Turquoise Energy Ltd. News #75
  April 2014 (posted May 3rd)
Victoria BC
by Craig Carmichael

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

Feature: Flat Panel LED Light: simple, low parts cost, looks nice (see Month in Brief, Other Green Projects)

Month In Brief (Project Summaries)

In Passing (Miscellaneous topics, editorial comments & opinionated rants)
* Prep Brochures, finances, gardening, aquaponics, the Spark of God Within

Electric Transport - Electric Hubcap Motor Systems
* Centrifugal torque converter - new version 1/2 made
* Bedini style bicycle motor-generator?
* Lithium cells are heavy?

Other "Green" Electric Equipment Projects
* Power Adapter Battery charging components printed circuit board
* Prototype flat panel LED light: promising design, simple, low parts cost.
* Thermoelectric Fridge & heat pumping experiments continued.
* Bane of Peltier modules: nearly all are rated 15 V max: 22-28 V gives much higher COP at 12 V.

Electricity Generating
* Flux concentration for magnet motors?

Electricity Storage - Turquoise (NiMn) Battery Project etc.
* Need pure graphite powder

No Project Reports on: Lambda Ray Collector, Magnet motor, Pulsejet steel plate cutter, CNC Gardening/Farming Machine (sigh, maybe summer... 2014... 2015?), Woodstove/Thermal Electricity Generator (may abandon), evacuated tube heat radiators, individual EV battery monitor (almost started on circuit board).

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

Construction Manuals and information:

- Electric Hubcap Family Motors - Turquoise Motor Controllers - Ersatz 'powder coating' home process for protecting/painting metal
- Preliminary Ni-Mn Battery Making book

Products Catalog:
 - Electric Hubcap 4.6KW BLDC Pancake Motor Kit
 - Electric Caik 3KW BLDC Pancake Motor Kit
  - NiMH Handy Battery Sticks, 12v battery trays
& Dry Cells (cheapest NiMH prices in Victoria BC)
 - LED Light Fixtures

(Will accept BITCOIN digital currency)

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

April in Brief

   I decided that April would be well spent if I made a 2.4v Ni-Mn battery cell that worked half decently without any notable self discharge, and made or at least progressed well towards making a new version of the centrifugal torque converter with a flywheel to give the "torque hits" the punch needed to turn the drive drum instead of slow the driving motor.

   I didn't get to the battery cell beyond thinking about it. The first idea was to use no graphite or carbon powder at all in the positive electrode, since impure powder (from an art supply store) is now the suspect in the self discharge, and accept a high impedance electrode. (The graphite felt and graphite foil, being made for battery use, are probably pretty pure.)
   Later I decided I should look up some supply and see if I could order some graphite powder of known purity, or else make some from the sugar technique. The big trouble with that last was, where does one get nitric acid? No doubt that too would be a special order from somewhere, not a local pickup. Later someone told me that you can make it from sulfuric acid (battery store) or hydrochloric acid (hardware store), so I'll look that up when I get there. Another potential for trouble is... the product is said to be 'pure', but might nitric acid not leave some nitrates behind? That's the very thing that causes self discharge. Perhaps there's some other way to make graphite. But it's all academic until I find time to dig into it.

   I started with the torque converter on the 4th, but didn't get far except for a theory and a 10" x 5" aluminum ring purchase. The theory went that many small slots would slow the motor less with each hit and reduce or even eliminate the need for a flywheel. The other part of this was that instead of having springs to hold the shoes back until the speed was high enough to want to engage, springs would push the shoes outwards against the drum at all times, in order to shove the toes farther into the slots than inertia would take them at any speed, and give torque hits with enough force to move a car. It would have to be low enough spring force that the motor wouldn't jam with the toes in the slots when everything was stopped.

   Then the fridge started cooling poorly (a peltier module had cracked - they're not well supported in my unit) and I diverted into re-examining peltier modules and how to use them to best effect. I re-examined using two stage cooling, but to cool by about 30° the highest COP still seemed to be a single stage run at much lower than its maximum voltage.
   Ironically the common 15 volt rated modules have needlessly poor COP at their typical voltage of 12 volts and a typical 30°c spread from the warm side to the cold side. People seem stuck on 'maximum pumped watts' and ignore the fact that they can do approximately as much cooling by driving them at lower power, owing to less internal heat being generated. It is especially surprising that camping coolers, hungrily eating battery power, shouldn't be made to be as efficient as possible. Modules rated for around 22 to 28 volts max, which are hard to come by, work much better at 12 volts.
   But I finally found some affordable "24 volt" peltiers, along with some "4 volt" ones, (from China with no datasheets, sigh!) and I placed an order. This was well. The cracked unit when reinstalled seemed to be working nicely... for a few days, then it quit completely. I put in a 6 amp module for the time being.

   In the middle of the month, I started thinking about the long overdue website that Jim Lawrence had created for me in 2011, but which was never quite usable. When it was started I felt it was best to let him do the rather complex coding, and anyway I didn't want to spend my time on it. But owing to the fact that the new site was theoretically being done, I stopped trying to update the old site, and everything got more and more out of date. Now everything was a snapshot of early 2011. After a last try to get him to do some updates which didn't appeared on the site after a couple of weeks, I finally decided I'd have to tackle it myself if it was ever to be made usable.
   It was a beautiful page format, but the coding was so complex that as I tried to change things, I soon had borders in the wrong places, missing pictures, and all sorts of headaches. After a second frustrating session of 3 hours, at 1 AM I put it up and linked to it anyway. At least I reorganized the menus, edited some text, wrote a little history of the development of the Electric Hubcap motor system, and put in a couple of newer pictures... one of which refused to show. The next day I got to the right hand column of the two column page. The content is still woefully far from updated and presenting things well, but at least it's on the move.
   I'll be taking Jim up on his offer of "any help I need".

   Getting back to the torque converter, cutting the small, thin slots - easily done with a bandsaw - into the inside wall of the ring/drum looked like a challenge.

Centrifugal Torque Converter drum with narrow slots,
mounted in the Chev Sprint transmission box.

   On the 17th I finally took an old, broken bandsaw blade (the sharpest one to be found in the bushes where they get thrown) and silver soldered it together looped through the 8" I.D. torque converter 'drum' ring, installed it on the saw, and cut 24 slots into the inside wall. For once, it all went more smoothly and easily.
   At first I thought I would try this smaller drum converter on the motorcycle and save the big one for the Sprint car. Since previously the motorcycle would only just start moving at 60 or 70 amps with a 4:1 chain reduction, getting a decent ride from a torque converter, at much lower amps, should at least prove the converter's efficacy.
    Then I started thinking about the walls: round plates, with pressed bearings in the middles, for both walls. The drum would turn independently of the shaft that would go right through it and turn the shoes rotor inside. A gear or pulley would attach somehow on one side. On the other hand... it would be simpler to try it on the car using the same configuration as before. If it wasn't quite there with the small drum, I could reconfigure it all and put it on the bike. If it did really well, I could put in another gear and drop the final ratio from 4 to 1 to 2 to 1, which should put it on the street RPM-wise, if not on the highway.
   I had to cut a big arc out of the side wall of the housing since this drum was too wide to go inside, and I ended up taking the motor apart to put a new and longer shaft in it. That was as far as I got for April - but it's good progress.

   Then, after a friend needed a battery charger adapted for NiMH cells, I did 'generic' printed circuit boards to put between AC power adapters or lead-acid battery chargers and NiMH (or other?) batteries to adjust the voltage and limit the current to what the adapter could handle. The component values will vary with the application, but at least they can all be mounted on a decent board instead of strung haphazardly on a wire. Not offering chargers has been a sticky point for those interested in buying NiMH batteries. This should simplify things.

   After that PCB and toward the end of the month, I decided to do the PCB for a flat panel LED light I'd been considering. Instead of one or two large LED emitters, it has a dozen or more smaller ones spaced around the board. Curiously, all of them together cost much less than single large emitters, and there are no especially costly parts except the 12 volt power adapter to allow 120 VAC operation. I made a nice 13 watt prototype light to test. I hooked it to the 12 VDC solar wiring. There are some things to change and improve such as to find a diffuser that absorbs less light, but with its practical simplicity, low parts cost and IMHO simple visual appeal, I see great product potential - especially as electricity costs rise.
   Some see more sales potential in getting specific colors of LED emitters and making them as grow-lights. It's not what I had in mind, but it might be pretty simple to offer both. Maybe I'll make a few and try them in the aquaponics setup. Year round lettuce would be great!

Flat-panel LED light, 1" x 6.5" x 7.5", 12 watts

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

Collapse warning pamphlets - finances - gardening and aquaponics

   I spent one of the first days of April writing a single page 'pamphlet', Are You Prepared for the Coming Collapse?, warning of the apparently inevitable financial collapse, which will surely lead to a period of global chaos and devastating supply disruptions. The media have kept silent about this biggest story of our time, perhaps hoping to delay it or avert panic. Who wants to hear it anyway? Few are old enough to remember the disastrous disruptions of the first half of the 20th century, and few will listen to an individual 'alarmist' or prepare, but they'll all remember who warned them and many may come knocking on the door when there is no food supply and none to spare. So those who are prepared or preparing are afraid to try to warn others.
   But the more who are prepared, the better for all. Thinking on this problem, I got the idea to distribute pamphlets anonymously, quietly dropping one or two here and there around the neighborhood, and hoping some would reach people who are listening. But somehow so far I've just put a few on park benches, on one evening.

   I've been burning through money with disquieting speed in the last year. (Why haven't I finished filling out my SR & ED investment tax credit forms yet?) Partly it's energy projects and some overdue home maintenance, and partly it's inflation. Fuel, various items of food, and even various materials and parts are up substantially.
   On the other hand, Turquoise Energy made a bit of money in 2013 - a pittance, but I trust the trend will continue and expand. Someone is buying an Electric Weel 'kit' as a large, low RPM generator. Another person (or two) is interested in an Electric Hubcap or Caik motor and controller for a boat. And since I think I have the answer to the last NiMn battery chemie problem - self discharge - there's the possibility of battery manufacturing agreements.

   One of the other projects is expanding the vegetable garden. I keep thinking of the CNC gardening/farming machine to make gardening easier, but without finding a moment to work on it.
   Then I saw some info about aquaponics and looked it up on youtube. Basically the fish waste is pumped around and fertilizes the plants, and the clean water goes back to the fish. It uses a few plastic containers piped together for the fish (tilapia are popular but need warm water) and vegetables, a water pump, and an aerator pump. It can be done on any scale and seems like a very good way to get both fish (protein!) and vegetables from a small area, with the inputs apparently consisting of fish food, vegetable seeds, and much less water than for a regular garden.

Spark of God Within

   Sometime around 200,000 years ago, a brilliant spirit being, administrator of several hundred evolving worlds, Lucifer, and his assistant Satan, decided to run things their way and to heck with the rest of the universe, preaching a doctrine of unbridled personal liberty. This caught the imagination of the spiritual overseers of about 3 dozen inhabited worlds - especially very primitive worlds like Earth was back then. After a very, very long while as we humans would view it and every opportunity to make their case to all, the rebels were incarcerated and replaced, but the 3 dozen planets have pursued stormy courses. 37,000 years ago Earth's rebel overseer, Calagastia, tricked the new physical representatives of higher order into trying a proscribed act - to speed up the painfully slow planetary progress by trying to produce a great leader for one of the more advanced tribes. (You guessed it: Eve had sex with an Earth native. Cain wasn't Adam's son.)
   Since then, we've been virtually disconnected at the humanly conscious level from all association with higher beings except for two brief periods. This planet has become so materialistic the very idea of there being some sort of spiritual oversight guiding our paths and planetary progress is thought by many to be silly, and the entire vital issue is little discussed and perhaps little considered.
   But the Lucifer Rebellion was, at long last, evidently adjudicated in or about 1986 Earth time. Only the rebels' way of thinking has been left behind. I wrote quite a while back how our civilization appears to have hit an evolutionary dead end, how the entrenched have seized power, and have politically and economically disenfranchised the inventive and those who would bring progress; how society thus can't make the self corrections and changes that need to happen to continue evolving towards a bright future.
   The turmoil now starting to make itself felt in many lands is the beginning of a wrenching and rather rapid evolutionary adjustment back towards the normal paths of love, life and light. This program planned on high will gradually result in the changes in attitudes and perspective of most individuals that will make humanity receptive to major changes in social systems and structures. Future generations will be contemptuous of 20th century civilization, and the social order won't settle down for a long age of moral quickening and progress.
   The program will culminate, perhaps long after all now living have moved on, in the adult physical appearance of a magisterial son, Serara, a spiritual relative of Michael of Nebadon (AKA Christ Michael). He may perhaps stay for some human generations, suggesting ideals of social and political organization - as leaven in the bread to foster human progress.

   It would seem that one of the key things if not the one key thing, that the celestials presently in communication via various volunteer contact persons and associated e-mail lists would have us understand, and which far too many are unaware of, is that every person of normal mind has within his or her mind a spirit fragment derived from the first source and center of all things, the uncaused cause, God the Father. By faith one can come to know and experience that this is true - through prayer and meditation, seeking, receiving and following the guidance of the still, small voice that ever says "This is the way!" This is the doing of the Father's will, which leads us to do better and to gradually become more divine, even to be all we can be.

Electric Hubcap Motor Systems - Electric Transport

Centrifugal Torque Converter

   In considering the previous torque converter attempt, with 5 very large torque hits per rotation, the motor was greatly slowed with each hit. This meant transference of much of the impact force to the motor, the transmission housing, and the body of the car instead of to the output drum. Again, it was like driving a spike with a tack hammer. The remedy to smooth out the force would be a fairly substantial flywheel on the motor shaft. With that, the shaft would probably have started turning instead of just the car shaking. I considered buying a new identical steel brake drum and making 10 smaller slots in the wall for somewhat smaller hits twice as often, and adding a flywheel, just to prove the point. This seemed like a lot of effort and again there was just nowhere a heavy flywheel could be conveniently added, just for a test.
   I then considered the clutch - centrifugal or otherwise - where the torque of the motor is evenly applied, turning the output shaft... and making heat. No flywheel is needed owing to the steady, even forces.
   If there were enough very small slots in the drum, there would be many small torque hits, and the weight of the motor rotor and the shoe rotor driving the drum could be sufficient flywheel weight for them. Even if no single hit was sufficient to move the vehicle, the rapid succession of lighter hits might do it, just like the 'infinitely' rapid succession of very minute hits in a clutch.

   A 'new' thought for the torque converter was that the 10" I.D. x 4.5" aluminum pan drum could mount on a car wheel, making the originally planned Electric Hubcap configuration, perhaps more easily than the Sprint transmission mounting could be modified. But that would mean that the converter would have to move the car with no following down gearing. But I hadn't been successful yet even with a 4 to 1 reduction following - why build in an extra challenge?
   Another idea that occurred to me was that with the smaller 8" I.D. x 3" pipe, I could try making a converter for the motorbike first, and then apply any lessons learned before hacking up the larger one. Since previously the motorcycle barely started moving at 60 or 70 amps with a 4:1 chain reduction, getting a decent ride from a torque converter, at much lower amps, would surely at least show the converter's efficacy.

The E-Hog, a locally (Victoria BC Canada) converted motorcycle, has a considerable
motor and a 4.125:1 reduction with a huge rear sprocket gear. The fixed ratio works
well once the bike is moving at street speeds, but in common with my attempts it has
rather low torque from a stop, making for slower take-offs. It's a fine bike as built,
yet it's another illustration of the desirability of creating a really effective variable
torque converter to handle all speeds optimally.

   I marked off the inside of the drum with 24 even lines about an inch apart. This convenient number would mean 24 torque hits per rotation, and the strength of the hits could be varied not only by the size and shape of the slots and shoes, but by the number of shoes, which could be evenly distributed if there were 2, 3, 4, 6, or 8. (There wouldn't be room for 12 or 24.)
   On the 17th I finally took an old, broken bandsaw blade (the sharpest one to be found in the bushes where they were thrown) and silver soldered it together looped through the 8" I.D. torque converter 'drum' ring. I installed it on the saw and cut the slots in the inside wall. Then I unsoldered the band to free the drum. For once, all went smoothly and easily and I wondered - except for having too many things to do - why I hadn't done it 10 days before.

Cutting the slots/slits in the drum.
This was unusual not only for having to put the band together inside the drum,
but for having to work from behind the saw to observe the cutting.

   I had the idea to put a long shaft on the motor, long enough to run through the converter, with the center disk with shoes attached to it, and the grooved drum and output sprocket or pulley on bearings to turn independently but perfectly centered on the shaft. Then I started thinking about the drum walls: round plates, with pressed bearings in the middles? The gear or pulley would attach on one side. It would have to have side clearance and shaft clearance and yet be perfectly centered. It would almost need a bearing on the outer end...
   As I thought about disassembling the motor and the considerable reworkings and remountings to make this, I decided to try this smaller drum in the Sprint car after all, with the same basic configurations as before. Faster to do. If it didn't quite 'have it' in the Sprint, I could try the other route and redo it for the bike.

   On the 24th I had my plans, but I decided to run them by the people at AGO. Well, nix on welding steel to aluminum! (I should have known - the two metals have completely different melting temperatures.) But for an drum end, I had happened to pick a used 10" rotor disk with six threaded mounting holes (from the first centrifugal converter). These turned out to be exactly outside of the drum diameter (wow, what luck!), and so could be used with long bolts and hooks to clamp the cylinder to the disk. (I used some small angle brackets, with one leg partly cut off, as hooks.) The machinist from AGO came over after work and we figured out a simple way to mount the rotor - any rotor with an SDS taper lock shaft bushing center - for turning on the lathe, where I dug a groove into one face to hold the cylinder centered. I didn't however turn it down to the size of the drum per the original plan. That would have turned off the mounting bolt holes. (The 10" rotor had about 1/16" clearance over the lathe bed - whew!)

   Next, the motor and transmission mounting box was too narrow to accommodate the 3" wide drum. I cut out a 10.2" diameter arc from one side for the drum (or even a 10" drum) to protrude through. It was well I cut it larger since I hadn't thought about the clamping bolts sticking out beyond the 8.75" diameter.

   To cut this 1/4" steel plate, I used a jigsaw with a metal blade, at quite a slow speed and pausing frequently to add more oil to keep the blade cool. It took a while, but I made a rather clean 10" long cut, with one (new) blade. The angle grinder with zip disks would surely have cut it faster, but I'd have spent more time afterward cleaning up the ugly inside-curved edge with a grinder. And it would have still been rough. Cutting a whole 7.5" Electric Caik motor rotor would be a 23.6" cut - tedious but doable with the jigsaw in a pinch, if you don't want to wait for waterjet people to get around to it or it's not available. The 10" Electric Hubcap rotors are 5/16" thick and a 31.4" circumference  cut - probably doable too, but would probably try anyone's patience and might consume several blades. But I digress.

   I didn't find a 1-1/16" center "H" coupling, and I finally gave up and took the motor apart to change the shaft to a 1.0" one. (I did learn about "boring bars" for lathes, and found there's a taper lock bushing size smaller and seemingly better than "H", labeled "JA". I may employ this size in the cramped quarters of the next Electric Caik motor.) I made it a very long motor shaft, not only to reach to the drum shaft through a wider drum, but also so a flywheel could be mounted on the other side of the motor if one should be required. Then I got onto other things and that was as far as I got, with the inner rotor and some chunks of plastic for the shoes lying on the bench.

   I plan to drill holes in the end of the motor shaft and the drum shaft, and insert a pin between them to force them into alignment.

"Bedini" style bicycle motor?

   With most 3-phase permanent magnet motor systems, the magnets are in a balanced configuration, two poles, north and south, for every three coils. In the 1970s John Bedini made a motor with unbalanced magnets, and he actually extracted electric power from it, charging a battery, even as it ran a mechanical load. It's said Bedini was beaten and threatened ("You'll use oil for the rest of your life or else!") and subsequently did little work, but with his coaching his system was demonstrated in the 1990s by a girl at a school science fair (the "Bedini SG"), confounding everyone.

   I haven't investigated this in detail, but as I understand it, the process goes something like this: Two like magnet poles are used at opposite sides of the motor, instead of one north and one south pole, and two coils. The iron coils and magnets are naturally attracted to each other, and being unbalanced, there is a half of the rotation that powers itself as the magnets approach the coils and they pull together. Furthermore, during this part of the cycle, the coils can be connected to an electrical load and generate power into it.
   When the coils pass the magnets, electricity is applied for a short period to make the repulsion to push them apart and continue the rotation. Sometime before the midpoint is reached, the coils are switched back to 'generate'. The energy stored in the magnetic field from the powered phase is released to the generator as the field collapses. This is where energy used turning the motor is returned. After that it's back in the regular attracting, 'generate' part of the cycle. The mechanical work done is less than the electrical input applied, and the electricity generated is also less, but evidently together they add up to more than the electrical input. It probably depends how hard the motor is working, but even under heavy load it's likely to consume less power than a typical motor system, owing to there being at least some returned electricity.

   I still have the bicycle wheel rim motor in my mind, and it has a lot of freedom of potential design at this point. Furthermore, since the bike can be started by peddling, having torque at all points of rotation isn't a strict requirement. Of course, a bike motor that uses less electricity or even keeps itself charged would be fantastic.
   I'm also not entirely convinced that the results can't be achieved with a regular permanent magnet BLDC motor, using a special motor controller and perhaps two separate wires to each phase - neither delta nor wye wired. The Electric Hubcap and Caik certainly have positions they magnetically pull themselves into - "cogging" as it's called. By hand it's hard to turn the shaft to break it out of these positions.
   I intend to study this further and then play with motor and controller design ideas for a while before actually building anything.

Lithium Ion Batteries are Heavy?

   When I got the eleven 'Thundersky' lithium-ion cells home (36v, 100 AH, 3600 WH) I weighed them as they seemed quite heavy. The bathroom scale said about 93 pounds. When I disassembled them, one cell read 3684g. For all the hype about the high energy density of this chemistry, that's just 89 watt-hours/Kg, not 140-170 per the usual published Li-ion figures. And I would mount them in some kind of box, so they'd work out to around 80. Probably the heavy plastic cases to protect such large cells accounts for much of the 'extra' mass, but it shows that the 60-66 WH/Kg I get for NiMH D cell assemblies is certainly not "eclipsed" by fabulously lighter weight with lithiums.

   There were 11 cells total. To get "18 volt" batteries I could use 5 and be perhaps a little low in voltage (16-19 empty-full) or 6 and be rather high. With 5, I'd get two sets with a cell left over, so that was the obvious choice. For boating, one set running low would indicate it's high time to turn around and head home using the other set. I got two tote boxes to put them in for in the boat, which measured slightly too small. Sure enough, slipping a battery into one it cracked in one corner, so I ended up with a totes considerably larger than required.

   The left over cell could go into the Mazda as a token lithium cell to bring it up to a nominal 135 volts. The Mazda would then have all three typical battery chemistries: PbPb, NiMH and Li-ion. 3-1/2 volts won't add much to the performance, but it should allow for some comparison between types... and add a challenge to making and programming the individual battery monitor.

   With a voltmeter, I noted that the cells weren't at equal voltage. Two were substantially lower than the other nine (2.4+v versus 2.7+ for the rest). When I connected a set directly to a 90w solar panel four were soon over 4 volts while the fifth was still down at 3.6. Evidently a charge controller/Battery management system is important even for a few Li-ion cells in series. Hmmf! I got an 18.5 volt, 3.8 amp power adapter as a charger, and some 3.9 volt zener diodes at the local electronics store to put across the cells to prevent excess voltage. But they're only 2 watts and would probably burn out with the solar panel and charged batteries.

Electric Mazda/Lead-Acid Batteries/Mixed Batteries Update:
battery watering, constant voltage float charging, pulse charging

   At the electric car show in February, Canadian Electric Vehicles showed a Mite-E-Truck dump truck with a lead-acid battery watering system installed. I've heard of these, but they seemed a bit of a luxury item to me. On the 13th it occurred to me that I hadn't checked the water levels in the Mazda batteries in a long time.
   Presently the Mazda has two 90 AH NiMh batteries composed of 90 D cells in tubes, one 100 AH NiMH battery as a soldered pack of 100 D cells (two stacked wooden boxes of 50AH), 5 'regular' deep-cycle lead-acid batteries with two pop-off triple cell covers, two sealed lead-acids, and one lead-acid with six screw-off cell caps.
   Nothing could be done with the NiMH dry cells and the sealed lead-acids even if it was desirable, leaving six.
   I had never opened the one with screw-off covers. The cells all looked as dry as a bone. Perhaps it's supposed to be that way, but I added 100mL of water to each cell to get the level just over the plates, which bubbled air as I filled. (It worked well afterward.)
   The five 'regulars' were still full to the max after all these months. Obviously a battery watering system here would be a waste of time. This seems like a vindication of sorts for my gentle 'constant voltage' (13.8 to 14.0 volts) float charging system. After an initial fill, watering is virtually a non issue.

   However, the glowing reports of the retention of performance and longer life span afforded by "pulse charging" (or perhaps "supersonic charging"?) at last led me to buy a pulse charger. It's a pretty slow charger that reduces the pulse strength as the battery charges, and it seems to me it would do as well (for lead-acids only) as my float units, and keep them in optimal condition. The only trouble with making them the regular PbPb chargers is that each one costs more than 100$ - substantially more than a "reconditioned" battery. I rotate it around every couple of days to occasionally "pulse" each of the several PbPb s in the car.

   That didn't stop the one battery that was in the car since the beginning, for a year, from losing mileage rapidly over a few days, in spite of pulse charging it. (...bought in 2009 or 2010 and it had sodium sulfate added when new. It sat around a lot until I got the Mazda.) I removed it on the 24th and the car was down to 120 volts again. I caught on fairly soon and I don't think I ruined it with over-discharge like I probably did to some others (must do that individual battery monitor!), and I intend to try renewing it.

   I found all the tire pressures were down a little, 25 to 32 pounds, and I filled them to "max", 35 PSI. Following this, and also with warmer weather, the amp-hours needed to go a mile went from 2.2-2.5 down to as low as 2.0, in spite of being short a battery. Wow!

Other Green Electric Equipment Projects

PC Boards to make AC Power Adapters into Battery Chargers

   When I speak of nickel metal hydride batteries, people ask about charging. I usually say I just charge them with a constant voltage float charge, but they seem to want more detail, like a specific charger and a price.
   The simplest way is to use a power adapter, generally followed by a diode or two to isolate it if there's no power and or to reduce the voltage to a desired figure, and a resistor to limit the current when the battery is low, followed by the battery. If the driver is a "real" battery charger being adapted for a slightly lower voltage, usually from PbPb to charge NiMH instead, it may need a sense resistor bypassing the diode in order that the charger senses the presence of the battery. Various component values are necessary for different setups. A fuse might be a useful addition.

   Finally it occurred to me to make "generic" circuit boards that can accommodate the common arrangements. Most power adapters have 'typical' jacks on them, so using the matching socket means no modifications to the adapter; just plug it into the board. For the battery end, I just used big solder pads to attach any cable. On the board are spaces for a fuse, one or two 5 watt resistors or one 10 watter, one or two large diodes, and a sense resistor across the diodes.
   I designed boards on the 26th, printed a sheet with 12, and etched 3. The patterns were large, but there was some pretty fine text in the copper, which came out more legible on one board than on the other two and showed the limits of resolution of the iron-on system.
   I found a couple of layout problems and I did a second board design on the 27th, but haven't made any so far.

   Now, for example, to charge my 18v lithium batteries, I have an 18.5 volt, 3.8 amp power adapter. If one plugged any such typical adapter straight into any appropriate batteries when they were low, it would be overloaded.
   I'll omit the diode on this one since the voltage is just right as it is. No sense resistor is needed. Since the cells may go down to about 16 volts: 18.5v-16v=2.5v. A .66 ohm resistor would thus limit the current to 3.8 amps when the cells are 'fully' drained. So I'll append on this one's circuit board just the power adapter socket, a .68 ohm*, 10 watt resistor or perhaps two of .33 ohm*, 5 watt, which I may have on hand, and a fuse. The solder pads will connect a cable with the battery's charging plug on the end.

* For the uninitiated, the reason seemingly odd resistor values are commonplace is that each one is 20% higher resistance than the last starting with one: 1.0, 1.2, 1.5, 1.8, 2.2, 2.7, 3.3, 3.9, 4.7, 5.6, 6.8, 8.2, 10, 12, 15, ... and as electronics manufacturing got better there came to be 10%, 5% and even 1% values in between.

Making a Prototype LED Flat Panel Light

   After I had done the first PCBs for battery charging on the 26th, I decided I should tackle my conception of a flat panel LED light using a number of small emitters apparently to be run at 1 watt: 2.9V, .3A, and 90 lumens. If it had 4 sets of 4 LED emitters in series (theoretically 11.6 volts), it should work at 12-15 volts, draw a little over an amp, and put out 1440 lumens. Using a bunch of small emitters is actually substantially cheaper than using a couple of big ones - they come in packages of 20 for about 5$, versus using two big ones for around 10$ each. And they're more amenable to PC board mounting, since the heat is spread out among the emitters and across the board area.

Note: One 'failure mode' drawback of my circuit is that the current is constant, so if one LED in one of the 4 legs fails, that leg will go dark, but the other 3 legs will be driven too hard and will brighten - until they too burn out. Well, best not to drive them hard enough that any will fail.

   I simply divided the main area of the board into 16 rectangles, one per emitter, each forming a big slab of copper insulated from all the other slabs by a small gap. This would do the maximum of heat sinking without adding components. Since my budget version of Eagle PCB layout wouldn't do a big board, I copied my latest LED driver board into one corner, and did the big blocks in a paint program.

   Of course I couldn't very well do a foot square board with the laser printer. 7.5" x 10" would have been about the printer's limit. I picked 6" x 7". When I printed it on the glossy paper, I could see that interiors of big squares don't have much toner. (It wasn't too noticable on regular paper.) This translated into blotchy or missing areas in the artwork when ironed on to the PCB. This is the first board on which the iron-on transfer method hasn't worked well since I learned not to rub the iron back and forth but to lift and move it, and to be really careful to press near the edge of the iron down on each and every point of the surface... but then I did start rubbing it too, owing to the large board size. I used packaging tape to cover the rectangles up for etching. That was a lot of work to do an obviously hand-made job. Production will probably need another method. Or perhaps non-solid line pattern squares can made -- which may print out better?
   Later it occurred to me I had another laser printer. It might do better. And I hadn't even set my usual printer to "darkest" - although its adjustment range is slight anyway. And then that it should be a good board to do with a dremmel router on the CNC machine, because most of the copper is retained and only strips need to be removed. I must make a holder to mount the dremmel on the CNC machine!

   I etched the board on the 28th - and I left it in too long and over-etched an already shoddy looking piece of work. I soldered on the 16 LED emitters and tried it out with the lab power supply. The brightness seemed good, but it took about 13.5 volts to get full brightness, instead of 11.6. Once again, it looked like "2.9 volt" LED s aren't necessarily 2.9 volts, or even close to it. Full current would take at least 14 volts with the driver voltage losses. Evidently I could only put 3 in series instead of 4, making a board with 12 emitters and 1080 lumens - or just accept that it wouldn't be entirely full brightness except when the sun was shining on the solar panels.

   Then again, I thought, the 16 emitters might total 1080 lumens or more even when it wasn't full brightness, and each emitter would be driven more gently. 900mA (225mA per leg) * 16 emitters, gives a value of 14.4. The full 1200mA (300mA per leg) * 12 emitters = 14.4. But the light value of the 16 is greater because the majority of the brightness comes from the first half of the current - the second half gives somewhat less. And emitters driven with 225mA will run cooler than with 300mA. But when I tried it, it still took over 13 volts to get 800mA. So much for that idea!

   On the 29th I removed one LED from each string and put wires across the gaps. As might be suspected, it now took about 10.5 volts to reach 1200mA (300mA per string). When the internal constant current power supply was wired and tested, 12 volts was exactly the lowest voltage that made full power. Power supply efficiency would be 10.5/12=87% at 12 volts and 75% at 14 volts. A good switching regulator could do better than the linear for 14 volts, but at 13 or less there'd be little point to it. If there's 14 volts it's because the system is being charged (ie, for solar the sun is out), and a few extra percent power to lights will hardly be noticed. I went back to the supplier website (dx.com) and found more exact specs for the emitters, which I hadn't discovered before. They still appear to underestimate the 3.5V forward voltage drop, but even the "3.2V" figure says four won't work well at 12 volts.

- Material: Plastic + copper
- With LED white light
- Power: 1W
- Voltage: DC 3.0V~3.2V
- Current: 300~350mA
- Brightness: 80~90LM
- Color temperature: 6300~6700K
- Contains 20 pieces a pack

   I still thought maybe a lower current in the emitters would be preferable. 5 strings of 3 LED s instead of 4 strings, run at 250mA each instead of 300, would mean a 1250mA total supply instead of 1200, and 15 emitters emitting at this somewhat lower current would probably be brighter than 12 at 'max'. Also on the 29th I got a piece of 3/16" 'frosted' plexiglass. It was actually pretty clear and probably let nearly all the light through, but one could see the outline of each emitter as a bright spot. If it was moved a couple of inches away, it was somewhat more acceptable. Somehow that didn't quite fit my vision of a flat panel. But maybe a "light box" would be a good adaptation.

   And herein came an idea with a new advantage: The commercial flat panel lights all had aluminum edges, and that had been my plan all along. But I could glue acrylic edge pieces, any height, to the acrylic front, and skip the metal entirely. These sides could be slotted to slide the circuit board into, and I could glue acrylic tabs to the edges for screwing the unit onto a ceiling or wall - a feature lacking in the aluminum edged lights. Most of the whole unit would glow: light would come out the sides as well as the face. The whole unit would be substantially simplified, the materials costs would be low, and it would look nicer and give a bit more light.

   I cut the sides of the box in white on the radial arm saw, but discovered the white scrap plastic I used wasn't acrylic - it wouldn't bond to the frosted face with methylene chloride. or probably with anything else. Rats!

   I visited the plastic shop the next morning, the 30th. A translucent milky white acrylic for all faces would be great. They only had 1/8" thickness of that unless I had it specially cut, so I got that for the bottom and some 1/4" thick "bronze" acrylic for the sides. I asked about the light properties of the translucent acrylic, and found one more and a serious inefficiency: the 1/8" was said to absorb 50% of the light, and the 1/4", 70%. (They didn't have 1/16".)

The flat panel LED light, mounted with a couple of screws

   The "bronze" turned out to look unexpectedly like "clear" with the bright LED s inside, and they glare out from the sides. Then there's the ugly circuit board with the missing LED emitters making for an uneven light pattern. However, it's a prototype for a potentially excellent product, which I will improve on for the next rendition... and a nice light regardless, which I will use. I anticipate these will be both less costly and more desirable than the LED globe lights have proven to be so far - they're another new form interior lighting can take. But rising electricity prices will make all LED lights more popular.

   For testing and temperature readings I mounted it in the solar PV system wiring closet where there was already a 12V CAT plug-in and it was within easy reach. I'll be looking for a bottom cover that'll transmit more light without allowing the glare of each emitter to show through.

The prototype flat-panel LED light printed circuit board layout

Thermoelectric Equipment - Experiments

First, the Conclusions...

   There's a mishmash of stuff below, essentially my 'lab notes' written at the times I was doing the paper research and the experiments. So here are the main findings, which most may not care to read beyond.

1. I couldn't find 'nanomaterials' Peltier modules with improved COP s for sale so far. There are a number of patents for them.

2. Laird has some 'improved'(?) peltiers rated for higher voltages for sale at electronic suppliers (Digikey, Mouser) that may be a little better than most - but they are costly. DX has some small (15x30mm), '4 volt', 5 amp Peltiers that can add their voltage to a 15 volt (5 amp) unit to raise its voltage rating, as well as a '24 volt' 6 amp unit and several '12 volt' units, for affordable prices. Since DX provides no datasheets with any of these, use is somewhat speculative. Undoubtedly the '12 volt' units are rated for 15 volts max. like all the others. Is the '24 volt' one really 30 volts? I ordered some anyway. They haven't arrived yet.

3. Owing to heat rise on the module warm side and the need for the cold side to be below freezing to freeze an ice tray, the temperature spread is likely to be around at least 30°c with a 20° ambient room -- and higher if the Peltier module generates much extra heat owing to excessive input power, or in warm summer weather.

4. To get a good coefficient of performance , Peltier modules should be run at around 50-65% of their maximum rated voltage. For a 12-14 volt system, modules rated for maybe 20-28 volts, or combinations of modules to attain these voltage ratings, use much less power (than typical 15v max. units) to provide almost as many watts of cooling.

5. Most Peltier modules, and nearly all the lower cost ones, are rated about 15 volts maximum, intended to pump maximum watts of heat with 12-14 volt systems. But this is an unfortunate choice for typical 12 volt cooling and heat pumping. The extra power used by the lower COP shows up as extra heat in the warm side, so except with noisy high rate fans the temperature differential is greater and it actually consumes those extra watts to pump the same amount of cooling to the cold side. Typically it appears there's no cooling gain at all over using a higher voltage module(s) at 12 volts with lower current (and hence consuming less power) to attain higher COP.
   In my fridge, power used with a single 15V, 8.5A module was about 85 watts, and the warm side temperatures were exceeding 40°c. For a 15V, 6A module, using 60 watts, the warm side was about 35° and the tray would only partly freeze over. It has a quiet fan, but a very good heatsink. With two 8.5A modules (15v+8v) making a 23V rating, power was 45 watts, the warm side was just 30°, and the ice tray froze as fast or faster as with the 85 watts.

6. Thus 15 volt units - almost the only ones available - don't take into account the need to conserve power in battery operated systems, energy conservation and efficiency in general, nor the extra heat that will raise the temperature of the warm side. Of course, they could be run at 7-9 volts quite effectively, but for 12 volt power that introduces power supply issues. At 6 volts of 15, the cooling capacity gets rather low with a typical temperature spread, so using two peltiers in series doesn't appear to work out very well at 12 volts either.

7. Peltier modules pump heat more efficiently across a low temperature differential, so I thought a two-stage unit with one Peltier on top of another, each pumping across 15° instead of 30°, might give a better COP.
   The problem is that the warm side module has to dissipate the heat made internally by the cold side one as well as the heat pumped by it. So far, for a 30° spread, it looks like things work out pretty much the same with either single or two stage cooling.
   However, this is not the last word. I've ordered more modules to get more balanced pumping between the stages and try more actual experiments. Sometimes there are things that don't show up in the datasheets - for example, when the unit is shut off, less cold will be transferred from the fridge to the warm side heatsink through two layers than through one and so it'll stay cold longer.

8. A technique recommended to maximize thermal conductivity is to metalize the warm side of the peltier module and solder it to the heatsink. Since graphite seems to conduct heat very well, an experiment I'm trying is to mix silicone heatsink grease and graphite powder, to get a much better thermal grease. Ideal contact with the heatsinks could lower the temperature differential at the peltier surfaces by a couple of degrees - and maybe double that for two stage coolers. It seems to be working well, but spreading it thinly and evenly is still vital - the difference is probably small.

Thermoelectric Fridge Update: another cracked peltier module

  I got on the web to see if perhaps anyone had started making higher performance peltier modules, and I read that they have a life span of some sort, with deteriorating performance over time. This was the first inkling I'd had of that, which would mean occasional module replacements would have to be figured into the cost of peltier driven appliances. At the same time seemingly just to emphasize the point, the fridge started working worse and worse, and after a couple of days, it would no longer even make ice, just cold water. (The copper bar right under the modules remained iced over.)

   I had been thinking I might change it from 15+8 volt peltier modules to 15+4 volt ones to get faster freezing of the tray, but I had only 15v and one more 8v matching modules. I thought of trying 8+8 (was that really better than 15?), and I took it apart on the 8th. But a corner of the 8v peltier was cracked. That was of course the reason it wasn't working well. I thought I had again clamped them too tightly, but on further reflection, they don't seem to deteriorate immediately, and pressing on a corner of the heatsink would stress out the edges and corners of the peltiers even if they were clamped loosely, potentially breaking them. It would seem I need to ensure that external stresses don't couple through to the modules, or they'll continue to get 'randomly' broken.
   The choices were to use 15v+8v [the other 8v], 15v+15v, or a single 15v. IIRC 15+15 didn't make enough cooling, and the large surfaces caused excessive transfer of cold to the heat sink when the unit was off. I put it together as it was, 15+8 with the spare 8, and considered getting more options (8v, 4v, 2v) next time I ordered electronic parts. It turned out after some perplexing experimental results, when I finally read the numbers on the part, that the "spare 8V" unit was in fact a 12v unit with the same smaller dimensions as the 8v one. I didn't have a spare 8 volter.

   I haven't found any nano layered materials, higher COP modules - so far. No doubt the lifespan of uncracked peltier modules is at least several years -- and surely much longer if used at around 1/2 the maximum ratings instead of up near the limit.

The Two-Stage Peltier Cooling Module Revisited

   In reading I was reminded that efficiency of Peltier modules increases rapidly with decreasing temperature differences. (The piece suggested that they might be used in conjunction with compressor and gas heat pumping to increase the efficiency.) I had early on concluded that using two stages instead of one didn't make much difference... but had I discovered the 'deceit' of the graphs when I estimated that? And might the result also be improved by the efficiency increase gained by running the modules at well under maximum voltage? It seemed to deserve another look.

(The document said:



Some observations on thermoelectric technology, especially those relevant to large scale air conditioning and refrigeration.

* Thermoelectric modules are solid-state electronic devices that directly convert electricity to temperature difference. Thermoelectric devices have no moving parts and therefore are inherently reliable and require little maintenance. Furthermore, the lack of refrigerants used in the systems provides many benefits to the environment as well as for packaging and safety.

* The use of thermoelectric devices and systems has been limited by their relatively low energy conversion efficiency. Present commercially available thermoelectric devices operate at about 10% of Carnot efficiency if used as home refrigerators, whereas compressor-based refrigerators usually operate at about 30% of Carnot efficiency.

* A broad search for thermoelectric materials with high efficiency has been conducted. Currently, there is no known theoretical impediment to significant increases in thermoelectric energy conversion efficiency. A breakthrough in thermoelectric materials could spark many applications that use thermoelectric technology as a safe, efficient, and reliable alternative.

* Thermoelectric technology is suitable for applications where its compact size, reliability, absence of moving parts, and silent operation outweigh its relatively low efficiency. Thermoelectric devices have been used in situations where the heat load is small (e.g., <25 W), the required temperature lift is small (e.g., <10°C [18°F]), or the variation of the heat load is large (e.g., train passenger cabin). It is important to note that the COP of thermoelectric modules increases significantly with decreasing temperature lift, as shown in Figure 4. [figure wasn't there]

* Instead of utilizing a fully designed thermoelectric cooling system, it is also possible to use a small thermoelectric system as a subcooler to improve the performance of a traditional system. This is a "hybrid system" since it combines a solid-state cooling device together with a conventional vapor-compression-type air conditioning and refrigeration.

   Now, if one took the 30° spread for the fridge or a heat pump, and divided it into two 15° stages, being run (especially the second stage) in the peak efficiency area per voltage and current, could a higher COP be obtained even with today's 'low performance' modules? If a COP of 2 could be obtained, a warm side peltier delivering twice the cooling could sit directly on top of the cold side one running 1/2 the power (assuming they were the same size or separated by a block of copper to fully contact both surfaces), and the temperature drops should be about equal.
   I dug into the CUI module datasheet graphs. Again the graphs were deceptive and difficult to evaluate, since they chart watts of heat transfer versus temperature for different supply currents instead of for different supply watts. 1/2 current (all else being equal) also means 1/2 voltage and hence 1/4 power, not 1/2 power. I finally printed out a couple of the graphs and started scribbling numbers on them.

   My estimate, which can only be a rough one, was that four 15V modules could be used to attain better COP. Two "8.5A" modules (40mm square), electrically in series and hence being run by 6 volts each, would sit on top of two "4A" or "5A" modules (same dimensions, same wiring) and at 12 volts deliver around 40W of cooling power. Total supply current would be 5.8 amps, for a 70 watts supply draw. (At 13 or 14 volts, all the figures would rise a little. I didn't work it out.) A single 8.5A peltier would use about 80 watts to make about 35 watts of cooling. So the improvement seems to be there: more cooling with less power. And there might be somewhat less heat leakage from the cold to the warm side when the unit is off.

   The fridge at present probably has only about 25 to 35 watts of cooling, but it would be nice to up that so it can cool faster while the sun is out, so 40 watts would be better. Otherwise, the peltiers could be reduced to say 6A and 4A devices, or 5A and 3A.

   Is it worth it? Maybe. Certainly for a heat pump in a battery powered electric car, all savings in power are valuable.

   Logically the next thing to try figuring out was how two sets of three stacked peltiers, each dropping 10° to make the 30, would fare. A quick look showed that with the small change in temperature drop, the 6 modules would use more electricity than 4 and pump less heat. If the temperature drop was, say 45°, three stages might become worthwhile.

   Being lazy, I stuck an 8.5A and a 6A peltier together, connected them in series, and tried them out. The voltage balance looked reasonable and I installed them in the fridge. I got the following rough readings:

Supply: 12.25 v
8.5A peltier: 5.20 v
6.0A peltier: 7.05 v
Current: 2.7A (33W in)

Or (unplugging the electric car so the voltage came back up):

Supply: 13.7v
8.5A (warm side) peltier 5.9v
6.0A (cold side) peltier 7.8
Current: 3.0A (41W in)

   Next question was how much cooling would that provide? If it made 20 watts of cooling it would be doing well, and that probably wouldn't be enough. After a few hours the copper bar had frost, but it took ages to make ice in the tray, and the coolest area in the fridge was about 7°.
   The next morning, the 9th, I tried wiring the two stacked modules in parallel instead of in series. With 12.4v in it drew 9.5 amps - 120 watts, and probably making about 30 to 40 watts of cooling to the fridge. That's a pretty poor coefficient of performance and at best no more cooling than the theoretical two pairs at half voltage using 70 watts. Furthermore, it's probably an overestimate, because with so much power going in, the heatsink temperature rose from the usual 27-33° to a high measure of 46°, negating the intended lower temperature difference effect. In fact, it made so little cooling the ice melted off the copper bar.

   It's frustrating that nearly all common (cheap) peltier modules are rated ~15 volts maximum. This would be of course to get maximum cooling from a 12 to 14 volt battery supply. However, a much better COP from those voltages is attained from peltiers rated about 20 to 24 volts. These seem to be considered specialty products and cost far more. A DC to DC converter might take the voltage down to 7, 8 or 9, but that introduces its own inefficiencies and complexity. That leaves putting two modules in series, which at 12 volts leaves them running at just 2/5 of their rated voltage, where performance is marginal for a 30° temperature spread. 1/2 to 3/5 is probably ideal.

   In the afternoon I put in the "full deal", two stacks of 2 peltier modules, 8.5A and 6A. (I'd have gone for 8.5A and 4A or 5A, but I only had one, 5A size.) Current was 5.1 amps, about 67 watts. After temperatures stabilized, they read 37° - 25° - 0° (warm side, between peltiers, cold side). It seemed the extra watts had mainly just raised the high side temperature, and the 6 amp units were somewhat too large to go under the 8.5 amp ones. Owing to the unequal temperature distribution it appeared there would be only around 26 watts of heat pumped from the fridge instead of 35-40. In mid afternoon the water temperature in the ice tray was 5° at the top and 3 or 4° near the bottom. I left it to see how fast it would cool. In an hour those were only down one degree. It did eventually make some ice, but most of the tray stayed water and the fridge didn't get under 6°.
   But again, really the main reason for low heat pumping, regardless of COP, was probably using the modules at 43% of their rated voltage with the 15 volt peltiers in series (6.5v each), with no intermediate voltage modules available. Where were all the other possibilities like 10v or 24v?

   I also went to DigiKey to look at peltier modules. Here (and then elsewhwere), I found Laird Technologies peltier modules, which had more thermocouples for higher voltage ratings.

"The UltraTEC Series is a high heat pumping density thermoelectric module (TEM). The module is assembled with a large number of semiconductor couples to achieve a higher heat pumping capacity than standard single stage TEMs. This product line is available in multiple configurations and is ideal for applications that require higher cooling capacities with limited surface area. Assembled with Bismuth Telluride semiconductor material and thermally conductive Aluminum Oxide ceramics, the UltraTEC Series is designed for higher current and larger heat-pumping applications."

"The ZT Series is a high performance thermoelectric module (TEM). The module is assembled with premium Bismuth Telluride semiconductor material that achieves a higher temperature differential than standard single stage TEMs.

   It seems to be vaguely implied in the bolded phrases that these are somehow more than single stage peltier devices, but there is no such indication in the datasheets. Three higher voltage modules looked appropriate, which would according to my readings of the graphs provide the following results with 13 volts and 30° temperature spread (presumably with hot side temperature 25°c) :

UltraTEC Series UT6,24,F1,5555 ("30V"): Supply 2.4A (31W), 28W pumped [84-89$]
UltraTEC Series UT8,24,F1,5555 ("30V"): Supply 3.4A (44W), 37W pumped [108-$]
UltraTEC Series UT6,19,F1,4040 ("24V"): Supply 3.0A (39W), 31W pumped [66-71$]
ZT Series ZT5,16,F1,4040 ("20V"): Supply 3.2A (42W), 28W pumped [57-$]
ZT Series ZT7,16,F1,4040 ("20V"): Supply 4.1A (53W), 35W pumped [79-$]

[Prices Canadian$ @ Mouser-Digikey, April 2014]

   The first three provide almost as much cooling watts as input watts, and the 24V unit is also very good. This is excellent for a 30° temperature spread - compare it to the 8.5A Cui "regular" peltier type:

Cui CP85440 ("15V"): Supply 6.8A (88W), 37W pumped [24$]

   But a comparison between 30v and 24v units at lower voltage and a 15v unit isn't fair.

With a small 15° temperature spread, the specs were even more impressive:

UltraTEC Series UT6,24,F1,5555 ("30V"): Supply 2.6A (34W), 51W pumped
UltraTEC Series UT8,24,F1,5555 ("30V"): Supply 3.6A (47W), 70W pumped
UltraTEC Series UT6,19,F1,4040 ("24V"): Supply 3.2A (42W), 52W pumped
ZT Series ZT5,16,F1,4040 ("20V"): Supply 3.3A (43W), 40W pumped
ZT Series ZT7,16,F1,4040 ("20V"): Supply 4.4A (57W), 54W pumped

   A worrisome feature of the 30 volt units, as with putting two 15 volt units in series, was that they would be operating at the very bottom of their voltage range for the temperature spread. If in the summer the 30° spread increased to 40°, the heat pumping would be cut in half, and also by 1/3 if the supply dropped just a volt to 12. The 24 volt unit wouldn't change so drastically, going from around 30 to 20 watts at the 40° spread, or to 25 at 12 volts.

   The prices were steep, even as much as putting together the four 'ordinary' peltier modules, but the design appeared to do exactly what I was trying to do, and better, in a single module. And the prices for all the modules were up sharply from the last time I ordered, a year or two ago.

   Then I went to figuring out how my 'original' two side by side peltier setup stacked up. The 15v and 8v modules in series (electrically) made effectively a 23v one, about 50x50mm. According to the graphs, it should do around 35 watts of heat pumping, with a supply of around 45 to 50 watts. At a glance it looks as good as the fancy setups except for one thing: it doubtless transfers substantially more cold from the fridge to the outside heatsink when it's turned off. Although two-stage pairs pumping across smaller temperatures pump more heat, some of the heat pumped by the warm side unit simply goes into counteracting the heat produced by the cool side unit. (Maybe if the warm side dropped 20° and the cool side only 10? ...Then the cool side would make less heat. But then the warm side would have to pump that across the 20° instead of 15°. Maybe it all comes out equal regardless of how it's configured?)

   In order to keep the fridge cool I put in a 6 amp module and awaited the arrival of the new choices. I'm still waiting.

    This isn't the final word. I plan to try out more configurations and two stage arrangements. There may be other details that affect performance that, like the off-state heat transfer, don't show up in the datasheets.

Electricity (Energy) Production

Magnet Motors: Magnetic Shielding and flux concentration

   It occurs to me that in considering magnetic shielding, one should examine ceramic "cup magnets". The rather thick-ish steel cups concentrate the flux around the edge of the front face of the magnet, and in that limited area, the field attains the force range usually associated with supermagnets. If such cups were applied to supermagnets, the flux should be powerful indeed.
   If a magnet motor is going to work, this would be the means of making it powerful enough to produce a decent amount of power for its size, rather than feeble power that merely keeps it turning itself with a small load or none at all. The way the cups are formed for ceramic cup magnets is thus probably the ideal way to make them for supermagnets and magnet motors. (Short lengths of steel tube with an end soldered or welded on might be a simple way to go?)

Electricity Storage

Turquoise Battery Project

Ferric Chloride

   I've mentioned painting ferric chloride onto/into the positive electrode a couple of times. What's it for? It converts to ferric oxide/hydroxide, and people have mentioned using this as a positive electrode active substance. I thought I'd add it on spec.
   But as someone on a list mentioned, it has another action. In becoming oxide, it turns copper into copper chloride - it's usual use is as copper etchant for making printed circuit boards. This means it eats away at the copper content of the monel, creating pits and rough surface area and exposing more nickel on the surface where it can convert to chemically active NiOOH and nickel manganates, both increasing the electrode capacity derived from the monel.
   Finally, Cu++ of the resulting copper chloride is the ion used in the electrolyte of an experimental battery which apparently recharges the cell from the ambient temperature of the room. If such an action can be obtained, it would be phenomenal. So far, the high self discharge of the cells has prevented any such action from being observable even if it is present.

Cell with no Graphite Powder?

   In order to test whether the graphite powder is the culprit in the self discharge as suspected, I at first decided to make a posode with no graphite, and accept whatever higher internal impedance it would have. I didn't get around to it and now I think I'd best find some pure graphite for a new electrode.

Victoria BC