H. That certainly compared well with the first test. It seemed like good improvement with cycling. As long as the zinc electrode isn't deteriorating, I'm happy.

   I decided the "expired" voltage for NiZn is about 1.6V. In this cycle test for example, the cell ran 15 minutes above 1.7 volts, 35 minutes before the voltage dropped to 1.65 volts, then it crossed 1.60 at 46 minutes, and it only ran another 7 minutes after that to 53 minutes, with voltage down in the 1.4 range and dropping like a rock.
   In an hour it had recovered to 1.744 volts. I could probably have run another short load test with it. Would it have hit 100 mA-H? Would it next time anyway? By now it was late at night. 3 battery tests in one day was time consuming!

Microcontroller for Cycle Tests, Cycling Cells?

   Perhaps I should program a microcontroller to do all the grunt work of recording times and figures, and have it turn off the load when the voltage goes under, say, 1.5 volts. And then maybe catches the recovery figures and then recharges it. OTOH, perhaps I should have done that years ago and now shouldn't bother? Then again, it shouldn't be that hard, and a lot more cycle tests, or just cycling to increase capacity, could be run - automatically - in minimum times. If I'm serious about production, it's probably a good idea.

How Many Cells for 12 Volts?

   I'm no doubt getting ahead of myself here, but if the minimum voltage with a reasonable load is 1.65 volts and "expires" is at 1.60, and the cells charge as high as 1.95 volts, how many do we want for a 12 volt battery? I've seen 8 Ni-Zn cells used, but probably the equipment was made for it.

Low charge
under load
(1.60 V/cell)
[1.5 V/cell]
Normal minimum
under load
(1.65 V/cell)
Fully charged,
no significant load
(1.80 V/cell)
while charging
(1.95 V/cell)
7 cells
11.2 [10.5]
8 cells
12.8 [12.0]

   While we call it 12 volts, 13+ volts fully charged is probably more ideal - at least, that's what you get with lithiums or NiMHes. Right around 11.5 to 12.5 is very similar to lead-acid voltages, except that the charging voltage is somewhat lower. (Beware: A lead-acid battery charger or charge controller would try its utmost to fry them! And beware of pulse and PWM chargers - they're very dangerous for any chemistry besides lead-acid. (Lithiums generally tolerate them.)) And while I was thinking it was a long way between 1.6 and 1.95 volt per cell, that whole range is tighter than lead-acid, which is usually considered to be about 14.4 volts while charging, down to 10.7 volts 'expired' (2.40 down to 1.78 volts/cell). And like 2.40 or higher for PbPb, 1.95 V for NiZn is a charge-and-stop voltage, not a continuous float charge. The maximum for that is probably 1.85 to 1.9.

   Anything over 15 volts is probably too high - it starts risking damage to equipment and appliances. And a nominal "36 volts" that hits (eg) 46.8 is heading toward hazardous (and would likely fool some equipment - like my PowMr solar charge controller - into thinking it's a 48 volt system instead of 36, which would be dangerous because it would try to fry the batteries based on trying to charge them to 48 volts). With 7 cells per section we hit almost 41 volts when charging.

   So, while another 1/4 to 1/2 a volt might be nice, probably 7 NiZn cells per each 12 volts is what we want. That's 21 cells for 36 nominal volts, compared to 18 for lead-acid, and typically 12 for lithium and 30 for NiMH. Considering the cost of lithium and the weight of lead-acid, that's doing pretty well! And 21/30 cells is only 70% as many cells as for NiMH and again the cells tend to be lighter.

   And I'll be trying salt electrolyte again. I don't see why it won't work with these zinc trodes if I make the positive electrodes right. The voltages will probably go up a bit in salt. Yep! - just 7 cells per 12 volt battery definitely looks optimum. Probably if they are well made they will be well over 100 WH/Kg. And so cheap compared to lithium, and hopefully 'forever' lasting.

Positive Electrodes & 3D Printer Troubles

(29th) Using the dry cell nickel oxide electrode pieces had been of great assistance for getting good zinc trodes figured out, saving all the work of making matching positrodes. I had gone through 3 or 4 sets of the pieces that ended up with ethaline or something else hard to clean off in them. But now I really wanted those "porous" plastic shells for positive electrodes. That was now going to hold things up until I could get a 3D printer working to make them.

[3D Print](30th) Okay, what about the new 3D printer, the "AnyCubic I3 Mega" ? Could I just connect it to the computer and make it work with my present software? It seemed unlikely. I plugged it in, but pressing "connect" in Pronterface didn't work. There was another possibility. Both printers had an "SD Card" socket. The "DemoOwls.gcode" were on an SD card. I found the SD card from the printer and copied my electrode G-code file to it. On the printer (the new one has a touch screen display!) I selected it and then "Print". To my amazement it started printing my electrode shell! And (using the same G-code files) it did the same funny things with fine line designs, leaving them all disconnected at the ends to maximize the chance of a piece or even multiple pieces coming unstuck from the bed. But here nearly every fine trace stuck flawlessly to the textured glass. (...also it was PLA plastic, which is probably more forgiving than ABS.)
   Furthermore, the beads were finer and it was automatically making the open "basket weave" that left little spaces - sort of, but not uniformly. Thus one could see it was possible to do it, and the "pores" were finer than with the old printer. There was obviously going to be more to it than that to be able to print both "porous" and "sealed" parts of the electrode, because the software (unless it had been changed, which seemed unlikely) wasn't set up for doing such a thing. But the essence was there.

Not very porous PLA shell

   What was really galling about the old printer - which will probably be the same with the new - was you could tell it manually, or insert at the start of a G-code file, a command to set the amount of filament to extrude per pixel, but if the command was inserted elsewhere in the file, eg, after printing the porous face, in order to print the rest solid, the program would recalculate how much filament "should" have been used and want to extrude a ridiculous additional length right then and there as if it had been making the thicker trace all along and "should have" already extruded all that. This results in the print stopping with a "long extrusion prevented" error message. Was there some way to re-initialize the length to start over? Maybe as if it was a new print on top of the old one?
   All this promised to be harder to figure out than how to get the old printer to work again. But could I just "fudge" it? I tried a new design. It left little channels, vertical this time for gas bubbles to rise to the surface, on the outside of the faces. And I only put one layer of diagonal 'weave' on that instead of two. It was .8 mm thinner and porous everywhere. The pores were a bit coarse and I would definitely put parchment paper inside. [I actually picked thin PP non-woven cloth: more open to let ions pass better.] It looked just fine. Then I thought of a couple more refinements around the edges: an interlocking rim to hold the two pieces aligned plus prevent edge leakage, a broader internal flat rim for where the edges of the cloth would have gaps.

   I was still wondering what to use for a current collector if I used salt electrolyte. Probably that urethane paint with conductive carbon black mixed in, on a sheet of copper? Maybe I'll take a page from the Comet airliner and round all the corners of the copper, hoping there'll be no bare protruding corners to start corroding.

   On the 31st I added the 'refinements' and made another shell of PLA. It pretty much snapped together. On each face, a vertical grille of single beads of plastic spaced 2 mm, .4 mm thick, was a space for bubbles to rise. Another single .4mm layer of diagonal beads made it a more solid but still "porous" layer, now .8mm thick. I'd have added a third layer, diagonal fill going the other way (that's how the printer prints layers of fill) to stiffen it, but it seems to fill in the pores and make it solid.

L: "Porous PLA Plastic" Shell done on new 3D printer. (Test cell electrode size)
R: with cupro-nickel current collector, bits of nickel oxide electrode from dry cell,
and a paper thin piece of PP cloth liner to prevent escape of oxide particles.

   I cut an existing piece of cupro-nickel down a bit to fit in, and put a few bits of dry cell nickel oxides electrode into it. Facing the outside I used the paper thin polypropylene cloth instead of parchment paper. It was more open to conduct ion flow better, and it was being held .8 mm inside the surface of the shell, so unless something is wrong, the chance of the electrode inside shorting to the zinc one (itself encased in agar) seems slight.
   I knew methylene chloride wouldn't 'glue' (liquify) PLA (poly lactic acid) plastic. I tried methyl-ethyl ketone with almost equally poor results. PLA seems to print nicer/cleaner, but I would have to switch to ABS for shells that would glue together.

 Well, next month!

(30th) Load Test #4

   Continued load tests would be needed to see if the zinc electrode did or didn't deteriorate with long cycling. In the morning I decided to try a lighter load test with 30 ohms. The cell sat at 1.769 V in the morning. It was high enough that I forgot something: I had not fully charged it after test #3 the previous night. I was simply pleased that it had held such a good voltage overnight. That it did so with hardly half a charge was even better.
   The voltages under load dropped fast enough to jog my memory. Well, it would be a good test anyway: half charged down to "empty". It was under 1.7 volts within 10 minutes, delivering 46 mA. It ran for exactly an hour before dropping under 1.5 volts. The 45 mA-H result showed that it held energy overnight, again proving I finally had a real, practical battery.

   Darn these load tests take time! The more successful the cells, the longer they will take and the more tests and cyclings I'll want to do. I'm looking more and more favorably on programming that microcontroller.

(31st) Load Test #5

   On account of the weather (finally a nice day!; why would I work inside?) the fifth load test wasn't until the night of the 31st. It ran a "20 ohm" load for 75 minutes, at which time it dropped under 1.5 volts and was plummeting. Current averaged about 80 mA, so that was 100 mA-H. It recovered lethargically, which again suggested that its energy was pretty much spent.
   One might hope for well over 1/2 an amp-hour from 20 sq.cm, which (assuming the zinc was the limiting factor) suggested the electroplating should probably be made thicker. That might also increase the maximum current drive. I wasn't being too particular as long as the agar jell wouldn't fall off a slick surface. Also, only one side of the zinc was in use. I doubt the rear side contributed very much. So we might say ideally it would have had 300 to 400 mA-H - and the same on the other side facing another nickel oxide electrode.
   But it was certainly a real battery! It was already good enough for stationary bulk storage and it can doubtless be improved.

(August 1st) Load Test #6

   On the night of Aug. 1st, not wanting to proceed too slowly, I did another load test. But it was late. I made it a 5 ohm load - 320 mA at 1.6 volts. The voltages were lower with the heavy load and the test barely lasted 10 minutes - really 9 - before the voltages were plunging down below 1.5. (By 10 it was passing through 1.25 V.)
   The important thing was it would be cycled and recharged, to provide another opportunity for dendrites to short the electrodes if they were going to do so.

   The recharging current, previously starting at 100-110 mA, this time started out at around 130 mA. It eventually dropped to around 20 as usual.

Aug. 2nd Load Test #7

   This test with a fairly heavy 10 ohm load (plus other circuit resistances as usual, apparently making it about 12 ohms) did something none of the others did. It was averaging about 140 mA drive. At about 34 minutes the voltage dropped under 1.6. By 41 minutes it was down to 1.500 volts. I confidently expected it to start dropping faster and faster as before and I started recording the figures every 30 seconds to catch the end. But instead of the usual "terminal dive", it began a very gradual "terminal decline". It continued to lose just 16 to 20 millivolts per minute until at 61 minutes it was down to 1.197 volts, still delivering 103 mA. After a couple more minutes, at 1.171 volts, with it still not "diving", I decided it was just too absurdly low for a 1.7 volt cell, and ended it. In 30 seconds it was back to 1.665 volts, and within 10 minutes it was back over 1.7 volts, hitting 1.725 in 18 minutes. That suggested there was still some small amount of energy in it.
   Energy delivered was about 137 mA-hours - 37% more than the previous measured test, #5, and with a heavier load. Apparently the cell was gaining capacity with every charge-discharge cycle. I still don't know if this was the zinc or the nickel oxide side's gain, or both. I rather suspect the chunks of nickel were forming better connections with cycling. After all, zinc is theoretically 820 amp-hours per kilogram and nickel oxyhydroxide is only 289. And that's in theory - I understand 90-100 is a realistic figure. (A correspondent said he was getting 110 - great!) I don't know what a "realistic" figure for zinc is, but it's surely much higher.
   Regardless If the cell continued to last, the improving readings bode well. The zinc either had or was gaining toward good charge capacity.

   Recharging started at around 150 mA this time, and the current dropped more slowly, commensurate with a higher capacity cell. If it recharged well after test #7 [it did], it would have outcycled all my previous cells.

Aug. 3rd Load Test #8

   This test was again with a 10 ohm load (again about 12 ohms total). It was very similar to test #7 with voltages a few millivolts higher for most of the test, dropping through 1.600 volts at about 37 minutes and 1.500 after 45 minutes. In the last 15 minutes it dropped off faster, probably owing to having delivered a bit more for most of the test. I stopped at 60 minutes which was also when it dropped under 1.200 volts.
   It averaged about 142 mA drive for 45 minutes, and maybe 116 for the last 15 minutes for an estimated/calculated total of 135.5 mA-hours. It would seem the limits to the capacity gain are approaching.
   In 30 seconds it was back to 1.666 volts, and in 4 minutes it was back to 1.705 volts. There was no doubt just a bit more energy still in it from not flogging it the few more minutes of test #7.

(Aug 4th) Charging current was up yet again, starting at around 190 mA. After charging instant short circuit current was up to about 3.75 amps, then 2.5 dropping to 1.75 after 10 seconds.

Aug. 4th Load Test #9

   I decided to pick a lighter load this time, but as I didn't want the test to run for hours I made it 15 ohms. That seemed to be about 17.3 ohms total circuit resistance, as it was delivering exactly 100 mA at 1.73 volts. It was 30 minutes before it dropped below 1.7 volts instead of 10 minutes. Again I ran it until it droped under 1.200 volts. It hit 1.7 V after 30 minutes, 1.6 V at just over 2 hours, 1.5 V after 2 hours 23 minutes, 1.4 V at 2h33m, 1.3V at 2h44m and 1.2V at 2h54m. (So much for it not running for hours! I was glad I started it earlier.)
   Current delivered averaged about 99 mA for the first hour, 95 mA for the second hour, 90 mA for the next 30 minutes, and 78 mA for the final 24 minute decline. This added up to 270.2 mA-hours. Being over 1/4 amp-hour in this small cell with its rather thin 'fuzzy' plating on the zinc electrode - and probably(?) not actively using much more than one face of the electrode - must be some sort of milestone. Even if it got no better, it was getting into practical range for producing stationary storage batteries.
   But both faces should then make it over 1/2 an amp-hour, and the full size electrodes are 4.5 times more surface. So counting both faces they would have at least 2-1/2 amp-hours each -- and I would electroplate them more thickly and evenly, so they would have more.

   The test must have used most of the available energy as recovery was very sluggish, taking 20 minutes to hit 1.700 volts again. Recharging current was higher again, starting out at 210 mA and again staying higher longer.

   When it was fully recharged instant short circuit current hit 4.01 amps: 211 mA/sq.cm. That was great current flow. After 10 seconds it was down to about 1.8 amps, still 95 mA/sq.cm. Then driving a 1 ohm resistor (apparently 1.2 ohms total) it dropped almost immediately to 1.04 volts, but it stayed over 1.00 for 2.5 minutes delivering just under an amp, about 45 mA/sq.cm. Real performance!

Aug 5th: Take Apart

   I decided I couldn't leave y'all in suspense for another month, and I took apart the cell to inspect the zinc electrode. The agar still coated both faces. It still had the same milky white color (with the samarium oxide), and it still had the same feel of jelled agar. It didn't seem to be deteriorating. That much made it a success in principle.
   What wasn't good was a lot of rough zinc poking out all around the edges. And it seemed to have started growing a few dendrites at those edges. My take on that is that after dipping the plate in the agar, as it cools and jells it shrinks back a bit and leaves the sharp edges exposed. So some extra needs to be painted on around the edges afterward to ensure it's all completely encased.

   The idea works. "Everlasting" zinc electrodes having good current flow and energy storage appear to be possible and practical. The agar protects the electrode from the electrolyte and the octavalent osmium doped film somehow helps the ions to pass freely through the agar to the zinc surface. (The amount of rare and costly osmium needed is very minute.)

Haida Gwaii, BC Canada