The Cape Girardeau ENDURO.

Of batteries and range and endurance and costs and complexities and batteries….

This week we continue A123 module obsession AND we do some minor work on the Cadillac Elescalade.

Apparently the A123 cell thing is of intense interest to at least some segment of our viewership. If developed, it would appear to offer some alternative cell strategies. However, these are not entirely without difficulties.

A couple of items. We continue to have lab anomalies I’m hesitant to report in detail. The reason I can’t share these with you guys is you guys. You shred me every time I do with a laundry list of how I SHOULD have done the test.

We are accustomed to testing large format cells – 160 Ah and 180 Ah units with the odd 400 Ah cell thrown in. It’s true I’ve done quite a bit of work in the last year in pairing two 90Ah cells for 180 Ah.

Suddenly dealing with 20Ah cells is just a different world. 1/2Ah or even 1 Ah isn’t anything on a 180Ah cell. It’s not really much of anything on a 100 Ah cell. It’s a LOT on a 20 Ah cell. And so our procedures and test equipment, which I am ALSO always testing, gets a little bollixed up on what I’m doing at any given moment.

IS the decrease in capacity from 19Ah to 17.8 Ah due to test equipment or have I damaged the cell? Or not fully charged it in the first place?

As everyone who has gone into real cell testing has discovered, defining fully discharged OR defining fully charged is actually a little squishy. You can fully charge the cell using anything you want. Let it rest a couple of hours, and it will take more using the same metric. Similarly on discharge. A “bit more” starts to get hard to define at 20 Ah.

But we have had some mysteries. Like bottom balancing four cells to exactly 2.50 volts and then charging the pack. When we discharge it, they are all out of b a lance again. So we repeat the process and they are BACK in balance again?

How about draining a cell to 2.50v and then leaving it overnight with NOTHING connected to it. To find the next day it is ruined at 0.85v. Wait a minute. Recharged it works fine again????

We do not see these sorts of things with the larger prismatic cells. I would characterize these A123 cells as about as stable as a burlap sack full of cats. But I don’t know if that’s a good thing or a bad thing.

[jwplayer file=”news021712-” hd.file=”” image=”” streamer=”rtmp://” provider=”rtmp” html5_file=”” download_file=””]

More reality. Yes, I know you all are all packaging engineering experts and I was made to WATCH the other kids during art class in the second grade. But whatever I do adds 1/3 to the weight and volume of A123 cells. You can pretty much figure that if it takes 10 cells and they weigh .496grams each that will be 4.96 kg. My package will be 4.96 x 1.33 or 6.6kg. So any increase in energy density you “imagine” from these cells is not only not possible, but if I’m doing the packaging there is a penalty here. A half sized pack doesn’t get you half the range. It probably gets you 1/3 the range. So your 100 mile car with a pack that costs slightly MORE than half a pack probably actually delivers 1/3 the range or 35 miles or so.

Similarly the costs. Yes, it’s $100 for resin. Let’s try polycarbonate at $150. Or straw, at $90. Whatever I do, at least in prototype, dramatically increases the cost of these cells which STARTED OUT very pricey themselves. As their price has fallen, the percentage devoted to connectors and resin and molds and so forth has risen as a percentage.

So despite my obsession with these cells and posing them as a low cost alternative because of their high power output, reality keeps intervening to make it clear to me that this will NEVER be my choice for cells in a car. The large format leggo block 180Ah CALB is the battery of choice for me in designing a car at the moment. And I would list A123’s as “problematical”.

But that’s not to say some effort making them LESS problematical isn’t in order.

On the good side, despite my totally ruined mold, the 13.2v 120 Ah battery is testing well at about 113 Ah without any real overcharging or over discharging. I have mentioned some oddities but the cells seem to still be ticking along. But from what I am seeing I can see new insight into why some are so focused on top balancing BMS systems. They are a brute force way of making these cells do your bidding or appear to. I suspect there is another better way, but I’m still looking for it.

And the 13.3v package is looking pretty good to me in camp green and black. It is a little heavy at 36 lbs was it? But obviously durable. All cells are available at the top, perhaps too much so. A dropped wrench on this one would be fireworks. I might put some effort into a soft rubber “cap” made out of the silicon rubber mold material. Pour it on, let it cure, and pull it off. Then you make your connections and mash it back on to secure the terminals. Kind of a soft rubber hoodie for a battery.

All this is kind of stalling out our larger projects. But I confess I am having some fun. ANd it feels like we are just doing some things very differently from the OEM’s and the very elegant high tech style so in vogue. But I kind of think “car stuff” out to be a little more rough and tumble along the lines of the SLI battery, which has evolved over 100 years – in fact we missed the anniversary of Kettering’s electric start automobile 100th anniversary this week.

The 1912 Cadillac Touring Edition was first to eliminate the hand crank and opened up driving to everyone. Cadillac founder Henry M. Leland, who had already pioneered electric lights and electric ignition on his cars, worked closely with Charles F. Kettering, the inventor of the electric starter, to incorporate the device into his cars. The electric starter also was GM’s first electric motor – a core business today anticipating the growth in the electrification of the automobile.

Point is, that battery has been “evolving” for a century now. It looks the way it does for a reason, or a million reasons. It arrived in Darwinian fashion to be exactly what it is. And so using that as a starting point is not a bad, if slightly blind, strategy.

But new ideas are good too. And so we are casting about. We have used several. We used the alternating cells on the nylon threaded rod. I liked that, but it buried our terminals. We kind of stress tested it, which on reflection may not have been good medicine. But it got us to a pretty gruesome failure quickly which prevents spending more time on such a thing and then having it blow later.

We then mimicked the individual cell thing we already get from China. A word about that. I’ve heard a lot about just using similar boxes and similar terminals from China. Well, I’ve looked for them on Alibaba and I can’t find them. If you can find who makes those, yes, I wold prefer to just buy the existing hardware and plastic extruded boxes, even if I had to cut them down a bit. Haven’t found a source.

And this to the concept of ideas. An idea is not a general piece of shit on a napkin. They work better with measurements, specific product recommendations, and sources. “Expanding foam” is fine. But there are a brazillian. Similarly urethane resins, polycarbonate, silicon rubber, etc. ALl have different cure times, shore hardness, tensile strength, exothermic reaction, temperature tolerance, etc etc. ad nauseum.g

This is why I can be a little short with some of the arm chair theorists. If you’ve really thought this through by looking up at the ceiling, you’re not in the same category as someone like Nabil or Peter who have hooked some of this up and then had an idea they don’t have the resource to implement. And then if you’re a Damien Maguire or Paul Holmes are someone in that category, just send me a list of parts you need and a delivery address and we’ll wait for YOUR video.

I did include a segment of Damien Maguire’s in this episode. If you were looking for someone to liven up the pace of conversation, Damien is more in my camp than out of it. But he does some interesting and of course tedious work on bottom balancing and shows you the ugly end of it on his BMW and so I included the entire video he uploaded to YouTube. I think he will find bottom balancing surprisingly effective in a lot of ways, and no easier than he thought.

I was reminded by the head of REAP systems, a leading BMS developer at the EVCCON that while bottom balancing might be more effective than top balancing, at least arguably, it was not very convenient. My response to him, and to Damien, is that in all the battery testing we’ve done, across several years and now locations, I’ve never once had a SINGLE LiFePo4 cell express ANY interest at all in my convenience. They are just curiously agnostic and apparently have no feelings for MY feelings whatsoever. But on the other hand, if I ask one to do the dive for the cause at my behest, they will head toward zero volts so fast you think they actually LIKE to give their lives in order to save mine. Valient. Just valiant. Splendid behavior.

This week I’m working on a new supur sekert module project. We are code naming this one the FLATENUM SERIES. It will be a flat pack designed to slide under Speedster Duh and ultimately get hooked up to it to provide 120v of juice at 60Ah. I have kind of a theory that this small Ah pack will act as a pack stiffener – holding the voltage up on acceleration. After acceleration, it would be restored by the larger prismatic pack to an identical state. This appears obvious. But it either may or may not be TRUE. We haven’t really done such a thing. I would THINK it will work and instead of plunging to 106v on a hard acceleration we should maintain up at about 114 or even 116volts. This means more power through the controller to the wheels.

it should add about 200 lbs to the car, but in theory another 25 miles as well.

To make all things fair, i’ll just quietly mention in passing that we are considering an addition to the Drag Race and Autocross at EVCCON this year. It will be a two hour race we will call the EVTV ELECTRIC ENDURO. It will run through a very SCENIC fall drive from Cape Girardeau up the hilly twisty U.S. 61 to Perrryville Missouri, thence across highway 51 to Illinois, where it will pick up Illinois highway 3 down the banks of the Mississippie before recrossing at the Cape Girardeau Bill Emerson Bridge back into town. And it looks like 127 miles on the map.

I think Speedster Redux can make it now and Speedster Duh likely will make it after the A123 addition if it works. But if you’re working on an extender pack, battery trailer, etc there may well be a showcase for it in the EVTV Enduro. Probably a timed race rather than a true road race. Time and the ability to complete being the issue.

View Larger Map

I may try to arrange some charging in Perryville for those who want to try half of it, and still get back to the festivities without a flatbed. Yes, the entire concept is for this very beautiful scenic country drive to be just beyond your reach. An electric road race no one can finish. That’s just the way my mind works. I would have been different had I been born a TALL instead of a ROUND. But I’m still around.

One other element that will apply to drag race, autocross, and endure – anti homolugation. If you have produced over 100 of these vehicles, you cannot compete other than exhibition only. It must be a custom conversion or VERY small run OEM with less than a hundred instances on the road. We remain TEsla Fanboyz too, but that’s not really what EVCCON is about.

Jack Rickard

131 thoughts on “The Cape Girardeau ENDURO.”

  1. Definitely some strange behavior on the A123 Cells.

    I occurred to me when you talked about using the A123 Cells as a pack stiffener that A123 might be doing the same thing at the Cell Level. If you assume (and we all know how well that usually works out) that 1/2 of the cell was designed for High current and the other half was designed for deep cycle durability, the behavior you are seeing would make some sense.

    Does the A123 recommended charge profile have several voltage step charges at the end of the bulk charge period?

  2. Since doing that bottom balance last september the car has been driven to a halt twice. On one occasion the voltage dropped so low that the victron bmv600 flamed out and the car started rolling backwards. Charged right up. No cells damaged. Bottom balancing is a pig of a job now made easier by the lightobject meter.

    Planning to add more cells next week so will be at it again. Should be interesting to see how close they are on the bottom after 7k miles.

    Damien Maguire

  3. The Flatenum project is the beneficiary of ‘prior art’ Intelligent Dual Energy Power/ iDEP employed in some electric bicycles and motor-scooters. The Tesla “brick” story circulating is a potential public relations disaster. Why can’t the parasitic load business be easily solved. Maybe Tesla needs an EVTV white board refresher and to migrate into LiFePo4.

  4. Hi Jack,

    Back in th “old days” of motorcycle enduroes in Potosi, MO there were checkpoints that had to be hit at certain times and penalty points were assessed for showing up early or late at a checkpoint. I think that would be a good thing to use to keep traveling times to regular speeds. You might check with the local dirt bike clubs for some guidance in this area.


    1. I believe Alex had mishandled the cells and didn’t know what he was talking about.
      if you think about it, if they were that volatile as to die after first use, we would probably have heard about it..

      iirc his discharge graph started from 4.0V. that’s a high value to hold stably for a lifepo cell. too high. I also seem to recall it contained some 25% more energy than they are supposed to. another bad sign.
      personally I thought it was unfortunate at the time to put out that message in a show and it was never retracted, even repeated a couple of times. I tried communicating it to Jack but my feedback gets mixed reviews.

    2. You can believe what you want Dan. Jack agreed. Said he wasted $10,000 on these plus a tab welder. Told of his disasters and said he was done with them and Alex can take them away.

      Alex said his 40AH Kokams are amazing and hold up to 60AH.

    1. I saved the show so its no problem for me. Jack doesn’t seem to of saved it to youtube. A great pity, its a landmark show.

      What Alex said on A123 cylindrical cells. (sic) After the cells were driven hard, we let off the load and as we expected the voltage rose back up but after a short while they would dive down on their own to zero. We thought we broke the cell or had issues with our equipment but this characteristic was found on cell after cell.

      Jack said some things on his experiences with these cells too.

      Tonight I’ve put my money where my mouth is and ordered 180. πŸ˜‰

    2. Thanks. I guess this means don’t drive one’s batteries too hard. Congrats on your order. As I recall, you are working on a trike/tadpole three wheeler. Do you have a blog page or website with a sketch of the frame? I think e-trike/bikes and e-scooters/motorcycles are going to take off unless Big Oil and the auto industry use undue influence on politicos.

    3. I’ll definitely consider blogging it up when it’s all compilable. This will save on endless pages of garbage. Keep it to the point but the debugging and findings will be ongoing there after.

      It’s based on the MEV E-Trike but I’ve departed from its basic mantra of being super compact.

    4. That sounds good to me. I think something between the MEV E-Trike [which I just looked up] and a ZAP Alias makes sense if it could be had for 1/3 price of electric car. [A Mango Velomobile on steroids?] An affordable runabout. I like the MEV’s roll bar. I think a vehicle with doors and/or lower entry would be better than an athletic Batman entry. I’m conflicted on side-by-side vs tandem seating.

    1. The entire thread sounded like typing themselves smart. We don’t know. We’ll try it and measure it. And see.

      But this doesn’t sound right to me. The further drown the A123 pack goes, the greater the voltage difference and the more recharge you would get from the larger pack.

      We may be able to set up two AH counters and see what the result is from a shared load.


  5. Hey Jack!

    Not to suggest how you spend your time or money, you might cobble up a test of one 19.6 AH A123 pouch paralleled with one 180 AH CALB, and compare it to the CALB alone in voltage sag under a 10 second 1000 Amp discharge, etc. Seems building a 200 lb. flat pack is a lot of risk and work to test feasibility. Can’t think of any reason the results of such a test wouldn’t scale up to the 120 V pack.

    Mike Kaindl

    1. Mike:

      I don’t have a way to do 1000 amps in the lab. What we MIGHT want to do, now that I think of it, is set up TWO AH meters one on an A123 leg and one on a CALB leg for example. Then hit them for 300 amps for 10 seconds. Do that 20 times. And see what the AH meters show.

      The A123 cell should have current OUT but then current back IN to balance it out. The CALB side will just have current out. Hopefully, they will both contribute according to their AH rating, instead of their internal resistance.

      We’ll see.


    2. From reading the elmoto thread (thanks Warren), sounds like the A123 side will lose charge quickly to hold up the load, but will recharge slowly from the CALB side because the V vs. SOC of the chemistry is so shallow (a lot like constant voltage charging at 3.2V). There isn’t enough delta V to move charge over from the CALB to the A123 fast enough to come to full recovery equilibrium, and the A123 SOC just ratchets down incrementally with each acceleration power cycle. Range is then roughly defined by the capacity of the A123’s, at which point why carry the CALBs?

      Or you could come up with some snazzy Buck-Boost DC/DC charge transfer/balance controller (I’m guessing neither cheap nor available) with isolation diodes, etc., etc., that could provide the delta V to recharge A123’s in a timely manner, but may have to be bypassed during charging both packs in parallel to avoid overcharging the A123’s.

      Or, maybe a 144V CALB pack recharging a 120V A123 pack, with a cheaper charge transfer controller between them (no buck-boost), and independent charging for each part.

      Admittedly typing myself smart, but simple passive pack stiffening smells enough like a magic smoke escapement that it would warrant a small-scale test.


    3. Jack,

      You’re right, what would be interesting is the recharge current out of the CALB after the load pulse. How long will it take at that rate (which is probably decaying as voltage balances) to significantly share the kW’s burned in the pulse?


    4. Not sure I would be interested in that or would know what it means. Things happen a bit faster than that I’m afraid.

      If we have an AH meter on each leg, then the recharge current should offset the outflow from the A123. In an ideal world, we would wind up using current from each battery in proportion to it’s capacity.

      So after 20 300 amp pulses of 10 seconds. I should for example have used say 6 Ah. Ideally 5 from the CALB 100 and 1 from the A123 20.

      If we measure this over a full discharge, they should arrive at the “bottom” at the same time. And the “balancing” should happen dynamically and at an increasing rate as the curve steepens.

      What is more interesting is what might happen in a CHARGE scenario. I assume the same thing.

      I just can’t picture cells in parallel out of step with each other no matter what the IR. I can’t get my head around it.


    5. Jack,
      Can’t wait for the tests. I am also curious about what happens on a long sustained run, lets say you are driving 50mph for 10 miles and constant pedal pressure. Will they still slowly equalize?

      One question, in the above example, if more power is the goal, wouldn’t the A123’s be using more AHs? I don’t know, maybe I am running my mind in circles and don’t quite understand how power is removed from parallel strings.

  6. I think we’re typing ourselves smart here. But perhaps. And perhaps there is a way to devise a meaningful test short of 1000 amps.

    It sounds very much like Yogi Berra’s comment on the restaurant – “Nobody goes there anymore – it’s too crowded.”

    The A123 pack depletes because there isn’t enough difference between the two to charge it? Then it must never get very low then.

    Let me pray on this. I think a pair of AH meters with small shunts and then a series of 200 amp pulses on a parallel CALB/A123 four cell might just be the ticket.

    Jack Rickard

  7. As long as were typing ourselves smart here, how about this test. I can only type myself smart as I don’t have any A123 cells nor a 40aH prismatic. But if you took two A123 cells and married them to make a 40aH set and put them at a voltage that was 50% SOC and then took a 40aH CALB that is set to 100% SOC and connected them with instrumentation on the connections, you could see how long it takes to transfer 10aH into the A123.

    1. I would consider another test in this proposed set-up, right after having run it – to charge this hybrid pack. They are of the same chemistry, but act differently. Reminds me of Luc Besson’s Fifth Element fiction story – of superior human-like-aliens.
      Although married performance wise, this booster A123 and CALB pack, shouldn’t they be charged separately?
      Before diving into big $ expense, such a small scale tests shouldn’t hurt. Typing myself smart here, because my experience with EV’s is equal to none, because driving LEAF and Prius is as war as I got so far.
      The idea to benefit from A123 and CALB characteristics is brilliant and rather exciting. I congratulate you upon reaching to this point for using it as a model for improving acceleration sag of the CALB pack. Just, make smaller steps at the time.
      EVTV’s show has become more interesting to me than ever, not to mention this blog.

  8. an A123 pack in parallel will naturally stiffen the voltage and it will equalize provided their discharge curves are substantially overlapping.
    but a pure A123 pack is simply better. piling on even more batteries is the wrong way.

    Jack, I know you like the PU casting and it’s fun to cast plastic but save it for some other application, it’s very expensive and it adds weight and it’s not needed.
    rather than encasing 4 sets in a block just put them in a battery box instead. a long row. either fiberglass or something else. that way you can take them apart, it doesn’t cost anything and it doesn’t add weight.

    and get over the fear of correction from the audience. it’s only natural that a 1000 freaks with no money to do anything themselves will have some suggestions that you haven’t thought of. roll with it

  9. Jack,
    An idea, I done it before in mold making and you would have the same opportunity with your SiO2 molds for casting your thermo-set urethane battery boxes. The idea is when your cast the SiO2 to take the form of the shape that you want to copy, hot melt a 3/8″ pipe nipple or union on to the plug,(the the shape of the battery to copy. Fill the pipe nipple or union with clay and have it extend past the thickness of the SiO2 glove.
    Why all of this? Because now you have the ability to use pressurized air to depart your urethane part from silicone glove.

  10. Concerning the Volt fires, GM is missing a golden advertising opportunity.
    They should have a gas car and a Volt crash into each other. In true Hollywood fashion the gas car immediately bursts into flames and calmly a voice announces “You have 3 weeks to get out of your Chevy Volt.”

  11. Recall the voltmeter with the control relay that we now use for bottom balancing? Recall that I also had an odd one that counts Ampere Hours and it appears to be very accurate?

    Well he got me 10 more…. We have amp hour meters….

    Jack Rickard

    1. In my professional work, we do put more than 70 strings of VRLA’s in parallel. Each string is more 32 or 48 times 12V/9Ah batteries. The only recommendation we have is that one string should be replaced as a whole and replacements must be of same age, charge condition and type (we have 3 different suppliers).

      I also have a colleaugue that have been driving a mixed battery pack (VRLA and A123’s) in hard parallel for over a year. The A123’s are getting all the hits and the VRLA’s are supplying the bulk capacity.
      He had great conserns about it a year ago, but I told to just do it as long as he could get fast out of the car… πŸ˜‰
      He is using a 1/20 ratio (4.6Ah A123 for each 72Ah VRLA’s – afair).
      After doing this, he has increased the regenrative braking and most of the current goes to the A123’s.
      The main reason for not going all lithium was a matter of $$ or lack of…


  12. We will probably be offering a kit/setup for the AH meter. The meter, a shunt, mounting the shunt in an insulator of some sort, and a DC-DC converter. I’m working on some documentation that will hopefully also allow you to use it in a car.

    Jack Rickard

    1. Good documentation: Value added!

      Where DC-DC converters are required, I prefer the idea of a small one per unit.

      If the shunt is boxed including the DC-DC supply so it is power in and sensor/supply socket out. Then a wired loom (choose ~3′ or ~12′ long). That would save loads of hassle for individuals messing around and sourcing for small quantities of bits.

      I like!

  13. An interesting conundrum with mixing cell types. Call me stupid if you like but surely, if nothing is taken to their ‘C’ limits or forced hard towards each end of charge then everything ought to be ok.

    I like to visualise cells as a series of (or parallel) reservoirs. Each pool fed underneath with a pipe. A123’s are a small pond with a big pipe.

    Could go further in defining these cells as a shaped tube that suddenly narrows at the top. Height is voltage which gives pressure and flow rate is current.

    The paper Jack posted from SAFT defines batteries used in the telecom business agree’s with batteries that are simply ticking over.

  14. The foils within a single cell are individual batteries and EACH has a different internal resistance guys. We can take this as far as you want, but I smell total BS out there wafting through my world again.

    But we will test….. you never know….

    I had bad dreams about this last night. I might ought to consider getting a life…..

    Jack Rickard

    1. Lead acid batteries are not for any EV at all
      I know from bad experience since the year 2000 when I built my Renault into Lead Acid.
      It could run MAX 50 KM in the summer and winter MAX 14 KM. and it was Trojan Batteries
      There is simply too large internal resistors in them, and apparently too much antimon. They were like a elestik voltage drop over a large area, they get warm and have a miserable efficacy rates and is far too heavy


      I do not think, no matter what you put together with lead acid you will have a bad reslutat and at worst destroy LiFePO4 batteries.

      I look forward to getting mine Thunder Skye cells in the car, expecting to be finished next month.

  15. I’ve got to figure out some way of describing the relay action that makes sense and finish the documentation. And I want to check out a 12v dc-dc to make sure we have isolation on a car.

    But yes, I should have something together by next week on the meters. The ONE I have in the lab is the most accurate thing in there by time and current. If I can document the setup, it isn’t really that hard. The book that is available just sucks really hard.


    1. Hi Doc,

      Lightobject made a run of 20 of the meters, Jack took 10, the remaining 10 were sold on the Lightobject website. (unsurprisingly, now sold out!) The website description states a blue display, so I don’t think you’ll need to swap out the LED display.

      Regards, Paul

  16. My intuition is getting confused about this whole mess. I agree its time to slow the theories and speed up the testing.

    But as our friends on the upside down part of the world say, Horses of Courses. I’m going to let those with the cells and test equipment cook some cells for my benefit. Meanwhile I’m sharing some of my experience with fabricating with Polycarbonate. Have a look and give me some ideas for what you’d like to see tested.

    I have adheased together three pieces to make an I beam. I’m intending to force it over – make it italiziced with brute force. We’ll see which material gets it’s wish!

  17. Hi Jack and all,

    An interesting way to test the pack stiffing effect of a paralleled A123 might be to make a series string of 1 ea. 100ahr, 1 ea. 130 ahr and 1 ea. 180 ahr cell. Add a 20 ahr A123 in parallel with the 100 ahr cell. Monitor each of the three cells, or cell pairs with three different volt meters.

    Now do your pulse tests(10 sec?)on the 10 volt string. This would give the relative stiffness of the different cells. Of coarse, the pulse load would have to be significant (500amps?)for best info.

    Hopefully the A123 pair would be stiffer than the 130 ahr cell and maybe as stiff as the 180 ahr cell under the same pulse load.

    Just a thought.


  18. Karl do you mean bench testing what Jack implied with paralleling A123’s with standard prismatic cells?

    Voltages tend to fade under load more with standard prismatics than A123’s. So when Redux hit 6C with CALB’s the voltage dropped some 20%(?).

    Volt/C discharge graphs…. Someone has done it. I’ve seen one somewhere.

    1. I enjoyed your video, though I am left wondering why the Warp9 was disassembled? I will be wowed to learn you took it apart a factory spec motor to check tolerances, etc. Is the heating/cooling of various parts to ease assembly a standard procedure? Did you heat/cool the parts to break the motor down? Your English is far superior to my second tongue.

    2. Thanks Perin,
      A also made a film about the disassembly but have not realised, that my memorycard was full at an important moment. I will see if it can work as an video or do a new one.
      For the disassembly you need two, three spezial pullers. The heat-thing is something that I dit in my moms oven when I was young, working on a 50ccm Motobike engines, and had no money for special tools.)
      The reason for the disassembly is that I want to know how things work, and also changed the bearings to some SKF vor a test. I am (starting to become) a dealer for NetGain Motors and realy need to know what I am talking about.

  19. With regard to flat packs, the strongest method that comes to mind is a symmetrical vacuum formed polycarbonate. In the voids between the tabs, you could have a depression in each side that would meet each other. A fastener would bind these spots together. The resulting clamshell would be very robust.

  20. There’s too much overthinking about the A123 pack stiffener idea. As long as each string has the same number of cells, it will work. You don’t even have to bottom balance each string at the same time before connecting them, or require them to be at the same SOC. When both have the same terminal voltage, go and join them. Eventually both strings will reach their bottom at the same time.

    The only concern is if eventually, several cells in one string go bad. Since the 2 strings (one A123, one CALB) in parallel form a complete circuit, the string with dead cells will eat current from the other. I’ve seen this happen several times with lead acids. When good batteries are connected in parallel to a battery with a shorted cell, the defective battery will drag the others down to 10.5 V. That issue is mitigated though in the Speedster since there are 40+ cells in series in each string. But it’s worth watching out for.

    1. I guess I was misunderstood. Yes, each *string* needs to be bottom-balanced, no question. But you don’t need to connect both strings to each other while they’re at the bottom. Jack can just partially charge the A123 string (after bottom-balancing), have them rest down to the terminal voltage of whatever state the installed CALBs are at, then slap them on.

    2. Yes, in truth these cells ARE the same chemistry, just different power levels. And this was of course nonsense. But it had me going int a circle and actually there ARE some interesting things that go one with different cells when charging and discharging.

      The bottom line is the they DO react differently, but in a short time sort everything back out quite nicely. THe concept that the voltage difference is too small for current flow to correct was of course the logic disconnect. Yogi Berra – “Nobody goes there any more – it’s too crowded.”

      If the voltage difference is too SMALL then what’s the problem. And if the voltage difference is large enough to BE a problem, it causes current flow and the transfer of energy to the point of problem correction.

      But this goes back decades and there have been reports of vastly dissimilar cell mismatches in parallel that work just fine even when in theory they should not.

      The admonition that you should not parallel these cells is just nonsense.

      INSIDE any of the cells, is a series of cells – each foil sheet being one. No two are actually identical. But when you pile them all up, you have a battery. That’s kind of what a “battery” is.


  21. One of my early misadventures in forum land was with a newbie who called himself DIMITRI on the DIY Forums. He was selling a device called a VOLT BLOCHER that connected across the top of your cells and purported to “balance” them. I ordered some and in fact it destroyed a couple of cells immediately. They were horrendously dangerous devices that went into thermal runaway IMMEDIATELY and left unattended would burn anything to the ground.

    I attempted to have a telephone conversation with the guy. He was by then already refunding the units but I really didn’t need my money back, I had only bought four for testing. I tried to explain to him what we had learned about how cell charging worked and what a “useful” type of BMS might look like. He reacted very badly, and despite having an almost TOTAL misunderstanding of the cells and charging in general, insisted angrily that he knew better. Went on to design a totally useless BMS that at least does little harm and a number of people ride around with this thing, which does NOTHING actually, but afford a false piece of mind, calling it a BMS.

    As it turns out, there’s a reason the EMW Android system came apart. They are in DIY struggling MANFULLY to assert it was my clumsiness that caused their totally under built board to FALL APART IN FRONT OF ME. Yes, the weight of the cables is an unnecessary abuse and if you mounted it correctly and lovingly in a vehicle, apparently they feel the STILL SOFT adhesive gluing the el cheapo hall effect current device to the board will NOT break away on its very first use. They also still insist that this tiny bit of copper will carry a thousand amps or so without danger.

    The hardware design skills of this guy are DANGEROUS. And now the 4.9AH inaccuracy at 100 amps (nearly 5%) is explained as well.

    I played with some very GOOD hall effect sensors made by HAAS a year ago, not the little cheapie they’ve selected here. And it is quite difficult to work out zero amps and account for temperature. Again, this is a case of a guy who doesn’t know, and doesn’t know he doesn’t know. We kind of struck meat here I’m afraid. The accuracy of this system will NOT be acceptable for even casual use until they get some new hardware in the game. Sorry.

    I like the pretty pictures. And I got all excited here. I didn’t realize el DIMITRI was involved and know that we know, that explains quite a bit actually.

    I would beware this otherwise attractive product. It is fragile, but also inaccurate and the guy designing the hardware has NO clue how lost he is.
    That’s not a good recipe for a fix.

    Jack Rickard

  22. Jack,

    For your next project I would suggest a range extending trailer. Local driving is completely possible with an EV. What isn’t possible is for conversion folks around the country to hook up a trailer to their conversion and drive it to the EV convention this year. It’s going to take a 10 to 20 kw generator hooked up directly to the battery and some circuitry to control it. Until ultra high rate recharging becomes available along the roadway a trailer/charger is what is missing from our list of available components.
    Everything else exists from one source or another at this point in time.

  23. Jack,
    Was it only at high currents that you observed the inaccuracy? I’ve only actually studied the current readout when on charge (max 40A) and running the heater (20A) and it seemed to be reading within 1A of the victron bmv. This morning lying in bed I observed the end of charge on the android which is programmed to occur at 3.2A and the display was hovering around 3.5A when the charger terminated. I don’t spend too much time watching the ammeter while driving as I’m usually too busy dodging the hollywood class stunt drivers that populate the roads in Ireland!

    Damien Maguire

    1. Damien:

      5% of 20 amperes IS 1 ampere. Kind of hard to dig out. We ran at 100 amps and 200 amps using Fluke meters, the meters on the loads and power supplies, and a little Chinese meter I’m becoming terribly fond of. The Fluke would read EXACTLY 100.0, the power supply 100, the other meter 100.0, and the Android 104.9. At 200 amps this was 210 amps that the Android read. And it worked to the same degree and the SAME inaccuracy on discharge.

      It’s a very regular 4.9% HIGH error. When I asked Valery how to cal it in the software, I didn’t precisely get an answer. Now the answer appears to be it is “coming’.

      If they are just starting to address this, I can save them about 18 months. It’s not that easy. I wound up with TWO hall effects, a positive correction AND a negative correction AND a zero connection for EACH of them and it was getting pretty good. But they were much better sensors than these guys are using.

      Bottom line is I just feel for something that sounded too good to be true, and it turns out that was because it WAS too good to be true.

      I still like the Android software/bluetooth idea, but this hardware won’t do it – if you CAN keep it together.

      Jack RIckard

    2. I’m trying to figure out why they went with a hall device in the 1st place. Since you must break the electrical connection to insert the magical copper bar that passes through the hall sensor, why not just use an appropriate shunt instead? A lot less complicated to deal with, it’s what gives the Light Objects AH meter a leg up on accuracy.

      Jack was responsible for breaking the hall sensor loose from the PC board but that’s not the point. The point is that Jack represents a typical customer and a good product needs to withstand typical customer handling and (ab)use.

    1. Good video’s.
      Transparent! πŸ™‚

      However, I’ll attempt boxing up with Correx as a first try. It’s cheaper, lighter, smaller and should be more serviceable.
      But the thought of donking the thing on a lower corner won’t be much fun.
      I’m warming to the idea of a pre-cast plate like Jack made originally to connect the box, tabs and bolts.

    2. Mathieu, thank you for the info!! I have many such sample packs of other materials at work. I feel very foolish not to have asked for such a sample pack to begin with. This is the best kind of input!! Thank you again.

  24. Jack, I got to thinking, and I think you hinted at it too, that maybe the problems you have been having with the A123 packs could be related to your bottom balancing? if the cells are more varied than you are used to with the block cells from china.
    if there is as much as 1Ah difference in capacity and you charge to a system voltage without watching the cells you might well strongly overcharge the smaller ones. could that be?

    if you critically overcharge an A123 cell they might go to zero (like Alex experienced) and then when you discharge the pack there is a perhaps serious problem. one that could lead to pressure and break apart a casting..

    1. Thinking oneself smart is certainly easier than typing oneself smart.

      Sorry, I couldn’t resist… I fall into both camps but when I do something I’m reminded that Murphy in my copilot.

  25. Pete, thanks for the explicit demonstration of making a mockup for our mockups. The thought had occurs to me, but I’ve been burned too many times with almost right information at the shop. The outline dimensions have been well documented as well as the thickness per cell. But could you do a favor and measure the combined thIckness of say those four cells? I just reel at the idea of multiplying a rounded number four or five times. Considering the taper at the edge, I’d say stick a straight edge flat across the cell and extend it out to overlap a tape measure.

    Thanks for the help and insight.

    1. What you’re missing is the current limits. Jack thought he was buying a 75amp Clipper Creak unit for $3k. It turned out later that is was a 32amp unit. The Voltec unit in the picture is only rated for 15amps hence the lack of a heavy contactor. It looks like they used a pair of Tyco T90 power relays to switch the current. The low current allowed them to run the power through the PCB where the current sensors are easily mounted as a 8 pin SOIC surface mount hall effect current sensor IC. Cheap and effective but limited and certainly not worth $1k.

      A quick check of the Home Depot website shows both a 30amp and a 16amp unit going for $1k. The Schneider Electric unit is 30amps for $800. So the prices are all over the place depending on how much profit they want to make.

      The open-EVSE project shows just how little you need to build a EVSE using a modern micro-controller. As soon as the competition heats up (aka the Chinese get involved) the prices are going to drop like a stone.

    1. Lad,

      Very nice. I will dig into that later. Since I already have two EVSE’s this will not be a high priority but I may try my hand at building one. Looks to be an excellent way to go for an EVSE if you need one. I mostly don’t have the time since I work full time and then some. So buying one is usually the order of the day for me. I do however play when I can.

      Pete πŸ™‚

  26. I’m seeing an opportunity for the DIY CNC guys to improve on the FLATENUM pack. The case could be milled out of a few of sheets of HDPE. No aluminum required. Just seal the edges with a thin smearing of RTV to keep water out.

    – Doc

  27. I think the Flatenum is great. I like that it incorporates several of the ideas that have been developed over the weeks. The whole thing is flexible and actually very simple to make. Great stuff.

    I’m looking forward to seeing these A123 cells in an actual car now.

  28. Interesting show, I liked the news piece about fuel prices. Even though the price of gas here in Ireland is $ 8 per US gallon, I’d guess that the average cost of running the average gas guzzler in the US is probably higher compared to one in Ireland. So right now the average American is spending more money on gas a week than your average European. Already, we are downsizing our cars. By we, I mean my family. I think it will be interesting to see how it plays out in the US in terms of the economy class, car sales in the next couple of years if fuel prices remain high. It’s certainly an incentive to build your own Li EV; especially if you want to flip the bird at the stranglehold the gas pump has on us.

    Kind regards,


  29. With regard to the “flatnum” pack holding up the voltage and increasing acceleration, I have a few questions and comments. I’m not sure how an AC controller like the curtis handles power to the motor, but on a DC controller, the acceleration is almost purely a function of the amperage you send to the motor. The only time it will help is once the controller has reached 100% on with the PWM cycle. As long as the controller is putting out the maximum, say 1000 amps, to the motor, it does not matter the battery voltage. Motor voltage will always be less until the PWM reaches 100%. Having a pack at a higher voltage will reduce the PWM% all along and then also extend out the point to where it reaches 100%, and this extension is where you will have a gain. Great for racing, but on a street car, you won’t see much difference. It will most definitely give you more range, but acceleration will more likely be hurt by the weight in 0-60, but 60-80 may be better. For example your third gear run in the speedster ( ) you hit 100%pwm at about 58 mph. My question is, does an AC controller handle power in the same way (limiting motor voltage but giving maximum amperage during acceleration)?

  30. We’ve been through all this guys. Let’s NOT do it again.

    As soon as we are 100% PWM that’s WHEN the pack sags. In Speedster Duh, we drop from 188 volts to 147 at 1000 amps.

    That’s 147 kw. IF we could hold it up to 160, that’s 160kw.

    Yes, it only applies to acceleration. That’s the point where you need the 160kw.

    If we could hold 170volts at 1000 amps, that would be 170kw.

    We cannot put out more power than our battery pack can deliver. With the A123’s we would be able to deliver more.

    After we’ve delivered it, we have to replenish it from the slower but larger CALB 180’s. A claim was made that this would not happen or would not happen fast enough to use 60Ah of A123 with a 180Ah CALB pack. I intend to disprove that.

    You type. I’ll test.


    1. Jack said: “After we’ve delivered it, we have to replenish it from the slower but larger CALB 180’s. A claim was made that this would not happen or would not happen fast enough to use 60Ah of A123 with a 180Ah CALB pack. I intend to disprove that.”

      I’m 100% sure you will disprove that. Since the terminal voltages between both packs will be nearly the same regardless of their relative SOC how long it takes is irrelevant.

      With bottom balanced packs it is impossible for one pack to go to zero volts without the other one going to zero volts, or any other voltage for that matter. Same goes for charging. No need to charge each pack separately. That is the beauty of the LiFePO4 voltage curve.

      As for connecting the two packs it would be best to have very little voltage difference between them. When I hooked a TS 40Ah cell at near 0%SOC with one around 90-100%SOC the initial current between the two cells was ~28A through a 50A 50mV shunt. Initially the high SOC cell was at 3.339V and the low SOC cell was at 3.021V. Both had been sitting for a day or more with no loads attached. In less than 5 min the current was down to 10A with one cell at 3.245V and the other at 3.228V. So a voltage difference of 0.017V still could sustain 10A through 18″ of 1 gauge wire and a 0.001ohm shunt. This is why I don’t think it will take very long for the A123 pack to charge back up from the CALB pack. A small voltage difference will translate into quite large currents between them. The current will just rise to the occasion. πŸ™‚

      David D. Nelson

    2. I start feeling this becoming a truly collective project. Thanks for sharing David. To me TS experiment is a “connected vessels” analogue experiment.
      I was hoping Jack would not go, as bold as he is, for the real thing marrying packs before testing with couple of cells. Relieved is how I feel having your confirmation this kind of marriage actually does perform as expected.
      Nickkkkk’s suggestion is, so to speak, my path. I would rather power the small van, or a golf cart, which are sitting around in Jack’s garage, and check how A123 pack behaves off the bench in the real life case.
      What Jack is doing, most of the time I am afraid thinking of. Not that I am chicken, I don’t know enough and dispose of no means. If the latest green 12V urethane cast was around $600, flatinum will be around 10x worth in dollars. It sounds like a K2 expedition, or rather trip to Mars for my budget.
      Is anybody in position to hook a single CALB 180A and A123 20A cells in parallel and charge them, and report the findings? I would like to read confirmed Jack’s teaching, just for the record, before he does his experiment.
      I am aware Jack is doing what he does because he can afford it. Andyj helped me prepare for the next friday’s show stating it will work. I am a movie lover and this is becoming a thriller series for me.
      What I wouldn’t like to see is Jack leaving A123 streak of experiments for getting tired of working with them, because to many attempts failed.
      Improving by doing is the path I could agree with, but Jack’s scale is blowing my little mind and inexistent budget. I am more of the Soviet school where math has to prove it all before the money could ever come to the stage.
      I am growing like a person following this effort.
      EVTV is something else. There could be lots of parallel broadcasting channels and for EV movement they are needed, as many and as good as possible. When I have things to share upon my EV conversions, I’ll run one.
      I would already like to give tribute to EVTV, Brian and Jack for this community. It is very hard for me to say, but I know I will miss 2012 EVCCON as well, most probably.

  31. John Hardy’s proposal for connecting pouch cells is beautiful. Hope his Ice Free book is a good read as well.

    I grasp the idea of A123 pack serving coupled in parallel as a CALB pack stiffener and find it curious. This is something Jack will soon put to test.

    Before these packs are connected to the load, should they be charged separately?! They will have different totals V and Ah wise. Although both packs are prismatic LiFePO4 cells based, should they be charged separately even if matching their V and Ah dimensions?

    My concern is safety. I fear stated differences could lead to uneven charging time for the CALB pack and A123 pack stiffener. If that is among the possible scenarios while charging them connected in parallel, perhaps shouldn’t be left unattended.

    I wouldn’t like to learn the hard way, nor anybody else, particularly volunteering and pioneering for the benefit of us who at this point cannot do much more but type, read and imbibe.

    I am hooked to EVTV πŸ™‚ Great show and even stronger prospects for EV world due to gas prices.

    1. Nenad, “Do no harm”. Everything you say here will. It refutes everything Jack keeps telling us.

      V = Have to be the same (parallel connection)
      A = Irrelevant, (parallel connection)
      Differing states of charge = NOOOO!!!!

  32. Hey Andyj,

    If there are no “differing states of charge”, why stiffen in the first place? Doesn’t the different discharge rates precisely mean different Ah and so different State of Charge, at least transiently?

    Mike Kaindl

    1. Not at all. Indeed, the issue between these two types of cell revolves around diffusion rate.

      Basically thicker layers of active material on the electrodes lead to greater capacity, but lower POWER output ability. Conversely, thinner layers of active material lead to lower diffusion rates and so the ability to instantaneously produce HIGH levels of current, but diminishes the overall capacity of the cell.

      This is the inherent tradeoff. People get stuck on internal resistance and I keep telling you there IS no internal resistance. IR is an explanation of something, not a thing itself. It is “apparent” and the proper term is EQUIVALENT series resistance. But it’s not really a resistance. It’s a diffusion delay in a chemical process to produce electron current flow.

      By marrying the two cells in a parallel fashion, we hope to produce a hybrid battery. Like any such marriage, it is bound to be part horse and part cow, but we’re hoping for offspring that will offers slightly HIGHER voltages under load, while still having a fairly large capacity.

      The price we must pay is a slightly lower energy density level if we look at weight and volume for kWh across our TOTAL pack.

      Jack Rickard

  33. Mike:

    Same time zone problem that has ALWAYS offered great success to electric drive. I noticed it first with airplanes. 99% of an airplane’s life is spent in a hangar, not flying at all. When aircraft ownership is shared, much effort is spent resolving how conflicts in use are dealt with. In practice, they don’t occur indeed the airplane winds up sitting most of the time, and the owners have to set up special events just to see each other once a year.

    Similarly a car. It spends 99% of its life sitting waiting. Once rolling, it spends 99% of its time coasting.

    This is how a 32 hp motor can make 400 hp. Its rated at 32 hp for an HOUR. If we accelerate for an HOUR we are at twelve times escape velocity and approximately 1/3 of the total distance to MARS. In practice, we accelerate for 12 seconds, not for one hour.

    On the freeway at 70 mph, I”m averaging 70 amps. Now I can draw 1000 briefly getting there, but I can’t sustain it long without being in felony ticket range. And what we will see there is 55 amps from the CALBS and 15 amps from the A123. When we take our foot off the accelerator for the exit ramp, we will immediately see 15 amps of rebalance between the two. And by the end of the ramp that’s cut in half. The stoplight at the bottom of the ramp will take care of the rest.

    Jack Rickard

    1. Thanks Andyj for sharing the link. It is the most inspiring endeavor. At least for me, Paul’s example is the proof everybody should do something about their own transportation needs.
      A conversion of my 1986 318i BMW can wait. An electric bike is something I could much faster get the budget for. Not immediately but sooner then for any car conversion.
      Flatinum might prove to be the right choice for such a light vehicle or a hybrid pack as Jack is devising.
      Thrilling stuff indeed; 0 pounds/year in taxes and less then 1 pound/year for insurance on top of zero maintenance for the motor and batteries after 50.000 miles all together.
      It shouldn’t necessarily end up as spartan as Paul’s vehicle is. I am amazed Andyj, thanks again, a lot!

    2. Hi Nenad,
      Thank you. πŸ™‚
      The bike belongs to Cedric Lynch. The very same guy who designed and developed the Lynch/Agni motor!

      It is still running with the odd modification here and there and must be about 12 years old now if I’m not mistaken.

      A123’s on a bike should do well. Most seem to prefer headway cells due to their shape.

  34. Jack, The next time you mold a part try appling air between the mold and the part to help brake the part lose from the mold and it will help eliminate the vacuem issue created when pulling the part out of the mold.

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Verified by MonsterInsights