Turquoise Energy Ltd. News #31
Victoria BC
Copyright 2010 Craig Carmichael - September 1st 2010


August in Brief (summary)
  * Workable design for Mechanical Torque Converter, and working electrochemistry for economical high energy Ni-Mn batteries - and even "dirt cheap" high energy Mn-Mn batteries: Now it's down to implementation details... though a few of them are still pretty challenging.
  * With the torque converter "in sight", I think it's time to start pushing ahead with adoption ideas - like getting motor controllers made, custom cut steel and aluminum parts, and motor making workshops.
  * Pulsejet steel cutter - potential project - or tool? Designed to be jet engines with no moving parts, pressure pulsed flame might make a great steel cutter. Use free-form for prototyping, or on CNC machine for automated cutting. (for the custom cut parts mentioned above.)

Electric Hubcap Motor Making Workshop
  * Finally Going Ahead. Self-paced, flexible hours per participant needs. Please Call!
  * Cost $1000 includes motor parts, supplies (further details on web site).
  * Accommodation here for out of town workshop participant, $150 per week.
  * Explore commercial possibilities for making motors, motor controllers, parts for them, install systems on vehicles, etc. (There have already been a couple of offers to buy finished units.)

Torque Converter Project: a Working Torque Machine
  * The mechanical torque converter is in fact a new type of machine to add to the "classical" machines:
     Lever, Wheel and axle, Pulley, Inclined_plane, Wedge, Screw,
     Hydraulic pistons, Torque converter.
  * Quick experiment & calculations indicates oscillating masses torque machine really ought to be feasible.
  * Weighted escapements experiment (136 grams versus 36) shows...
  * It works! "Clock Escapement" type torque machine provides motive force.
  * Ideal shape of converter drum: a "bowl shaped" convex bottom? - EH unit would pivot more smoothly when car wheel hits bumps.
  * Gears... all these problems and considerations, bypassed by the torque converter!

Turquoise Battery Project
  * Nickel plating dissolves off collector screen! - Inquiry reveals "brighteners" are added to the plating, so it actually is not pure nickel!
  * Sodium Silicate and Calcium electrode binder?
  * Graphite Collector Plates: Cheaper, more available than nickel, simpler than nickel plating.
  * Commercial lightweight "expanded graphite" sheets seem to work without corroding!
  * Hard electrode shells with "caramelized" egg white?
  * Mn-Mn: DIRT CHEAP 'everlasting' high energy 1.8 volt batteries.
  * Expanded graphite sheets, carbon (graphite) fiber & mat, carbon powder: the keys to a great, very conductive positrode!
  * Alkaline pH turns neutral, Ni-Mn voltage rises: ~~2.25 V - 15% more energy.
  * Bolt-Box Electrode Compactor Mark II - "how to" instructions.
  * Design ideas for "production" batteries.

Newsletters Index/Highlights:

Construction Manuals for making your own:

* Electric Hubcap Motor
(latest rev. 2010/02/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 battery longevity treatment - "worn out" battery renewal procedure.
* Simple Spot Welder for battery tabs, connections (in TE News #30)

all at: 

August in Brief

Electric Hubcap motor with prototype torque converter escapements.
Behind, drum with angle-toothed ring gear.

   Tic-toc! The "clock escapement" type of torque converter design seems to work! Last month's unit was re-tested on the car with just 100g of weight added to each of the three escapements (as shown above), just to get an indication of whether the design was worth pursuing. Unexpectedly, the torque was sufficient that it would have got the car rolling on level pavement if the wheel hadn't been jacked up. With heavier masses or six escapements instead of three, it would surely start rolling even up hills. But the big ring gear teeth folded up quickly when it was run in reverse, ending tests without an actual try to move the car.
   A somewhat more robust "fanfold" gear was made. It held up much better but the forces seemed lower. Put to the test, the car just moved - it was on the verge. Improvements as to details and proportions are needed. Then it should move cars - probably even squeal tires once it's really optimized.

   Near the end of August I realized this is essentially a new type of "basic machine" to add to those mostly known since antiquity - lever, pulley, wedge... ...the torque machine. I think I like the name. "Torque converter" is a very fitting name, but of course everyone immediately thinks of those horrible hydraulic things perpetrated by the auto industry.

   It was disappointing to find my nickel plated mesh and wire dissolving away in the battery positive electrode. I found out that the electroplating shop doesn't use pure nickel. I thought "Every metal disintegrates, and I can't seem to buy pure nickel or even nickel plating. I need a whole new approach." Then it struck me: Edison used graphite powder to increase the conductivity of his earlier nickel hydroxide positives; so did the recent India research... that meant graphite must survive where all the metals don't. Couldn't one use a graphite sheet for a current collector plate and forget the nickel metal entirely? Sure enough - it works! Plus, graphite powder can be had at a local art supply, and ready made graphite sheets are available on line. I bought some powder, made a couple of sheets in the electrode compactor using sodium silicate as binder, and ordered samples of commercially made "natural expanded flexible graphite" sheets. These proved to be excellent, with low electrical resistance and very light weight. They even look metallic silver gray like a metal instead of black (the image below caught them in a bad light), and unlike pencil graphite leave no smudge. I tried one out and it seemed to solve the corrosion problem. Graphite (carbon fiber) mat seemed a good solution for an internal mesh, and again is available locally (@ Industrial Plastics).
   It also struck me that in addition to 2.3 volt Ni-Mn batteries, one could make 1.8 volt Mn-Mn batteries with no pricey nickel in them at all and still about 80% of the energy density. Such "dirt cheap" everlasting, high energy batteries have broad implications for energy storage.

Conductive Graphite - sheets, fiber mat, powder:
my key material for non-corroding 'positrodes'.

   While there are details to sort out before having practical, working models, after a year and more of setbacks, "duds", and the feeling the main projects had all become hopelessly mired in mud, it is heartening to finally have what appear to be workable designs for both a mechanical torque converter suitable for the Electric Hubcap system, and two 'green', economical, high energy battery chemistries with workable materials to make them.

   I think it's finally time to resume the long delayed workshops for building the motors (including the torque converters) idea, with the assumption that by the time they're actually underway, a practical torque converter "production prototype" will have been built and tested, and can be duplicated. I'm working on that now...
   For those interested, current plans are that $1000 gets you a self-paced workshop including parts and supplies (over $400 worth), and you leave with your Electric Hubcap motor and torque converter, ready to install on your car. I expect to have the motor controllers available for $500, or there'll be a motor controller making workshop if your soldering skills are good. Times and dates will be flexible. I have a room to rent for $150 a week to crash in should anyone wish to come from out of town and perhaps devote full time to doing it as a "crash course".
   Even with the motors being made by hand, this is much the most economical and practical way to turn a car electric. Expect small business opportunities to open up for making and selling the motor systems and perhaps the controllers - there are already interested customers - and also for installing them in vehicles. I'll probably be offering various components to facilitate the process.

   Speaking of "various components", custom cut motor stators and rotors of 3/8" plate steel would be great. My plan was to take designs to a CNC waterjet place, but someone has introduced me to pulsejets, which were designed as jet engines with no moving parts. However, it seems to me that a with small one it may be possible to make a pulsejet steel cutter burning propane or MPS gas that would do the job. The jet pressure pulses would drive the flame through the steel, and oxygen would be unnecessary.

A small pulsejet device, two variations from a you-tube video.
Needed: transfer most of the heat from inside to the nozzle to cut through steel plate.

A Mechanical Torque Machine:
At Last, Torque Leverage Without Gears!

Electric Hubcap motor with torque converter having weighted escapements,
output drum behind with sprocket toothed gear.
The hookes on the escapements fit just between two tooth points
(three teeth apart), causing the escapements to oscillate back and forth
as the motor rotor turns relative to the drum.
The continually forced changes of momentum of the escapements'
masses provides the torque that drives the drum
- independently of the motor's own torque.

   This torque converter essentially works. I'll start with a couple of thoughts on torque converters:

1. When the car speeds up with a geared motor, the gears as well as the motor speed up. With this torque converter there are no static, stationary parts; everything spins with the motor and wheel. The speed difference, or precession, between the motor and the wheel is dependent only on the torque required regardless of vehicle speed. Thus, where gears may be spinning madly and making a lot of heat on the highway, the spinning torque converter is loafing along (precessing) little faster than at low speed. The transmission gears need to be in a bath of oil to prevent overheating; the torque converter (hopefully) just needs grease, solving all those problems of inefficiency generating heat, and then oil lubrication to cool the mechanism and generating fluid friction and more waste heat.

2. How can a torque machine provide more torque at the output than the motor has at the input? It works by oscillating masses. In the "clock escapement" type of torque converter, the masses that oscillate are the escapements. These drive the ring gear and hence the car wheel by their inertia. The power to spin the escapements comes from the motor, but it's the mass and speed of the spinning, oscillating escapements that provides the torque to the output drum, independently of the motor's own torque. This provides a variable 'n' to 1 "gear ratio", where 'n' is always greater than one. At the risk of being repetitive, the more the torque required, the greater will be the difference in speed (not a ratio of speeds) between the motor and the wheel.
   The ratio difference will depend moment to moment on the speed difference of the motor and the wheel, and hence will gradually drop as the vehicle speeds up, being a very high ratio at very low speed and a low ratio at high speed - the exact desired effect.

   The day after I sent newsletter #30 saying "Torque Converter Design #5" had way too little push, I considered that the light aluminum escapements had very little inertia. They could have a "pendulum", the real inertia piece of any clock.
   The force was so light in that test that it seemed simply making heavier escapements and doubling them from 3 to 6 could by no means bring the forces to the required order of magnitude. But if an arm with a weight could be "suspended" on a pendulum - in this case "suspended" perhaps meaning "flung outwards by centrifugal force" - from each one, the required "vastly more" force might actually be forthcoming.
   An escapement piece weighed 36 grams. If the weight weighed, say, 720 grams, that would be 20 times the inertial moment. If also that weight was out on the end of an arm, it might average 4 times as far from the pivot center, so it would oscillate 4 times as fast (average). E - 1/2 MV^2, so that would be 16 times the inertial energy. 20 * 16 = 320 times as much force. Doubling the escapements from 3 to 6 could then make it 640 x. Shorter escapements, spanning 1-1/2 teeth instead of 2-1/2, could cause greater pivoting angles and hence greater oscillations of the weights. Thus a gain of up to 3 orders of magnitude more force could in fact be realized - however much was needed, grams to kilograms. That just might bring the forces into the range of "car moving".

   However, adding so much weight and fitting arms on was easier said than done. How could I get pendulum arms through from inside the drum to outside?

   As a preliminary experiment, I took an escapement and ran it loosely around the rim of the drum, holding it by the pivot bolt. It swiveled back and forth as it went around, but the force was small. Then I contrived to stick on about 200 grams of steel pieces. The force was far greater. It readily turned the drum on its axle against friction even at quite low speed, which the escapement with no added weight barely did at a considerably higher speed.
   Perhaps sufficient weight could be attached to the escapement pieces on the inside to at least make a worthwhile live motor test on the car? If the jacked up wheel spun with sufficient force, then I'd have a feel for whether it was worth trying to devise a real design.

Speeded-up Masses?

   On the 17th it occurred to me that if one edge of the escapement had gear teeth, and it was made to turn a small weighted gear back and forth at high speed, the inertial effect of the mass of the gear would be tremendously magnified. Which reminded me that I'd been intending to study mechanical clock and watch mechanisms in more detail. But perhaps this was the key right here. Sounded complicated, though, and would make some heat.
   This whole idea proved to be unnecessary, superfluous.

Experiment August 18th: Surprise - The design concept seems sound!

   I bolted some little steel weights to the escapement pieces "as-is" to see how much force that could make with a live motor test on the car. The escapements' mass went from 36g to 136g each, with the extra 100g being two 50g weights near the outside edges. (per the photo above.) When tried on the car, the jacked-up wheel had far more push than without the weights - not enough to move the car, but then that wasn't expected - it was just to get a feel for it. It seemed that with the heavy "pendulum" mass, or the geared speeded-up masses, the force could be made sufficient.
   However, the batteries I used were low, especially one of them. I thought I should charge them up and try again.
   With fuller batteries the force was 5 or 10 times greater. I couldn't hold the car wheel back by hand when I turned the motor right up - that was quite a surprise! I estimate the torque I felt was greater than that needed to turn the wheel by hand and move the car on level pavement (which I did later for comparison), ie, that it would have got the car rolling. (I really should figure out some way to measure foot-pounds at the car wheel!) After the several designs I've tried that had much less push than expected, this one was the opposite!
   I was already concerned about the sprocket teeth - I could see tiny flecks of aluminum coming off. Next I tried running it backwards. That proved to be a mistake: the 45º teeth all bent away and let the escapements spin freely past. I tried pounding them back with a hammer and screwdriver, but I didn't get them all the way back and now even going forward they bent out again, after providing one more burst of considerable torque. If I bent them again they'd surely break off. That ended the experiment, but the "clock escapement torque converter" concept has been proven to work. Later inspection revealed the the sharp points on the leading sides of the escapements had also been worn to unsharp, just slightly rounded.

Drum - ring gear with bent-in teeth.

   So! from almost nothing without the weights to car-moving force with 300 added grams. More mass and more escapements are probably needed for the desired "tire squealing" push, but from the good force seen and felt in this experiment it appears the amount or quantity needed is maybe 4x to 8x rather than 25x or 250x -- the design seems practical with the configuration almost "as is".
   The challenge now will be to design and build an improved unit that's robust enough to handle the forces and last a long time.

   The aluminum sprocket teeth were wearing quickly, and as noted got bent in running in reverse, where the escapements pushed against the face rather than against the edge. But the escapements' contact faces were only slightly blunted. The escapements were .1875" (3/16") thick aluminum and the sprocket teeth only .08". But if the escapements were, eg, .375 or .75" thick, that would spread the contact with the sprockets over two or four times the width, reducing the force of the actual contact by 50% or 75% - assuming everything contacted squarely.
   Part of the problem is doubtless that there was too much play. The excessive slack allowed parts to hit each other with force instead of 'meshing' smoothly with each other.

   More precision is required, and more strength. At first I thought I'd better use steel. But that seems less smooth. It rusts (making it abrasive), it's heavy, and it doesn't dissipate heat as well. I decided to stick with aluminum until and unless it definitely proved to be impractical, in which case stainless steel might be necessary.
   So then I thought I'd try thicker aluminum for the gear, say .125" or .15" instead of .08" so it wouldn't bend, and thicker aluminum for the escapements, say 3/8 or 5/8" to spread out the contact pressure.
   Then I thought of taking a strip of the .08" aluminum and folding it up in accordion folds one inch apart, each fold 90º (± the rim curvature) to make the 45º teeth. I could adjust the angles minutely to set the precise distance between each two teeth. Regardless of direction the escapements would press on an area wider than just a thin edge, and also against the end of a wall, which would take a lot of force to buckle.

I bought this vise-mounted sheet metal bender to make the fan folds
The usual types of benders won't work for these bends, and even
this setup had to work around the center of the vise.

   The exact shape and configuration of the contact areas of the escapements probably need to be more carefully determined so as to eliminate regions of 'slop' in the mechanism so it will all mesh fairly smoothly. (It looks to me like the points should be slightly closer together and I'll try that on the next one - the 6th one.) The prototype made a lot of noise, so I don't think it was there yet. The steel stand-offs (oversize nuts) in the middle adjacent to the pivot points were probably hitting the ring gear teeth. However, judging by where the wear was, this didn't account for much of it. But I changed to thinner stand-offs of brass pipe, and sanded down the center area of two of the three escapements a bit more. Another factor I found in the "poor fit" was that the frying pan isn't quite round or quite centered, and this actually made escapements that went around 85% of the rim fine jam on the other 15%, so they had to be filed back more - making a sloppier fit on the other 85%. An accurately made drum would naturally be better.
   Then of course the optimum amount and distribution of mass on the escapements still needs to be determined. However, it would seem from the experiment that the weights required will easily fit within the drum, 'under the hood' - no "pendulums" sticking out, no geared-up masses, will be needed.
   I am also concerned about the small nylon bushings the escapements pivot on, and the small 1/4" bolts the bushings themselves pivot on. There isn't, as made, room for ball bearing races. That would require moving the pivot point away from the gear and, at the very least, changing all the angles and curves already rather laboriously worked out. Perhaps the thing to do is to make plates that extend over the tops of the escapements, so that the bolts are supported at both ends.
   Or, I could simply switch to fatter bolts. I might get away with 5/16" - maybe - but 3/8" or more would certainly mean slightly moving the pivot holes and modifying the escapement shapes so the center doesn't hit the teeth. On the other hand... none of the bolts or nylon bushings have actually had any problem so far. Maybe I should quit worrying about it unless they do!

   On the 27th I had the new 'ring gear' made, with accordion folded aluminum. (or is it fan folded?)

The fan-fold ring gear.

   The torque was fair but it didn't seem as impressive as the first time. The only explanation I can think of is that although every third tooth is bolted to the rim of the drum, there was doubtless some "spring" to the other two which might have limited the force transmitted, whereas on the first gear in the forward direction, the teeth were pretty solid. (Perhaps that also explains why the teeth of the first gear were more worn - indented - where the escapements hit?) I drove the car to level pavement to try it on the ground. The car moved sluggishly but didn't accelerate, and I turned it off after a foot or so, nervous about the ring gear's teeth. The drum could be seen to vibrate, spring, against the car wheel, which doubtless absorbed some of the force. Well, I hadn't originally expected this light setup to move the car. And yet, what happened to the higher torque I'd felt when the first ring gear was used?
   One thing about this as opposed to the directly connected motor moving the car sluggishly: I can readily modify this to increase the forces. I can double the number of escapements, or I can double the weight on each escapement, or both. Getting more complicated, I can try different shapes and sizes of escapements. With the motor, the only choice was to make a much bigger motor or to install four of them... or better yet, eight.
   It turned out that this gear was in quite good shape after several tests: no gouges, dents, or bends - a surprise after the first one. There were just slight marks where the escapements first hit on each side of the point. It might go a few miles at the very least. On the other hand, it hadn't been receiving really car moving forces yet. The question was whether to make three more escapements and try again, or to try some of the ideas below with all-new tackle. I thought of an escapement shape that I thought would be smoother and decided to try three new escapements of 3/8" aluminum first, total six. The shape proved to jam as envisioned, but when sanded down to flat outer sides seemed somehow to have improved fit over the previous ones with curved outer edges. But somehow I didn't get all three made by the end of the month. Since I've just noticed (Sept. 1) that the points seem just a bit too far apart, the last one might be better than the rest.

Some main design considerations

* is 25 teeth in a 12" diameter gear is anything like optimum? For all I know, anything from perhaps 10 teeth to 50 with appropriately sized and shaped escapements could work better. Come to think of it, the more the teeth, the lighter the force needs to be at each tooth to turn the wheel, and the longer the aluminum is likely to last. Only 10 'hits' around the wheel would each need to be five times as heavy as 50 hits to provide the same force in one rotation.

* Balance of forces & number of teeth: When I picked 25 teeth, I figured, without thinking about it very hard, that 2, 3 or 6 escapements would each be in a slightly different position relative to the teeth for the "most even force". However, each escapement pushes on exactly one tooth at a time, distant from the tooth each other escapement is pushing on. So to have one tooth at a time being pushed does nothing to minimize the force on each tooth, but it does minimize the maximum pulse of force delivered to the wheel. It also makes the forces to the wheel unbalanced at all times. If instead for example three escapements 120º apart all hit at once, the momentary force to the wheel would be higher and also balanced around the rim, both aspects helping to turn the wheel and get the car moving. This suggests that the number of teeth should be divisible by three, if not six, eg, 24, 27 or 30 teeth instead of 25. With 27 teeth and 6 escapements 60º apart, there'd be two alternating sets of balanced pushes. With 24 or 30 teeth, all 6 would strike at once. With 26 or 28 teeth, three sets of two teeth at opposite sides would strike each time. I'm definitely leaning towards 27 - or maybe 33, 39 or more if the gear wears out very fast.

* The escapements only hit the outer half of the teeth. The fanfold ring gear could be made with 90º angles for the teeth as shown, but two 45º angles on the outside against the drum rim instead of one 90º. Each tooth would then have a small flat spot against the rim where the bolt would go through. Joining two sections of folded aluminum would be easier. In addition, this would slightly increase the effective diameter. (I'd have to change everything including the escapements and the pivot hole positions, so I won't do it with this prototype.)

* Would "ripples" for teeth and gentle curves work as well as or better than sharp angles and pointed 'hookes'? One expects they would last longer, but a "smoother" oscillation might produce less thrust than one with quick reversals of direction -- which seems to be working. It might need considerably more mass on the escapements, probably eliminating the expected 'less wear' advantage.
   To form the curvy ring gear, one could make a jig with some zig-zag bolts going across between two flat pieces, and bend the aluminum around each bolt, each next bolt being inserted in turn as bending proceeds. The diameter of the bolts determines the radius of the curves, and accurately placed bolt holes would result in uniformly spaced and shaped teeth. (Or one could simply make a single vise-mounted bender with a rounded punch instead of a sharp one.)

* Drum diameter. Of course, the larger the drum, the more leverage against the wheel the forces will have. I picked 12" O.D. mainly because I had the 12" frying pan - it was convenient, and I thought the steel 9" I.D. brake drum was too small. I think 12" is fairly minimal. A larger diameter should probably have lower losses and less wear. Of course, if the diameter is too large, it is likely to be damaged in the event of a flat tire, and this size depends on the wheel rim size. Note that the motor rotor diameter needs to be large enough to mount the escapements, or it must have extensions added to mount them.

* Is a toothed ring and escapements the best idea in the first place? There may be a number of possible configurations of in-plane oscillating masses that would work. (That said, having finally found an arrangement that does work, I'd have to have some very brilliant flash of an idea for something better to want to fundamentally change it!)

* Are 45º triangle teeth the best shape? Of course, these could be replaced by pins where the apex is. But also the teeth could be, for example, just slots in the rim of the drum. If the holes took up at least about 2/3 of the rim with 1/3 solid between them, it should be possible to design escapement shapes to work with them. The effective diameter would be maximized. This might be a good thing to try out, since it makes doing a ring gear from sheet metal simple, with a punch.

* Keeping the maximum "pulse" of the forces to a level that won't hurt the aluminum is very helpful. Evidently, this depends on the fit, the number of teeth, the geometry and the hardness of the metal used.

Harder Aluminum?

   Of note to that last, people had been saying I should get a blurry DVD player for my new hi-def TV, and I went into Sears Home Center to look at them. (Future Shop having gone missing from the complex.) A show was on (on about 25 TVs) about BMW's diesel engine plant. They were making diesel engine blocks out of aluminum and had 'new techniques for making aluminum strong enough to use as a diesel engine'. The last aluminum engine blocks I saw were in the Pontiac Astra/Vega in the late 1970s, and those weren't on the road long - the engines only lasted about 40,000 miles. But the salesman, consulting a small piece of fruit in the palm of his hand, said you can temper aluminum with ultra-fast cooling. That sounded good (also the "we won't be undersold" price), so I bought the DVD player. BMW is probably using further newer metallurgical techniques on the aluminum, which might potentially be applied to aluminum torque converter parts. How much equipment investment it would require beyond the DVD player, or whether pieces could be "case hardened" (or whatever) at a local facility somewhere in most cities, is another question. But then, even if aluminum torque converter parts needed replacing every 10 megameters, it would only be a nuisance, not a Vega engine disaster.

   Seemed to me I could get the escapements made by CNC abrasive waterjet cutting, along with the motor stator and rotor and perhaps a couple of other custom parts. Once the CNC program is set up by the contractor for the machine, the parts can be made beautifully, uniformly and inexpensively. Now I'm wondering about pulsejet cutting, on a certain CNC router I know of, made by the owner.

   Of course the best thing is to continue hand prototyping to determine what works best before designing escapements that will soon need to be modified.

Drum Shape vis a vis Vehicle Handling

   Another thing that occurred to me was that the ideal shape of the torque converter output rotor drum might be a convex base rather than a flat one. That way, when the car wheel hits a bump, the whole Electric Hubcap assembly would more smoothly rotate so as to minimally affect the unsprung mass. If the "bowl" continued as a smooth curve all the way to the rim, the rim wouldn't "hit" the wheel anywhere when twisting with larger road bumps.

Flat outer face of converter with 'pins' that link to the car wheel:
change drum to smooth convex bowl shape?


  The section below was written only days before the first 'successful in principle' torque converter test. (It shows my slipping level of confidence after so many failed converter designs.) I now have no use for gears, but this can serve as a good indication of all the troubles the torque converter eliminates.

   Given that I still, after considerable time, have no working torque converter, I also started thinking about making some sort of planetary gear setup just to get a car moving, even if only for a demo. I estimate about a 7 to 1 ratio with an EH motor would get a car going uphill. That, however would have the motor doing 2000 RPM at under 30 Km/H.
   Something that could actually hit the street could be two motors, left and right, with 3.5 to 1 gears, which would give as much start-up torque and allow speeds 55-60 Km/H. Here once again we see the "oversize" motor (2 motors) needed to gain sufficient torque for start-up with direct drive, or in this case with a fixed ratio gear. To allow highway travel we'd need motors on all four wheels at 1.75 to one - or seven motors (somehow) for direct drive.
   Two motors would obviously increase the battery current requirements to get the car moving as well as doubling the parts, but it should work. Also the motors would have to effectively be disconnected from the wheels before the car was to be run on the highway to prevent over-revving them.

Two planetary gears
Right: a starter motor with three planetarys, ratio about 4-1/4 to 1.
Left: A car transmission gear with five planetarys. (input gear missing.)
The 'momentary' starter gear is lubricated with grease, while the 'continuous' transmission gear is sited in a transmission oil bath. To use these (transmission gears) with an Electric Hubcap, shafts must be replaced, plates welded on, etc, and how long would the gear last if not immersed in oil?

   Someone once told me a single planetary gear can make a 3-speed transmission, but on looking at one I'm puzzled. I can only see how it could be 1 to 1, the designed ratio, and "neutral". Unless I'm missing something, I would go only as far as calling that two-speed.
   It's just possible that one might use the designed ratio and 1 to 1 to get the two-motors car onto the highway. Perhaps with some sort of centrifugally actuated gearshift. Perhaps also they could work at different ratios to avoid both hitting weaker performance zones at the same time, eg 4:1 and 3:1. (Or with 5:1 and 2:1? -- the 2:1 could stay that way at all driving speeds.)

   Getting more complex, a single motor might be used with a three speed transmission, (eg) 7:1 to get the car moving, 3:1 for city driving speeds, and 2:1 or less for the highway. A question then arises whether a single 5.5 HP motor has sufficient power for acceptable performance, with the losses of gear frictions and less than optimum motor speed and power at various vehicle speeds. These factors will also up the energy requirements for either a one or two motor system, increasing the battery capacity needed (...towards that of other electric vehicles).

   But the above was written before the surprisingly strong push of the "clock escapement" torque converter on the 18th when the little weights were added. At this point it will be easier as well as far superior to get to a practical version of the torque converter than to attempt to mount car transmission type planetary gears - which themselves might not last very long without being in an oil bath. There are now probably fewer remaining design challenges to work out for the desired advance on the state of the art than for the "common" solution.

The Torque Machine

   At almost the end of the month, I suddenly realized, wow!, this is actually a new type of 'simple machine' to add to machines mostly known since antiquity: pulley, lever, inclined plane, wedge, screw, wheel and axle (presumably including gears), the more recent hydraulic pistons, and now the mechanical torque converter machine. It does follow the usual machine definition: it converts mechanical energy, ignoring frictional and other losses, from one form or amount to another form or amount, such that:
   FORCEin * DISTANCEin = FORCEout * DISTANCEout (minus any applicable losses).

   The most unusual feature with the torque converter is the time or velocity factor as related to the force. If the rate is slow, little force is required to turn the rotor and slowly oscillate the masses - FORCEin~='0' and (with small losses) the output may not move at all. As the RPM speed increases, FORCEin also increases, exponentially, to cause the masses to oscillate faster and faster. The work done at the output - which is in fact done by the oscillating masses - is still equal to that done at the input (minus losses).

Turquoise Battery Project


   We have become conditioned to think of batteries as being either crappy or super expensive, and having limited lifespans. But there are 1.2 volt Ni-Fe batteries that have been in continuous service for decades, a few for almost a century. And I have Ni-MH 1.2 volt "AA" cells that I bought in about 1996 when they were first coming on the market that still work like new. Ni-Mn (2.2 V) holds the same promise. With the higher voltage and energy of manganese negatrodes unleashed via a bit of egg albumin, there's much potential for astounding transformation of key energy systems.
   Electrically powered vehicles can perhaps be given enough storage for a full day's travel with Ni-Mn, to be recharged overnight.
   In addition, it seems to me Mn-Mn 1.8 volt batteries could be made. They'd be "dirt cheap" with only about 1/5 less energy density and again 'indefinite' life, perhaps for transport but also for off-grid homes and other applications where weight isn't critical. I plan to make a MnO2 positrode and try it out - it may even provide more amps by size.


   The new electrode compactor was finished on the 4th. "Recompacting" an electrode that had ended up about 7.5(?) to 8.75(?)mm thick with the old compactor thinned it down to 7 to
8.3mm, the thinnest dimension being the edges and the thickest the middle.
   That was better compaction, but I was disappointed with the amount of bulge in the 3/8" top and 1/2" bottom plates. Evidently even thicker steel is needed! And that compactor took several hours to make.
   Let's see, I could turn the bottom into a 1/2" top, drilling out the threaded holes to 1/4". That would salvage a top and sides, and only a new... 3/4"?... bottom would be needed. It already is a 2-handed carry!
   Or maybe I'll just weld another 3/8" or 1/2" piece onto the bottom. As long as the bolts don't go in more than 1/2", I won't need to drill and thread it. A 1/2" press piece will do better than the 3/8" one, and it doesn't really matter if the top bulges.

   I had hoped better compaction would give lower resistance readings - it seemed almost inevitable. But in compacting an existing electrode, dry, it became brittle and fell apart, and any attempt to push the meter probes firmly in caused each piece to crack. Thus the megohms readings were unrepresentative.

   I wanted to disassemble that electrode and redo it with nickel plated mesh anyway, and add some monel. Where I had planned to use it as a negative, I decided it was actually a good positive formula (60:40 Ni:Mn) and that I'd use less nickel (even very little or none) and more manganese in the negative. If more conductivity was required in the negative, evidently copper powder, cheaper and most conductive, should work, since the copper screen and leed do.
   It took a lot of crunching with mortar and pestle to reduce the clumps to powder again, and it isn't 3/4 the volume of the unprocessed powder.

   Thinking of how sodium silicate "water glass" hardened even in a closed jar in a ceramic mix got me thinking of the problem of a good battery electrode binder. It certainly made good ceramic mix binder! What would it do in a battery? Could it be permanently solidified by adding calcium (or barium) as my brother had suggested, to make an electrode solid?  Calcium also had some good redox potentials for use in a positrode. I decided to try it out.

   I took last month's battery apart and put a nickel plated screen behind the positive electrode. The nickel-brass plate and the stainless steel mesh were about dissolved and conductivity had dropped to "not much". The manganese negatrode looked about as it had when I first made it. I started to charge it again.
   Once again it could handle currents up to an amp and more like when it was new, but those currents dropped off overnight, and the nickel plating had turned green and dissolved off the copper screen! That's just not supposed to happen! And we've already seen that copper screen quickly dissolves without protection.
   I phoned Victoria Plating to confirm that the plating was pure nickel and was told there were "brighteners" in the solution that would end up in the plating. "We're a decorative plating shop" he said. SO! The quest for real, pure nickel or nickel plated materials, as required for alkaline positrodes, has been thwarted again, unexpectedly, from a hidden cause! I'm getting really, really sick of dissolving screens and leed wires. How can a whole new battery technology founder on something so seemingly trivial, whose solution is well known?

   Perhaps a whole new idea is needed? Graphite powder has been used to improve the conductivity of nickel electrodes. Presumably then it's a rare conductive material that doesn't degrade and remains solid in the charging positrode? It's about the most stable form of carbon. So: what about a solid (non-porous) graphite collector sheet instead of a metal one? One with a plate or screen of metal behind it - or embedded in it - connected to the terminal but out of the electrolyte? Graphite has electrical resistance, but it would only be across the thickness, over the entire plate area - not much overall. As long as the electrode material was well pressed against it, it should work.
   And rather than trying to find ready-made graphite sheets or plates, I could just buy graphite powder, add just the right amount of sodium silicate/barium/calcium "glue" (rendering it non-porous yet still conductive, I trust - or some other technique to accomplish that if this doesn't work), and press graphite/metal mesh sheets the same way as electrodes. (Hmm... I wonder what heating the product in the kiln might do?) On checking, I find graphite powder is readily available at art stores, about 30-40 $/pound retail in small quantities.

   Suddenly this seems like much the best plan!

   In fact, some cheap metal will do for the screen or plate, so it replaces the pricey "unobtainium" nickel or nickel plating with cheap, readily available graphite. The only remaining nickel in the battery could be the monel powder (which I hope can be retained - we'll see) and the active nickel hydroxide.

Graphite Collector Sheet Experiments

   Obviously the next task was to try making a graphite collector sheet or plate, since I seem to have most of the other things working reasonably well except for that worrysome self discharge - which may itself be a symptom of corroding collector plate.
   I bought some graphite powder at Opus art supply, mixed in a little sodium silicate and barium sulfate (a few percent), and tried to make a sheet. The powder looked much the same as several black powders I've been using. The sheet turned out too thin, and very brittle, and with plastic underneath it curled into and arc shape as it dried. But the resistance measured only ones of ohms - great!
   Okay - four teaspoons of graphite instead of two, more wetted with sodium silicate. (Maybe I should measure these things?) That made a sheet I consider to be about the right thickness, a little harder and reading about 10-20 ohms.

   Also, I found www.GraphiteStore.com on line and decided I'd hedge my bets and order a few ready-made "flexible" graphite sheets, .06" thick and .12".
   These turned out to be 'natural', 'expanded' graphite, '99% pure'. They had the desired low electrical resistance typical of graphite. I expected dense, black, brittle, smudgy stuff, but they were very light, silvery, could flex some, and weren't smudgy at all. In appearance, they could almost have been some dull metal, though they dented and scratched easily. The light weight will subtract almost nothing from the "energy density by weight" specs.
   I'm pretty sure this is 'the right stuff', ideal unless something in the 1% impurity proves to be a problem, which it didn't immediately seem to be on trying it out. The remaining problem will be making a seal to have the electrolyte stop at the graphite so that the metal mesh or whatever behind it, and the leed, going to the "+" terminal, doesn't corrode away.

   When I tried one of the sheets out, on the night of the 21st with the manganese electrode (still not in bad shape except for a couple of small broken chunks) and a remaining chunk of nickel electrode that I peeled the rotting stainless steel mesh off of (only about 2-3 square inches - the whole electrode was breaking into pieces), it quickly charged to a higher voltage than before. It was soon 2.66 volts on 100 mA charge, and started dropping from about 2.2 volts (later 2.3) when the charge was disconnected instead of from about 1.95, closer to my original rough voltage estimate of +1.0:-1.36 = 2.36 volts for Ni-Mn in neutral salt electrolyte.

   Seeing that, I checked the pH of the fresh electrolyte. The electrolyte had originally been turning alkaline around 12 or 13, and it was now in fact neutral - 7 or 8. Although the original mix was Mn powder and MnO2 powder, not Mn(OH)2, I had added Ni(OH)2 to the mix to derive nanocrystalline conductive nickel metal from, and this would have charged first, releasing the OH-'s that quickly alkalized the original solution.
   I may mention the possibility also that in neutral pH with salt, it may discharge from Mn to MnO rather than Mn(OH)2, and the positrode from NiO2 to NiO rather than Ni(OH)2, and the pH will remain neutral. With the KCl salt electrolyte, the O-- ions might be carried on the chlorine as potassium hypochlorite, KClO. I could be wrong. (In fact, the electrolyte hypochlorite ion idea seems suspicious.)
   But this would have interesting implications if it's true. It could also mean (a) that nickel oxide would be the 'correct' form of nickel to buy rather than nickel hydroxide and (b) the Mn negatrode would also work with a chelated lanthanum perchlorate/chloride positrode, which also moves oxygen ions. ...which would mean nickel electrodes in any form are entirely unneeded, and super high energy densities would be attained. However, I could be wrong and the presence of water even at neutral pH will result in hydroxides on discharge rather than oxides. Really though, this works and looks like a fantastic battery either way. It's better than the best and will put electric drives on the road - and for the moment, that's all I really care about!

   The next morning (22nd) the battery, though charged overnight, wasn't holding its voltage as well as it was at first, and was taken apart and cleaned, the electrolyte being replaced. There's still something causing self discharge. I suspect the monel, and I'll try an electrode with graphite powder instead.
   On inspection the silvery graphite sheet seemed unchanged. A slight discoloration where the electrode had been in contact readily washed off and it proved to be unblemished - it wasn't possible to see where the electrode had been. (...except the expanded graphite's surface was so soft there were indents from the scrub bush used. Brushing another area left it with exactly the same appearance.)
   By evening the voltages were marginally better yet, dropping from around 2.3 volts (and somewhat more slowly), around 20% improvement over the original alkaline 1.9 volt levels. I had a similar improvement when I used the commercial Ni-Cd electrodes with salt electrolyte: from 1.32 to 1.6 volts or so. (Wow... 2.3 volts open circuit! - perhaps 2.1 volts nominal, 1.75 times the usual 1.2 volts!)

   It would seem at least some of the awful self discharge I've been seeing all along throughout all my experiments was indeed from corroding positrode metal, and that the graphite collector sheet has improved the problem.
   This battery with the chunk of electrode supplied 12 to 9 mA (gradually dropping) into a 100 ohm load for an hour and a half, and then recovered to 1.2 volts. Since the open circuit voltage started at about 1.7 volts, 1.2 / 1.7 = .5 V drop, so the internal resistance of the battery was about 42 ohms, which is definitely not as good as x1's of milliohms or less. This points to the need for better conductivity within the electrodes. I ran this a few more times, drawing a load for an hour or more once or twice a day, and the figures gradually improved, running up to 17mA (with 100 ohms = 1.7 volts) which dropped to 14 in a few minutes and then gradually to 10 in an hour, to 8 in a couple of hours. (Even without a load, the self discharge brings it down to about 1.2 volts in an hour, 1.1 volts in two hours, and 1.0 volts in three. Obviously this is by no means satisfactory.)

Graphite Powder - Carbon Fiber

   Graphite powder should greatly increase the conductivity of the electrode, and the heavy monel powder could be eliminated or reduced. Fine, short bits of carbon (graphite) fiber might be even better, or both together.

   Again, when I first put the chunk of electrode to the graphite sheet, it was (for once) dry, and I measured the conductivity. The resistance was in the "x 100 K Ω" range. As long as its not an insulator it should charge, but that's not exactly fab conductivity.
   Hmm... I did add cobalt powder and monel to that electrode. I think I should also try adding graphite powder - or using it instead - to help it to conduct better. In fact, perhaps the remaining self discharge would be eliminated if I eliminated the last copper and nickel bearing metal from the positrode. (Monel will doubtless still work fine in the negatrode if it also proves to have low conductivity and needs it. Monel or nickel might also work fine in a lower voltage positrode, eg Mn - the voltage might be too low to corrode it.)

   On the 23rd I had the idea to try carbon (graphite) fiber rather than, or in addition to, plain powder. The idea is to improve conductivity, and electrons can doubtless pass along fibers more easily than between doubtfully connecting powder granules. If, in an ultimate example, very fine closely spaced fibers, "nanorods", all ran like the bristles of a brush in parallel from the font surface of the electrode to the graphite backing sheet, the whole electrode would have "ultimate" "short circuit" conductivity. Some university can figure out how to accomplish that later. Random fibers in the electrode mix would be the next best thing. Carbon is much less dense than monel, so the energy density by weight would be improved, and the "crackly microphone" effect Edison got from graphite powder in his electrodes would surely be eliminated with fiber.
   Suddenly - and at last - I think I can see obtaining the sort of amps needed to start a car engine or run high power electric motors efficiently with very small batteries!

   I tried cutting carbon (graphite) fiber into short bits to help improve electrode conductivity on the 29th. However, it bunched up and didn't mix into the electrode powder well, and dust came off it, mostly while cutting. It was like working with fiberglass or asbestos and I was soon feeling itchy needle pricks just from the little bit I did - yuk! I'll stick with the plain graphite powder in the electrode mix, I think, and a graphite fiber mat - with minimal cutting - as the closest thing to an inner metal mesh that's available.
   Or I could try buying pre-chopped fibers, which come as short as 2 or 3mm lengths. That should be easier to deal with if it doesn't clump up during mixing.

Chopped carbon (graphite) fibers, much finer than human hairs - and dusty.

Egg Layer - hard electrode shells?

   If one were to smear or paint egg white all over the electrode, some would wick in but much would dry on the surface. If the surface was then torched, it would be left with a "caramelized" hard but microporous surface layer, to prevent the molecular changes that occur with charging and discharging from occurring right at the separator. That sort of surface "casing", keeping the reactions and "stuff" inside, might make the battery last far longer - virtually indefinitely. One could look at it as a simple and in many ways superior replacement for the metal electrode housings of the pocket battery.
   I tried this on one face of the little crumbling piece of Ni electrode, though not a very thorough coating. I didn't do the edges, and it seems to work, or at least not to do anything bad, whereas it continued disintegrating around the edges. The next thing will be to try "painting" and "caramelizing" a whole Ni electrode, and then see if it gradually crumbles in use like the last one or stays in one healthy piece.

Manganese-Manganese: DIRT CHEAP High Energy Batteries

   As well as being the new negatrode material, MnO2 is of course the positrode chemical in both standard and alkaline non-rechargeable batteries. In neutral KCl solution the reaction might be:

2 MnO2 + H2O + 2 e-  <==>  Mn2O3 + 2 OH- [+.5 v],

or it may take the same form as in alkaline solution, except with the .5 volt figure applying rather than the alkali +.15 volts:

MnO2 + H2O + e-  <==>  MnOOH + OH-  [+.5 v]).

   Whichever variant applies, it seems certain one could make a Mn-Mn/KCl battery. It would be .5 volts less than the Ni-Mn/KCl battery, about 1.8 v open circuit, but otherwise similar. It would probably have similar almost indefinite lifespan. The lower voltage, but with slight compensation for it making a lighter electrode, would mean about 1/5 less energy density. (Hmm, actually, that's better than I was thinking. ...and still higher than lithium types!)

   A bit lower voltage and energy... what's the advantage?: Low, low cost!

   The nickel, as oxide or hydroxide, is the main cost of Ni-Mn. The Mn-Mn battery would be "dirt cheap" - similar to all those dry cells that are used once and thrown away!
   For transportation, higher energy density is worth a certain premium (though I don't mean the sort of absurd "premiums" being paid for lithiums), but for stationary applications like off-grid homes very low cost means one can afford sufficient battery reserve to store oodles of energy when the sun, wind, waves or tide is providing it, for the times when it isn't. Or a generator could be run intermittently to charge the batteries rather than running full time. On the water where weight is less of an issue than on land, boats and other vessels could probably employ Mn-Mn to advantage for their electric propulsion. (That could be combined with wave, wind or solar power to recharge whenever conditions were good.)

   The more I consider it, the more other potential advantages I see, too:

* manganese positrodes might be more conductive and so provide higher maximum amps than nickel. (In fact, the thinner electrode - see next - is bound to be more conductive both to electrolyte ions and to electrons.) That provides superior power even for electric transport, and faster charging.

* The nickel electrode has a lot of manganese already added to it to allow the nickel to charge to a higher valence - a considerable portion of the manganese needed for a Mn electrode is already there. Without trying to work out exact proportions, assuming MnO2 only changes by one valence state (to MnOOH), twice as much MnO2 is needed in the positrode as in the negatrode, which changes two states (Mn -> Mn(OH)2).
   My sense says that the denser MnO2 electrode will be much thinner than the fluffy Ni(OH)2 one - half - allowing the whole battery to be almost 1/3 thinner. This saves on all materials except the negatrode, saving weight and space - probably enough to compensate for the higher voltage of nickel. It seems likely that in spite of needing a few extra cells because of the lower voltage, eg about 21 cells for 36 volts instead of 16, the energy density might not be lower than with nickel after all!

* The lower voltage might allow use of the most conductive conductivity additives such as monel and nickel, which probably don't corrode below a threshold voltage (about 1 volt), again raising potential current capacity and reducing charging times.

* The same substance is in both electrodes should the ("green") ingredients be deemed worth recycling. And it's one less chemical to stock.

* The Mn positrode is more "old hat" with no(?) recent developments, so there would be little for vested interests to bite on to attempt to prevent manufacture by patent challenge.

* The low voltage would have no problems with oxygen overvoltage during charging.

   Wow! Considering all this, it may well be that it's not worth bothering with Ni-Mn and I should switch entirely to Mn-Mn! But I am now beginning to trespass on September.

Next Battery(?)

   For the "production" battery, I wish to make bipolar cells that can simply be stacked together to obtain the desired voltage, almost as I had originally planned except this time these will be individually separate, like a giant version of stacked "button cells" used to make small higher voltage batteries. They will then be stacked into an appropriate case that is then sealed and holds under pressure - the giant buttons themselves aren't expected to hold against pressure.

The battery cross section:

- End wall with flat-head terminal bolt sticking through the middle, tapered hole. This must seal. Red plastic ring around bolt identifies "+".
- Sheet metal or mesh collector sheet, soldered to the bolt and against the wall.
- "n" stacked bipolar cells to obtain desired voltage, eg 36 V. (details below).
- Sheet metal or mesh collector sheet, soldered to hex head bolt.
- Sponge piece. This should compress to about the thickness of the bolt head. It holds all the cells and connections pressed together.
- Far end wall. This is assembled last, and is preferably gasketed and bolted or screwed on rather than glued, making it is easy to replace individual defective or weak cells.
- A nut holds "-" terminal bolt tight against wall for seal. Black plastic ring identifies "-".

Cross section of each cell:

- expanded graphite sheet (positive connection)
- a painted layer of calcium carbonate on the inside of the graphite
- the active positrode briquette (details below)
- carbonized egg white layer - painted on electrode, dried, and then torched (also on edges)
- microporous cellophane sheet
- separator paper
- the active negatrode briquette
- copper or stainless steel screen
- nickel-brass sheet.

   These cell assemblies, 3" x 6" (20 amp-hours) or 3" x 9" (30 A-H) by about 8 mm thick, will be sealed around the edges by dipping them in some sort of slightly flexible paint or RTV cement. This seal is not expected to be perfect or to hold pressure: that's the job of the sealed case that the cells are mounted within. It is also the job of the chemical additives to convert hydrogen and oxygen gasses generated back into water to keep the pressures low.

Before I do this full size, I'll think I'll try one in a 1.5" x 3" format - 1/4 of the 3" x 6" size and only about 5 amp-hours. Since each cell is to be edge sealed, I want a close fit. I think melt-forming round pipe is out. I wish I could find a source for square pipe with sharp corners.

Why haven't I made it yet? There's only 44,640 minutes in a month, and many things to do! I did find time to make inserts for the new electrode compactor for making 1.5" x 3" electrode briquettes.


I planned the next positrode mix as follows:

* Collector sheet of expanded graphite.
* A piece of carbon fiber mat for a "grill"/"mesh".
* A coating of CaCO4, calcium carbonate, on the electrode side of the graphite.

* NiO - 60 wt%
* MnO2 - 40 wt%
* Graphite powder or chopped fibers - TBA wt% (of above)
* Sb4O6 - 1 wt% (of Ni+Mn)
* Co2O3 - 1 wt% (of Ni+Mn)
* NaSiO2 - 2 wt% -- best "glue"?


For the negative side:

* Nickel-brass collector sheet
* Copper mesh

* MnO2 - 97 wt%
* Egg white - 0.1 wt%
* NaSiO2 - 2 wt%
* Sb4O6 - 1 wt%

This is to be "pre-charged" in a tank of KCl (or NaCl - cheaper).

Bolt-Box Electrode Compactor Mark II

   I considered that 9 or 10 inches tall would be a good height for batteries placed somewhere in a car, eg in the trunk, and that 6 inches seemed a useless size limit, so I made the new compactor for 3" x 9" electrodes instead of 3" x 6". If desired, a spacer(s) can be inserted to make smaller electrodes with smaller press plates, but none bigger than the whole compactor are possible. I made pieces for the original size with the original 3" x 6" punch plate.

Parts List:

Bottom: 1" x 4" steel bar, 12" long. [1]
Sides: 0.5" x 0.5" steel bar
   1 @ 12" long[2]
   1 @ 10: long
   2 @ 3.0" long[3]
   (or one piece about 28.5" long to cut the above from)
Top: 0.5" x 4" steel bar (3/8" x 4"), 10" long.
Press Plate: 0.5" x 3.0" steel bar, 9.0" long.[4]
Bolts: 38 - 1/4"-20 x 1.5" hex head. (-20 = 20 Threads Per Inch.)
          38 washers might also be nice. I haven't been using washers so far.
          7 - #8-32  x 3/4" flat head ("machine screws")

A 0.25" or even 0.5" thick plate bulges considerably, making rounded, poorly compacted electrodes. 0.75" might be enough. I used a 1/2" piece, then after trying it I decided I should weld another 3/8" piece onto the bottom to strengthen it - total 7/8". For 9" long electrodes, the length of the box is 10". A 11.5" or 12" base length provides space on each side to clamp the box to the worktable with C-clamps, which is necessary. (Mine, at 11.375", could stand to be a little longer.)
[2] The extra 2" of length allows a holding bolt at each end "outside the box". When one flat-head bolt is removed, and one end piece (having one flat-head bolt) is removed, the side can be swung open to easily remove the finished electrode.
The press plate needs to press flat all the way across, and there should be virtually no side or end slack. Maybe allow a few mils to fit a sheet of thin plastic. I had to grind slightly rounded edges off my first press piece and the compaction area ended up about 2.9" wide instead of 3.0". The 3" end pieces were ground down to 2.9" to fit that before the bolt holes were drilled.
[4] I think this is thick enough. If the electrode isn't quite flat, switch to 1/2"... and probably to 1.5" long bolts. Also a 1/2" plate will press all the way to the bottom without a spacer for thin electrodes.

1/4", 20 TPI threading tap
7/16" hex socket (=11mm) for powerful, variable speed electric drill to do up bolts
Any spacers and smaller press plates desired for making smaller electrodes.

Naturally I'm only going to make one (more) compactor. There are more things that could be tried out, some mutually exclusive:
- fine thread 1/4" bolts instead of coarse (20 TPI), or other size bolts entirely.
- Other bolt head types besides hex, eg socket heads.

State of Nickel

   I don't understand why nickel sulfate or nitrate is used to make nickel hydroxide. If you just put nanocrystalline nickel oxide, NiO, into the positrode and charge the cell, when the OH- ion comes from the negative, would it not attach itself to the nickel oxide to make nickel oxyhydroxide, the same product as the regular charging reaction except no water byproduct?

 NiOHOH + OH-  <==> NiOOH + H2O + e-
 NiO + OH-  <==>  NiOOH + e-

   The nickel oxyhydroxide would then discharge to nickel hydroxide like usual, gaining HOH in the overall process, which could easily be made up or allowed for with extra water initially.

Nickel oxide is a very fine black powder. I don't understand why it isn't used, so I bought some to try out. Since it's sold over the counter at pottery supplies, it's much more accessible than nickel hydroxide, to purchase which I had to import it in a 10 Kg pail (minimum order), provide a company name for, and sign something.

The Price?

   The nickel is the most costly main ingredient in the battery, but even at $18 retail for just 113 grams of NiO (higher quantity would be somewhat better price - wholesale would be better yet), it worked out to 560 $/KWH:

18$ / .113 Kg = 160 $/Kg (...that's surely WAY over the wholesale price - Pure lanthanum rare earth metal ingots are less!)
[98 (weight of Ni(OH)2) / 72 (weight of NiO)] * 420 AH/Kg (for Ni(OH)2 with MnO2) = 572 AH/Kg.
572 AH/Kg * 0.5 volts = 286 WH/Kg
160$/Kg / 286 WH/Kg = 560 $/KWH.

   However, the nickel is only 1/2 of the of the battery. Manganese oxide is 1/3 that price. Of course, that doesn't account for other materials or the cost of manufacture, but it's also for small quantities of the active chemicals at retail price. Ballpark, I think these batteries should cost somewhere around the same price per effective kilowatt-hour as lead-acid batteries, with effectively 1/5th the weight and size and at least 10 times the longevity.

   Without going into details, logically it follows that Mn-Mn batteries should cost considerably less than lead-acid.

Victoria BC