Turquoise Energy Ltd. News #35
Victoria BC
Copyright 2010 Craig Carmichael - January 3rd 2011 - Happy New Year!
http://www.TurquoiseEnergy.com
= http://www.ElectricHubcap.com

Contents/Highlights:

Month In Brief (monthly summary)
3 Years in Review (a look back at developments: wave power, EH motor and car drive, batteries, Department of Progress, other projects)

Electric Hubcap System & Motor Building Workshops
  * Metal Stator: had unexpected eddy current losses
  * New Plastic Stator: molded composite of strong, lightweight polypropylene mat & epoxy resin - non-conductive so no losses.
  * Toroidal Iron Powder Coil Cores: Commercially made "T200-26B" core is EH's exact size! Will greatly simplify making coils, boost efficiency and performance.
  * Testing motors to determine specs.
  * Calculations with above improvements indicate EH motors will attain 95% efficiency.
  * Motor making workshop price reduced with easier to make motors: $650 (Includes parts/materials for motor, $300).
  * MC33033 motor controller deficiencies uncovered.
  * Making new A3938 motor controller (some problems/adjustments remain, but controller works).
  * A3938's current limiting drive works differently to typical PWM - and more like a gas pedal.

Torque Converter Project
  * More mechanical torque converter theory.
  * 5-Star Escapement Torque Converter: best design yet.
  * Testing Problems: Motor controller deficiencies and sloppy mechanical coupling to car wheel have been hobbling torque converter tests.
  * simple coupling change for next tests.

Turquoise Battery Project
  * 9V battery disassembly shows simple yet exacting construction techniques
  * Ni-MH battery price reductions: for $200 I may replace my car battery with 30 "D" cells. (If you want some Ni-MH batteries too, a bigger order would get us better quantity discounts.)
  * Latest Turquoise battery design overview (details here)



Newsletters Index/Highlights:
http://www.TurquoiseEnergy.com/TENewslettersIndex.html

Construction Manuals for making your own:

* Electric Hubcap Motor
(latest rev. 2010/09/xx)
   - the only 5+ HP motor that can easily be made at home?
* Turquoise Motor Controller
(latest rev. 2010/05/31)
   - for the Electric Hubcap. (Probably there are commercial controllers that would work, too.)
* 36 Volt Electric Fan-Heater
   - if you're running your car on electricity, you'll want a way to defog the windshield and keep warm.
* Lead-acid batteries: Sodium Sulfate 4x longevity additive - "worn out" battery renewal.
* Nanocrystalline reflective rear electrodes to enhance DSSC Solar Cells.
(in TE News #28, 29)
* Simple Spot Welder for battery tabs, connections (in TE News #30)

are all at:  http://www.TurquoiseEnergy.com/



December in Brief

   Thinking I had the EH motor design pretty much finalized and having been given some motor equations and techniques, I spent two or three weeks trying to measure and derive performance specifications, hoping to be done with it and move ahead on other projects. But some of the numbers weren't making sense, and soon a fairly serious deficiency I'd glossed over came to light: the steel plate of the stator disk was causing a rather strong electromagnetic drag load on the motor, so the no-load currents were rather high, reducing efficiency and overall power out. I had expected the distance of the plate from the magnets (over 1.5") would make this effect trivial, but it didn't.
   Putting a 3/8" gap with pieces of plastic between the stator plate and the coils helped quite a bit, but I wasn't very happy with the arrangement. Then I unearthed the very first prototype EH stator, 9 "reduced fill" iron coils cast in a polyester 'donut' ring, and contrived to couple it to a new magnet rotor, for the first time with no backing plate. Now there was no steel except the coil core laminates, and it was obviously better. With the steel plate, if you give the motor a good spin by hand, it stops in about one turn: electromagnetic drag. With the plateless ring and partly unfilled cores, it would go about a dozen. A few measurements showed the low no-load currents I was originally expecting and aiming at... but didn't actually have with the steel plates.
   I decided a new stator design using plastic composites (polypropylene mat and epoxy resin) would be necessary. So much for getting on with other projects! Being able to make a stator of the exact shape and size I wanted would be an improvement, though. The motor is thinner, it protects the coils better (the wires stuck out past the rim of the metal one), and it's also an end cover of the motor enclosure. It's also lighter, and the bare bones motor weight is down to 25 pounds.


The new lighter, more efficient EH motor with the cast composite stator.
(Stator spray painted red with hi-temperature motor enamel.)

   The old stator had gained some of its high efficiency from a special "hollow" core. I also found a source for off-the-shelf toroidal "hollow" iron-powder cores of the exact right size and shape. Samples are coming. These will not only improve efficiency, they'll make the coils, and hence the motors, much easier to make.

   Now no welding is needed to make the basic EH motor, and the machined bearing hubs, now molded in, are easy to make on any metal lathe - or for me to supply through the mail - eliminating a prospective motor builder's need for a lathe. The coils are also much faster and easier. Only casting the stator - probably - takes more time than the extra turning of the hub and welding it to the metal stator did.

   I did get a couple of other things done. One of them was the small amount of work needed to finish the "5 Star" version clock escapement torque converter. I took it out to the car and tried it with the wheel jacked up, but I also took out the current probes now that I had the DC probe, and found out the motor controller wasn't letting the motor get enough amps to develop more than around 1/3 power. In addition, I finally realized that play in the coupling to the car wheel has been robbing the wheel of receiving much of the force from the pulses of torque the converters have put out. It's been "hobbled" like that for all of the torque converter tests from day one! When I figured that out I was disgusted and I disassembled everything and took it all inside, including un-installing the motor controller. The torque didn't feel much greater than last time, but later I regretted not at least trying to move the car on level pavement. I think now it's a very good design - it might have done something relatively inspiring like take the car across the parking lot.
   Along with that, I started visualizing better more refined designs for a similar torque converter that would last a long time - this prototype is probably good for a few miles at best, seeing hte aluminum dust it emits when it's going.

   I burnt out the shop motor controller during motor testing owing to a motor short circuit problem, and I put together a PC board with the new A3938 motor controller to replace it, which was after all on the project list. In testing it, the chip got burnt out and I didn't get it working. In light of that, the controller from the car was needed for further motor performance testing. More and better motor controllers are now assuming a high priority. Tristan removed the minuscule dead 38 pin chip and soldered in a new one, and on the 31st the new controller was working, but soon burned out again when highest current was demanded. Some adjustments to the maximum allowed current seem to be needed.

   Another thing I did was to try baking another gooey carbon pitch/graphite sheet, and I found out that if you cool the sheet off in cold water, the aluminum that's glued solid to the sheet at room temperature peels off relatively easily. Putting it in the freezer should work even better.
   Being able to liberate the pliable sheet probably makes it a workable electrode contact sheet, and probably much the best solution yet. Workable salt electrolyte batteries seem a good step closer. However, I didn't have time to try to actually put together a battery, much less run tests on it.



3 Years in Review

   As I begin the fourth year of the Electric Hubcap and Turquoise Battery projects, started almost from "Happy new year!" in 2008, perhaps a look behind is in order.
   I started my 'green energy' inventive career by creating some designs to generate power from ocean waves in spring 2006, improved over designs I'd seen in March, and as usual in trying to create something new, I got enthusiastic lip service from some quarters, incredulity and scorn from others, and no support at all from anywhere. You would think the government would enthusiastically assist, and businesspeople would grab up, a new type of product that nobody else had previously made, to be there first. But most of my fine inventions, even when completed to at least a good demo state, have fallen victim to complete indifference on the part of those with resources to help develop, foster or commercialize them. It seems instead like no one today believes anything really new is even possible - no wonder the middle ages lasted so long! However, I made a wave power test unit, and a generator with supermagnets for it. I got the idea that the generator would make a great electric car motor.

   In January 2008, I decided to pursue the electric plug-in hybrid idea, and to make designs that could be made at home or in small shops so that those who liked the ideas could do them as "DIY" projects, regardless of the indifference of government or the hostility of capitalists. For the first time, great ideas could be published and spread via the internet and anyone, anywhere, could access and use them.
They could spread despite my personal semi-poverty. Others better placed and more business oriented than I might turn the inventions into businesses and then fortunes, and electric transport would grow from the grassroots. The products I was about to lavish my time and energy on wouldn't go to waste like so many of my others.

   I finished the proof-of-concept wave machine on the cheap regardless, but mechanical defects prevented a successful test. Ironically, the generator I'd made had heavy magnetic cogging and was part of the problem. I substantially re-did the mechanics recently and got a non-cogging lawnmower motor for a generator, but since then I haven't had time to finish it up and try testing it.


Dual-float wave power test unit with mechanics mounted on bouncy trailer -
one of the sources of slack in the mechanism that prevented a successful test.


   As originally conceived, it seemed to me that supermagnets, used in an axial flux motor, should have so much torque at such low speeds that it could drive a car by direct connection to the wheel, and so the main components of the system would be the motor, the solid state motor controller, and, hopefully, some batteries that were better than lead-acid. But though the first good motor - after some poor but gradually improving prototype designs over the summer - moved the car in October 2008, the torque was insufficient to be practical. It finally dawned on me that it was actually limited by the weaker field of the electromagnet coils, not by the supermagnets.


Car moving with directly connected Electric Hubcap motor, Oct 20th 2008

   I made a better motor, but by spring 2009 it was evident that driving a car would take something more than a fairly small motor running at much too low a speed, and I got the idea of a magnetic or mechanical torque converter. A workable magnetic design proved elusive. The oscillating masses theory of mechanical torque converters and workable mechanical designs proved to be two different things (also for others who've tried recently, though Constantinesco's 1926 design worked well), and it wasn't until the last months of 2010 I found a workable design concept - the "clock escapement" inertial masses torque converter. Even then, continuing motor improvements and tests, and other needs, kept the construction and testing work down to a snail's pace and prevented a working design until December 2010. It sits ready to test, a couple of hidden problems that have been in the test setup all along having just been identified. One is 'fixed' for a prototype test, the other will soon be solved by a new motor controller using a more advanced motor controller chip (A3938), finally working December 30th.
   On that subject, when I started I didn't realize there were off-the-shelf 3-phase brushless motor controllers, or chips for the purpose, so I designed my own from six various logic and driver chips. I proudly put the design on the web in the fall of 2008, only to be shown the MC33033 controller chip. I made a new controller out of that and it seemed great. But a few weeks later I got a message from Digikey where I'd bought it, saying it was to be discontinued. From somewhere (web search?) I found the A3938. It turns out the MC33033 has some quirks, and I belatedly recognized them at the same moment that I had designed the A3938 PC boards and was ready to assemble the first one.
   In the course of all this I tried a number of layouts and finally hit on a double row format with the high power components bolted to heat sink bars and hard wired, with the logic/gate driver PC board for whatever controller was in fashion bolted on top with 'standard' wire connections to the power components. The bars couple heat through to bars on the outside with fanned-out thin aluminum roofing flashing heat dissipation fins clamped beneath them, eliminating the need for expensive aluminum extrusions.
   And I did a complete integrated aluminum chassis layout wherein all the wiring related to the motor system started or ended. The controller occupied one side wall and could be removed for repair without dismounting the box and cable clamps, and the whole box is an an extension of the heat sink. The transistors hardly get warm.

  
Motor controller with controls on the chassis, for testing EH motors in the shop

   Concurrently with the car drive system, in early 2008 I started trying to create nickel-metal hydride batteries that could be made at home, after finding that economical NiMH batteries big enough for electric transport have deliberately been made unavailable - suppressed - commercially. But I also had in mind the possibility that being new to the whole field, "unindoctrinated" as it were, I might stumble across something great that others had missed. In my ignorance I started seeing seemingly great 'overlooked' possibilities, some of which were impossibilities, but I became convinced the whole field wasn't very well explored and that there were much better possibilities than any current battery technologies.
   As time passed my knowledge of batteries broadened, and my electrochemistry ideas started coming into line with things that might actually be possible. I was still trying things nobody else was.
   By early 2010 I had the outlines of a fabulous battery chemistry, novel techniques for doing vital but seemingly very difficult parts of the work at home, and the outlines of an excellent battery construction. It seemed I had solved nearly all the issues... at least in theory.
   I decided to use dissolved salt (KCl) electrolyte instead of acid or alkali, as the average reaction voltages are highest, including the overvoltages, and it's a fast electrolyte.
   I created a way to make -1.4 volt manganese negative electrodes work, which it appears no one else has ever done, by using egg albumin to raise the 'hydrogen overvoltage' so it can charge - an idea gleaned from an obscure piece of 1962 research. By comparison, iron is -.9 and metal hydride is -.8, and manganese is lighter.
   Instead of just nickel for the positive electrode, I put in a mix of nickel and manganese. The manganese charges from Mn(OH)2 [II] to potassium permanganate, KMnO4 [VII], an astounding valence change of 5 at +.9 volts for fantastic energy density. The nickel, charging from Ni(OH)2 [II] to NiOOH [III] or NiO2 [IV], mainly adds stability instead of being the main ingredient. Graphite powder increases the conductivity of the oxides, just as with the cheapest dry cells. A couple of other treatments also prevent the permanganate from migrating.
   Thus the battery's electrochemistry provides a fabulous 2.3 volt cell with incredible energy density from mostly cheap ingredients (except for the nickel). They should be cheaper than lead-acid and shrink them at least fivefold with even more weight reduction - there's the real battery revolution!

   The last and most thorny problem - perhaps more thorny than the electrochemistries - was that with salt electrolyte every metal I tried dissolves - oxidizes - in the positive electrode. The common dry cell has the same problem and from its carbon electrode rod I finally realized that carbon/graphite conductive sheets were about the only thing that would work. Making such sheets and getting a battery that truly worked properly was elusive. Again everything came to a near halt in the fall as I grappled with the motor and motor controller development, testing and improvement issues, but before the end of 2010 I had the concept for a somewhat gooey, pliable carbon sheet that should solve the problems.


Conductive, sealing carbon battery layer

   As time went by working for progress I began to realize that the chief problems with switching to clean, cheaper, renewable and non-polluting power, including for transportation, were not technical but social. Existing, proven technologies that were known to work and could easily do the vital jobs weren't being employed or further developed. I identified two main culprits that are preventing us from rapidly achieving the world of our dreams.

   The first culprit was vested interests. In the last century and more, a very small number of unscrupulous and manipulative people have exploited quirks in our social and political structures to seize economic power and build giant economic empires. These are, for example, the less than one percent - and probably far less than one percent - that "earned" 25% of all income in the USA in 2009. That only left 75% for 2010 for everybody else, and 75% of that 75% for 2011, and so on. These parasites control the multinational corporations as pawns in their monopoly game, and they are sucking the life blood out of our economies. We are at war with internal enemies we don't understand and only vaguely glimpse, so they win most of the battles. These people like things just the way they are, because they know any and every truly progressive change is going act to cut their power and influence, and they stoop to anything to prevent change, including both using the legal system to their advantage and violent crime if nothing else works. Many people vaguely sense that something is wrong, but it's hard to put a finger on it. I personally was oblivious until I started working on electric transport myself. The documentary "Who Killed the Electric Car?" plus the fact that I couldn't find the car-size Ni-MH batteries those cars used available for any price (much less a fair price), and
subsequently finding Chevron Oil has acquired 125 metal hydride patents and killed the technology as far as transport is concerned, started to open my eyes.
   Their supporters all the way down the line likewise instinctively realize that their jobs, businesses and lives will be changed if they embrace new things, and react against them. Perhaps typical of this gut reaction is one local business owner - himself an inventor and in some areas more progressive than most - to whom I showed the sodium sulfate salt packages for quadrupling the life of lead-acid batteries, who said "I sell ten batteries a week. Why would I want this stuff on my shelf?" (Remember that next time you have to buy a car battery from a smiling dealer!) The factories are all controlled by the gangster "elite", and every automotive shop I tried sells batteries. No one in the whole transportation industry wants them to last longer, or wants better types of economical batteries to come on the market -- only every consumer.

   The other culprit was that there is no department, branch or agency in government to oversee what is going on, correct problems or recommend legislation to correct problems, and to work towards desired change, to foster even the government's own stated long term goals and the future on behalf of citizens and the public good. All the departments are concerned with today and the immediate future. The effects of the errors in our governing systems aren't being monitored. No recommendations are made to the legislatures for fixing loopholes that hinder progress, and that allow power vacuums to occur and to be filled by greedy and power hungry manipulators.
   For example, the patent system has been routinely used to kill priceless new transportation technologies for over a century. And inventors usually have no means of getting paid for their work, either while they work or after successfully creating a new product or technology. (Less than one percent ever get paid -- and even more rarely because they took out a patent.) The whole idea of the patent publication system was so that inventors would get paid for their successful work, while their fine new ideas could spread rapidly to industry. It has never worked, and it has never been repaired so that it does. Blatant abuse of a dysfunctional system continues unabated,
the latest affront (AFAIK) being the murder-by-patent of a fine new lithium battery technology, years in the making, just weeks ago (again with Chevron as the "hit man").
   In addition, anything that seems too far off is dismissed by government as being "too visionary", however simple it may be, for example wave power: That's for the future -- it won't make money or solve problems starting in the next year or two. No action is initiated to traverse routes to proceed from the present to the desired future, even when that future is a stated goal of the government, and even though developing the routes may in fact cost pennies today for future rewards of millions. And often that desirable future could be much nearer term than anyone thought once the needed development effort is applied - look how fast everything changed in world war two, all simply for fear the enemy might get it first. We can progress in peacetime too. The answers are so often right in front of us.

   What is needed is a "Department of Progress" or "Department of the Future", charged with dealing with such matters, working towards the government's long term goals, fostering creative endeavors if they are deemed worthy, connecting product development with research and with industry, and making progressive corrections to the procedures and processes where things aren't working right.
   I first started with an idea for an "Inventors Society" which might allocate funds to worthy projects, funded by the government. I corresponded about this idea with my MP, which led to contacts with the minister of finance, in which I conceived and developed the idea of the Department of Progress. I elaborated on it and the reasons behind it. Finally I realized that, in the absence of the very department I was now proposing, there was no one in the chain of command between me and the prime minister of Canada, so I wrote to him with a by now fairly well developed idea, and urged him to appoint an able minister to start the department. Unfortunately I have had only an acknowledgment that the package was received. If he doesn't wish to be the hero, I may have to send another package to his successor and even the successor after that until someone in power understands that there's a gaping void in his administration - in every administration today - that needs to be filled.

   Along my way, I also digressed into a number of shorter, seemingly simple projects, and also various techniques that could be employed in installing the motor systems in cars.

   I found that by making a little fuse assembly one could plug terminal lugs the same size as fuse lugs into the car's fuse box and re-route wiring as needed without changing any wiring in the car. In particular, the turn signals could be run off "ACC" instead of only with the engine turned on. That, probably an electric vacuum assist for the brakes, and an electric heater-defroster, seemed to be the main issues for running my own car with the engine shut off.



   I made a 36 volt electric car heater for the 36 volt motor/battery system after seeing just one overpriced unit on the web earlier. When I had finished it, I searched again and found a number of choices for lower prices. Perhaps I was wasting my time on that one!



   I had an idea along the way little related to energy: to create a web site where Canadians could initiate informal 'polls' or 'referendums'. Currently governments claim whatever they do is what the public wants, but there is no way to find out what the public does want. Often those screaming loudest about something are taken or mistaken to represent public opinion. Sometimes these are just a small, highly opinionated minority. Other times they are even organized vested interests with their own agenda, as when big oil puts out one-sided propaganda and claim it comes from 'a concerned citizens group'. Either way, they should not get their way by claiming to be the voice of public opinion, and if the more polite silent majority are allowed to speak, many things might start turning out differently.

   On DIY_EV_CARS@yahoogroups.com discussion list (which last summer mysteriously vanished without a trace -- could be innocent, but it fits an all-too-familiar pattern for most anything to do with non-petroleum transport), I learned that people were renewing lead-acid batteries with alum. That sounded weird, but I looked up alum and found it was a salt of sulfuric acid, which started to make some sort of sense. I started experimenting, my battery work giving me some knowledge of what I was doing, and I finally realized that sodium sulfate was the 'correct' salt, diluting in the acid to sodium bisulfate, a sort of acidic baking soda to perpetually keep the battery clean. Alum, the first sulfate salt long ago found to work, also contains aluminum or magnesium - 'impurities', and may contain potassium instead of or in addition to sodium.
   Battery renewal was possible and good, but simply adding the salt to a new or near new battery will make it last 3 or 4 times as long. I worked out how much to put in. Of my four new or not-very-old batteries that I added sodium sulfate to well over a year ago now, all are still working great.
   With sodium sulfate in my car battery, I've turned the idle down very low - the battery is only charging when the car is moving. Letting it run down a bit while waiting at red lights is now no problem. This simple step has reduced my car's gasoline consumption about 10%! (in mostly city driving.)
   As mentioned, I found out that the battery companies have known all about it for many decades, but they don't add it to consumer batteries - the last thing they want is to sell better batteries. I found some clever patents for impregnating it into electrode separator sheets for the highest priced batteries. From there it dissolves into the acid when the battery is filled. That way, no one even in the battery factory knows it's there except a trusted few.


Sodium sulfate additive: battery load/cycling test

   And somebody mentioned nanocrystalline ceramic cores for motor coils, and that they could be better than iron cores. With little knowledge I attempted to create some in the kiln. I thought I could improve on the most promising results and find what worked and what didn't... but I got no decent results whatsoever to improve on. (I think a reducing or oxygen free atmosphere is probably required, which can't be attained in an ordinary electric kiln.) I also tried to buy nanocrystalline alloy core strips, but e-mails went unanswered and I got no farther at that than trying to make the ceramics myself.
   But there was an unexpected positive result: in trying out nanocrystalline glazes, I discovered I could make leaded borosilicate glass with transparent nanocrystalline titanium dioxide, which could be very useful for a reflective rear electrode for dye sensitized solar cells, improving their efficiency by altering and reflecting back the light that got through the dye, so it was more likely to be absorbed on the way back. This could improve the efficiency of DSSC solar cells by at least 25%. I haven't given up on this idea, but I haven't had any time to develop better solar cells as they deserve to be developed, either... or even to make a whole cell.


Nanocrystalline borosilicate glaze DSSC reflector electrode prototypes

   I also saw videos of pulsejets in fall of 2010, and I thought a small one might make a great steel plate cutter that needed only propane and no oxygen. I started to put one together but only got it half done before the motor improvements claimed priority over most everything else.

   Testing the motors this fall has led to considerable improvements, including a composite stator of polypropylene mat and epoxy resin to replace the steel ones. The steel turned out to be causing considerable electromagnetic drag.


Motor with the new PP-epoxy stator (painted red) mounted on the test stand

   Then in the last days of 2010, I got a link to www.micrometals.com, who it seemed actually made various cores of various powders and especially iron powder. I downloaded their catalog and found a cheap, off-the-shelf core of exactly the EH motor's outer core dimensions. It was also a toroid with a hollow center, but I started to realize from one of the tests that its shape seemed beneficial as to flux lines and would have lower losses than a solid cylinder, and that such an iron powder core should provide great efficiency perhaps equivalent to what I expected to get from the nanocrystalline ceramics. Nine are currently en-route to me to use in a test motor.


New toroidal iron-powder coil cores - same 2" O.D. x 1" tall size.

   I started the motor project (as well as the controller and the battery projects) doing most everything either wrong or some needlessly hard way. After 3 years of ongoing improvements, I've changed almost everything about the motor except the supermagnets and the basic magnet rotor configuration, having nine coils, and the basic axial flux layout with its resulting 'pancake' motor shape. Copper and core loss figures now lead me to expect 95% peak motor efficiency with the latest exciting changes. It'll be a very hard motor to beat - and it can be made at home!



Electric Hubcap System & Motor Building Workshops

A Motor problem and solution

   I hadn't bothered to ground the frames of the motors - after all, they're only 36 volts. No one would be hurt if there was a short to the frame, and once the motor was mounted in a car, it would be grounded through the chassis.
   It turned out one motor had a short to the frame through a coil mounting bolt, the coil having been mounted a little off-center, which there was nothing to prevent. A piece of the ground wire of the magnet sensors was bare (hey, it's ground anyway, right?) and it touched the frame of the in-shop motor mount. Zap! the coil voltage on the frame, coming along the skinny ground line into the circuit board, fried the motor controller. Absolutely nothing fries solid state circuits like a voltage appearing on the 0 volts reference ground line feeding virtually every component.
   Now I'm specifying insulating sleeving over all the coil mounting bolts for all motors, I ground the shop's mounting frame to the controller box... and I may put a small resistor on the circuit board feeding both the ground and the 12 volt supply wires to the motor sensors, to eliminate or at least limit damage from a similar event in the future. And if a motor isn't to be grounded through a chassis, a fourth wire on the motor cable, grounding the motor to the controller chassis, is probably in order.

   Later it developed that the stator plates are to be plastic. It might be a little tough grounding that, but no shorts will conduct through it, either!

The new A3938 motor controller

   In fixing the motor controller, it appeared that the chips on the PC board were probably toast, along with a few of the mosfets. Since I had just finished designing and having made the new A3938 controller PCB boards, I decided that rather than make another MC33033 board, I'd kill two ostriches with one brick and get the new design working.  But one ostrich tossed the brick back. The tiny chip blew and I set it aside.
   On the 30th, Tristan with his young eyes and fine soldering skills came over and replaced the minuscule chip while I worked on motors. Some problems were found - a mistake in the schematic, and a backwards plug socket. After some troubleshooting I figured these out and got the new controller working. It worked great until a mosfet blew for no apparent reason. After repairs the next day, it worked again. I saw currents up to 35 amps during acceleration. Then I tried to turn the control up quickly from a stop, and transistors blew again, along with the chip. The current limiting should have prevented destructive currents, but evidently there's something odd at work and it needs to be set to a lower level.

   The A3938's system of operation is interesting. To compare: the MC33033 essentially provides a constant voltage dependent on the input control setting. This means one must hold the 'gas pedal' farther and farther down for faster and faster speeds. This is the depth of push to go that speed, not the depth for a certain level of power. But if the load is too high, the current limiting cuts in. This simply cuts off the rest of each PWM cycle when a fixed maximum current is reached. It doesn't drive the motor again until the normal start of the next cycle, leaving however long an "off" period until then. The switchover between voltage PWM and the current limiting causes some weird effects, like sudden reduction of power. (I discovered this as a cause of low motor starting torque and low power that was going unnoticed in the torque converter tests: all the converters were being driven by a motor whose power input was being stingily "rationed" by the controller. It puzzled me that a few times in more recent tests, converters seemed to suddenly "take off" and have surprising torque -- but only once the wheel was spinning, so it didn't help get the car moving.)
   The A3938 works on current limiting alone. A variable maximum current is set by the amount of press on the 'gas pedal'. When the motor current hits this value, the controller shuts off, but only for a short, fixed period. Then it turns on again. The "PWM frequency" varies in accordance with the current requirements, dropping for higher currents as the "on" period lengthens.
   Thus the average current can be much higher - almost to the maximum limit, but without ever exceeding that limit (if it's adjusted right). It seems like a better system.
   It's more like the gas pedal: the depth of push regulates the flow of current - of gasoline or of electrons. With the current being regulated instead of the average voltage, the motor just keeps on gradually accelerating until the maximum RPM for the current or for the total battery voltage is reached.

Magnetic Friction!

   In trying to determine motor performance I found there was a big discrepancy between the DC and AC figures. At 500 RPM, 36 volts times 3.0 amps said the motor was using 108 watts of power... somewhere... with no load. But 6.8 amps squared times .064 ohms is only 6.0 watts of copper losses. Somewhere, it appeared 102 watts was being eaten. I found out where.
   I already knew the motors stopped pretty quickly if spun by hand, but not really why. After running some tests for a while, I discovered that although the coils seemed little warmed, the steel stator plate holding them was very warm - even "hot"! Most of that 102 watts of no-load energy was going into heating the stator plate!
   So I took an empty stator (no coils but with bearing hub) and put it on an axle. I gave it a spin and it turned around about 40 times before coming to a stop.
   Then I put it on the axle with the magnet rotor, with a 1-1/2" gap. This time, it took only 5 or 6 turns to come to a stop with about the same starting force. In spite of the considerable separation, the magnets were slowing it that fast. I turned the empty stator plate over, which resulted in about a 2-1/4" gap, and it turned about 12 turns, certainly an improvement.

   The first thing to try would be to get the stator plate farther away and measure the idle currents again. Presumably they would drop significantly. I got longer bolts, dismounted the coils and replaced them with 3/8" of flat nylon spacers stuffed under them, adding that much gap to the stator plate. This was a considerable improvement in no-load power consumption, and the motor would spin 2 or 3 turns by hand. But now I was sure it should be better yet.

   Meanwhile, it occurred to me that the pancake motor (minimum 2-1/2" thick in my original theoretical concept) was already getting fatter, with 1/2"+ flux gap. If I had to add another, say 3/4" more space between the rotor and the stator plate, it might be more like a cake cake rising than a pancake! Was there a another approach?
   Both the wide flux gap and the need to move the stator plate back are a result of the large depth of field of the 1/2" thick supermagnets. Thinner magnets would have a  lesser depth of field. That would mean both a narrower flux gap and a reduced effect from the stator plate "only" 1-1/2" away from the magnets. As a bonus, the magnets and rotors would be *somewhat* less dangerous to handle. On the other hand, would this be enough magnetic force without going to 18 magnets?
   I decided to buy twelve 1" x 2" x 3/8" magnets and try out rotors made with that magnet thickness. The 3/8" thick size cost more than the 1/2", but the prices will come way down for whatever the chosen size is when lots of the motors are being made.
   Another technique would be to build the stator plates from transformer laminates where they cross the magnet sweep. Fine for a factory, but not for DIY.
   Finally, one could make them from an insulating substance, perhaps a high temperature plastic like bakelite or nylon. What would that do to the magnetic fields? I decided to try this instead and never took the pricey thinner supermagnets out of the shipping box.

Stator Losses and improvements

   That seemed to help. I calculated efficiencies up to 68%. It was still way below my expectations. But now the plate didn't get very warm, nor did anything else, so it seemed there were no more 'outstanding' problem areas.
   The 'best compromise' stator plate gap, between perhaps 1/4" and 3/4", remained to be worked out. More 'radical' solutions were conceivable: the thinner supermagnets will give less depth of field. But that might mean less power from the motor. They might well be better matched by shorter coils, eg 3/4" thick instead of 1". Then a gap to get the stator back to 1" distance might be required!
   Or the opposite: thicker coils might be employed. This would move the stator farther away, but the magnetized coil iron would on average be farther from the supermagnets, again reducing the power.

   But perhaps that's all thinking inside a box? Another way to change everything, and just by making a mold, would be to make a stator plate out of entirely non-conductive material. This would be the ultimate of low-loss materials.

Composite plastic stator with no steel backing plate

   However! With this thought, it occurred to me that the very first Electric Hubcap stator prototype had the coils cast in polyester resin, a polyester donut ring like the windplant generators it was originally derived from. Although the coils had minimal steel laminates, I could run it long enough to measure some of its performance characteristics to see whether the idea was great, not worth trying, or a 'maybe'. During the night of the 18th-19th, I came up with a way to set it up: Turn a wooden piece to make a center for the stator "donut", with a hole in the middle for a bearing hub, and turn a bearing hub to match. This occupied much of the 19th, but finally it was ready. I gave it a spin by hand and it went a dozen turns (instead of one with the original setup or two with the 3/8" spacers under the coils). I used the hall effect sensors Tristan had wired up, removed from their mountings and screwed to the stator. It wasn't perfect, but it ran. Boy, did it run! I didn't need any more tests to tell me that it was a vast improvement, but I took some readings anyway. The flux gap was perhaps 17mm, a bit wider than in the earlier tests before spacing out the plate (16mm), and considerably wider than the 13mm in the tests with the spaced plate. Reducing it would have required more wood lathe turning, and with a new lathe setup.
   Even allowing lots of room of error in comparing different varieties of apples in the various tests with somewhat different coils, wiring, and flux gaps, the figures seemed conclusive: It took about 22 watts from the DC supply to turn at a steady 500 RPM instead of the original 108, or 72 with the spacers. The AC motor current (total) was 4.1 amps instead of 9 or 10 amps. At 1000 RPM, it needed 52 watts instead of 259 (original) or 206 (spacers) and again the AC current was way down - 6 amps instead of 13 or more.

   Here I thought again of the polypropylene cloth with resin that made such a nice, strong cover for the rotor magnets - whether used with polyester or epoxy resin. (It would take a lot of resin, and epoxy is pricey.) As a bonus, the stator could be sized and shaped exactly to suit. I wasn't quite sure about attaching the bearing hub. It could be done by welding a flat steel plate to the hub, say 5" diameter, that the plastic plate is bolted to. But it would be nice to figure out a way that would eliminate welding to the hub piece. That would require that the center of the plastic piece be strong enough to simply press in the shaped hub... or to simply press the bearing rings into.
   Another nice touch might be if the stator plate was sized to cover everything and fit in or on the outer shell-ring, hence forming the outer end wall of the motor. I went with this idea, and with molding the bearing hub into the plastic. as part of making the stator.

   This led me on another quest for polypropylene cloth on the web. Searches showed that it seemed to be used for everything and present everywhere (including those re-usable cloth shopping bags) - except for sale as cloth. A zillion third-party advertising sites with no products of their own begging you to take their useless links to nowhere utterly dominated the results. (Simply saying ' -alibaba ' in the search reduced the number of supposed "hits" by over 500.) Finally I found it's used as landscaping fabric and could be had as cloth or nonwoven mat for low cost at any local hardware or gardening store. The only problem was that the material isn't identified on the package. The "landscaping fabric" could have been made of fibreglass, PVC, rayon, or hemp for all I knew or anyone in the store could tell me. It should be identified 'PP' for recycling purposes if nothing else.



Mold

Laying sheets of PP mat in the mold

Pressing the stator - PP mat and epoxy resin, with bearing hub in the center, in the mold with plywood above and below.
I did the first ten layers of mat and the lower half of the hub, then the second 20 layers and the top part of the hub later.
Later I finished by slopping on the rest of the thickening resin (kept in the freezer 2 days) and spray painting with engine enamel.



   I found that it took 30 layers of polypropylene "landscaping" mat to get a 3/8" thick stator plate. When I got it molded, it seemed considerably less stiff than I had hoped. It seemed it would need 'ribs' at right angles to impart some stiffness to it. But when I checked it again the next day, it had become very stiff and seemed okay to use as-is. Perhaps a good idea anyway, but I'll skip it for now - it's not vital. It's a prototype and the design may change, which might make any new jig immediately obsolete.

   The amazingly light new stator, after discounting the cast metal bearing hub which is molded into it, replaces the 6 pound disk brake rotor with under 2 pounds of plastic composite. That should bring total motor weight down to about 28 pounds.


Stator before painting

   Now it's tempting to think of shaving off another 4 by changing the magnet rotor to plastic as well, but the rotor's inertia and solidity are needed for the torque converter and probably for other motor applications.
   The motor case outer barrel (whose 1.3 Kg isn't accounted for in the current 32 pounds measure), could be lightened by a switch from PVC to lightweight PP/Epoxy, saving at least a pound - even two by making the wall thinner.

   On the 30th I got the holes drilled and assembled the motor with the new stator and ran it with the A3938 motor controller, also completed that day. Somehow, the bare bones motor weighed only 25 pounds.


New motor runs great!

   I drilled the holes with the CNC drill program. I painted the rotor with high temperature motor spray enamel. I'll also be adding washers and a layer of tarpaper between the coils and the plastic stator plate.

   It seems excellent, so I'm now putting together another one. In the middle of cutting out all those PP mat circles, it occurred to me that if I 'marked' the circles with an xacto knife instead of a pencil, it would save a lot of snipping with the scissors. I zipped off another 18 circles that way in under 15 minutes. By the nature of molded composites, I can probably use the (PP) bag of PP mat scraps for another stator, too.
   I've also turned the bearing hub, so It's ready to mold. The first bearing hub wasn't quite straight though (I'll even the gap by adding washers under each coil to suit), so I'm trying to come up with a good jig to hold it perfectly centered and straight.

Powder Cores

   I'd been wanting to find a way to avoid having to piece together those cores of nail gun finishing nail strips - they add a serious amount of labour to making a motor.

   In talking in a discussion group I learned a bit more about powder coil cores, and there was a link to micrometals.com, who appears to be (or to own) the chief manufacturer of ready-made cores of various compositions. I downloaded a catalog, and discovered that although there was no cylindrical powder core around the EH's 2" diameter and 1" length, there were toroidal cores of exactly that size, "T200". I decided that a big hole in the middle of an otherwise perfect core shouldn't stop me from trying them out.
   I phoned for technical assistance, and it sounded like the basic "26" iron powder would be, though perhaps not the ultimate in permeability, high saturation and low loss, a good compromise - quite serviceable and economical. (And doubtless better than the nail strips.) I was asked if I wanted a couple of samples of 26 and a couple of 52, which had a little higher permeability. I said I needed 9 to make a motor. He said he'd send 9 "T200-26B" cores. It turned out these are only $1.88 in single lot pricing. I wouldn't be ordering less than 50's (only 5 motors worth), I'm sure.

   If these work okay - and there seems no reason to doubt that they will - the coils will be far easier to make. I envision mounting the core on the coil winder with 3" diameter non-stick plastic side 'washers' to hold the wires in place, mixing some epoxy and winding all nine coils in well under a couple of hours, painting the epoxy onto each layer of wire as it's wound. I might wrap a layer of polypropylene around the outside. The entire coil would be dismounted from the winder chuck complete with the center bolt that holds it together and left for the epoxy to set, and there'd be nine sets of winders to do all the coils with the same batch of epoxy.
   An additional, and large, benefit is that the powder cores appear to be very efficient (the manufacturer's specs are in the catalog), where the nail strip cores were "good enough" but had significant losses. The composite plastic stator and these cores together with the other design advantages should boost the EH motors into the singularly ultra-efficient (95%+) category I didn't previously dare hope to achieve.

   There's one more twist: Considering the magnetic interactions, the toroid form of coil core may actually perform better than a solid cylinder. The copper will magnetize the core in an area more concentrated towards the sides - closer to where the lines of force from the magnets actually are as they approach or leave the coil's vicinity. When the magnet pole is right over a coil, that coil is turned off anyway. And, the hollow core has 40% less iron to have iron losses from. As long as the 60% iron that remains doesn't saturate, that's a further considerable efficiency boost.

   As an additional bonus, the toroidal cores will weigh only 220 grams each where the laminates weigh 265 plus whatever the motor varnish adds. The motor will weigh about another pound less. Bare bones, the motor to drive a car (now just 24 pounds) will weigh about the same as a typical mounted 13" car tire.

   The one thing I think I need to watch is that evidently the iron powder cores are only good for 70ºC, so there's not a lot of point using high temperature motor varnish to allow 150º or more. (In fact, baking varnished coils in an oven at 225ºC with 70º rated cores seems like a very bad idea.) I will use a high temperature rated epoxy, though. If temperature does prove to be a problem, they have the same toroid cores made from higher temperature spec materials. A motor that can go up to 175ºC can run at higher power longer than one that only goes to 70. On the other hand, the ultra high efficiencies will make this a very cool running motor, so 70º parts may well be just fine. And with the toroid shape, cooling air gets to the inside of the coil as well as around the outside. The polypropylene stator's epoxy is also only good for 70ºC, but the coils will be mounted slightly off it with tarpaper and washers, which again allows cooling air under the coils.

   $1.88 each, super high efficiency and performance! ...and all that work I did earlier trying to make nanocrystalline ceramic cores! The professionally made iron powder cores have very low losses per cc at EH motor frequencies, and the toroid form would appear to reduce cc's and further increase its efficacy, probably accomplishing as much as I had hoped to get with the ceramics. And they'll also accomplish the other vitally important part of the job: making the coils, and hence the motors, way easier to make.

Motor testing and performance categorization

    I did considerable measuring of Electric Hubcap motor parameters in late November and the first part of December, and through the wonders of math and equations came up with some seemingly bizarre and highly suspect figures for efficiency and power (which also resulted in the plastic stator mentioned above). I've heard brushless motors can have "over 90% efficiency" and had high expectations, but I wasn't seeing anything like that for the EH motors. Here I thought I had created a great motor, and I was embarrassed to find I was calculating efficiency figures like 55% and even 45%.
   My test setup was primitive, and I only tried a couple of (unknown) loads at few RPMs, usually 500, 1000 and 1500. The "load tests" were done by pressing my foot against the spinning rotor and increasing the 'throttle' until I read some considerable DC amps coming from the battery to maintain roughly the same RPM. Other than being unsteady, providing no torque readings (to compare with the theoretical), and the smell of burning rubber from my running shoe, it seemed to work okay. It was expedient but not elegant.

   But upon reading more 'stuff' on the web, I started realizing some things that were good to keep in mind. The first was that efficiency was so high only under the best conditions, and that motor manufacturers evidently usually rated their motors at the peak point of efficiency: at the ideal high RPM and light load. (Why wouldn't they?) I found a document and graph that showed more real life conditions:



Sample: performance graph of a small commercial DC PM motor.
Efficiency is highest, 75%, at very low torque, eg 10% of full.
At the highest output power - at about 1/2 speed, 1/2 torque, 1/2 current - efficiency is 47%.
(Source: http://www.micromo.com/n390432/n.html)



Later I found this performance graph,
probably the most similar graph to what's now expected from the EH motors except for RPM range.
  It's a 36 V brushless DC PM model airplane motor (with just .06 ohms motor resistance!).
At very light output power the efficiency is 92%.
At very high power where push tries to shove, it's just 50%:
almost 11000 watts in, under 5500 watts out.
EH operation will be restricted to the left half of the graph - under 5000 watts in -
with 95% peak efficiency, dropping to 75% at full load.

   After seeing the graph and accompanying stats, the figures I was getting began to look more understandable - and reasonable.

   At no load, a motor's efficiency is de facto 0% - whatever energy is used is used just to keep the motor spinning. On the EH motors, the current at no load with a steel stator was high enough to prevent approaching great efficiency figures at light loads. At high loads, the same extra current becomes part of a heavier I^2 loss figure, causing considerable extra copper resistance losses. The EH with the new PP/epoxy plastic stator is much more satisfactory, and with the iron powder coil cores it will attain the highest possible efficiency standards.

   Another problem was that "power in" is battery volts times battery amps, and the DC current clamp was wildly off at low currents and still reading 15% high at 10 amps, above which point I have to pull the Fluke DVM amp meter out of the circuit before it fries (which is why I bought the clamp in the first place)
, so I have no other comparison. Also someone on a list said that with PWM you had to use meters with a high bandwidth - that low frequency meters could read "wildly" off - possibly the problem with the DC (& AC) current clamp, although it also reads higher on AC currents compared to the other clamp and to the Fluke meter on amps. At first I was also just assuming the battery voltage was 36 volts, but at high currents it was often under 33 volts - 10% lower. Power out is calculated from the motor AC amps times the torque constant times the RPM: it doesn't care about the DC supply. If the figure for power in is, eg, 20% high, and the output power figure is right, then an efficiency of 95% reads as 75%, and around 55% comes out as being around 45%. The error is further compounded if the AC and back EMF readings are on the low side. Without a motor load that will tell how much torque the motor is actually producing, I have no good check that  the Power Out side readings are consistent with the Power In side.

   I have an idea for a magnetic brake to give me this load: take an EH magnet rotor and get it near to the rotor end of the motor, in line with it. Put a 2 meter long stick on it, sticking out 1m each side - balanced. The magnets will attract to the back of the rotor and will try to spin with it, but this rotor can't spin because of the sticks. Put a weigh scale under one end of the stick. Each 102 grams the scale weighs is one newton-meter of torque the motor is putting on the rotor. An aluminum plate 'coupling' near the load rotor magnets (spinning with the motor rotor) may work best, and will drag but not grab to the magnets. Adjusting the gap adjusts the loading. Having the essential idea and taking the time to figure out all the construction details and make it may however prove to be to different things. Without even looking it up, I doubt I could afford a ready-made test brake.
   Another person has suggested a simpler "prony brake", simply an adjustable circular clamp around the rotor with an arm sticking out, with a scale on the arm as above. The work done heats the brake, so it may smell as bad as the running shoes.

   While having no figures for other motors of similar size and construction to compare with, and figures of suspect accuracy for the EH motors, after some weeks of all these frustrating measurements, calculator banging, and frying a motor controller, I had:

a) actually improved the motor's efficiency by changing to plastic stators, and made it more reliable by putting sheathing on the coil bolts. (Now the sheathing will be made redundant by the toroid coil cores.)
b) found that now nothing was getting even very warm, indicating there were no more exceptional losses (and certainly saying nothing bad about the magnets as cooling fan blades - you can feel the air they blow).
c) a bunch of actual performance figures and calculations, done in the first half of the month. These were to follow on next, but the change to the plastic stator, not completed until after Christmas, makes most of them obsolete, and the new ones will soon be obsoleted by the iron powder cores. Hopefully I'll have some excellent figures next month.

Performance measurements and calculations

   One calculation isn't rendered obsolete by the changes: The coil resistance between motor terminals is a crucial parameter and measurement, and I measured it with a power adapter to supply a small power across the coils. The measurements were plugged into R = E / I: .070 V / 1.10 A = .064 ohms... virtually the same as the calculated resistance based on tables of the resistance of #14 copper wire, the coil connections, and the lengths of wire in the coils. Evidently one simply measures the resistance between any two of the three terminals to get "Rm". They all read about the same, but what happens with the third leg when you only measure across the other two, is unspecified. Should the resistance be multiplied by the square root of two like the current... or should it be divided by the root of two because the legs are in some sense in parallel? I think and infer that one just uses the actual reading, but it would be good to not be left guessing about these things!
   I also think the EH motor is best here: lower than others of the same power and size because of the round 'donut' coil windings, magnetizing the most nearby iron, in closest proximity to the rotor magnets, with the least length of copper wire. The resistance would have been .072 ohms for square coils, and more on standard coils with "overhangs". The copper losses - the biggest efficiency loss - are directly proportional to the resistance -- motor resistance times the square of the current.

95% Motor Efficiency

   The change to the plastic stator, and the coming change to iron powder cores, pretty much blots out the meaning of most of the figures measured earlier, except insofar as those figures actually prompted those improvements. Supermagnet motors often peak at over 90% efficient and axial flux is the best, and we can calculate or estimate the following factors that will make the Electric Hubcap's newest design better than others:

* low loss iron-powder toroid cores that will also put the maximum magnetic flux right where it's needed.
   Eddy current losses in powder cores are trivial, only the hysteresis losses being significant.
   Iron losses, based on the manufacturer's printed specs, will be only about 10 watts at 1500 RPM.

* the new composite plastic stator eliminates most of the magnet rotor drag caused by metal parts
   Only internal nuts and bolts, and more distant mounting hardware, remain.

* the wide (and adjustable) magnetic flux gap is known for high efficiency (the axial flux advantage).
   It also prevents gradual demagnetization of the supermagnets, making the motors 'everlasting'.

* The lowest coil resistances attainable for the voltage and size, .064 ohms.
   At low loads, the copper losses are again only 15-20 watts at 1500 RPM.

   The total no-load losses at 1500 RPM will thus be only 10 or 20 watts (Cu) plus 10 watts (Fe) plus bearing friction plus air friction, just a few tens of watts. By contrast, they rose to several 100's of watts at that RPM with the steel stator and nail gun strip coil cores.

   15 or 20% load is likely to be the most efficient range. 15% load is 700 watts and losses should be only about 35 watts, so it seems we can expect 95% peak efficiency. At my intended maximum watts, about 4600, 75% might still be attained, where other motors show graphs dipping to 50% or less at their maximums. This is partly the built-in efficiency, and partly that I'm limiting the maximum power to the best operating area so as not to generate too much heat or drain batteries needlessly quickly. (The second graph above ("E-Flite") would also be minimum 75% efficient at full load as shown if they limited full load to 150 amps of current.)



Mechanical Torque Converter (MTC) Project

More Theory

   The oscillating masses torque converter is something like a rotary version of a hammer striking a nail. There is a longer period where kinetic energy is gained, the down swing, and then a shorter period where that gained energy moves the load: hitting the nail.
   In the torque converter, the energy is imparted in such as way as to not stop the motor each time but only to slow it a bit so the rotary motion can continue - a 'glancing blow' or slip, which may either be a sudden 'impact', or a relatively rapid but non-instant change.
   Like simply pressing a hammer against a nail, the average torque of the motor isn't enough, eg, to start a car rolling up a hill. But that average is divided into longer periods of little or no torque, and shorter periods where that accumulated energy is spent to produce much higher torque. The high torque hit starts the car moving, and each subsequent hit gets it moving a bit faster.

   In this, one must consider the periods and magnitudes of the forces. If the unit were, for example, to have a 20 second period to build up momentum and then a 2 second period where that built up force is applied to turn the wheel, no one would question that the car would start moving, and indeed it might well be doing 'street speeds' after the two seconds was up. But to store up this much energy the unit might have to do something like start a massive flywheel spinning and then let out a clutch to engage it.
   If on the other hand the unit had just 20 microseconds to build up speed and then 2 microseconds to dissipate it, perhaps all that would be noticed would be a high pitched vibration. The minuscule stored force would be averaged out with the same effect as a steady force, perhaps too little to start the car rolling. On the other hand, no heavy parts would be needed to hold the minute, short period energies generated within such a system.

   Something in between is needed. The smaller the forces and the shorter their time periods, the tighter everything has to fit. At some point, even the rubber tire will absorb the tiny torque hits and present to the road the average, too small, force. We need forces of sufficient magnitude and length of period that the car actually budges with each hit of torque, but not so large and slow that excessive masses have to be employed, with the car moving in perceptible lurches.

Hidden torque converter testing problems

   I've become aware that there are at least a couple of problems in the torque converter tests that aren't the fault of the torque converter.

   The first one is the coupling between the torque converter output rotor and the car wheel. A while back, I made a plate for the car wheel with four 'extra' holes and some tapered pins for the output rotor that would fit into them. But when I changed from the smaller steel drum to the 12" aluminum drum, those pins didn't fit and I used four 12mm bolts. They fit in the holes, but there's a lot of loose play. I can see it in the videos I've taken: The output rotor bounces back and forth relative to the car wheel, when they should move as one. The escapements must 'hit' the output rotor, but then the bolts hit the edges of holes in the coupling to the wheel, then bounce back. Larger, longer period torque hits than I hope to employ would be required to overcome all that slack. (The new "5 Star" escapement converter design has the strongest, longest period escapement hits so far.)
   On the 23rd, I took the plate off the car wheel and set it on the converter. On looking it over, I realized it could easily be made to work. I had in fact drilled 12mm holes that just fit 12mm bolts. The original sloppy, oversize holes were coupled to the converter. If I simply bolted the plate tightly to the wheel using the oversize holes instead of the small ones, the small holes would be left for the pins to the converter, and the play would be virtually eliminated.

   The second problem I discovered on the last test, on the 15th. On a whim, I took out the AC current probe and the new DC current probe to check the motor currents. (...along with all the heavy things I have to lug out and set up for each test.)
   Much to my surprise, the highest the AC motor current readings got was 40 amps, and the DC supply 25. I've seen 80 amps AC on the same motor with my original motor controller.  (Multiply reading by square root of 2 for actual 3 phase motor current: 40->57 & 80->113 amps.) I got another battery and boosted the voltage from the usual 36 volts to 48 - a bit risky with 60 volt transistors and having seen 12 volt spikes coming from the coils on the oscilloscope. Even more to my surprise, the maximum AC reading was still just 40 amps, and the unit had no more power!
   The motor controller was set up to limit the current to '150 amps'. But that's peak amps. It seems that the average amps is far lower. 40 amps instead of 80 means 1/2 the torque, and 1/2 the torque means lower speed with the same load. I'd noticed it didn't seem so energetic before with various torque converters for a load, but with all I was trying to do whenever I ran a test, somehow it never occurred to me to check the motor performance out. Thus my EH car motor, ever since I put that controller in, has been distinctly limited in maximum power. Since that was before I started on the torque converter project, all the torque converter tests from day one have been done with unduly limited motor power!

   It's funny how things like these can affect everything for so long. I should of course have tested the new motor controller's operating parameters when it was first working, and I should have looked into some sort of tighter coupling to the car wheel. My only excuses are that that's what can come of having too many things on the go at once!...
and that I didn't have a good meter for high DC currents.
   Now I'm putting together the new A3938 motor controllers. It seems like a better chip, and the low current problem is likely to disappear. And I'll probably head out on a new quest for a better flexible coupling system between the converter drum and the car wheel - I still don't see the four pins as being optimal. Some sort of spline system might be better.

   The new version of the converter is surely still the best design... but none of them had really fair trials.

The "Five Star Clock Escapement" Torque Converter

   In the escapement torque converter, the motor rotor starts to spin, spinning the escapements at its rim along with it.


EH Motor Rotor with 5 Escapements

   But owing to the points on the escapements meshing with the teeth on the output rotor, they have to oscillate back and forth in order to spin.


Output Rotor (behind) with 25 'Teeth'
'5 Star' version finished December 13th


At low speeds they move freely and easily, but as speed increases they have exponentially more and more resistance to being oscillated. They are continually pushed in one direction or the other as they spin, a bit like pendulums on steroids.

   All five escapements now operate at the same time: they move alike, in unison along the 25 rotor teeth. In the previous version, six escapements around the rim all operated 60 degrees out of phase with each other. The new way gives a single quintuple strength torque pulse (the total 665 escapement grams) just 1/6 as often as the previous six single strength pulses (133 g each) - a longer period with correspondingly stronger hits: the better to couple the peak force to the tire and road with less averaging out of the torque.
   Where one or the other end point of the escapement hits the next fan tooth is where their direction of swing, the oscillation, suddenly reverses. For most of the oscillation, nothing happens and the motor spins freely, but at this point five hammers hit the nail. The average torque of the output rotor is the same as the motor's torque, but it's divided into the free swings with no torque, and the sudden "5 hammer" torque "bumps", hopefully at least ten times stronger than that 'average'. It's a better approach to "gearing down" the motor than gears, because it's continuously variable depending on torques and speeds.
   Ways to improve on this basic design now are:
 (a) make it a larger diameter. This is much the best improvement, giving all the forces more leverage. It also makes (c) or (d) (below) more feasible.
 (b) reduce the number of teeth. This reduces the number of hits per revolution. Changing from 25 to 20 teeth (if practical, still with 5 escapements) would mean 20% stronger hits, 20% less frequent. (I'm pretty sure reducing to 15 would be impractical. Then again...)
 (c) increase the number of escapements. (with the number of teeth being a multiple of the number of them so they're in phase.) This increases the mass of the hits.
 (d) increase the mass of each escapement. Wear to the unit from heavier hits needs to be considered.

   As the car starts moving with the repeated strong 'hits', the output rotor speed starts to catch up to the motor speed, and the torque hits (and noise) strength and frequency are gradually reduced as less torque is needed. One could add a spring to one or more of the escapements so that below a certain torque, the converter would lock to the motor speed.

The Test

   The "5 star" version was completed on the night of December 13th, and on the 14th it poured rain, so I worked on the new A3938 motor controller. On the 15th I did the test with the wheel jacked up that uncovered the motor controller problem. After finding the problem, I took everything apart in disgust. In retrospect, I should have tried to move the car on level pavement while it was all together - it might have traveled and made a good video.
   These tests, with carrying the heavy motor and batteries, and jacking up the car wheel, are quite hard on my back, so I don't do any more of them than I can help. Everything also takes time to set up and dismantle, and the weather has to be good.

Next Unit

   I see no need to make another unit for the next tests. however, in making one prototype one sees its deficiencies and improves the next unit. This is certainly no production design. Two things of note are that (a) it's noisy and (b) bits of aluminum dust are coming off it when the motor is spinning, indicating it'll wear out rather quickly.
   I think the noise level can be much reduced (a) by using thicker metal and (b) by gluing on thick rubbery sound deadening strips like car makers glue to body panels of cars - maybe a ring all the way around the outside. Anyway, the muffler doubtless was invented after the infernal combustion engine, not before.
   The wear is of course less than on the first escapement converter that only turned a few turns, and is improving. Probably a good way to fix it is, instead of having the thickness of a piece of aluminum hit the teeth of the ring, turn the escapement hook pieces vertical and have the full inch height of the hooks hit the full inch height of the teeth - or however tall they need to be to have low wear.

Motor Currents Torque Converter?


   At the same time I was putting together the 5-star, I also had been working on figuring out the performance of the EH motors. An interesting equation was for the 'stalled', 'locked', or otherwise stopped rotor if the full voltage of the batteries was applied unmodulated: I = V / Rm. 36V/.064ohms=562.5 amps. Obviously this would burn out the motor controller or the motor coils in short order. One could hear in my 2008 car move test that the motor was being driven beyond normal limits to get the car to move. It would have soon overheated. So, the powerful field of the supermagnets couldn't be used directly to turn the car wheel because making the electromagnets similarly strong to push them with, would burn out the controller or the motor. I've said it before: only a much oversized motor could supply the needed torque directly at 1 to 1 RPM with the wheel.

   But what if high power or even full power were applied for, say, 1/20th of a second, twice a second, and then it was turned off for the rest of the time, a 10% duty cycle? That would give momentary pulses of high torque just as the torque converter gives, without the danger of overheating the coils with a too-high average current. The mosfets are theoretically rated for 840 amps of momentary surge as well, or 1680 amps together. They might just handle a burst of over 500 amps then as well.
   In effect, the motor and controller could - possibly - act as an electrical torque converter. The strong field of the supermagnets could be matched for a moment by the coils with high current, and the car would start moving by pulses.

   The questions to be answered before building the circuitry for this and trying it out would be:

a) Roughly how much torque could the motor provide? That one seems easy to answer in theory: The torque constant worked out to .171 N-m/A. Multiply by 500 amps: 85.5 N-m.

b) Roughly how much torque is required to turn a car wheel for decent performance -- an actual figure in newton-meters or foot-pounds. The web is quite vague on this point. After all, it all depends on the car, how steep a hill it is or how fast you want to accelerate, etc. But it appears that at least a few hundred would be required.

c) How much current can the controller, and the motor, take, for brief blips of time, without burning out immediately... or much too soon?

d) Are there negative factors, such as magnetic saturation of the coil cores or simply insufficient force to be had in any case between the supermagnets and the "super" coils, that would reduce the force available to less that the mathematical answer to question (a)?

e) Losses increase with the square of the current, and the current is extreme. How significant would be the extra energy used to run the motor in short, high current pulses instead of at "steady state"?

   It would appear that the last three questions are redundant in light of the answers to the first two. One anticipates that it would give the same sort of lazy crawl across the pavement that I got in 2008. Even if all else is in order, it's likely that the very high currents and consequent very low motor efficiencies will drain the battery needlessly quickly. But if the mechanical torque converter (now that I've finally made one that should work well), should in fact be unnecessary, redundant, that would be worth knowing!

   The next day, it occurred to me that there was a simple, if somewhat unnerving, way to test the idea: Put a motor on the car wheel, put the car in neutral with the parking brake off (chocks to limit travel), bring out 3 batteries, and "spark" two leads at a time with the batteries. (gloves, safety glasses.) If one or more combinations made the car lurch forward or backwards with some energy, then the motor is physically capable of making it work. If it just went "gzzzt" and didn't move, or moved lethargically, it isn't. I recall doing something like this once trying to measure actual torque, but the wheel was jacked up and all I did then was break the hooks off my fish scale.
   A complication for these tests is that the motor that's set up for the car wheel is also set up for the torque converter, not to connect directly to the wheel. The other motors don't have the right mountings for the car wheel. In view of the dim prospects this will work, or that it's a good way to do it even if it does, I'll await some fortuitous circumstance.



Turquoise Battery Project

Disassembling 9V Battery Discloses Interesting Construction Techniques

   I had disassembled one multi-cell battery previously and found it simply contained a pile of individual button cells. On you-tube, someone took apart a 6 volt lantern battery and showed it was full of typical "AA" cells. However, this Mn-Zn dry cell was different.
   Around the outside was a sheet metal body. Inside that was a single formed tube of shrink-wrap plastic, isolating the body from the workings. Solid plastic end pieces did this job at the top and bottom, the top piece also having the metal terminals, and a thin insulated metal strip that ran down the side to connect the bottom cell to "-".
   Inside the shrink-wrap was a layer of wax to help seal the cells inside. Breaking the wax separated the individual cells.
   Each of the six cells was wrapped in its own little piece of shrinkwrap, with some more wax inside. They consisted of: a thin carbonized plastic sheet, so thin it was flexible, adhered to and covering a zinc sheet (the "-" electrode), then a piece of thick paper that wrapped up a manganese/carbon electrode briquette (the "+") on five sides out of six, leaving the top open except for the shrunken wrap, which covered much of it around the edges. But the middle was open with the briquette contacting the carbonized sheet of the next cell above.
   The bottom cell had no carbonized plastic sheet, and there was a double sheet between the top electrode and the terminals at the top to keep the electrode and electrolyte away from the "+" terminal's metal.

   This construction is somewhat along the "stack of cells" lines I've been moving towards to use for multi-cell batteries. Mine would be much larger in area.

   The MnO2 electrode briquettes were about 5-1/2 mm thick, the first ones that seem to verify my ideas that 3-9 mm is a good electrode thickness. Of course, these are low current cells, so for high current batteries with similarly thick electrodes, the KCl had better turn out to be a very good electrolyte!

   The resistance of the carbonized plastic sheets seemed to be around 12 ohms, but if the thin sheet was wrapped around the meter leeds, this dropped as low as 2 or 3 ohms, verifying that most of the resistance measured is due to the limited contact points between the meter leeds and the carbon.

   Later I wondered if similar carbon impregnated plastic could form one or both faces of a cell, sealed with MEK in the usual way to the ABS plastic sides like I've been using? With the faces held closed by an outer shell, that could make a nice multi dry cell layout!
   But I decided my sticky, pliable sheets (below) would be easier when I finally found a way to get the aluminumfoil off after baking them wrapped in foil.

Ni-MH Price Reductions - Car Battery Experiment?

   Perhaps a year ago I wrote that the price of Ni-MH batteries seemed to be being kept artificially high, 1000 $/KWH, which was probably hefty royalties to Chevron/Cobasys and their 125 acquired MH patents, imposed not to make money but to keep the prices of the batteries from becoming economical for electric transport.
   While many sources and sizes are still about that price, I now see 3 or 4 sources on the web selling Ni-MH "D" cells of 9 to 12 amp-hours capacity for $6-7 in tens quantities, or 600-700 $/KWH. (In fact, I also see NiMH "AA" cells for around 450 $/KWH - worth considering, though 4x more soldering to make the same capacity pack - and I don't see them with solder tabs for that sort of price. Theoretically you shouldn't solder straight to battery terminals, but by heating quickly, getting the wire on in a couple of seconds, and getting the heat off at once, I've almost always had good results.) In 2003 people were talking about making Ni-MH's for 300 $/KWH, but for larger batteries. Since then, metals have gone up considerably, and these are small cells (the biggest available are "D" and "F" ('stretched D') thanks to the vested interests). $450 to $600 is at least headed towards more economic levels.
   On my car I measure about 200 amps as the starter motor starts up with the voltage dipping as low as around 9 volts, and 155 amps during steady cranking. I previously thought that the Ni-MH "D" size was probably only good for around 20 amps discharge rate, thus it would take perhaps 6 or 8 parallel sets of ten in series to replace a car battery, ie, $600 or more. But I see radio control groups on the web talking about 50 amps from Ni-MH "D" cells. I think it should only take about 3 sets; maybe 4 -- perhaps one more in freezing weather. Most of the economical cells seem to be rated around 30 amps, but starting the car doesn't normally take long and they'd surely put out more for a few seconds. Voltage will be just a little bit higher than with lead acid with 10 cells in series (or a bit lower with 9), but I think car alternators should be able to handle it as-is. 30 of these cells would be about the same effective capacity as a (new) 35 pound lead-acid battery and weigh about 8 pounds, lightening a car by over 20 pounds.
   Now with the lower prices and realizing fewer batteries are required than I thought, I'm minded to try replacing my car battery with 30 Ni-MH "D" cells for around 200 $US list price just as an experiment. (or 120 "AA" cells for $175) They are likely to last longer than lead acid even with the sodium sulfate added, and being lighter would continually save just a bit of gas on every trip, so they would probably pay for themselves. The only trouble is that the lower cost battery sources are all in the USA - in fact, here in town I've never even seen a Ni-MH "D" cell for sale. (That's disregarding "AA" cells in placed in a "D" cell package. You can tell these imposters by their light weight and the specs on the package.) Tacking on shipping, UPS border charges and perhaps duty will add $75 to $100 to the price - more than a new lead acid battery, locally available. Maybe I should wait for the US$ to drop some more... but the value of doing the experiment sooner is probably greater than a lower cost later.
   With the lower prices and high currents also I could conceivably use Ni-MH "D" cells for the electric car: As a minimum, 120 of them would give 36 volts at 120 amps (at 30 A max rate), and 40 AH (1-1/2 KWH), for $800 (++). Then 10 AH increments for $200 each. It's not cheap, but less than the $2000 Ni-MH starting price I'd calculated previously. The 40 AH performance would be better than for 70 AH of lead-acids and they're much lighter.

   If anybody else (in Victoria BC) wants to get some Ni-MH batteries at the same time, please call or e-mail me... we can get the best quantity discount and lowest shipping charges per item with a bigger order.

Carbon Sheets (meanwhile, back at the ranch...)

   My attempts to make flexible, impervious graphite sheets using tar or pitch and powdered graphite had been foundering on the fact that they really should be wrapped in aluminum foil (or something) to keep the air away as they were baked in the oven, but the aluminum stuck quite solidly to the work and they couldn't be separated without wrecking it.
   On the 17th I tried again with a piece that had also had hexadecane mixed in. The trouble was no different. But that same night I was (finally) replacing some leaking O-rings in my bathroom sink, and I suddenly thought: it's more gooey when it's hot - it turns to liquid on the stove... what if I run cold water on it?
   Sure enough, much of the aluminum peeled away easily, even suddenly. If it hadn't already been somewhat shredded, it might have come off nicely in one piece. Putting the sheet in the freezer for 10 minutes would probably work even better.
   I also thought of another expedient both to decrease the stickiness and improve the conductivity: paint the sheet with graphite power before wrapping it and putting it in the oven. It should adhere, especially to the stickier areas that cause the most trouble.

   I put the sheet of tarry graphite into a battery box I'd prepared a while back.


Compacted and oven heated sheet of carbon (tar, hexadecane, graphite)
as a conductive end layer behind terminal post in battery box.
I may re-compact the next sheet after baking it to smooth it out.

   Behind the sheet and in contact with it is a terminal that goes to the outside. Presumably, the tarry sheet will seal against the side of the box, preventing any electrolyte from getting in (and air leakage), so a terminal post of any metal at the center underneath, in electrical contact with the sheet, should work fine. And a copper or other mesh on the outside side could up the conductivity.
   Building the battery from the "plus" terminal up, the next layer on top of the impervious carbon sheet would be the dual-valence nickel-manganese-graphite electrode briquette. With the tarriness of the carbon sheet, I expect good electrical contact would be made between it and the briquette without any sort of grille or mesh.

   This seems to me to be the most promising construction idea yet since I finally realized that every common metal would dissolve in salt electrolyte. (Jungner really hit it lucky when he found pure nickel would withstand strong alkaline electrolyte!) But the resistance of this particular sheet was too high and I pulled it out again - I didn't add enough graphite when I mixed it.
   Naturally, I couldn't put much time into the project while working on motor evaluation and then design revisions.


A Battery Design Concept

   I've expanded on the above and placed an overview of my latest ideas for a good battery design here: www.TurquoiseEnergy.com/batts.html .

   In this I envision using the conductive pliable carbon layer to seal separate cells in a multi-cell flat plate battery, eliminating the shrinkwrap plastic of the 9-volt cell mentioned above. The pieces would be cut very carefully to seal against the edges of the case and prevent electrolyte from conducting between cells. A ring of wax could also be painted around the edge to help ensure there were no leaks.

   As I now conceive it, the battery in its simplest form would simply be sheets or plates of various materials, stacked as layers within a tube, that are pressed together as it grows.

   Hopefully I'll be able to try out this design in January or February.



http://www.TurquoiseEnergy.com
Victoria BC