The White Lightning 818e thread

[John Rubicon]

Hi Gary,

I am using motors that were used in the Fiskar Karma sports car. It is very hard to get these motors since Fiskar has been resurrected by the Chinese company that bought them, but I have connections. My company matched them to our own inverters for use in heavy electric trucks. They've also been matched to a chinese inverter and a swiss inverter, but these last two are really not rugged enough for trucks. Or supercars ;-)

I haven't started a build thread yet. But I do have one that could become one called "So I just bought a partially complete GTM...now what?" in the Intro forum, and one at the "other" site under the same "handle". It'll be a couple of weeks before motors and batteries arrive.

JR

[Speedy G]

That's why you're talking about resolvers... You deal more with PMACs where startup torque is more difficult. There's really little reason to use a resolver in an induction motor since flux orientation is not dependent on the position of the rotor. Induction controls usually depend more on just 2 phase currents, voltage, drive frequency (an encoder is also required for DTC).

I'm surprised to hear the encoder wasn't accurate enough for shifting. I was doubting the accuracy of the subie tranny encoder since it's only 2 bit (4 pulses per rev), but I'm thinking of using a hardware counter scheme at high frequency so I can get more precision. I'll have to see if I can do the same for the motor encoder if we have the same issue.

I'm using Powerex IGBTs. I looked at the Semix stuff (specifically the SEMiX603GB12Vs IGBT). They seem nice, fast(er) and cheap! The reason I chose the Powerex (CM600DX-24S1) is because they also sell the gate driver circuit (VLA536-01R), which is what I called "conditioning circuit" earlier. I also found the cooling plate easily on the web (cp-1001-xp easily good for 3.3kW of dissipation). That makes for a really compact power stage for the power it can handle. Since I'm going for a simple solution, I'm not sure the replacement power stage will have all the bells and whistles that the original had (regen, charge w/ same IGBTs, etc). The reason for that is that the snubber circuit may be in the way of some of those, we'll see. The advantage of what I'm doing is that if I can find the pwm output at the microprocessor of the original drive, I can pretty much connect it straight to (but isolated) the new power stage without too much trouble. Also, it was a package deal so I might as well use everything I can.

I may have a board or two left over from the original drive (gate drivers probably for sure) after I replace the one with the powerex. Also I think 1 of the IGBTs is still good as well as capacitors (but they ARE 17 years old). I'll let you know what's left for reuse after I'm done.

I'm curious as to what switching frequencies they used, since 17 years ago (or even 20 since mine was number 2048) IGBT losses must've been much higher and since that's related to switching speed, I'm pretty sure they switched at lower freqs than current controllers. That, without taking into account DSP processing speeds available back then. Back then it was more art and faith than the lego style design that is possible today. You are inline with what I was thinking in terms of frequency, but there's nothing like a real life waveform to validate assumptions. I just need the max switching speed for the dimensioning of the stubber caps, so it can wait. I still have to analyze how they did the regen circuit, etc to see how much I need to redesign.

As far as the electrical isolation of the rotors, there's little room, so I'm thinking of using ceramic bearings for the coupling pulleys. The documentation shows me some phenolic disc coupling with the tranny, but since my talent isn't in high speed rotating assemblies, the less I do the better. The truth is that although the rig could probably take the 12,000 RPM for a bit, shfiting earlier doesn't hurt acceleration, 8-9k is fine. It's not like I get more torque at a higher rpm. The coupling pulley really drops efficiency above 5kRPM anyway.

BTW, I'm hearing sooo much stuff being done in electric trucks. It sounds like a lot more than regular EVs. I've spoken to a few that are doing truck trannies for electric motors as well. What's going on? Are trucks in the next decade all going to be EVs?

[John Rubicon]

Speedy,

I apologize for writing to you in such overly simplistic language about a subject that you clearly know well. Electrical Engineer? Power systems is rarely taught in Universities anymore, so most EEs are not well informed on these subjects.

I've met almost all the ACP Caltech brain trust, great men, Al C even took me for a spin in the original TZero. That was some real hocus pocus back then - you've seen inside the product... I guess the 7 Amp discretes were just good enough and cheap enough to make the product work. They took the same approach when they built the proto-pack that became the Tesla Roadster pack. Lots of small cells.

Those PowerEX half-bridges are rugged product. Little lossier as I recall. You know the Semikron folks make awesome 6-packs with integrated driver and heat sink, right? Assembled, tested, bulletproof - more expensive...See http://www.sindopower.com/Products-a...al6/S33/?cur=1

I should get my batteries in the next 2 weeks, and among other things, I want to resurrect my ACP unit and spin it up. I'll capture the switching frquencies for you under small load if you can write me a procedure showing me where, and how in the circuit. Would this be on the motor leads? It might be 2-3 months though. OBTW, wouldn't the phase board cap set teach you what the snubber should look like? Or is that more dependent on the IGBTs than the motor inductance? I had ACP replace all the stock 400VAC electrolytics with 440VAC versions. Running close to 400V battery voltage in the heat caused a bunch of their inverters to blow up. This was the fix.

Electric trucks? Yeah, check out www.transpowerusa.com. Batteries are just good enough to make the proposition feasible. These are the best available, IMHO. You be the judge. Full disclosure: I'm a founder.

So you really live in Columbia, the country known for such beautiful women? Or is it Columbia , South Carolina. I've been looking for a reason to visit Columbia, maybe we can have a "technical" meeting in person sometime and write that trip off :-)

[Speedy G]

Yep full gEEk. I actually worked on an earlier project as a coop (intern) engineer. We developed the craziest highest budget (probably) hybrid engine ever, the patriot project. That was at SatCon Technology back in '93. The contractor was Chrysler and the engine was a dual turbo-alternator (essentially two mini jet engines each coupled by shafts to alternators), flywheel storage, a water cooled induction motor and all the electronics to control all that, including the IGBTs. I think they applied for patents ranging from igbt designs, to turboalternator designs, to control design. The power electronics managed something like 3.5MW and only weighed 140lbs. The motor controller was also vector controlled (or FOC), so the thing was really on the edge of tech for the time. Everything was really high speed. The lowest speed thing in the engine was the 4 pole induction motor running at 24,000 RPM (750hp). Up from that we had the low speed turboalternator @ 50,000 rpm, the flywheel storage weighed 147 lbs and ran at 60,000 rpm, and finally the high speed turbo alternator ran at 100,000 RPM. All this fit in a racecar that was going to race at the then new American Le Mans series. There's a lot of lies about that project on the net, but the truth is Satcon did build it with around 30 engineers (mostly MIT), the budget was $35 million spent not $5 million as most web sites say, and I think 2 full engines were delivered (5 were the target). There's also a misconception on the net, they say the engine was 500hp, but they somehow didn't count the flywheel stored energy. The turboalternators were able to supply 500hp constantly, but the flywheel could supply another 250hp coming out of a turn, so effectively it was a 750 peak hp engine, with a nominal 500hp. Starting torque was 525ft lbs, with a flat 350ft lbs to 12,000 RPM.

The project was scrapped because the explosion of the flywheel was not contained by the container. Theory, and practice at SatCon said that carbon fiber flywheels turn into a kind of foam when they disintegrate at high rpm. For some reason this flywheel didn't disintegrate correctly. My though is that it was just larger than what we usually dealt with (notice I said larger, not higher speed).

These days I'm more on the software side of things, and I haven't touched high voltage in decades, but the reason for getting my feet back into electric motors is that I remember that project as being crazy awesome. It's really sad that a company that was able to do all that in '93-'95 still hasn't found its niche. Really brilliant and crazy cool bunch. Kers... yeah, we did that back in '94...

The thing about the ACP controller caps is that cap technology has also advanced a lot in the last decade. You just told me they used electrolytics, which is something I wouldn't use today since now we have high current film capacitors. That means they didn't have some circuit topologies available to them back when the drives were designed. Now you can just put a snubber cap directly on the IGBT, simplify the circuit, and improve efficiency. Back then they were probably forced to use a dissipative resistor in there somewhere, which just turned precious electricity into heat.

In any case. Cool to hear about your startup. It makes all the sense in the world, here even more. We probably have more hydro power than any other country, so going electric here is truly green. It is Colombia, South America. Booming place btw, and yes the current Ms Universe is Colombian, and Obama's security detail hooked up and got caught when they came down.

Btw on Linked-in, check out the Hybrid, Electric, PEV, and PHEV Engineering Professional Group Hybrid, Electric, PEV, and PHEV Engineering Professional Group. It's really relavant to what you're doing.

[Bob_n_Cincy]

Hi Guys, This is kids stuff, JK
I could not help myself. On my build thread you have probably seen my son Michael Working on and driving the 818.
Here is Michael doing final assembly on an 80KW flux vector AC inverter. It is headed for delivery truck application in Mexico City. I think this was in 1999.
mike 150 inverter.jpg
He's a little bigger now.
Mike and bob at Pikes Peak.jpg

I had 3 or 4 of the controller is these races. ACP was in Ohio State and Oklahoma.

[Speedy G]

Well, your drive looks waaay smaller than the ACP Gen1.

[Bob_n_Cincy]

Well, your drive looks waaay smaller than the ACP Gen1.

I had a liquid cooled version with 800 amp IGBT that I rated at a peak of 160kw used in 22 passenger shuttle buses. Size 12x24x7 I will probably use one of these in my 818e. I also might use a brand new controller out of an EV1. I should donate it to a museum.
lq.jpg

[Speedy G]

Nice Bob! Those look really cool btw. What did you do about the control algorithm? Did you program it all yourself? What about motor parameters and modelling? I love the name "Fluxdrive".

When do you think you'll start you e-build?

[Speedy G]

Hey John, I just looked at your website, NICE! You wouldn't happen to know Frank Falcone? I had an interesting discussion with him in the forum I told you about, so in fact I may have been talking to people in the same company about ev trucks!

An interesting idea popped up in that discussion. The prius uses something similar. The idea is to use a single planetary gearset as a transmision for high efficiency. If you could just use clutches to couple the wheels to either the ring gear OR the planet gears OR the sun gear, you could have a 3 gear tranny as simple and as efficient as can be. In fact that's how the powerglide tranny works, only with a bunch more stuff for autoshift, and the torque converter. If I were an ME I'd be all over that. I'd also do a dual planetary version to replace the differential and use 2 motors.

[John Rubicon]

Hi Bob,

Impressive! I like how its a family passion. You and I and Speedy seem to have been bitten by the electric performance bug. That feeling is so powerfully different from infernal combustion that it lasts for decades.

Are you selling inverter drives? If so, have you matched to an IPM machine?

JR

[John Rubicon]

Hey John, I just looked at your website, NICE! You wouldn't happen to know Frank Falcone? I had an interesting discussion with him in the forum I told you about, so in fact I may have been talking to people in the same company about ev trucks!

An interesting idea popped up in that discussion. The prius uses something similar. The idea is to use a single planetary gearset as a transmision for high efficiency. If you could just use clutches to couple the wheels to either the ring gear OR the planet gears OR the sun gear, you could have a 3 gear tranny as simple and as efficient as can be. In fact that's how the powerglide tranny works, only with a bunch more stuff for autoshift, and the torque converter. If I were an ME I'd be all over that. I'd also do a dual planetary version to replace the differential and use 2 motors.

Do I know Frank Falcone? Yeah. He was my MSME student while I was a professor, and I liked his skills so much he was my first hire. I'm looking at him in the flesh right now....small world, eh?

The idea you guys discussed is one that came out of my lab a decade ago when we were working with racing transmission builder Lenco. Their stuff is all planetary. I think this is in essence what Wrightspeed does now. Planetary units are lossier than spurs gears when not in 1:1. And the Lenco unit of the day was not meant for more than a dozen trips down the track between rebuilds. I passed on spending time on that due to the investment required and the absence of a performance market at the time. I know of a Powershift product that converts all the auto functions to manual for racing use, but I still wouldn't want to put 1050 ft-lbs through it.

JR

[Bob_n_Cincy]

Nice Bob! Those look really cool btw. What did you do about the control algorithm? Did you program it all yourself? What about motor parameters and modelling? I love the name "Fluxdrive".
When do you think you'll start you e-build?

The deep FOC code was already done, I did write the all the frontend code. I think I still own Fluxdrives, fluxvector, and electricmotorsports web domains.

Hi Bob,

Impressive! I like how its a family passion. You and I and Speedy seem to have been bitten by the electric performance bug. That feeling is so powerfully different from infernal combustion that it lasts for decades.
Are you selling inverter drives? If so, have you matched to an IPM machine?
JR

I been working on motor controller for the last 35 years. I'm having a blast working on my turbo 818.

I sold off my business in 08 and now work in the ev industry.
Yes, A lot of stuff I've done in the last couple of years has been with IPM motors.

Talk about family. This is after we won the 2010 TTXGP national championship. My daughter, son and me on the right. If you guys know the name Jeff Major from the EVDL. He is a close friend in the back row.
VIR2010.jpg
Bob

[Gary Livingston]

Hey John,
Very kewl. Thats some serious power there. Should be a nice package by the sounds of it. What battery are you using? The Fiskar A123 pack perhaps?

[rtz]

Maybe EnerDel?

http://www.diyelectriccar.com/forums...les-86089.html

[Gary Livingston]

Maybe EnerDel?

http://www.diyelectriccar.com/forums...les-86089.html

This is Jeff Major, Bob's friend. I learned a lot from him. He helped me to design/build a 13" series wound motor from parts used from a couple other motors. I think I have some 2400+ posts on that forum also. The enerdels are used on Richard Hatfields Lightning bikes. (the team Bob and Jeff are on).
I was referring to to John Rubicons' post where he said his motors and battery were coming. His motors being from the Fiskar, are the Jing-Jin ones I think and the battery pack from the Fiskar cars was A123. So.... was wondering if he is getting one of the batteries from the Fiskar builds also. Nice 20KWh pack.

[John Rubicon]

This is Jeff Major, Bob's friend. I learned a lot from him. He helped me to design/build a 13" series wound motor from parts used from a couple other motors. I think I have some 2400+ posts on that forum also. The enerdels are used on Richard Hatfields Lightning bikes. (the team Bob and Jeff are on).
I was referring to to John Rubicons' post where he said his motors and battery were coming. His motors being from the Fiskar, are the Jing-Jin ones I think and the battery pack from the Fiskar cars was A123. So.... was wondering if he is getting one of the batteries from the Fiskar builds also. Nice 20KWh pack.

I have some experience with Enerdel's packs. Overall good tech, No, I'm using an excess of Nissan Leaf modules. Alligator versions. 6C on their best day, but if you know you're going to parallelize them, then that is not an impediment. A123 i a great drag racer choice, but I think the Leaf module is better when energy is also a consideration. I'm shooting for 140 miles range - which by the time I finish the vehicle, may be child's play for new tech cells in development now. Still hard to argue with high-volume OEM cells, If I can fit 1500lbs in the vehicle :-0

JR.

[Gary Livingston]

I have some experience with Enerdel's packs. Overall good tech, No, I'm using an excess of Nissan Leaf modules. Alligator versions. 6C on their best day, but if you know you're going to parallelize them, then that is not an impediment. A123 i a great drag racer choice, but I think the Leaf module is better when energy is also a consideration. I'm shooting for 140 miles range - which by the time I finish the vehicle, may be child's play for new tech cells in development now. Still hard to argue with high-volume OEM cells, If I can fit 1500lbs in the vehicle :-0

JR.

Oh ok kewl. These packs are starting to come available from wrecks also. The NMC (lithium nickel manganese cobalt oxide) are a better chemistry than many for energy density. The trick with these cells it seems, is to limit the the charge voltage and high temperatures if you want to preserve cycle life. Seems Nissan may not have fully captured the effect of one or both of these influences since many owners in. California have experienced a 25% loss of capacity over 1 - 2 years. If you are picking up a used pack, might be better to source one from the more northern states.

[John Rubicon]

Gary,

you're right on capacity loss. I have the newer "Alligator" gen cells that have better cooling. Also, the 2015 pack I received is pristine ;-) Charge voltage is a little more under my control than in the Leaf, also. I'd be interested in a life vs.SOC at max charge kind of graph if it were in the public domain. I expect that the last 5-10% charge does most of the life-reducing damage.

[Gary Livingston]

Gary,

you're right on capacity loss. I have the newer "Alligator" gen cells that have better cooling. Also, the 2015 pack I received is pristine ;-) Charge voltage is a little more under my control than in the Leaf, also. I'd be interested in a life vs.SOC at max charge kind of graph if it were in the public domain. I expect that the last 5-10% charge does most of the life-reducing damage.

Oh, that's good. I've heard them termed "lizard" cells and have some more secret sauce added to preserve high temp service. Far as I know, 4.1 volts is the magic number. Terminate the CV phase of charge below 4.1 and it should make the difference. Did you get one pack? what's the plan for configuration/voltage and what do you suppose you can get out of them peak?

Cheers!

[Speedy G]

Ok, guys since this turned into a battery discussion, you guys got me wondering about what I should do. I need ~1000 amps for drag racing and >120 mile range would be cool. Motors were rated at 164kW and 165kW @ 350V. I was originally thinking of getting just 120 100Ah CALB cells and hope for the best. Now you guys got me wondering if I should go with something like 120 70Ah CALB cells and 30Ah of A123 cells or Enerdel cells.

Gary you talked about the A123 cell pack as a sprinter pack and CALB as a range pack. Is there a ratio of CALB Ah vs A123 that you'd see as ideal? Is there a way to calculate or is it just looking at graphs and guestimating? What would that ratio be for Enerdel? Let's say the budget is 20k for the battery pack.

[John Rubicon]

Oh, that's good. I've heard them termed "lizard" cells and have some more secret sauce added to preserve high temp service. Far as I know, 4.1 volts is the magic number. Terminate the CV phase of charge below 4.1 and it should make the difference. Did you get one pack? what's the plan for configuration/voltage and what do you suppose you can get out of them peak?

Cheers!

Gary, you're right, it's Lizard. Though Alligator sounds way cooler ;-) I have strong rumors that they really don't like more than 375A discharge (not even enough for 1 motor at peak), so I am planning around that for peak power. If you know of data that says they can deliver more amperage, reliably, please do share.

Just one pack to start, to get it proofed for function. I may just have to go higher than 450V to get what I want from the system.

JR.

[John Rubicon]

Ok, guys since this turned into a battery discussion, you guys got me wondering about what I should do. I need ~1000 amps for drag racing and >120 mile range would be cool. Motors were rated at 164kW and 165kW @ 350V. I was originally thinking of getting just 120 100Ah CALB cells and hope for the best. Now you guys got me wondering if I should go with something like 120 70Ah CALB cells and 30Ah of A123 cells or Enerdel cells.

Gary you talked about the A123 cell pack as a sprinter pack and CALB as a range pack. Is there a ratio of CALB Ah vs A123 that you'd see as ideal? Is there a way to calculate or is it just looking at graphs and guestimating? What would that ratio be for Enerdel? Let's say the budget is 20k for the battery pack.

Hi Speedy,

I wouldn't push the small Calb's that hard. That's 10C for the 100Ah units. Calb sells some metal jacketed cells that can go above 6C and likely 10C for short bursts without too much drama. I think they are called CAM-series cells. More $, but they have a fighting chance because they can cool.

On the subject of ACP power, you know that the Gen 1 peak power might last 15-20 seconds, right? That was my experience, at least. Then the hardware must cool for quite a while before it can make another burst. Derate I estimate was to half of peak power or less on a warm day. This is one reason I went with the liquid-cooled IPMs I have coming. They derate to about 75% of peak power, and that takes minutes to happen, not seconds.

If you have $20k to spend, look into whether the folks at Plasmaboy Racing http://www.plasmaboyracing.com/whitezombie.php will build you a pack from the Dow-Kokam's they were using. They were good at this electric drag business ;-)

IF you really want 120 miles AND drag racing capability, then vehicle weight, volume and balance start to be a challenge, as the pack gets larger. That's about what I'm looking to build. So we are in a similar boat. I thought the 165kW ACP drives flowed 550A each, but not sure if that was DC or phase current. My IPMs require about 450A DC each.

JR

[Gary Livingston]

Hey Speedy. Sorry for hijacking ur thread I guess it's all a common pursuit. My initial thoughts are similar to John's on most points. (One point there is that the CAM series cells are the same chemistry as the CA's except for the one CAM model CAM 25P - power cells that haven't quite hit prime time yet)
Personally, I like to back into the battery selection from the controller and motor capability. You say your motors will do 330 kW peak. So if you want to max them out, and you can feed them 500 amps each, you will need to maintain 330 sag volts at 1000 amps. Now what upper voltage will your controllers take? If it's 425 for example, and you want to use LiFePo chemistry, you must determine number of cells based on charge voltage. If you charge to 3.65vpc, you can only use 116 units. Voltage settled after charge will be about 3.35x116 or 388 best case. It will be lower soon as u draw from it. Getting a battery to sag only 50 volts at 1000 amps is a lofty goal.
I don't know if these numbers represent your case, but it gives you the method to determine.
This is what leads to the selection of higher c rate cells as they tend to sag less as well as drag racers looking for light weight packs that will enable them to race 1/4 mile not 120 miles. Clarify some of those points and we can get to the next step.

[Speedy G]

I figured you guys know so much about batteries that leaving some of that info on the board would be good for the community. It's not a hijack at all.

Yeah, I know the ACPs Gen I's are cheetahs, not marathoners. That's also why I'm looking at improving cooling as well, but at first look, the motors look pretty difficult to water-cool. I know mine are modified, but I don't know how they were modified to be able to handle high power longer (2x Bonneville @250mph, 90 s for each run). I know the insulation is still good so Ed didn't cook them. There is one trick, however. If you run them in the upper RPM range, less current will flow through the motors, and they'll run cooler. Also peak efficiency is above 6000rpm and below 120kW (lower than 100kW at higher rpm).

Max voltage in the controller is 420V with original specs so let's assume 110 cells initially, 70Ah pack CALB only (initially for a single motor @500A max or 7C. The upgrade to the second motor would also include a sprinter battery pack. The sprinter pack would probably need to supply at least the other 500A, while only having 30Ah. That would mean running the sprinter pack @ close to 17C. The problem is how to determine the sag of the CALB pack vs the sag of the A123/Enerdel pack to determine how many cells in parallel I'd have to run in the sprinter pack. Is there a graph/equation out there with 5C, 10C, 20C, 30C sag info for those cells?

[John Rubicon]

Speedy, do you intend to keep the two packs separate or to Siamese them together cell by cell to be able to use just one BMS? I know of one build in the siamese design, but it's a cabling and packaging non-starter for me.

In my experience, the cell info you're looking for isn't released by manufacturers because it's outside the recommended operating envelope. Even for A123. And, our suppliers don't share anything like a map of internal resistance v. SOC. We have to generate our own data to have any faith in how a given string of cells will behave under extraordinary loads, and to be sure that connection resistance is low enough at these currents.

Mixing two chemistries, in parallel packs, with an end goal of controlling the current split across a range of SOC, is a difficult problem for which modeling can only get you so close. I've modeled it, you can too, but testing will be needed to fine tune it. I'm staying away from the split pack idea if I can, for just this reason. If I can fit enough capacity in the vehicle at 6C to meet the max current need, I'd rather.

JR

[Speedy G]

Ok, we'll see what I end up doing, but there's time, since I'm only buying the batteries after everything else is done. Hopefully CALB and others are working on some super cheap, ultra lightweight and ultra powerful batteries in the meantime. My kit is getting picked up on the 24th so it should be in my shop maybe a month later due to import delays, etc.

One thing that may be of use is the fact that I'm using 2 separate controllers and maybe the second controller can start pumping serious current only after the throttle is over 50%. Since the second motor will be using belt drive, this may be the most efficient strategy anyway. I'm using a control system to read the throttle and act accordingly, so the added complexity is just a couple lines of code and an extra wire. I can probably play with regen as well to keep the two separate packs somewhat balanced in terms of charge, or even use an extra IGBT to have the pack with more current to supply current to the first motor. In any case, there's options other than siamese and two connected parallel packs. I also looked at ultracaps btw, but surprisingly they can't hold the current. I thought that's what they were good at.

On more question, battery related. How to cool the batteries? I was hoping to find some kind of aluminum extrusion that I could use for water cooling the batteries that could use water cooling, but no go so far. Any ideas on cooling CALB batteries? Would cold air being circulated above the batteries be enough, would air circulation on the sides do anything? Any ideas?

[John Rubicon]

Hi Speedy,

the two separate ACP controllers may have GFD circuits that fight each other on the same DC bus. Two separate packs, one for each controller, may be the only easy way that can work.

If you can, find a way to modify the Gen 1 pre-charge hardware to be friendlier. Right now you have to spin-on and spin off a fine-threaded MIL connector with an active mechanical interlock. Using the solenoid mechanical interlock signal to close a modern contactor on the outside, seems much friendlier. The Gen 1 had an early precharge failure mode that stressed the caps and caused them to fail prematurely. Maybe that was fixed before yours were built. Either way, a purely electrical pre-charge would help a lot. You do not want to leave the pack connected when the vehicle is to sit for long periods.

I have used a small film resistor to allow paralleled packs to come to equilibrium with each other when not connected to the inverters. May not work with the ACP GFD though.

I'll try again to dissuade you: the 10C discharge you seek from plastic-cased 100Ah cells (Calb, Winston, others) will kill them pronto. Even 5C pulse is pushing it a little. If you can afford them, the CALB CAM series with the metal jackets, non-conductive mesh cell spacers and a stirring fan will probably be OK. Water cooling the surface of a polymer jacket cell really doesn't work very well. Too much thermal resistance. Plus you have to trust your plumbing a lot to risk moistening the high-voltage :-0

EV West may have some liquid cooled battery modules made by Tesla. 18650 clusters with a cooling loop built in. These would be used items taken from dismantled EVs.

What sort of BMS are you planning?

JR

[Speedy G]

I think Ed might've just taken out the cap GFD circuits. The guys at ACP said "the capacitor control system" was missing. I still have no clue what that is, but it might include the GFD circuits. I'll have to take a look.

Actually I'm right there with you on the CAM cells, which is why I was thinking of just getting 110 CAM (70Ah) cells for the first motor and for the second motor, get some A123 cells(30Ah or less). I was looking at the ANR26650M1B cells (like 550 or more of them). I'm really hoping for the new CALB cells, and maybe the new pouches from A123. I can have several gas pedal profiles, so on in "insane" mode I'll be hitting the CAM cells to 7C and the A123s to 20C (~100A per cell). On the regular profile, I can do 3C-4C on the CAMs, which will still give me almost full horsepower, just not full torque, and drive the A123s at 10C-12C. I'll have to do some calculations to see what the optimum gas pedal profiles would be.

As far as the BMS... Sounds like a complicated proposition. I really like the ligoo displays, but honestly that's like saying I want to buy a car, a red one. The rest of the details are just beyond me, especialy since these systems are expensive, but they're also Chinese. So am I going to spend 3Gs on something to make my batteries more reliable, but it may not be reliable? That's why I was thinking of just monitoring voltages and manually bottom balancing. If you or anyone have any input, it would be greatly appreciated.

[Bob_n_Cincy]

As far as the BMS... Sounds like a complicated proposition. I really like the ligoo displays, but honestly that's like saying I want to buy a car, a red one. The rest of the details are just beyond me, especialy since these systems are expensive, but they're also Chinese. So am I going to spend 3Gs on something to make my batteries more reliable, but it may not be reliable? That's why I was thinking of just monitoring voltages and manually bottom balancing. If you or anyone have any input, it would be greatly appreciated.

I will be top balancing during every charge cycle.
That is the best way in my opinion. Been doing that way since my first lithium truck back in 2008.
The harder you push cells, the more important balancing and monitoring is. As you push them harder, the center of your packs will get hotter.
Bob
trucks.jpg

[Gary Livingston]

Some chemistries are a bit fragile when it comes to balance, like RC Lipo. My opinion for LiFePo4 is to bottom balance and undercharge. I set mine to 3.45vpc. This will also prolong the life of the cells (under charging that is) No BMS (balancing) but BMonS system can be as simple as a split pack comparison. If you have a well balanced pack of cells, this will work best. You can go either way however, bottom balancing has an advantage related to running them to empty. If they are bottom balanced, you can drive the car till it wont move any more and since they all hit bottom at the same time, there is no damage. "bottom" can be defined. Top balancing will cause them to run out at slightly different times since capacity varies a little bit. At this point you will be driving several hundred amps through an empty cell and kill it. The top of the charge voltage adds almost nothing to the capacity if you study the charge curve. It just isn't worth it. Additionally, and these points all get hotly debated, BMS's have been responsible for he majority of thermal run away events, particularly when they control the charger cut off. Over charging is the most dangerous thing you can do. If you choose to use a BMS and actively top balance, please have a redundant control circuit to monitor the pack voltage and disengage the charger. This will work ok. IF not, don't be parking it in your garage.
Most of the original thinking has been BMS and top balancing. . .but now, most folks are not going this way. . . and there have been no issues. I've run my CALB pack at 10C many many times and never had balancing issues. I use to be paranoid about it and checked the cells constantly. It's gets rather boring after a few years. They are just stable.

[Bob_n_Cincy]

Hey Gary,
Thanks for your opinion. Lets leave the BMS/balancing discussions to the EV forums. Do you Agree?
Bob

[Gary Livingston]

Hey Gary,
Thanks for your opinion. Lets leave the BMS/balancing discussions to the EV forums. Do you Agree?
Bob

Hey Bob.
Yes, for sure. There has been plenty of discussion in other places. You are quite the gentleman.

Gary

[Speedy G]

Haha, that's an interesting discussion. The thing is I'm more interested in the how to balance than whether top bottom, left or right Do you guys have preferences in brands of BMS's or do you guys build your own BMS or BMonS?

Gary what do you use to bottom balance? Do you balance them by hand (down to say 2.75V)? What do you use to discharge the cells? What's your bottom hard limit when loading the batteries? 2.8V? Also, since you've used the A123 cells how do they compare in terms of balancing?

Bob, what systems or procedures have you used to top balance? Do you balance by hand? Any preferred BMS?

John, do you have any input on the subject?

[Gary Livingston]

Haha, that's an interesting discussion. The thing is I'm more interested in the how to balance than whether top bottom, left or right Do you guys have preferences in brands of BMS's or do you guys build your own BMS or BMonS?

Gary what do you use to bottom balance? Do you balance them by hand (down to say 2.75V)? What do you use to discharge the cells? What's your bottom hard limit when loading the batteries? 2.8V? Also, since you've used the A123 cells how do they compare in terms of balancing?

The first time I ever did it, I simply connected them all in parallel and left them for two weeks. It may not be the best way, depending on state of charge. While voltage isn't a great indication of state of charge for LiFePo4 in the middle of the charge state particularly, it still is somewhat relative if measured to a couple decimal places on batteries that have not seen and load or charge for a while. The next time, I ran them fairly low in a series pack and used a large resistor to bleed them down. 2.7 volts is a good number and nicely into the knee of the discharge curve and close to empty. Now I use a Powerlab 8. I have two of them actually. You can also automate the process with a small Lightobjects voltmeter with programmable outputs and a large resistor. Jack sells this stuff on the EVTV web store if you want to get it all in one place. The voltage is a bit "spongy" meaning you may have to cycle them a couple time to get them to stay where you want. I like to use a volt meter that reads three decimal places. The powerlab is nice for checking capacity as well.

I have top and middle balanced the same way also, manually.

The only BMS I ever used was on AGM lead acid. Floodies dont need anything, but sealed lead does. . . as you do not want to boil them to balance. In fact, a lot of the beliefs on this subject are a hang over from lead acid days. Ive heard good things about the Orion and the Mini-BMS . . but honestly, I'm not the best one to ask. A BMS will eventually balance your pack after cycling it some number of times. It may take a while though as most of them do not shunt a lot of current. One other point to be aware of if you use one is the parasitic draw. There have been a number of packs ruined by leaving them unattended over long periods and the BMS itself draining the pack. (Sorry Bob, just raising the point for awareness sake)

As for load voltage. . . it's an interesting topic. The voltage when there is heavy charge or discharge isn't really the voltage. It is, but it isn't. I don't charge more than 3.5 vpc, actually usually 3.45vpc. That said, if the charger was still in CC mode, I wouldn't worry too much if the voltage was higher. Of course to get it higher means a lot of current and there could be limits there. Same on discharge. The voltage can sag quite low, but the "real" voltage of the cell will show when at rest. So.... if you are not already confused.. . . all batteries will give their peak power at 50% of their nominal volatge . . . as a general rule. So, if you the ability to program voltage under load and you are using LiFePo4, I would set it to 1.6 or 1.7 vpc. Resting voltage limits are quite handy also, to progam a "limp mode" meaning. . git ur arse home.

The good news is that the cells now tend to come much closer in capacity than they every use to.

Cheers,
Gary

[Bob_n_Cincy]

The thing is I'm more interested in the how to balance than whether top bottom.

Bob, what systems or procedures have you used to top balance? Do you balance by hand? Any preferred BMS?

John, do you have any input on the subject?

I have used 3 systems during the years. The first one was a purchased system. The last two I designed and were proprietary to the companies I was working for.

On a new vehicle (96 cells) we would charge the up 3.6v in parallel before instillation into the vehicle.

One of the hard things to do in any EV was replacing one 700 AH cell into a half charged pack and trying to match it to the rest of a pack.
The replacement cell could be off by 100s of amp hours and take weeks to get in line.

To answer your question on how to balance:
To balance, you must bring each cell to the same state of charge. (SOC)

Bottom balancing is done by individually draining all the cells until they are empty. (2.7v)
Typically using a controlled bleed resistor on each cell while the pack is not being used.

Top balancing is done by individually charging all the cells until they are full.(3.6v)
Typically using a controlled bleed resistor on each cell while the pack is being charged at very low current.

This website has a lot of good information on BMS systems.
http://liionbms.com/php/index.php


Bob

[Speedy G]

Awesome info guys, thanks a lot.

Gary, good to know about the load voltage, it's something I was thinking about. So it's hard to be 100% safe on full load, since sag takes you way below safe full discharge voltage. The ACP controllers shut down below 220V, so with 110 cells that would be 2V, we'll have to see what we do about that. Btw, the powerlab 8 looks very helpful.

So I'm thinking it's possible to conceive of a BMS that could do bottom or top balancing with the same hardware. After all, it sounds like you're always draining to a voltage whether you're top or bottom balancing. Also, I can see how a BMS can kill the batteries either way. The easiest way to power a BMS drain circuit is with the battery it is "clamping". So if you're bottom balancing and forget to charge, oops, your battery's dead. Also, on the top side, it's easy to exceed the drain current used to "balance" a battery top side using any charger. I just saw that the Ligoo BMS uses 3 channels of 300mA battery each for balancing... 300mA and I'm charging with 20kW? Even at 360V, that's 55A, balanced by 300mA?? No wonder it takes weeks to "balance" a new replacement cell. I realize the voltage is "spongy" due the hysteresis in the charge-discharge cycle, but still, I would think balancing current capacity should be higher... Interestin topic, we'll see what I end up doing. I'll probably bottom balance initially, and later see what I can do to automate the balancing.

[Bob_n_Cincy]

Awesome info guys, thanks a lot.

I would think balancing current capacity should be higher...

Hi Speedy
I have designed BMS boards with 2 amp max bleed.
If one low cell, you might have 95cells x 3.6v x 2a = 684 watts you need to get off the circuit boards and out of a sealed battery box. Can you say, Easy Bake Oven.
Bob

[John Rubicon]

Hi Guys,

I inferred from Bob's post a hesitation to open up the BMS religious discussion, so I waited. It can get testy ;-)

I am not a fan of bottom balancing, but then again, I've been disappointed by batteries too many times, and want to have information to use in vehicle control. And I'm the kind of guy that drives his gas car down to the "E".

Plenty of people make bottom balancing work for themselves just fine. And it's inexpensive(!)

Bob, you've used 700Ah cells too? Cells the size of briefcases seldom have the same characteristics in my experiences. There is no practical accelerated life testing for large cells like these. One thing the Tesla approach does have going for it - massive parallel redundant current paths and uniform cell resistances.

I've experienced about a dozen systems and they all have their kinks. I did like the Ligoo system's UI that Speedy pointed to, but I couldn't tell a thing about how it works. bit resolution and accuracy, connector type, CAN chip, isolation, I needs to know. Maybe I didn't look enough.

I'm currently working with a couple different folks on a wireless BMS systems. The tech seems to work very well.

JR

[rtz]

I say keep it simple and no BMS. When you buy your batteries; drain down each one to the exact same amount. Now wire them up and charge up your pack. Now you are set. Never, ever, discharge below a certain amount, and never, ever charge above a certain amount.

Read this in full for the exact details to the one time procedure: http://media3.ev-tv.me/cellcare.pdf

Any more questions call, email, or stop by Jack Rickards shop(see webcam) and talk to him directly about it: http://evtv.me/

[rtz]

the kind of guy that drives his gas car down to the "E".

Keep in mind that most(all?) in tank fuel pumps use the fuel to keep the pump cool. So in the hot summer when that pump is running in air compared to if the tank was full?