Even Newer 4 to 24-cell Battery Management System (BMS)

[The Mighty Volt]

Could somebody fill me in, using basic English, on the nature of the improvement in the new boards? Thanks!!

 

[GGoodrum]

Could somebody fill me in, using basic English, on the nature of the improvement in the new boards? Thanks!!

There are a number of issues with all of the v2.x BMS variants, including noise generated during shunt operations, voltage drop issues, no workable end-of-charge detection function and just a general "whackiness" when the shunts operated. Richard and I started down what became the v4.x path, to address these issues. Actually some of the problems we didn't even know were problems, like the noise and voltage drop issues. These became more pronounced with the v4.x variants, because we were trying to double the shunt current, to 1A. Anyway, it took a boatload of iterations, be we finally tackled all of these issues, and latest shunt circuits are rock-solid.

The other big difference between the v2.x and v4.x versions is how the shunts are used to balance the cells. The older v2.x scheme, which is the same way virtually all shunt-based BMS designs I'm aware of work, is to set the HVC trip point right at the voltage where the shunts are fully on, just below the point they go into overload, or get "swamped". The "charge to" point is set a bit above this, and then current limiting throttling logic is used to keep the voltage right at that point. The net result is that once all the cells are full, the shunts will all be in full bypass. Having these shunts "cook away" at the end is one thing, when the max shunt current is only 200-300mA, but at the the 500mA the v2.x is designed for, things get pretty toasty. At 1A, it is more than toasty, it will get bloody hot! Fans are mandatory.

With the latter v4.x variants, a totally new shunt/balancing scheme is used. Instead of having all the shunts "cook" at the end, what we do now is only have the shunts come on for the first cells that get full, and then only enough to keep the cell from going over the balance point. The "charge to" voltage is set to be the same as the balance point, which is the voltage where the shunts first start to come on. This "by exception" shunt scheme means that at the end, when the cells are all full/balanced, none of the shunts are on at all. This contrasts sharply with the v2.x scheme where every shunt is cooking away at full power at the end.

Another problem with the v2.x design is that there is no definitive end-of-charge detection scheme. You can't just wait for the current to drop to something like 1/20th of the max current, the way many RC and even some SLA chargers work, because the current will never drop below the max shunt current. What the v2.x circuit did was simply wait for all the shunts to come on, and then shut things down. The problem with that is that the last cells that cause the "ALL SHUNTS ACTIVE" signal to trip, have just gotten to that point, and those cells won't be as full as the rest.

With the shunts "by exception" scheme, in the v4.x design, the shunts are off at the end, if the cells are full/balanced, so the current will drop all the way down to zero, eventually. Now we can actually just measure the current, and do a shutdown when the current drops below some selected point. We have a pot that let's this low-current shutdown to set from 0-2.5A.

The final difference in the two designs is how the HVC signal is used. As I said, in the v2.x design, the HVC signal is set to the shunts-fully-on point, and is used by the throttling logic to keep the shunts cooking, but not where they swamp, and allow the cell voltage to rise above the HVC point. The HVC signal is used differently in the v4.x variants. In this case, the HVC trip point is used as a "failsafe" point, or in other words, a safety net to keep errant cells from getting overcharged. This HVC trip point is set a fair bit above the balance point, enough that significantly out-of-balanced, but otherwise healthy cells won't ever have their voltages get this high. The 1A shunts wil handle significant imbalances, just on their own, but if a weak cell, with a lower capacity, gets full a whole lot earlier than the rest, it can still swamp a 1A shunt. This higher HVC trip point will keep it from going over this point by engaging a very simple "throttling" function that will let the shunt work to get the cell back in check, and then the process will continue on. If the shunt swamps again, and the HVC point is reached, the cycle will repeat. Eventually the shunt will be able to hold the voltage on its own and the process will continue on to completion.

Anywa, these are the major differences.

-- Gary

 

[The Mighty Volt]

Could somebody fill me in, using basic English, on the nature of the improvement in the new boards? Thanks!!

There are a number of issues with all of the v2.x BMS variants, including noise generated during shunt operations, voltage drop issues, no workable end-of-charge detection function and just a general "whackiness" when the shunts operated. Richard and I started down what became the v4.x path, to address these issues. Actually some of the problems we didn't even know were problems, like the noise and voltage drop issues. These became more pronounced with the v4.x variants, because we were trying to double the shunt current, to 1A. Anyway, it took a boatload of iterations, be we finally tackled all of these issues, and latest shunt circuits are rock-solid.

The other big difference between the v2.x and v4.x versions is how the shunts are used to balance the cells. The older v2.x scheme, which is the same way virtually all shunt-based BMS designs I'm aware of work, is to set the HVC trip point right at the voltage where the shunts are fully on, just below the point they go into overload, or get "swamped". The "charge to" point is set a bit above this, and then current limiting throttling logic is used to keep the voltage right at that point. The net result is that once all the cells are full, the shunts will all be in full bypass. Having these shunts "cook away" at the end is one thing, when the max shunt current is only 200-300mA, but at the the 500mA the v2.x is designed for, things get pretty toasty. At 1A, it is more than toasty, it will get bloody hot! Fans are mandatory.

With the latter v4.x variants, a totally new shunt/balancing scheme is used. Instead of having all the shunts "cook" at the end, what we do now is only have the shunts come on for the first cells that get full, and then only enough to keep the cell from going over the balance point. The "charge to" voltage is set to be the same as the balance point, which is the voltage where the shunts first start to come on. This "by exception" shunt scheme means that at the end, when the cells are all full/balanced, none of the shunts are on at all. This contrasts sharply with the v2.x scheme where every shunt is cooking away at full power at the end.

Another problem with the v2.x design is that there is no definitive end-of-charge detection scheme. You can't just wait for the current to drop to something like 1/20th of the max current, the way many RC and even some SLA chargers work, because the current will never drop below the max shunt current. What the v2.x circuit did was simply wait for all the shunts to come on, and then shut things down. The problem with that is that the last cells that cause the "ALL SHUNTS ACTIVE" signal to trip, have just gotten to that point, and those cells won't be as full as the rest.

With the shunts "by exception" scheme, in the v4.x design, the shunts are off at the end, if the cells are full/balanced, so the current will drop all the way down to zero, eventually. Now we can actually just measure the current, and do a shutdown when the current drops below some selected point. We have a pot that let's this low-current shutdown to set from 0-2.5A.

The final difference in the two designs is how the HVC signal is used. As I said, in the v2.x design, the HVC signal is set to the shunts-fully-on point, and is used by the throttling logic to keep the shunts cooking, but not where they swamp, and allow the cell voltage to rise above the HVC point. The HVC signal is used differently in the v4.x variants. In this case, the HVC trip point is used as a "failsafe" point, or in other words, a safety net to keep errant cells from getting overcharged. This HVC trip point is set a fair bit above the balance point, enough that significantly out-of-balanced, but otherwise healthy cells won't ever have their voltages get this high. The 1A shunts wil handle significant imbalances, just on their own, but if a weak cell, with a lower capacity, gets full a whole lot earlier than the rest, it can still swamp a 1A shunt. This higher HVC trip point will keep it from going over this point by engaging a very simple "throttling" function that will let the shunt work to get the cell back in check, and then the process will continue on. If the shunt swamps again, and the HVC point is reached, the cycle will repeat. Eventually the shunt will be able to hold the voltage on its own and the process will continue on to completion.

Anywa, these are the major differences.

-- Gary

Hey- thanks for taking the time and effort to spell all that out for me and the other members who are behind on this kind of thing. You deserve every dime you make and I for one would be happy to put as much $$$ as I can afford towards your operations.

If you kindly keep me filled in I will not be shy to buy these new boards when they come out.

Thanks again, best of luck with it.

TMV.

 

[ejonesss]

will this eliminate the issue where the negative most cells are lower than the rest?

so if i let it charge overnight then in the morning i unplug and replug in the charger all 16 cells for example will show full and not have 4 still charging?

Could somebody fill me in, using basic English, on the nature of the improvement in the new boards? Thanks!!

There are a number of issues with all of the v2.x BMS variants, including noise generated during shunt operations, voltage drop issues, no workable end-of-charge detection function and just a general "whackiness" when the shunts operated. Richard and I started down what became the v4.x path, to address these issues. Actually some of the problems we didn't even know were problems, like the noise and voltage drop issues. These became more pronounced with the v4.x variants, because we were trying to double the shunt current, to 1A. Anyway, it took a boatload of iterations, be we finally tackled all of these issues, and latest shunt circuits are rock-solid.

The other big difference between the v2.x and v4.x versions is how the shunts are used to balance the cells. The older v2.x scheme, which is the same way virtually all shunt-based BMS designs I'm aware of work, is to set the HVC trip point right at the voltage where the shunts are fully on, just below the point they go into overload, or get "swamped". The "charge to" point is set a bit above this, and then current limiting throttling logic is used to keep the voltage right at that point. The net result is that once all the cells are full, the shunts will all be in full bypass. Having these shunts "cook away" at the end is one thing, when the max shunt current is only 200-300mA, but at the the 500mA the v2.x is designed for, things get pretty toasty. At 1A, it is more than toasty, it will get bloody hot! Fans are mandatory.

With the latter v4.x variants, a totally new shunt/balancing scheme is used. Instead of having all the shunts "cook" at the end, what we do now is only have the shunts come on for the first cells that get full, and then only enough to keep the cell from going over the balance point. The "charge to" voltage is set to be the same as the balance point, which is the voltage where the shunts first start to come on. This "by exception" shunt scheme means that at the end, when the cells are all full/balanced, none of the shunts are on at all. This contrasts sharply with the v2.x scheme where every shunt is cooking away at full power at the end.

Another problem with the v2.x design is that there is no definitive end-of-charge detection scheme. You can't just wait for the current to drop to something like 1/20th of the max current, the way many RC and even some SLA chargers work, because the current will never drop below the max shunt current. What the v2.x circuit did was simply wait for all the shunts to come on, and then shut things down. The problem with that is that the last cells that cause the "ALL SHUNTS ACTIVE" signal to trip, have just gotten to that point, and those cells won't be as full as the rest.

With the shunts "by exception" scheme, in the v4.x design, the shunts are off at the end, if the cells are full/balanced, so the current will drop all the way down to zero, eventually. Now we can actually just measure the current, and do a shutdown when the current drops below some selected point. We have a pot that let's this low-current shutdown to set from 0-2.5A.

The final difference in the two designs is how the HVC signal is used. As I said, in the v2.x design, the HVC signal is set to the shunts-fully-on point, and is used by the throttling logic to keep the shunts cooking, but not where they swamp, and allow the cell voltage to rise above the HVC point. The HVC signal is used differently in the v4.x variants. In this case, the HVC trip point is used as a "failsafe" point, or in other words, a safety net to keep errant cells from getting overcharged. This HVC trip point is set a fair bit above the balance point, enough that significantly out-of-balanced, but otherwise healthy cells won't ever have their voltages get this high. The 1A shunts wil handle significant imbalances, just on their own, but if a weak cell, with a lower capacity, gets full a whole lot earlier than the rest, it can still swamp a 1A shunt. This higher HVC trip point will keep it from going over this point by engaging a very simple "throttling" function that will let the shunt work to get the cell back in check, and then the process will continue on. If the shunt swamps again, and the HVC point is reached, the cycle will repeat. Eventually the shunt will be able to hold the voltage on its own and the process will continue on to completion.

Anywa, these are the major differences.

-- Gary

 

[fechter]

Yes. The voltage regulator is coming off the charger, not tapped into cell#4.

 

[AndyH]

Boards have been ordered, and Andy should have the first of these early next week.

I need to clarify this. I've been informed that I will not be receiving these boards. Because of this, I can no longer report on progress or effectiveness of these new variants.

As to being 'backward'... Many of you know which end of a soldering iron to chew on and many can design and/or tweak circuits to meet your needs. Most of my customers cannot. They simply want a BMS solution that works for their application - whether building their own pack or retrofitting to an existing EV. While some of my customers are ebikers, many are in the 'maxi-scooter' and car worlds. Most cannot assemble a board. This group of customers cannot handle being 'alpha' or 'beta' testers - they're paying for a 'toaster' that just works - and I have to buy replacement cells if/when the BMS fails. In addition, I will not sell a product that is not completely tested and that I haven't personally built and used.

I still use a V2.2 BMS on my bike and find that it works beautifully with a bench supply and my Thunder Sky charger. I've not had any challenges with the V2x shunt method when the BMS is limited to the 500mA shunt current limit - and I've used this size successfully for my own charging for nearly two years with no cooling fans.

The V4.1.5 shunt channels do work very very well and I look forward to a full BMS solution that incorporates those channels. But until that solution is 'on the streets' and proven to be a 'toaster', I will not sell units with the new layout.

Andy
Rechargeable Lithium Power

 

[The Mighty Volt]

PM'd.

 

[GGoodrum]

The new v4.2.2 BMS boards arrive today. I'm going to build one up to use on the 24-cell PSI pack on my wife's bike, and then Richard is going to be testing one on his setup as well. I've tested the new control section already, with my latest CellLog setup that I've been testing, and it works quite well. At 100V, though, the 12V regulator FET gets a bit warm, so it might need one of those small finned heatsinks, but there's plenty of room. At 75V, and below, this won't be required.

As Andy points out, the latest shunt sections work quite well. I'm using these in the CellLog balancers as well, and they have worked flawlessly, to this point. The only test left to do is using these shunt circuits with the new control circuit I've been using already. I'm not expecting any surprises at all, so I think we are about as close, finally, as ever to having a "release" candidate.

-- Gary

 

[MitchJi]

Hi,

Boards have been ordered, and Andy should have the first of these early next week.

I need to clarify this. I've been informed that I will not be receiving these boards. Because of this, I can no longer report on progress or effectiveness of these new variants.

Would anyone mind providing further clarification (why you will not be receiving these boards)?

 

[GGoodrum]

Hi,

Boards have been ordered, and Andy should have the first of these early next week.

I need to clarify this. I've been informed that I will not be receiving these boards. Because of this, I can no longer report on progress or effectiveness of these new variants.

Would anyone mind providing further clarification (why you not be receiving these boards)?

Andy is simply going to wait until Richard and I have had a chance to test these thoroughly ourselves, before he goes forward with this new version. He's suffered losses as well as Richard and i, in this continuing effort, both in time and money, so he'd prefer to wait until we are done "crying wolf" before proceeding. Completely understandable, as it makes good business sense. I'm a lousy businessman, however, or I would've cried uncle when my "investment" went north of 20-grand. :lol:

Anyway, once we get this latest version tested, properly, I'm sure Andy will jump back in, and start offering assembled versions. It really shouldn't be too much longer (if I had a nickel for everytime I've said that... ).

-- Gary

 

[fechter]

Yes, I'm working with Andy to get the bugs out of everything but it's not quite there yet. It is definitely on the front burner though (hmmm... shouldn't use the word 'burn').

 

[The Mighty Volt]

Cheers chaps, we appreciate the effort taken to keep us informed. Many thanks.

 

[AndyH]

Hi,

Boards have been ordered, and Andy should have the first of these early next week.

I need to clarify this. I've been informed that I will not be receiving these boards. Because of this, I can no longer report on progress or effectiveness of these new variants.

Would anyone mind providing further clarification (why you will not be receiving these boards)?

As already outlined, I need to stay a generation back with my RETAIL boards until the V4 is stable and fully tested. I requested a lot of V2.6c boards; the order was refused. I was advised afterward that the next V4 variant scheduled to ship to me had been redirected.

As Richard made clear, I'm still working with him on V4 progress, and have been assured of continued access to earlier BMS boards as necessary.

Andy

 

[joe tomten]

I will admit to not having read the 130 some-odd pages in this thread.

But I've read a bunch of them. And like many people; a 16s bms with limited shunt current (or multiples of 6 or eight ) is all i really need.

Better and more sophisticated is nice, but not really a requirement.

I like Diy projects and learn a lot doing them, but come nowhere even close to the knowledge and capabilities of many others in the sphere concerning electronics (or anything else).

There is something going on here that I just don't get regarding continuous design and development, that never gets to into application. I understand that it is totally the prerogative of the developers to deny the distribution of their design and hardware.

But it feels funky, and I have watched and waited for a long time "to participate".

At some point I just have to decide to do (buy) something else.

Oh well.

 

[NorCalTuna]

I think it's a very prudent attitude.

Richard et al could market a product- albeit with a buyer beware style warning, like "Caveat Emptor: LiPo Batteries are flammable and dangerous; this equipment is experimental, all risks are incurred by the user in any circumstance of use" but that won't really keep every body happy and him out of court. Court sucks. Not burning up stuff on your custies, delivering flawlessly designed products, thats what ultimately assures the success of a company or endeavor.

I know that I sweat at night- I build and mod tube guitar amps for local guys. One day I could get a call... "Hey, I was drinking a beer last night when somebody knocked it over and it spilled in my amp... it looks like I was playing it outside in the rain, cuz theres mud all over it, but it's beer, I promise. Anyway, I got shocked and have brain damage (despite being shocked from the foot to the hand) and my lawyer wants your lawyer's number..." I'm gonna win, there's not a question there, the guys a dumbass, and as long as I warn dumbasses, in writing, i'm supposedly good. If you burn down your house with an aftermarket product you assemble, the judge will likely not listen to your claims that it's the EE who designed the thing's fault. But it's gonna cost me-or gary in the battery case- everything I have to defend myself and whereas I may recoup that through legal proceedings, I honestly rather wouldn't have started the process to begin with, i'd rather make a good product and spend my time doing that instead of defending myself.

So, I'm thinking that the fletchrum partnership is really just trying to produce the best, least risky, longest lasting, most flexible and awesome system available. If you don't wanna wait for that, I can understand, but it's not a conspiracy to only let three guys have the cool stuff; it's to keep everybody as happy as they can eventually make errbody. I'm on that list too, and I feel lucky that somebody is bothering to do this without corporate levels of overhead and such. I don't have the f'in time or money to proto the stuff... It's a huge gesture, really. What would, say, brammo pay for this work? They'd likely salary somebody for a year or better... or have a permanent team even.

making daring, "it's good enough" style decisions might be fine for House, MD (who incidentally loses patients routinely as a plot device) but I doubt our local EE community feels the same way. You can vote with your wallet, thats how the market works- and gary won't be bent because he'd rather (Shut me up if I'm wrong, anybody) lose a nominally profitable "customer" than you accidentally burn your house down with a half baked BMS. Could you imagine your bike catching fire violently (as in "vent with fire") riding at 40mph? Or wondering if you can get the electric garage door up in time and wheel the f'in thing out before that pile of oily rags (EVERYBODYS GOT ONE I"M LOOKING AT YOU LFP) goes up- if you're lucky enough to be there that minute? f that.

 

[fechter]

Well, the new version was supposed to be ready about a year ago (maybe longer). This has been one of the most frustrating and brain-draining projects I have ever worked on. Just when we think 'it has to work now', some issue comes up that casuses a problem. It's like playing whack-a-mole. Combine that with very limited resources and you get a long development time. We also have a lot of self-imposed constraints, like it can't be too expensive, too big, etc. If it weren't for the research fund donations and this insane passion I have for solving the problem, I would have given up long ago. I have a lot of other things I could be working on.

The shunt boards are solid and have passed validation testing, but the control circuit is still a hangup. We also have some packaging issues to work out, something we never even worried about with ver2.6 (it was up to the user).

 

[AndyH]

There are a lot of BMSs on the market. They run the range from "BMSs" that don't 'manage' they just 'monitor'; to computerized 4-channel boards (with 100mA or 200mA shunts) that work on the bench and for back-up power but aren't reliable on a road vehicle; to a DIY device that shunted about 1A but was pulled from the market after destroying a bunch of cells; to other BMS devices that provide LVC protection and shut the charger off - but don't shunt and cannot balance. At the other end of the scale, we have computerized devices with small shunts, computers, and CAN communications - and $$$ price tags.

That's what drew me to Gary and Richard's devices - even the V1.5 board I bought was more capable and more reliable than any commercial unit available at the time. My 'daily driver' BMS has been a V2.2 unit used multiple times a day for 18 months. Even thru trials with a couple of V3 devices and early V4 prototypes, I kept coming back to V2.2 because I knew I could connect a charger to the pack, go to sleep, and the battery would be fully charged in the morning and the garage wouldn't have burned down overnight.

As far as I'm concerned, while the folks at Nissan must have some sort of unobtanium BMS for their lithium, E-S is in the lead with devices we can buy, build, and rely on for our packs. And we know the pair that's turning the forum brain power into hardware.

Andy

 

[GGoodrum]

Research fund donations? :? I haven't seen any money from a research fund. I've spent as much as a what it would cost for a used Prius on this latest version over the last year+, and haven't seen one penny of "income". :lol:

I have no problem with Andy, or anybody else using the v2.x boards. I just don't want to sell them myself (if you want one, contact Andy...) because I don't want to be answering the inevitable litany of questions that will surely come, just like before. If I'm going to be answering questions, I want it to be on the newer version, that's all. Compounding the v2.x vs. v4.x issue is the CellLog-based variants that I'm trying to do at the same time. The latter is primarily for LiPo-based applications, but can be used with LiFePO4 configurations as well, hence more confusion.

As far as I'm concerned, the full BMS board is done. the controller works perfectly, as do the shunt circuits. The only remaining task is to test the two together, as a complete unit. I've been using the controller, in this configuration, with the CellLog-based test units for some time now, and it has proven to be quite reliable (100%, so far...). I've also been using the same shut circuits in the CellLog-based balancer sections, and again, they have worked flawlessly. The latest full BMS board has 24 shunt channels, and it has this latest charge controller section. The only difference between what I've been testing with the CellLog-based units, and this new full BMS board, is where the opto HVC/LVC "Alarm" signal is generated. In the full BMS, the opto signal is generated within each shunt circuit. The CellLogs generate this signal which is then opto isolated via the extra logic we provide. Functionally, the two signals are identical. The opto line is pulled low whenever the LVC or HVC signal trips in any channel.

In any case, it is prudent to test the full BMS as a single unit. Finding the time to do this, however, has been difficult. My first priority is the CellLog-based units, frankly because that's what I need. You will have to excuse me if I'm a little selfish that way, but after 20-grand, I still don't have a a BMS/CMS solution that is optimized for my own setups. Anyway, Richard has a day job that limits his test time, and Andy chooses not to waste anymore of his resources, time included, which is perfectly fine. He's spent a lot of time and money on this as well, and is making the prudent "business" decision to sit this last dance out. I seriously can not fault him for that. What we need to do, I think, is recruit from within here, to help finish off the testing. Before committing to that, however, I will talk to Richard, and see what he wants to do.

-- Gary

 

[AndyH]

In any case, it is prudent to test the full BMS as a single unit. Finding the time to do this, however, has been difficult. My first priority is the CellLog-based units, frankly because that's what I need. You will have to excuse me if I'm a little selfish that way, but after 20-grand, I still don't have a a BMS/CMS solution that is optimized for my own setups. Anyway, Richard has a day job that limits his test time, and Andy chooses not to waste anymore of his resources, time included, which is perfectly fine. He's spent a lot of time and money on this as well, and is making the prudent "business" decision to sit this last dance out. I seriously can not fault him for that. What we need to do, I think, is recruit from within here, to help finish off the testing. Before committing to that, however, I will talk to Richard, and see what he wants to do.

If I pulled myself out because I don't want to waste any more of my resources, why in the world am I still testing the V4? I'm not sitting this dance out - I simply requested that I no longer be the one to validate the PCB layout before building a test item.

View attachment 415_control.jpg
I agree completely that it's prudent to test the controller with the channels. I bought a pair of stand-alone charge controllers from you, and while the one I built works fine separately, it didn't work when fed the V2.2 HVC signal, and destroyed itself when I connected to the V4.1.5 HVC signal.

I'll continue to support Richard, and will continue with V2.6c until there's a replacement. And as I've already said, I'll continue to support you. I'm glad you're almost there with the CellLog version - I wish you great luck!

 

[GGoodrum]

What, you call that a hack?? That looks "production-ready" compared to some of the messes Richard and I have cobbled together, to test new ideas. Our most recent favorite is this one, where we had to remap all of the pins on a higher power opamp Richard decided we needed to use:

View attachment Brucie-01.jpg

I told him it looked like a gay spider, not that there's anything wrong with that ( ), and he named him Brucie. :lol:

-- Gary

 

[ejonesss]

looks like dead bugging.

you may want to check there may be a 4 pin version of the same chip.

 

[GGoodrum]

looks like dead bugging.

you may want to check there may be a 4 pin version of the same chip.

Naw, this was just to test that the part did what we needed. The next PCB iteration had the corrected pinouts.

 

[wookey]

What we need to do, I think, is recruit from within here, to help finish off the testing.

As I believe I've said before, I'm happy to join the fun on that basis. I know which end of a soldering iron and schematic is which. And have a headway pack that unbalances itself every week with meanwell charging. The scooter/pack is currently getting daily use (although my wife is doing the driving). And I won't be suing anyone if my bike and shed burn down one night.

 

[Gregb]

do what my elation battery makers did. Don't put a name on it. The supplier says you don't need one on any of the parts he sells . Like the charger as well . or who makes the controller. But I didn't know that until I had bought it. I am a retired tech and would be happy to use my suspect battery as a test element. 24V lifepo4 10AH.
Greg

 

[julesa]

looks like dead bugging.

He's not dead, 'e's just resting.