Turquoise Energy Ltd. News #89
  covering June 2015 (posted  July 4th)
Victoria BC
by Craig Carmichael

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

Highlights: New Configuration Electric Caik BLDC Motor Runs great, with superior new "Single Ended" or "Unipolar" Motor Controller! (see Month in Brief, Electric Transport)

Month In Brief (Project Summaries)

In Passing (Miscellaneous topics, editorial comments & opinionated rants)
- Clean burning plastic - Chem spraying continues - Eating crow - TPP, TPA or whatever it is - Greece: First Domino? - New Horizons Approaches Pluto

- In Depth Project Reports -

Electric Transport - Electric Hubcap Motor Systems

* "BLDC4:3" Electric Caik Motor & unipolar/single ended/half wave motor controller: making, running, testing, reports & ideas.
* Axial Flux Switched Reluctance Motor? (AFSRM)

Other "Green" Electric Equipment Projects
* Vertical axis lathe?
* Aquaponics & LED Plant Lighting update - Manual aquaponics?

Electricity Generation (no reports)

Electricity Storage - Turquoise Battery Project (NiMn, NiNi), etc.
* Making cylindrical Ni-Ni cell (but it didn't work well - what am I doing wrong?)

No Project Reports on: Variable torque converter transmission, Magnet motor, Lambda ray collector, evacuated tube heat radiators, CNC gardening/farming machine, Electric Weel.

June in Brief

    Development and testing of my new type of 'unipolar' (AKA 'single ended', AKA 'half wave', AKA 'SRM') motor controller, along with the new 4 rotor magnets per 3 stator coils configuration of "brushless DC" motor ("BLDC"), was the main focus throughout the month, and there are still a couple more things to try. It runs great! I'm sure it's superior to the usual BLDC motor and controller configuration that's out there now. Failure of BLDC controllers seems to be only too common (almost inevitable with e-bikes?), and it looks like my controller in addition to being inherently more reliable should have lower losses and generate less heat. The motor is actually the same as all my previous ones except the orientation of the rotor magnets, and there's a fourth power wire, common to one end of all three coil sets.

   It might be of interest to go over how my motor and controller development went. For the motor, at first (2008) I copied the Hugh Piggott axial flux windplant magnet arrangement, with 12 magnets on the rotor: 'NSNSNSNSNSNS', and with 9 phase coils on the stator, three sets of 3. Then I found the IR2130 3-phase coil driver chip and looked at the application notes for it. It showed "standard" motor driver configurations. When I tried it, the motor wouldn't turn - at every point of rotation the turning force was (seemed to be) matched by a counter turning force. I concluded (falsely, I think) that in order to run with that driver circuit, my motors had to be 'NNSSNNSSNNSS' - 2 magnetic poles per 3 coils instead of four. And every BLDC motor diagram that I saw showed the 2 poles per 3 phases arrangement.

   I started by copying the Piggot rotor. If I hadn't found the "standard" 3-phase H-bridge motor controller configuration, I just might have used my imagination and come up with the 'unipolar' configuration in 2008 to run with the 'NSNSNSNSNSNS' magnets as built. Instead I did what we all do: I copied what was done before, and changed the magnet rotor to copy the "standard" type... in spite of some doubts that having transistors that could short out the power was a reasonable approach. (...but if that's the way it's done...) It took until 2015, some odd experimentation with unipolar rotor and then SR motor designs, and the unexpected circumstance of the Weel being made as a 4 magnet poles per 3 coils generator, and then realizing it would have to be run like a motor to provide enough output, to put together the new and better "BLDC4" configuration. Not just a copy! AFAIK nobody else anywhere had/has got to that point yet, simple tho it seems in retrospect! (Any such configuration is certainly well hidden on the web. Wherever I look at BLDC motor controllers, I see the usual 3-phase bipolar H-bridge drive transistor configuration and coils connected "Y" (or occasionally "delta") with no access to the "Y" center, and for the motors, always 2 magnet poles per 3 stator coils. I've found no suggestion anywhere that there might be any alternative way to configure a 3-phase BLDC motor.)

   Similarly, if I had known the controller for a "unipolar" magnet motor controller would be the same as a "SRM" controller, I would have simply looked it up and seen "how it's done", and copied that, and I probably wouldn't have come up with the new idea for saving half the transistors and diodes, with their need for floating high-side gate drives, by using a coil instead. (While I had seen the name "reluctance motor", they weren't discussed in my Electric Motors course at BCIT, nor had I ever even taken a look at them since. To me the name sounded like a specialty or curiosity rather than a serious power motor type.)

   Similarly again (if somewhat digressing), I've seen many lathes, but only huge machine lathes will do larger diameter workpieces. I suddenly thought up the idea of a vertical axis lathe for making larger diameter flat disk rotors, pulleys and whatever. Again I had to break through what all my past said was "normal" in order to conceive of something new. And so it is: to create something new rather than a copy of some sort, it must first be imagined, and simply knowing the familiar way something is currently done actually stifles that creative imagination.
   It turns out there are vertical axis machine lathes, tho they too are very costly, huge, and seem vastly overbuilt compared to my 'compact', 'economical' vertical lathe idea - an idea I've written up FWIW under "Other Green Projects".

   To get to the actual story, I put the additional magnets on the formerly "N-N-N-N-" "unipolar" rotor, making it NSNSNSNS with four magnet poles per three phases instead of two. (It occurred to me that it should work exactly the same, but it would have more torque since the coils would attract a south magnet at the same time as they repelling the north magnet. It would be better.) Then I made a hall sensor circuit board and installed and troubleshooted(?) it, and assembled the motor.

   Then I hooked it to the semi-functional unipolar motor controller. I bought a fancy new digital oscilloscope and started troubleshooting the controller, and by the 20th I had everything running quite smoothly, on straight PWM modulation. I don't think I'd have got there without the scope. It least, it surely would have taken much longer with my crappy old 1970s Heathkit scope. The next day I entered in some "version B" stuff on the motor controller schematic and circuit board with Eagle PCB CAD, before I would start to forget what the changes were.

   Then I also gave thought again to doing an axial flux switched reluctance motor. After all, it's the same motor controller for it! I started thinking of doing one more along my original lines than the ones in the AFSRM research papers. It'll probably work better than I was thinking, and it can be "tweaked" in several ways. I hadn't got the iron powder, but I had many parts for my version including the ready-made iron powder toroid cores. I could make a limited amount more powder by grinding iron. I fiddled with some bearings and parts to determine what might be workable mechanical configurations, which will be perhaps the aspect most critical to initial success and performance. It can be tweaked and other materials and parts tried out from there.

   On the 21st, I gave thought to the perennial problem of trying to mount a rotor over 'the gap' on my present lathe, which is the only spot where larger diameter objects can be turned. The diameter is half the problem - mounting the piece is the other. The regular 3-jaw chuck is so fat it extends past the gap, so it's useless. Backing plates don't hold things exactly centered. All very frustrating! But most of the rotors I do fit on a 1" shaft. Now I thought of turning a 1" axle so that one end of it would fit inside the lathe axle, which is #2 morse taper. I found a short piece of 1" shaft and spent a couple of hours milling it down. I figured it would work its way out, so I center drilled the inside end, then I drilled and tapped it for 1/4" threaded rod, to stick out the far end of the lathe axle (about a foot long) and put a nut on. This is how the milling machine holds its tools securely, except it's a 6" hex head bolt.


   That worked pretty well, and I spent just a few minutes shaping the AFSRM rotor to mount a thrust bearing, which is to be one of the main things that hold that motor's moving parts in very close alignment. It was close, but it spun slightly crooked and I need to come up with some way to adjust the mounting for absolutely perfect alignment of the rotor being turned. The best way I can think of is two rows of set screws 120° apart on the rotor hub itself. Or perhaps I should turn down a larger shaft and have a machined shoulder on one end. (And how to hold the part securely against that?...)

Turning AFSRM Rotor thrust bearing flat

Thrust bearing on rotor

   Somewhere in there, while I waited for the new oscilloscope to arrive, I made the new nickel 'negode' for the cylindrical nickel-nickel battery. It should be a great electrode! I did a couple of versions of a cylindrical battery by the end of the month. The first one had a problem (bad graphite in the mix) and the second one, finished July 1st, [also] has poor current capacity and likewise doesn't seem to hold charge properly. What am I doing wrong?

   The Mazda RX7 EV update this month consists simply of saying that after the NiMH battery swaps in May, I drove it 9.3 miles on June 21st with only a bit of a charge at one point. (only because I forgot to plug it in after the second trip.) The batteries weren't bottomed out yet, either, and I could have gone probably about another 3 miles if I had had somewhere to go. That'd be almost 20 Km - a big improvement over 10 to 12! And by replacing just 2 bad cells, I have 7 tubes of left over cells (70 AH @ 12 volts) to use for something that doesn't draw as much current as the car - and to run experiments on re-hydrating the cells, which could possibly restore their current capacity. The lithiums are still doing great. The four remaining lead-acids are working well and have outlasted all the others by a wide margin. If only I knew which ones were the ones to get when I go to buy them!

   All the while, while doing all these very exciting projects, exciting and successful as they were, it's been gnawing away at me that I still haven't completed my SR & ED application, and that I still don't have my bitcoin miners, which I got in February, running. I figure that's "costing" me hundreds of dollars a month in income I could be getting. That just might be the money I could be leveraging into funds to commercialize some of my products.

   Much to my amazement after so long with no sales except the long-running Electric Weel kit/project, someone ordered two of the stackable 12V, 10 D cell 3D printed battery cases by e-mail. He paid and I sent them. Say, isn't this the way business is supposed to work?

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

Plastic Debris Clogging Oceans - Burn your grocery bags?

   Perhaps you've heard there are "islands" of plastic debris "the size of Texas" floating in the oceans? This
is becoming a critical issue to Earth's ecology. It keeps getting worse, and no one is dealing with it. Whether it's related to the "dead zones" isn't clear.
   I met someone, a qualified ship captain, who had proposed creating a clean-up ship operation that would profit from selling the recovered plastic. Of course that would require capital, and he said he couldn't find any interest or support. (It seems to me this is another illustration of how those who run our society have abrogated their responsibilities in ways that seem to make no sense whatsoever except that they must take some sort of satisfaction in doing harm - You can't get good fruit from a corrupt tree, even when it seems to make sense by their own "anything for profit" philosophy.)

   Now, people keep saying "Don't burn plastic. It creates toxic smoke." But there are many kinds of plastic, and this isn't true of all of them. In particular, polyethylene such as is used in grocery bags, soft transparent food bags, and many small plastic containers, creates nothing but carbon dioxide and water vapor. It burns as cleanly as gasoline or cleaner. And a light plastic bag is probably composed of less petroleum product than you burn to back your car into your driveway. If you drive a gas car, but worry about burning plastic bags, it's like worrying about an anthill and ignoring the mountain. And if you can stick bits of paper, sawdust or whatever into the bag first, it's doing double duty in containing stuff that would otherwise be messy to burn. And plastic as 'kindling' of a sort can help get a fire going. (Even so, I often re-use my grocery bags. They're easy to stuff into your pocket, and to keep a bunch in the back of your car.)
   Likewise, polypropylene is polymerized cyclopropane, and so has no more burning emission than propane, again just carbon dioxide and water - a cheery flame with very little smoke. In addition to many small plastic containers, cloth grocery bags and landscaping fabric are made of polypropylene.
   Look for the recycling symbols identified as "PE", "LDPE", "HDPE" (low & high density) and "PP" for these clean burning plastics. Ideally all these can be recycled, but except for larger, heavier pieces, the truck carrying all that fluff probably burns more petroleum product than what it's carrying. Thus it's my feeling that the best way to deal with them is to burn them in a woodstove or fireplace. At least that way you can be sure they're not adding to the mountains of floating debris in the oceans. (Some plastics that DO make noxious smoke include: PS, ABS, PVC, PETE.)

Chem Spraying Continues

   Many if not most days, now that I know what I'm looking at, I see chem trails and sometimes see or hear the aircraft making them. On the 15th extra-heavy chem trails criss-crossed the sky all day.
   They are known to prevent precipitation, and rain clouds have passed by overhead without any rain on several occasions this spring. (Probably the clouds finally release a deluge to cause the many flash floods seen this year in the interior of the continent.) A local fire chief says it's "very abnormal" for it to be so dry this early in the summer.
   One evening there was, finally, a light sprinkling that continued into the night, which filled a 220 liter drum from the roof downspout 2/3 full. I moved the hose to a second drum before I went to bed, which filled about 1/3. Since it hadn't rained in quite a while, one might expect a little dirty water off the roof at first. But this picture is the second drum (which has an open top). It was foamy at first. Later it was just this dark reddish brown color. Just what are they dumping on us? Yuk! (Later I found some water off the galvanized garage roof in a garbage pail, that wasn't too bad. Still...)
   Is it any wonder there's huge die-offs of sea life? A video about "extreme weather" of last winter showed masses of dead fish floating in lakes and washed up on ocean shores as reported on the news in various places around the world. It's been widely said that there is "a dark cabal" (as president Kennedy warned decades ago) manipulating from "behind the scenes" who would like the population reduced to that of the stone ages, and to enslave the remainder. Destroying the major food sources of the world is certainly a way to kill people. Can that be the motivation for all this?

   On about the 23rd someone told me there has been a massive algae bloom on the west coast of Vancouver Island that makes all previous algae blooms look like peanuts. I was told the entire fishery was shut down. I haven't heard any more about it. If it's true and really all that bad, logically we may next hear about another vast fish and sea bird die-off. Even non-toxic algae blooms, in sufficient density, cause oxygen starvation in the water as the algae dies, which kills the fish. Lack of food, or as seen recently the chem foam itself on the water, can kill the sea birds and other creatures. It's unsettling to look up and and realize that half the clouds hold not moisture but unspecified and unidentified 'nefarious' chemicals. Many can be identified as chem trails spreading across the sky and sometimes the planes are seen, until one becomes suspicious of all clouds.

   July 3rd: Last week the newspaper said Victoria had plenty of water in the reservoir in spite of the lack of rain and heavier usage recently - nearly 90% full. Hurrah! Today, just a week later, severe water restrictions are being imposed. Huh? Who makes this stuff up? My own rain water collected mostly from 36 hours last March collected in the "swimming pool" is almost 3/4 used up. It shows the value of storage capacity. I suppose it won't rain again here until mid August - if then. On July 3rd also open fires have been banned, and it's said things are tinder dry. This part is understandable. Water rationing with a full reservoir isn't.

   Why aren't there diplomatic protests, seeing the US air force is now chem spraying in Canadian territory? Why aren't more people protesting? And how anyone has the gall to perpetrate such atrocities on the planet they live on I don't understand.

   Please stop.

Eating Crow

   On the 16th I went out to feed the goldfish about 10:30 AM and found a drowned crow in the 'trench' pond, apparently a young one. Making the pond with no shallow place to stand and a drop-off from the rim was intended to keep predators out. It wasn't there the previous evening, so it must have just happened that morning. (Later I found a young crow in the garden that couldn't seem to fly over the fence and leave. Perhaps it was its sibling that fell in?) Notwithstanding the derisive expression "eating crow", or maybe just because of it, I plucked and cleaned it and boiled it (about 10 minutes - 5 was too short) for lunch, including the heart, liver and gizzard. It was a little tough, but good meat.
   This isn't a big livestock region, and the small island deer that today wander around town eating peoples' gardens won't last long when and if the food supply chain gets cut off. People may well need to do such things to eat, so, strange as the idea seemed at first, I thought I'd get in a little practice. Somebody somewhere does it with all the chickens, pigs and cows we eat, and I saw in a National Geographic once that the Basques in Spain would put up big nets to catch various small birds for food. Lo and behold, Joy of Cooking even had a section on "small game birds". And when I was little in Edmonton I helped pluck ducks, partridge and pheasants that my dad used to shoot, so it wasn't totally alien territory. (Wild game birds are delicious. But I once had a pretty tough partridge. Beef can be tough too, and after all, crow is just cow with an 'r' in it.)

   Joy of Cooking said blackbird and crow, if eaten of necessity, should be "parblanched" before cooking, which involved putting it in cold water and bringing it slowly to a boil. That's when I figured I might as well just boil it. The part about "if eaten of necessity" by German authors reminds us that food shortages do sometimes occur in tough times, and not just way back in history or in the "third world".
  But I want to get in some practice fishing this summer, in my electric outboard boat, also trying out the new motor and controller in real use. Most everybody likes fish! (I trust there still are some in the strait.)

TPP, TPA or whatever it is...

   How can a law or public agreement be a secret? The terms are "classified". No one is allowed to see them, or having seen them, is allowed to talk about them or make copies. Any US congressperson who tells the public what the agreement is all about will be thrown in jail. (I thought congress made the laws!?! How could they have decided such a thing against themselves?) Evidently the public isn't even to be told the terms for four years after it's been implemented! The government says "Ignorance of the law is no excuse", but we're not even to be permitted to know what the law is? But from leaks we know that corporations can take governments - ie, the citizenry - to court to sue for profits they they could have made if a decision they don't like had been in their favor. The people can never reclaim any property that has been 'privatized'. And legal actions aren't to be contested in real courtrooms, but in private hearings behind closed doors. We may not know what the rules are, or what was decided or done in the name of this "deal", but we are expected to abide by whatever we are told, or take whatever penalty is meted out for any transgression of the rules we haven't been told, by a secret kangaroo court hearing with no appeal.

   This so-called "trade deal" effectively takes the power to make laws out of the hands of elected lawmakers and gives it to global corporations and banks. Is that who we want running the world? Democracy is to be officially subservient to corporatocracy. Let the implications of that sink in! Are these, especially the banks, not the lampreys which have attached themselves to the unwitting shark, sucking the economic blood of productive citizens and of society as a whole? And now they are demanding total control and reserving all rights to themselves? Will they not kill the society that gives them life?
   Apparently the madness must continue until most every surviving human says "Enough is enough! No more!" Then they will reassert their God given individual sovereignty to create societies of strong families that will have more moral sense and the courage to act on it against threats and coercion, who will trust in humanity but individually and together rein in all those who make outrageous demands and perform outrageous acts trying to gain unfair advantage over others, keeping in mind the cosmic core values (quality of life, provision for growth, equality) in their acts and decisions.

"Someday love will rule this very world." - Jesus

   I recently saw a great video on youtube titled All Wars Are Bankers' Wars. While I don't agree with everything it says, it gives insights as to how the insidious process of behind the scenes control of societies, of their finances and economies, has worked over the past several centuries. But with the internet, people are gradually learning how these hidden things work, and soon and in the future will no longer tolerate them. For the few most stubborn and deluded individuals working insidiously and tirelessly behind the scenes to enslave humanity, we have a celestial promise that they will be "unceremoniously removed from the planet" when the time is ripe, their remnant Luciferian philosophies no more to disturb planetary progress.

Greece: First Domino?

   The latest bank runs closing the banks in Greece are only the tail end of around 35 billion Euros withdrawn this year, from a total of around 160 billion Euros on deposit in Greece in January. Since banks keep only a small fraction of the currency placed on deposit available, European central banking has (if I have it right) given emergency "liquidity" money to them five times to keep them afloat. 4 billion dollars withdrawn in 5 days at the end of June has brought things to a boil. The financial institutions don't like the new Greek government, which outright stated the obvious: "Greece is bankrupt.", shortly after being elected. They refuse to offer any terms besides "Give us Greece's pensioners' money, right now.", which wouldn't help Greece except to buy them another five months, ending with them right where they are now. They wouldn't even grant a few days to await a national referendum on whether the Greek people want to accept this temporary "solution". The referendum is to be held on Sunday July 5th and it's expected that the Greek people, in spite of the banks being closed, all the fears, and apparently an organized [read, paid for by the bankers] campaign to convince people to vote "yes", will probably vote to reject this latest imposition of further "austerity". (If they accept European terms, PM Alexis Tsipras and finance minister Yanis Varoufakis have said they will resign.)
   The Syriza party had only a tiny fraction of the vote in the previous election, but ran on the promise of ending European bank imposed austerity - the cause of much very real suffering and dozens or perhaps hundreds of protests in Athens. It's said the banks care little for anything now beyond punishing the government unexpectedly elected by the Greek people, and certainly they have no sympathy for the people of Greece or Greece's plight.
   Most of the immense "Greek bailout" money since 2010 went to financial institutions in Greece. They were bailing themselves out. The government received very little of it, yet is expected to pay the huge debts. (Remember Thomas Jefferson said there are two ways to conquer a nation: With armies and with debt.) The new Greek government has pointed out that several European nations are in little better shape than Greece and proposed a conference for a general write-down of all debts in Europe. The banks are apparently furious at that idea - that someone is working against their financial tyranny. (Will they dig out some patsy malcontent to try to murder Tsipras and or Varoufakis? Blame it on "terrorism?" Asked how it felt to be elected to power in Greece - and one might add, to go from being a Texas university professor to the center of the world stage - Varoufakis replied "Scary!") I found a good briefing video by Varoufakis on youtube on July 2nd, explaining (in English) what was - and wasn't - offered and why they were recommending people vote "no", but that that recommendation would change to "yes" if a Greek deal offering a ray of hope for the future was brought to the table - there were two full days left for the bankers to make an offer.)

   Missing the payment due June 30th is clearly a loan default by Greece, and no realistic alternative or compromise has been offered by the "Troika", the ECB, IMF and 'Eurozone Commision'. The question now is whether this hard attitude makes Greece knuckle under so they can steal the remaining wealth of its citizens, or if it may somehow blow over for now, or if it may prove to be the first domino in the collapse of the corrupt, insolvent, unsustainable ponzy scheme that is the global financial system. If it is the latter, events could move quickly. There are said to be 100 trillion dollars of outstanding Greek "interest rate derivatives" and "credit default swap derivatives", which the major banks have guaranteed.
   On the other hand, economies are bad everywhere since so many people are competing for the world's resources, and some like Greg Hunter (USA Watchdog.com) feel the decisions made at this point will make little difference to the final outcome. Bix Weir (always with a different take on things!) thinks the banks want Greece to default, and then they'll start printing the truckloads of money Europe already needs to keep the system afloat, and blame it on Greece.
   As I said after Cypress, the next time a country uses "bail-ins" - theft of depositors' accounts - to save their banks, people world-wide will probably start withdrawing their accounts and hiding their cash. It might take until "bail-in" #3, but #3 will probably closely follow if #2 gets away with it. (...which has led to the proposal to ban cash entirely to prevent bank runs - and to further interfere in peoples' lives!)


   My version of the "trickle down" theory of economics: Wealth from the productive starts circulating through the economy, but as it comes in range, it is sucked to the bottom into giant traps to add to the immense fortunes of the bottom feeders, never to re-enter circulation. Thus everyone else is sucked dry and economically enslaved, working harder and harder stay in the same place.

New Horizons Approaches Pluto

   The New Horizons space probe, after 9-1/2 years of flight, will fly by the 2345 Km diameter sphere called Pluto, around which orbit an even smaller 1250 Km diameter moon called Charon and several "oids", also titled "moons" per the prevailing misleading custom that gives equal billing to worlds half the size of Mars and to little whirling chunks of rock. Pluto is about 1/3 the size (by volume) of Earth's moon. For more on the mission: http://www.nasa.gov/newhorizons and http://pluto.jhuapl.edu . For my take on how Pluto with its associated bodies as well as Triton formed (think of the breakup of comet Shoemaker-Levy 9 near Jupiter): http://www.saers.com/recorder/craig/ (scroll down to find article link).

   BTW, I've seen nothing definitive on Ceres. The views are getting closer, and some ice extrusions are visible. Of course, I should be searching for spectral findings.

Newsletters Index/Highlights: http://www.TurquoiseEnergy.com/news/index.html

Construction Manuals and information:

- Electric Hubcap Family Motors - Turquoise Motor Controllers
- Preliminary Ni-Mn, Ni-Ni Battery Making book

Products Catalog:
 - Electric Hubcap 7.2 KW BLDC Pancake Motor Kit
 - Electric Caik 4.8 KW BLDC Pancake Motor Kit
  - NiMH Handy Battery Sticks, 12v battery trays
& Dry Cells (cheapest NiMH prices in Victoria BC)
 - LED Light Fixtures

(Will accept BITCOIN digital currency)

...all at:  http://www.TurquoiseEnergy.com/
(orders: e-mail craig@saers.com)

Daily Log
(time accounting, mainly for CRA - SR & ED assessment purposes)

1-3: Finished May TE News (#88)
4: Added 4 magnets to rotor to make Electric Caik "Unipolar" motor into "BLDC4" bipolar motor (so called because it has 4 rotor magnet poles per 3 phases instead of 2 poles). Edited & Printed Electric Caik Hall sensor PC board, and an LED flat panel light board.
5: Boards etched badly. Made a 2nd Caik board (satisfactory). Epoxied in the PP strapping to hold the new magnets securely. Moved the 2 tilapia from the mini-fridge into a barrel and shut off the aquaponics pump.
7: Drilled PCB, soldered parts and cable on. 2 Tilapia died - aquarium water bad. (High ammonium nitrate?) Transferred remaining large female to a 10 gollon aquarium.
8: Installed Hall sensor PCB, Assembled motor (not without a few hicups and adjustments)
9: Emptied and cleaned out 30 gallon aquarium.
10: Brought it back and refilled it.
11: Put the 2 tiplapia from the barrel into the aquarium. One didn't look well and died within a few hours. The other seemed fine.
12: Put the last tilapia into the aquarium, the big female. She didn't look well either, but was still alive the next day, with a bit more color. Rewired the motor controller sensor wires with two 3-wire plugs to match the motor. Then I changed both motor and controller to match the PC board socket, ie, 123456 = Ground, +12Vsupply, SenseA, SenseB, SenseC, Temperature.
13: Tested motor with unipolar controller. (Controller still gets hot.)
14: Ordered 'modern' oscilloscope to check waveforms in motor controller, and a few other parts.
15: Made inside-fitting end caps for 3/4" PVC pipe batteries. Painted battery negode with micro nickel flake powder in gum arabic.
16: Oscilloscope arrived. Learning how to use it... OSC signal on motor controller not right.
17: Motor controller testing and troubleshooting. Got it running well! (but not perfectly)
18: More motor controller testing and troubleshooting.
19: More motor controller testing and troubleshooting. (Main problem: high pulse current was overdriving apparently inadequate 'flyback' diodes!)
20: Redoing motor controller schematic & PCB layout in Eagle PCB.
21: Made morse taper shaft for mounting rotors on lathe; turned AFSRM rotor to hold thrust bearing.
24: Made new bar & mounted heavier 'flyback' diodes. (They are STILL inadequate, but not quite as badly.)
25: Consult with AGO about wobbling motor shaft. (Got new piece of shaft owing to slightly tapered end on old one.) Made NiNi battery 'posode' & assembled battery (wrong mix - contained art store graphite! Self discharge.)
27: Talking with prospective partners.
28: (Played a concert with a concert band, on my own invented instrument, the Supercorder - playing flute and oboe parts.)
29: A few more motor controller tests, with diodes, and a writeup about that for the OSMC motor controller e-mail group.
30: Made 2nd nickel manganate battery posode.
July 1: motor test: bypass coil. Sure enough, watts to run motor goes WAY up. More converse with OSMC: someone told me about "FERD" diodes, which may have lower Vf for the recirc diodes (if I use enough of them in parallel).

Electric Hubcap Motor Systems - Electric Transport

"BLDC4:3" Electric Caik Motor and Unipolar Controller

   Having decided the unipolar BLDC motor wasn't the way to go seemed to leave me with options of:

a) buying BLDC motor controllers to go with my motors.
b) trying to improve my controllers so they wouldn't blow up at high currents, so they'd be a product along with the motors.
c) making axial flux switched reluctance motors (AF-SRM.s) that would use an inherently more reliable type of controller.

   Option B wouldn't give me anything better than anybody else has and I'd probably have to charge more. In option C was the possibility of making regular SRM controllers or, if I could make it work well, my new type with half as many power transistors and half as many diodes, the remainder being replaced by a coil (to allow return energy from the motor coils).

d) Then along came the Electric Weel generator, needing a motor controller... but it had been made with 4 magnet poles per three phases instead of 2, and (I thought) a regular BLDC motor controller won't work with that. But it sometime dawned on me that the more reliable "unipolar" or "SRM" controller that I designed and made last winter would work instead, and better still, that it looked like it would be superior to today's standard BLDC configuration. With 4 rotor magnet poles going by where there were two, the electrical frequency is doubled and each coil is ON right between a north and a south magnet, where the motive force is highest attracting one and repelling the other. The torque per amp should be high and the torque ripple lower than a SRM.

   To me, given my many motor controller failures and seeing so many elsewhere as well, the reliability of the motor controller is the most important aspect. Of course I would do my best work, but even if it was "cheap" and "glitchy" the controller wouldn't suddenly fry at high currents, so here was my best chance to make highly reliable controllers in-house to go along with the motors.

   On the morning of the 4th I prepped the existing unipolar rotor by scraping off the paint and epoxy where four more, opposite polarity magnets were to go. Milling four more slots would have been awkward and left countless filings stuck to the existing magnets. So instead, I decided to slit the straps and fit each half through the existing four magnet slots, which were fat enough to pass another strap through once excess epoxy and strapping were cut or filed away. Not ideal but hopefully good enough for at least 3000 RPM without magnets working loose and flying off.
   Although the magnet placement jig couldn't fit around the magnet strapping, I bolted it over top and used it as a visual guide for placing the new magnets to what I hope is within a couple of hair breadths of where they should be. I was afraid it would be hard to get the 2" x 1" x 3/8" magnets to stay in place without the jig, and not jump over and glom onto the other magnets, but they stuck down to the rotor steel well enough to prevent that, even with wet epoxy under them.
   I forgot to sand the slick epoxy coating on the new magnets. This is where the original Electric Caik had failed on its first run, the magnets simply sliding right out of their (admittedly bottomless) pockets at no great RPM. But these didn't jump off when hammered... but then the epoxy wasn't fully set yet. I decided to wait and see how they fared once it was. (Then I forgot all about it. I've run it up to 1200 RPM so far with no trouble.)

   The next step was to make a regular hall sensor circuit board as used in the other Electric Caik motor, because the optical sensors were crowded enough to bother me. So if I didn't need them for this motor, I'd change it. I started that on the 4th too, and etched it (along with another LED light board) on the morning of the 5th. I thought I finally had the toner transfer method circuit board making down pat. But these boards seemed to take ages to etch. After well over an hour, I thought it must be really weak etchant. I looked again, and suddenly wondered if the unusual yellow board material looked the same as the copper in the yellowy ferric chloride. I pulled one out and rinsed it, and found it was not only done but badly overetched, with the traces being undercut and thinned. There were some breaks in the runs.
   I didn't quite have to start over since I'd printed two copies of the sensor board, but it was afternoon before I had a board, with just 15 minutes of etching in the same etchant. (Again it didn't look done until the etchant was rinsed off, but I was wise to it this time!)

   While waiting for the first "slow" etching, I managed to work the PP strapping for the new magnets through the slots, after widening them by filing with very thin files to get some of the old epoxy (and maybe some of the PP) out. In the afternoon I epoxied the straps on. It looked a little messy, but I trust it'll be good for 3000 RPM or so once it's balanced. I didn't bother with balancing for now.

Motor open, showing Magnet Rotor from the back

   On the 7th I drilled holes in the magnet sensor PCB and put the components on. I had a lot of trouble with mirror images because this board has the copper on the top instead of the bottom, so whatever I thought was right was wrong.
   On the 8th I turned the (oops, mirror image) hall sensors over and then soldered the cable to the board. The fact that 6 connections are needed is annoying for using "trailer lights" connectors, but I can't think of any other connectors I'd rather use. Instead of using a 5 pin "trailer lights" connector as I had been doing, and then needing one more pin for the temperature sensor, I used two 3-pin connectors as follows, with "F" and "M" applying to the controller side, the motor being opposite. (There is a 6-pin "trailer lights" connector, but it's two rows of 3 pins and rare - special order. And they might be hard to plug in and unplug.) I didn't use two identical plugs because they would be bound to get connected to the wrong wires, so I did one the other way around. I pinned it one way, then later I realized it would be confusing that it wasn't the same as the motor controller's header pin pinout, and I changed it to match: 1--6 = ground,+12V,SenseA,B,C,Temperature.

The first plug has one male pin on the controller side...
1. M Ground
2. F  Power - +12 Volts
3. F  Magnet Sense A

(The motor side is of course opposite: FMM)

The second plug has the two male pins on the controller side...
1. (4.) M Magnet Sense B
2. (5.) M Magnet Sense C
3. (6.) F  Temperature (AD590, 10mV/°K)

(The motor side is FFM)

(Now I just hope people don't get these confused with identical 3-pin plugs for speed control and a forward-off-reverse switch when they're installing a system. Anyway, it reduces the variety of the inventory!)

   Then I assembled the motor with various trials and tribulations. (It sure could have used another 1/4" of height... and 1/8" greater radius... for clearance in the rotor compartment!)

Piles of Testing, Troubleshooting and Correcting - New Oscilloscope

   I tested the sensors board with the lab power supply, a prototyping board (for connections and pull-up resistors), and a voltmeter, turning the motor to see the signals change with the rotor magnets. I was out of hall sensors and I used one labelled "?" for the third one. It didn't work. Neither did the temperature sensor. Had I got it mirror image after all that?
   Then I thought to use a hall sensor from a left over circuit board. I rooted through the drawer and there were some. I disassembled the motor stator compartment to access the board. The temperature sensor was squashed under a coil. That was fixed by straightening it. When it was all back together it all worked.

   Next was to hook up the 'unipolar' motor controller and try running it. Since it ran (barely) as a unipolar motor with only N-N-N-N- magnets on March first, I anticipated no problems at least getting it to turn with NSNSNSNS magnets. The question was how much energy would go into heating the transistors in the controller instead of turning the motor. For this, I had the other coil, with twice the inductance (400µH), to put in series with the power. The trick should be that the series coil should have much higher impedance to the coil turn-off spikes than the motor coils (90µH), so that the return energy would all be back to the supply by the time the " B+' " coil voltage started to rise much.
   It seemed to me it shouldn't matter what frequency the PWM/CRM was at since the reverse spike comes after the pulse turns off, regardless of how long it was. But that's probably not exactly true. The longer the pulse is on, the more energy may be in the coil.

   I got tied up with various things, and it was the 12th before I got the sensor cable wired. I ran the motor on the morning of the 13th. (@17 volts) I connected the coil phases per TE News #85 - I *thought* the magnet sensors were the same way around as back then, and it proved to be right. It seemed a little stronger with double the magnets, but still pretty feeble, and the driver transistors still got hot fast with just a few amps of drive. Just when I thought to try it backwards and see if it ran the same that way too (to verify that the phases were connected right), a transistor burned out.
   The energy return coil idea didn't look good so far! Then I checked over my wiring. The energy return 'flyback' diodes, on a copper bar hidden underneath the controller PC board, hadn't been connected to the common supply side! Well, that would explain a lot! (In fact, that may have been the problem last winter, too, seeing the symptoms were the same.)

   In the afternoon I replaced the blown mosfet (phase C), connected the diodes, and tried again at 18 volts - the voltage that had caused a quick blowout in March. The transistors still got hot pretty fast, but this time I could turn the motor up a little higher, and didn't get an immediate blowout. Phase C seemed to get hotter than the other two. The flyback diodes got a little warm. It was still only 5 amps and not much power, but more than "barely turning". I turned the current limiting up a little higher to about 7 amps and the voltage to 20, and it was still okay and the motor turned faster. But I wouldn't give the hot transistors a ghost of a chance of surviving 2 seconds at 200 amps!
   The coil had only about twice the inductance of the first one. Maybe it needed a still much larger coil to keep the energy from going back into the transistors? I didn't have any bigger ones, but I had more the same... would putting two inductors in series improve things?

   But when I thought about that... even if the energy was being shorted, it should short through the diodes across the coils, not through the transistors. So why were the transistors heating up? Ah... maybe the power supply needed more capacitors to absorb the flyback voltage? If the voltage was going above the transistors' zenor point, they would absorb the excess energy themselves and heat up. I thought it had two 4700µF capacitors, but on inspection they were were 270µF... at 100V rating because of the high currents and hence in rather large cans. Probably not enough. I added two more for a total of 1080µF. The transistors still got hot pretty quickly, and again phase C faster than the other two. Well, on the other side of that coil was the motor supply, with only 640µF. I soldered another capacitor on the bottom side of the circuit board, making it 860µF. That didn't seem to help either.

   Time to get out the oscilloscope and look at the waveforms. ...Or maybe buy a new oscilloscope that would be reliable, correctly calibrated, nicer, better, and easier to use than the old 10MHz, 1970.s Heathkit? I opted for the latter and ordered a BK2530B digital LCD storage oscilloscope the next day. I also ordered 50 more A1203LUA hall/magnet sensors, and while I was at it, a few 555 timers and LM339 quad voltage comparators. (After chancing to see the US$ prices before I saw the CAD$ prices at Digikey, I thought the exchange rate seemed off. I looked elsewhere and ordered them from Mouser instead, for about 650$ instead of 750$. The chips were cheap, but I could probably have done much better on the hall sensors if I'd ordered them from China. Evidently sometimes it still pays to shop around!)

   It arrived on Tuesday the 16th. Aside from the typical functions of old types of storage scopes, it has some great features including storage, "playback", printing on-screen the volts/division, time/division, offset voltage and trigger frequency, and saving the screen to a USB memory stick. The range controls are optical(?) dials instead of rotary switches with a zillion contacts to get dirty and wear out, and there were several menus to set things up with a minimum number of pushbuttons, which matched the menu placements beside the display. And with no CRT tube, it's compact and lightweight: 6" high x 12" wide x 4.5" deep. Even the probes are easier to use, and clip onto wires better than the old ones.

   After taking a while to look at the manual and familiarize myself with the controls, I put the probe on the OSC signal on the motor controller chip. It was wrong - staying high instead of oscillating. Amazing how a very foggy picture of what's assumed to be happening starts to clear up with proper test equipment! The old scope could have told me this too, but as testing went on, the new storage scope quickly proved its worth, showing things clearly by getting a snapshot at the critical time, for later
viewing including expanding very short period pulses across the whole display in sub-microsecond intervals.
   I could have cleared the picture by looking more closely at my schematic, too. I had somehow reversed the connections between OSC and ERRAMP pins. No wonder I've been having problems. OTOH I found had already corrected part of it last February. And I saw a better way to rapidly end the modulation cycle if it shuts off on overcurrent.
   But the next day (17th) I decided for testing to reduce the number of variables by eliminating the overcurrent variable modulation completely, and going with simple fixed high frequency PWM. This all started messing up the circuit board, with components upended and vias that had to be cut, all rerouted by messy wires. It starts to get pointless to go into all the details. I'll want a new PC board for the "real" controller.
   There seemed to be a lot of very glitchy signals, with an annoying thread of 3MHz oscillations at an amplitude of up to several volts. Even the magnet sense inputs had it, to the point where the scope couldn't trigger properly off them. (indicating I should be using shielded cable, and filtering them on the board!) I was amazed it worked. But when I looked again later (after having added some capacitors) it seemed not bad. It depended on where I connected the probe's ground clip.

A Hall Sensor Signal - the second time with
the ground clip near the signal.
42Hz * 15 = 630 RPM

   At the current sense input, the signal was supposed to be filtered with an RC filter (330Ω, .01µF), which seemed to do nothing - it had the same RF noise at the same amplitude as the direct connection. In case it needed a larger capacitor, or in case there was a bad through-hole connection on the capacitor, I soldered on a .1µF on the back of the board.
   When I did that, the motor magically ran properly and smoothly with the control, and the transistors didn't get hot: it was working! Instead, the flyback diodes got warm. Roughly, the motor seemed to use about 150 watts to idle at 1000 RPM. This didn't compare favorably to the original Electric Caik measured at 101 watts. (TENews #60) I thought it might indicate inefficient energy return - especially with the diodes heating up. However, if there was no energy return, currents could be expected to be at least 5 times higher than the other motor, if not 10 or more. So the basic system was essentially working. (In fact, later some of the load was shown to be the bearings, which were stuffed full of thick grease, and the shaft, which wasn't square to the lower bearing.)

   I thought that was good accomplishment for one day and it was 4:30PM, but a while later I remembered, then went back and dug out, the device I made a few months ago for measuring static "locked rotor" torque with a scale. Of course the rotor wanted to cog to certain positions. From the center of where it naturally sat, I got the approximate readings below. I could only get it up to about 3 amps if it wasn't turning, partly no doubt because of the disabled current modulation leaving only PWM, which I had set at 12KHz.

0Amps: 0g
2.0A: 140g
3A: 190g

If I set it to a rotation where it had the maximum forward turning force with no power applied,

0A: 115g
2.0A: 230g
3A: 280g

The scale arm is 6" long, so divide by 2 for feet. 454 grams in a pound, so divide by 454 to get pounds. The highest relative reading was the first, 70g/A (below), and the lowest ones, after subtracting the cogging force, were just over 55g/A. (In review it was probably a little lower than 140g, maybe 130-135, but the readings weren't at all steady.)

140g * .5ft / 454 g/pd / 2A = .077 foot-pounds/Amp

   As current is raised, the cogging force should become a smaller and smaller component, since it doesn't increase, which would seem to indicate torque ripple will prove to be relatively low. But until I can drive it to some higher static amps, which should attain to greater accuracy and precision of measurement, these figures seem to indicate somewhere around .6 to .7 of a foot pound per 10 amps, as opposed to the original Electric Caik which seemed - as measured very vaguely since I didn't then have a good system - to be around 1 foot-pound/10Amps. This reduction seemed in keeping with the 50% extra no-load watts.
   To end the day I ran it up to 1200 RPM with about 8 amps, in both directions. That was about all the lab power supply wanted to put out. The diodes rapidly got hot but the transistors stayed cool. Then 600 RPM in both directions took only about 2.75 amps, and the diodes heated slowly. Apparently it's wasting more power to run it higher power and speed. 'Why?' Anyway it finally was running well, and a lot of things could be tried or adjusted.

   The next morning I turned up the PWM frequency to 22KHz. I thought that would raise the 'locked rotor' amps and torque, but it didn't seem to change anything. The diodes still got hot. As I began to investigate with the scope many unexpected things were seen. On one end of a wire things weren't always the same as at the other end. (I probably wouldn't have noticed with the old scope.)

   The gate drive signals, short tho the traces were, instead of sitting peacefully at ground, had big glitches up to 4 or even 7 volts when a different coil was being turned on or off. These are just the thing that blows up a typical BLDC motor controller, causing "shoot-through" currents from the supply to ground. All they do here is start to activate the wrong coil. That doesn't of course mean they shouldn't be cleaned up - only that the controller is still working, which makes figuring out the problem much simpler.

An "off" gate drive as another phase is being switched ON then OFF.
Both traces center on 0 volts. The lower trace is straight at the controller chip,
the upper one is at at the mosfet gates, past the gate resistor.

   The scope screen shows the problem: At 5 volts per division (when about 8 volts turns most mosfets fully ON) these are serious spikes, and could well be the cause of the extra power that seems to be needed to run the motor. Why are they there, and why are they worse at the gates than at the driver chip?
   But checking the mosfet common showed almost the same waveform, so it wasn't really turning the gates on. This gets even more perplexing, because if mosfet common goes above the 100mV reference on the driver chip, it should shut off the cycle. How can it be going up several volts? But wait... what about that .1µF capacitor I put on the current sense pin, replacing the .01µF, and which made the controller start to work? I put the probes on the current sense and reference pins, but the simple presence of the scope probe on the reference pin caused the controller to malfunction.

   The next morning I increased the flux gap in the motor itself, and the current to idle at 1000 RPM dropped to about 4.75 amps, or 118 watts at 25 volts. Now it didn't seem so different from the first motor. Perhaps the controller was just fine... but then why were the diodes heating up? I also added a 270µF capacitor from the mosfet sources to the supply to try to dampen down transients on the current sense line and the gate drives. This seemed to be somewhat successful.

   To digress from the sequence but complete a topic, the motor wobbled at higher RPM.s. I thought the rotor wasn't on quite straight. A few days later (21st) I took the top off the motor to fix it. The lower bearing still held the shaft on solidly. So I ran it up to 1000 RPM with the top off and only the one bearing. Instead of drawing 4.75 amps it drew only about 3.5, or 87 watts instead of 118. In other words, substantial power was being used to overcome friction. The whole shaft wobbled, so it wasn't the rotor, and this (more than the thick grease coning out the gills of the bearings) was probably the major friction cause. This was doubtless the case in the previous motor too, and I noted in TE News #60 that it seemed to take less current and power as the motor warmed up - and perhaps as it expelled some of the excessive grease, or the set screws relaxed a bit. (I could see in both cases it wasn't getting the 95% range of peak efficiency of the larger Electric Hubcap motors, where I used lightly greased trailer wheel bearings that didn't care if the shafts weren't 'perfect' and IIRC was using around 70 watts for 1000 RPM. And I did attribute it to the bearings, which I discounted as being a problem because they'll 'wear in' and improve. But running with one bearing showed the effect starkly. I got a new shaft to try and turn more exactly, but that's another job to get around to.)

Diode Issues

   If you want to shorten your reading, skip this down to "Diode Report" below, where I summarized the findings.

   Next I checked on a coil (at the anode of one diode) and the supply power (at the cathode of all three diodes) and found that for the first 1/2 a microsecond, the forward voltage across the diodes was about 5 volts instead of .5 volts, even at the low currents I was using. (I might never have figured this out with the old oscilloscope!) They were power schottky diodes, but apparently they weren't the least bit fast diodes. The amps being conducted as each diode switches on, with the high voltage 10000 times per second would probably heat the diodes up and cause extra watts to be needed to turn the motor - the observed symptoms. It seems it needs fast switching diodes (short Trr) that can also handle very high current.

The coil voltage (cyan) as it switches off exceeds the power supply voltage (yellow) by as much as 5 volts.
The two probes are directly on the anode and cathode of one of the 40 amp schottky rectifiers, which should clamp it at ~.5 volts.
Conclusion: fast reverse recovery time diodes are needed.

Yellow: ON-OFF switching spikes on the power buss,
TOP: supply side of the coil (almost 10 volts peak)
BOTTOM: at the supply filter capacitors (about 5 volts).
These are the two ends of the same 9" piece of fairly heavy wire:
  with higher frequencies and currents wires aren't just wires!
Cyan: the motor side of the supply bus coil, damped by multiple capacitors. (under a volt)

   In the previous controllers, the diodes were the built-in body diodes of the power mosfets. Here the configuration is wrong for that, and I used the heaviest diodes I had, dual 20 amp schottky rectifier diodes. I checked and compared the specs:

IRFP3206 MOSFET body diodes: Trr = 33nSec (typ), Vf = 1.3v (max, @75A) If(max) = 200A, Ifm(max) = 840A
TRPS40M120 dual schottky rectifier diodes: Trr = not specified!, Vf = .79 (@20A), If(max) = 2 * 20A), Ifm(max) = 2 * 220A

   The datasheet claims that the diodes are good for "high frequency switching power supplies", but doesn't even specify Trr!?!? But the reverse capacitance graph up to (2 * ~.001µF) seems very high to my uneducated eye, so it looks like it's an awfully slow recovering diode. Not what's needed for capturing the energy from a short pulse!
   I didn't think much of the 1.3 Vf for the MOSFET body diodes, either. That seemed like a notable energy waste and heat source, that I was unaware of until now. (What happened to "typical" .5 Vf for schottkys?) I thought they might work better for now than the schottkys until I figured out what to order. Then I went to Digikey.com and found that simple diodes with the specs I wanted were in fact far more costly than the IRFP3206 with the built in body diodes - they started at over twice the price and went up from there! And the only spec that was better for most of them was Vf, mostly about .95 to 1.1 volt. But that was "typical" and the 1.3v was "maximum", so they probably weren't much different. The body diodes in the mosfets I already have are superior in every other way. Anything with faster Trr than 500nS was considered "fast". Only one was faster than the mosfet's diode.
   Then I checked the body diode specs for the other power mosfets I have, IRFS7437.s, and they looked even better, and cheaper. Again Vf(max) was 1.3 volts, but Vf(typ) was 1.0 volts. 1000 amps momentary pulse. 30nS Trr. The one minus was that they were only 40 volts reverse voltage. That's okay for the 24 volt controllers. I'll have to find something else for 36 or more volts. I started this motor controller thinking to cut the number of mosfets in half from 12 to 6, but now it looks like I'm going to add three just to use them as diodes! (But other than tying the gate to the source so they never turn on as mosfets, what's the difference? Lots of the diodes are in the same or similar package styles.)

   On the 24th I changed the flyback diodes to IRFP3206 mosfets (gate shorted to source). On the scope they still appeared to be rising substantially above the supply voltage, but not as much, maybe 3 or 4 volts instead of 5, and for a shorter time like 300nSec instead of 500. The current to idle at 1000 RPM, with the motor still open (unchanged), dropped again to about 3.25 amps from 3.5; 81 watts instead of 87. The transistors stayed cool. The diodes still got hot - probably not as fast. Wow, how much of the heat in a regular controller is from the flyback diode action of the mosfets, rather than from the actual mosfet conduction of power?

Coil pulses with single IRFP3206 'diodes'
Yellow: the coil. Cyan: power line.
1: An entire coil ON pulse - a little under 20µSec.
2: Turn-on, 250nSec/div
3: Turn-off, 250nSec/div (Energy recirculation when coil > .8 volts above supply voltage.)

   This means a design change: the diodes will have to be firmly heatsinked to the case. (But electrically insulated.) If they get hot on a copper bar with 3 amps of motor current, the whole case will get hot with 200 amps. And that's with 1/2 as many diodes as a regular controller.
   A possibility to eliminate much of the heat is to have "active rectification", where mosfets turn on whenever the diode would be on. The forward drop of a mosfet when fully on is much lower than a diode, so its power dissipation is also lower. But it gets complicated.
   Silly as it seems, it should be useful to parallel two or more diodes so they handle the spike of current, short as it is, without such high forward drops. If two diodes lowered the drop to 2 volts instead of 4 volts, they would not only split the heat, there would be half as much of it. I made the copper bar for this on the 28th and ran tests then and the next day.

Torque Inconstant

   I checked static torque per amp by turning the supply to 1,2,3 and 4 amps with the torque measurement arm attached. Kt (torque per amp constant) is supposed to be linear, but (come to think of it), that's the "AC" current, presumably as measured on any phase, not the average DC current from the supply.

1A: 100g
2A: 150g
3A: 180g
4A: 215g

   Perhaps the thing to do would be to apply a DC current straight to a coil from the lab power supply and leave the motor controller right out of it.

Diode Lab Report

   I found the diode stuff important enough that I explored it fairly thoroughly. On the 29th I wrote a "lab report" on it and sent it to OSMC@yahoogroups.com [Open Source Motor Controller], where I had been conversing with knowledgeable people about the new motor and controller. I noted the surprising amount of heat - energy losses - in the diodes. Of course, the "half wave" controller has only one set of forward diode drop losses instead of two, so that source of heat and energy loss is actually halved.

Here is a copy of the report, where I summarized the main points/findings.

For those interested...

I said I'd have more on the 'flyback' diodes that take the pulse of energy stored in the motor coils and feed it back to the power supply. So here's my "lab report". The whole issue seems more noteworthy than I had expected.

In this sort of controller, often used for SR motors, the diodes are in the wrong place to be the body diodes of the mosfets. That lets us view the diode activity separately from the mosfet switching aspects.

In this particular "single ended", "half wave" or "unipolar" motor controller, with no high side mosfets and only one direction of motor coil activation, the complication is that the spike is up above the supply voltage, but the other side of the coil is connected to that same supply (instead of to high side mosfets) and not referenced to ground in any way except when a coil is turned on. So a 'flyback' diode to "+" would simply short out the motor coil, wasting the energy.

Rather than doubling the components by adding high side mosfets and low side diodes as is typically done, I put in the supply line coil, to dynamically decouple the motor coils from the battery for an instant while the turn-off pulse dumps back the dynamically stored energy, into capacitors or the battery.

The system seems to work. The supply current is much lower with the diodes than without and the driving mosfets stay cold (at the lower currents for an unloaded motor). But, unexpectedly to me, the diodes heated up badly.


Being used to seeing and hearing ".6 or .7 volts" for diodes, I was surprised to see forward drop ratings for really high current diodes to be on the order of 1.0 to 1.3 volts. I was even more surprised to see on the scope 5 volts across the ones in the motor controller.

Evidently the forward drop can be much higher in a short, high current pulse. With 40A rated schottky diodes I was seeing 5 volts for 500nSec, and they got hot pretty fast.

So I changed them for IRFP3206 mosfets, connected backwards and with gate connected to source, just to use their body diodes as diodes. (Rated at 200A, 840A pulse, Vf max 1.3V @ 75A) This brought it down to around 3.5 volts for 300nSec, and they still got hot, but more slowly. The no-load power to run the motor at 1000 RPM dropped from ~87 watts to ~81, suggesting the reduced heat was less wasted watts.

Yesterday I doubled them, two parallel diodes per phase. This brought it down to about 2 volts for 300nSec, and they heated more slowly still, partly because of lower internal dissipation and partly because the copper bar mounting was twice as long with twice as many diodes. Current seemed to drop a little more to around 73 watts. (around 2.9A @ 25V... I have to say say "around" because the analog current meter on my power supply is something like 10% off, so when I see "2.6" onthe needle I have to add ~.3. I did however get a new BK2530 digital storage oscilloscope for this project. I don't think I'd have got it all figured out with my old 1970s Heathkit scope with just one decent probe!)

I'm satisfied with this, at least for now. (I suppose I could triple the diodes and lower the losses a bit more.)

They surely need to be heatsinked to the outer case. (but electrically insulated.) After it's in a case and not spread out on the bench.


I doubled the PWM frequency from 10.5KHz to 21KHz. The switch-off waveform looked identical to the 10.5KHz, but since it was switching twice as fast, the diodes got warm faster, and the power to maintain 1000 RPM went up a few watts. But there's more torque/current at 0 RPM - 6A instead of 4A. This shows there are a few savings to be made with variable frequency or "CRM" modulation, where the switching speed drops with increasing RPM and or reduced load.


Throughout, the actual coil switching mosfets have remained virtually cold. So the only significant heat in the controller is from the pulse diodes.


BTW for the astute... the reason the power is still relatively high to idle the 4.8KW(?) motor at 1000 RPM is that its needle bearings are stuffed with pretty thick grease (you can feel the resistance when turning them by hand), and the shaft isn't held straight, so it wobbles. I have a new shaft to turn, but that's another job to get around to. The grease will eventually thin out.


   The conversation continued with feedback on this message. I got a couple of insights from it, and someone mentioned "FERD" - "field effect rectifier diodes" with low forward drop, which I had never heard of. This sort of unexpected information is why it often pays to run things by others. To demonstrate that the circuit works (there was a "skeptic"), I bypassed the coil. And sure enough, the current and power to run the motor went way up. I'll order some FERD.s and try them out, but as the highest rated are only so high, I'll still have to parallel some. Maybe even 3 or 4 pairs per phase. Any more than that I'll stick with the IRFP3206es. Rather than rewrite about the same topics, here's the message I sent.

Thanks everybody for the feedback!

The circuit does work. The watts to turn the motor drop with lower Vf through bigger/more diodes. If the diodes were merely shorting the motor coil spikes and not returning the energy, the watts would remain constant.

But to see it better I've just tried bypassing the coil. I turned the PWM up to 21KHz instead of 10.5, but I could still only get the motor up to about 650 RPM, where it was drawing ~5.4 amps; 135W. (@25V) That's where the PWM is limiting on overcurrent with the present sense resistor. And the diodes get hot faster. (The driver mosfets still run cold.) With the in-line coil, the same RPM takes about 1.5A; 38W.

I already know that if you try to run it without the diodes even at the lowest power, the mosfets get hot fast, until one shorts out - also the motor barely turns.


>I think using diodes to steer energy will always be a lossy method.

At least with this controller there's only one diode drop instead of two! And the coil voltage is drained to zero instead of just to the power supply voltage.

>If you want to switch currents from here to there it seems like you should use a switching device not a blocking device.

I'm trying to keep it simple. This would re-introduce potential for shoot-through currents, and require floating high-side gate drives, both of which I wanted to get away from. Of course I never suspected such high diode losses. But the FERD diodes Chris P. mentions might be just as good, which is exciting.

>With large diodes you will always have a significant turn-on time (and turn off).  That is why you see the large forward voltage.  I would wager that you are seeing an average voltage.  With a high spike until the device turns on then the voltage will drop to the normal forward voltage.

I think slow turn-on is probably why the pulse was 500 nSec with the Schottky diodes I started with, and then it became 300 nSec with the fast (35nSec?) IRFP3206.s.

Oscilloscope screen pics (in next Turquoise Energy News #89 in a few days) show the switch-off waveforms seen, at 250 or 500 nSec/div.


>So some simple math. If the forward voltage is 0.7 at 10A and you are PWM 40A the current the diode has to handle is the same as that forward current. So lets say the recirculation current is 40A 40*.7 is ... 28W? lets say each set of diodes is handling that at ~50% of the time. so a total of 42W is being disapated in those diodes.

The math looks good, but it looks like the diodes are recirculating at several hundred amps for 300 nSec each PWM turn-off, for the presently active coil out of 3 coils. At 21000 Hz it seems to work out to just 2.1 mSec ON out of each second. The power dissipation will average out somewhere, but the Vf is much higher than .7V owing to the high current, leading to excess dissipation compared to reasonable current over a reasonable time period.

Looking at the graph of typical forward voltage versus current in the IRFP3206 datasheet (instead of just the printed number of "1.3V max @75A"), the current for these very brief periods seems to be in the upper hundreds of amps, if not over 1000, considering I had to double them up to get Vf down to 2 volts.

One could wish the coil spikes were a lower current, spread over a microsecond or two, instead of 300nS.


>If you do have a lust for diodes the ST Micro FERD diodes seem about the best.  I built a 3 phase rectifier with them for a high efficiency generator project.

Thanks Chris! These look very interesting, and I had never heard of them! It looks like the highest current rated is the dual 60 amps at 45 volts (FERD60U45CT). I'll probably try those out for a 24 volt controller. The fact that the graphs only go to 60 amps per and look like the Vf voltage goes higher rapidly above that makes me a little nervous.

I probably have to double them (4 individual diodes since they're already dual) or even more, but if the number needed in parallel isn't impractical (possible), it looks like they'd doubtless have lower drop than anything else.

I'm a bit nervous about the 45V spec for a 36V or higher controller, and none of the higher voltage ones are rated above 30 amps, which looks like it would need quite a few in parallel.



   I'm very happy with the motor and controller. I think they'll be the best BLDC product on the market - or at least the best design, with the least wasted energy and heat. The one thing left for the controller (that I can think of) is to get the variable "CRM" part of the modulation working, which wasn't working right and I disconnected it for the tests up to this point. Then it's "pre-production" PC board design and full layout design for the motor controllers.
   As for the motor... there's always little design tweaks. I put in a spacer so I could loosen the set screws on the lower bearing that were pushing the shaft off center, and with the top back on the motor, the current to idle at 1000 RPM dropped a little more to about 4.3A or 108 watts -- almost the same as the original Electric Caik motor with a "standard" BLDC controller. (The variable frequency CRM modulation should lower it further.) I've mentioned the slightly larger store-bought "disk brake" rotor to use next time. I'll have to see that it actually fits okay with the epoxied PP magnet strappings wrapped around the outside.

The initial "Version 2" motor controller schematic, with corrections from version 1.
Note that the variable frequency modulation circuit section (4025D gate C et al)
was disconnected during testing, and is untested at this point.

Full Circle

   Also on the morning of the 19th, when cleaning up I ran across my magnet rotor diagrams that I'd used in figuring out whether the new Electric Caik phase sequencing made sense. In 2008 the first motor wouldn't run and I had concluded that the magnet configuration had to be two magnet poles per three phases instead of four per three, and I had never since questioned that conclusion. It seemed to be supported by the fact that every diagram of a BLDC motor I've ever seen has two rotor magnet poles per three electrical phases. But somehow as I looked this time I saw that this was wrong: there was no reason four per three shouldn't work if the optics and power phases were in sync. Apparently I'd been laboring under a false pretense all this time!
   It would switch coils twice as fast, and it should have a higher Kv and probably more torque per amp since the rotor magnet field changes were closer together and closer to the active coils. To verify this would work, I'd have to make up an adapter cable for the magnet sensors and perhaps use some aligator clips for the power wires, and hook up a regular motor controller to the new 4:3 Electric Caik to try it out.
   Still, the new controller has 1/2 as many driver transistors and diodes, so it should make less internal heat and be more efficient, and of course it's also intrinsically more reliable with no chance of shoot-through currents. It's surely better. So just what was the point of this realization? Academic, perhaps.

   How balanced is the question? With the bipolar controller, current is flowing through two active phases of the three, while with the unipolar, it's only through one. For current "X" the bipolar one creates twice the torque. But heat is the limiting factor, and in the unipolar controller each phase gets to rest 2/3 of the time instead of just 1/3, so the current in the on phase can be higher without the motor getting any warmer. One thinks that "I squared R" losses must be higher with higher currents, but the coil resistance is for a single phase instead of for two in series, so "R" is half, so current can be square root of two higher with no more losses. I wonder if all this doesn't approximately or even exactly cancel out to yield roughly similar or even theoretically identical motor performance. In that case, what's left is the reduced motor controller losses, and improved reliability. And the new controller will run a switched reluctance motor as well - "two birds with one stone" as they say.

Now, about that Switched Reluctance Motor

A steel bottom plate & AFSRM rotor with thrust bearing
   The successes with the new configuration of BLDC motor and the unipolar controller didn't stop me from thinking about the SRM. I decided the designs in the papers I'd read weren't quite what I wanted. The Japanese one had huge torque to direct drive a vehicle wheel, and low RPM. But the real advantage to the SRM, it seems to me, is in having a solid rotor that can spin effortlessly and safely at many thousands of RPM. The need for huge torque is then eliminated by reduction gearing. If a vehicle like the Chevy Sprint needs 200 foot-pounds, a wheel motor with a 10 to 1 toothed belt reduction needs only 20 foot-pounds. (Or two motors with as little as 10 foot-punds each.) And if designed as I envisioned it so the reluctance changes more smoothly instead of going for a maximum torque at a single point, surely the torque ripple could be relatively small. If a motor is directly connected to a drive train (no clutch), the minimum torque is where it'll sometimes have to start moving from.

   I decided I would build one, and that it would be the sort of design I had first envisaged, as modified by realizing that the flux gaps must be tiny instead of huge, and that therefore it all has to be constructed differently, all precision built and everything must align perfectly. Successful performance will hinge largely on a carefully planned and well implemented physical/mechanical design.

   I like the two rotor idea, but it's much more difficult because it's hard to make a solid stator when it's full of wires and cores that extend the full height, and yet make it solid enough to mount from the outside edge, and perfectly flat on both faces. For a single rotor design, it'll start with a flat metal platter for a base - not just randomly "flat" plate steel but unless it's already truly flat, hammered flat (if that works well enough) or machined flat on the lathe. The toroid coils sitting on these will be carefully wound with two layers of #11 wire (20 turns) so that the wires don't protrude over the top or bottom by even a fraction, anywhere. Their sheet steel outer rings will be epoxied around the wire windings, held in place with cable ties until the epoxy sets. The inner lips of the toroids curve smoothly out at the top, and I would machine pieces (plastic?) that would fit over these curves to hold the coils against the platter without themselves sticking up above the coils, using flat-head machine bolts. If that's not stiff enough, it could all be glued together with epoxy.
   Once finished, this entire assembly would be sanded flat on a belt sander to ensure nothing was sticking up. The center of the platter could be machined to hold a needle bearing dead center at a fixed height, holding the rotor 1/2 a millimeter above the coils.

   I got started with slightly turning down a curved edge that almost fit the thrust bearings I had bought so that it did. I played around a bit with other pieces, seeing how they might be fitted together. Then I moved on to other equally worthwhile things and I may not get back to this before next winter.

"Green" Electric Equipment Projects

Vertical Axis Lathe for Large Diameter Pieces

This is just an idea that resulted from answering an e-mail...

Sounds good.
What do you think of the design of the alxion rotor and stator? longer magnets and coils in a drum configuration.
Getting a centered drum for the magnets may be tricky.

I thought they looked very nice. The drum would probably be done on a lathe,

   Here as I typed I thought of what a huge machine lathe would be required to turn a drum of the required diameter, and that something else might be rigged up... probably with the workpiece horizontal on a table.

or some device that would be used as a lathe.

There it was: Take a vertical axle that could mount and spin the workpiece. Add an appropriate lathe tool rest that sits by the outside rim, and a sturdy metal bed or frame connecting them, and one would have a vertical axis lathe that would be used for turning large diameter objects like drums, disks and large pulleys! Where a huge horizontal axis lathe is needed to do large diameters, the working length of a horizontal axis lathe is the radius that can be turned. So if it was even just 18", one could turn a disk of up to 3 feet in diameter. Even the Electric Weel motor, much the biggest thing I've done, is only 28" diameter with a 26" rotor. Surely some small and simple vertical lathe for relatively narrow rotor, disk and pulley shapes could be easily made?

   It wouldn't be needed for small disks that can be done on a small regular lathe, so it would only need slow turning speeds with lots of torque, just a few pulleys with a very large one(s) under the axle. Perhaps flat belt pulleys?

   In essence, initially I was only thinking of a fairly simple tool mounting at the outer rim to shape and center the outer rim area, not to profile the whole top face or work near the center.

   But that could be done too. A fixed gantry could fit over the whole unit not only for milling the top surface, but to allow tools to be positioned at any radius over the top surface. The part could accurately be rotated to any angle and clamped, to drill holes centered on the axis, for example mounting holes for car wheel or other 'stud' bolts. A large diameter could be hard to stabilize to the required extent for milling, so the area under the tool, just inside the outer rim, might have an adjustable but very stiff wheel (or two wheels, one before and one following the tool) contacting the bottom of the workpiece to prevent it from being pushed down by the tool. As I see it, the tool working from the edge would initially mill a small machined-flat lip underneath for the wheel(s) to run true on. (The Electric Weel project has shown me how much a bit of play magnifies as the diameter is increased!)

   I had never heard of such a thing, but I checked on the web. Wikipedia just mentioned vertical axis lathes (not even a picture!), but e-bay had quite a few for sale. Lathes need to be very sturdy and inflexible, but everything I saw looked way overbuilt and overpowered compared to what I was thinking, and wouldn't fit in my shop without removing most everything else. Only one unit was under 10000$ and they worked their way up from there to 350000$. A Fuji unit I looked at was about 6' x 6' and taller than long, was about 25 horsepower, and only did pieces up to 16" diameter (but quite tall). Others could do gigantic pieces up to 9 meters diameter.

  Gearing down, as with the belt and a very large pulley, should be the key to reducing horsepower requirements for milling the large diameters. The stabilizer wheel(s) is probably the key to reducing the requirement for a long, fat axle and huge bearings to attain the required precision.

   As if I needed more projects! But maybe somebody else will like the idea... or just maybe I'll have to make one for doing motor or drive work myself. My own lathe will turn up to a 10" diameter workpiece if it's narrow. (larger than many!) If I can just stay under that I should be okay. It's just larger flat belt pulleys (which I'm pretty sure I want), and potentially larger motor rotors, that pose a problem for which this could be a good solution.

Aquaponics the Manual Way?

   After the dreams of mopping up a flood, I disassembled the upturned fridge aquaponics system and transferred the two remaining tilapia to a 220 litre HDPE drum with the top sawed off, which made a better fish tank.
   The water in the aquarium started to smell bad, and it had got so murky you could hardly see the fish. Being busy with other things, I didn't do much about it. I put in some "aqua-clear" and changed a little water. One morning two of the three fish were dead. (The pH was 7+.) Not much of a fish farmer, I! I quickly transferred the survivor, the big female, to a 10 gallon aquarium and added fresh water to it. Over a couple of days I changed most of the water.

   Then I took the whole (30 gallon) aquarium outdoors and cleaned it out. I brought it back and filled it with tap water (and a little fish drum water to get bacteria started) since there was no really clean rain water. (The chem spraying saw to that.) My brother said chlorine dissipates in a day, so after a day I put the two fish from the drum into it. One of them I thought didn't look very well even before I moved it, and it died in a few hours after moving it. I don't know if it was already that unwell or if it was the move. I had fish burger dinner the next day. The other was fine. I was now down to two large tilapia, the female for some reason twice the size of the remaining male, and the one remaining offspring of the largest.
   A day later I put a coated wire fridge-shelf barrier in the aquarium and put the big female in the other end. This seemed to get the male very excited, but the female hadn't looked well in the 10 gallon tank and didn't look at all well now. I feared it would die like the one the previous day. The next time I came in, the fridge grill had been shoved aside and both fish were together. The female seemed to recover after a couple of days, but for whatever reason was much less aggressive than before, and the two fish behaved like great pals.
   The small tilapia in the small tank beside the big one was now 6" long, and on the 18th seemed to be trying to get through the glass to join the other two fish. So I put the fridge grill back in, towards one end, and transferred him. (From the bucket to which I was slowly adding water from the target aquarium to acclimatize him, he took a jump, but cleared the whole aquarium and landed on the floor. I wet my hands and scooped him up. He flopped out when I'd almost reached the water and it took another try. Hmm... I could have used the net.) He was okay. Now I would just have one aquarium to deal with until and unless there were more young.
   The fish still pushed the grill aside and the small one was often seen with the others. But they had a couple of pipes to hide in, and none seemed aggressive like before, for whatever reason, despite, or because of, being three distinctly different sizes.

   In the aquaponics setup the water gets cleaned of nitrogen compounds as it circulates through the plants, but in the aquarium they just build up until toxic levels are reached. The obvious thing to do is to replace some of the water frequently with tap water (or clean rain water if available), and use the removed "fertilizer water" either in a manually watered aquaponics grow bed or outside to water the garden. "Manual aquaponics." Something just doesn't sound right about that.

   On a related note, birds or some animal ate the wheat seeds I planted after tilling where the pool had been, and nothing came up. I replanted part of it and buried the seeds a little deeper. The wheat is now nearly a foot high. I planted some quinoa in a couple of rows, but I don't see anything forming rows. It doesn't help that I don't really know what it looks like.

Electricity Storage

Turquoise Battery Making Project
Cylindrical Battery

   I finally got back to this on the 15th while waiting for a new oscilloscope for the unipolar motor controller project. Not wanting a big ugly pipe cap on the end of the battery, I found a hole saw that I had already made for something long ago, to cut disks that were just about the right size (with just a little filing) to make internally fitting end caps for the 3/4" PVC pipes for the battery cases. I cut one out of ABS, filed it to fit, dripped methylene chloride on it and hammered it into a piece of the PVC pipe. This pipe was then a bit too short because the internal cap took up its thickness, 3/8". So I did another with a slightly longer pipe. I decided for this cell at least that both electrode posts would come out the top to minimize any chance of leaks at the bottom.

   I then took the gum arabic mixture and some of the fine nickel flake/powder and mixed them together into a paste. I painted some onto the inside of the etched cupro-nickel sheet metal tube. It seemed like a pretty thin layer, so I let it dry and repeated the process, working quickly enough that [I hoped] the first layer wouldn't soften again.
   With the nickel-rich micro-convoluted etched surface on the sheet of conductive cupro-nickel, and the mass of conductive micro-fine nickel flake jelled in gum arabic, I think it should be a mighty fabulous electrode!

   For the 'posode', I shredded some graphite felt by hand, which I planned to add into the powder to improve conductivity. Then I felt little bits pricking my legs right through my pants, as if I had been working with fiberglass or carbon (graphite) fiber. A magnifying glass confirmed that the felt fibers were long and fine, same as carbon fibers. How did I expect them somehow to be soft cloth of some sort?

   Then I cleaned the carbon terminal rod off (from an 'F' dry cell), and went to paint some osmium doped acetaldehyde on it to improve the surface conductivity. But the tiny remnant in my little test tube had finally dried up. I'll have to look it up to make some more, since I've only made it once, in, um... 2010?

   That was as far as I got before my new oscilloscope arrived and I went back to working on the new configuration BLDC motor and unipolar motor controller. And I didn't want to deal with the carbon fibers.

   Leonardo e-mailed on the 8th that he had sent me the 'ethaline' DES! It'll be great to try it out and see how it works in a cell.

   On the 26th I decided to make the positive electrode, without the shredded graphite felt. I just didn't want to touch it. And forget the doping - just get something that hopefully works together, even if the current drive is lower. People want it!
   It took about 40g of positive powder, to which I added around a gram of Sunlight dishsoap and 3 grams of Dieselkleen and mixed it up. I pressed it to 2 tons, then put it at an edge and pressed the whole thing out of the pipe. There were problems with pieces breaking off, including the bottom but mostly at the top, and more of them when I tried to insert the carbon rod. I filed the little bumps off the rod and it went in freely. I wetted a piece of Arches watercolor paper to make it pretty limp, and wrapped about 2 complete turns of it round the electrode. Then I cut part of the bottom away and folded the rest over the bottom, covering the bottom of the electrode.

   When I went to insert it into the outer nickel metal electrode, it was a tight fit, and I spread the sheet metal cylinder apart some more with a spoon; then it slid in okay.

   And the entire assembly slid into the outer PVC pipe shell okay. That left the top still open, and the 'stuff' stuck up right to the top. I decided that instead of trying to fit any solid top, I would just paste in RTV cement or heat glue over the whole top, which wasn't a very big area. Or maybe pour in melted wax?

   But I was too impatient for any of those things. I added (15ml?) KCl electrolyte and measured the voltage: .50v. Then I hooked it to the power adapter to charge through a 50 ohm resistor (~20mA), and the charge voltage was soon over 1.5v. I changed to 100 ohms (~12mA) and then 200 (~6mA) to keep the charge voltage down. It seemed I had one more poorly conducting cell. Then I melted some wax and covered the top.

   Soon it occurred to me that I had used posode mix I made before I had got the carbon black. (It even said "2013/10/20" on it. I should date them all!) It had the old impure art store graphite in it instead - at least, the carbon content was simply labeled "C". With no qualifying explanation, that would be the old art graphite. Oops! Wasn't that a main cause of self discharge? (This is of course what happens when you work on something so seldom - the finer details are forgotten!)
   As now expected, over the hours of charging the self discharge wouldn't go away. I guess this was just another unsatisfactory test cell. The remainder could go out to the compost pile to add trace minerals to the soil. I found another jar with a similar posode mix but having no "C" on the label. (Rats, no date.) Presumably I could add carbon black instead of graphite to that.
   I decided to scrape off the wax, grab the cupro-nickel sheet with small pliers, and take the first cell apart to re-use the shell and the negode. I couldn't get a grip on the sheet metal around the rim, but finally I grabbed the carbon rod, which had been loosely inserted. It was now quite tight and I pulled the assembly out of the pipe with it. The part about the posode swelling and tightening up around the carbon rod, inserted rather loosely, seemed to have worked great.

   On the 30th I got back to it. I put in 26.7g of the powder with no carbon, intending to make it 40g with conductive carbon black. But the carbon black was so light and fluffy that I only got up to 30 grams, and visually it looked like over 50% carbon black by volume. No doubt it compacts far more than the active materials. I added a 1g of Sunlight, 2g of Diesel Kleen, and 4 or 5 grams of water to dampen it further. I used it all in the pressing, which again didn't go smoothly, cleanly or nicely. When it came out of the press, again it crumbled and broke when the nail was pulled out. It lost maybe 5g of crumbled material off the top. I can see there's a number of changes and additional tools and jigs needed before making any quantity of cells. This time I decided to leave it to dry before trying to make a cell.

   On Canada day I torched it for a few seconds, wetted a piece of watercolor paper and wrapped it, slid it into the outer negode, and slid it all into the PVC pipe. It didn't seem to want to go in much farther than flush. That was annoying and there seemed to be no reason for it. I left it that way, paper sticking out, and added a little salty (KCl) water. I hooked it up and started charging at ~10mA, soon at 1.35-1.4V. Once again higher currents seemed to overpower it and the voltage went way up, so I couldn't charge any faster.
   After 12 hours, the center post seemed solid. But when I disconnected the charge, the voltage still dropped off markedly and continuously. It didn't look very promising, and stayed that way. What am I doing wrong? I was doing better with the NiMn, at twice the cell voltage!

Victoria BC Canada