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

[Bruce_Wayne]

Gary,

Check your Pm's once in awhile! I can get these built for you right here in the states immediately. This is a passion of mine, as well as others on this forum. I think very few people involved in this right now are doing it for the money. At the cost of these parts and getting the boards from express pcb or the like, I doubt your even breaking even, I know with 100 percent accuracy that you mean what you say.

My thoughts are, after seeing what happens in the GM board of directors meetings, is, this is up to the ordinary blue collar workers to get done. GM or any large vehicle maker is not really in any big hurry to get this done. There's always some BS excuse , or it's a couple years off, just wait. Well I've waited... Now for ten years to get my hands on an EV.... And still nothing.
I'm not waiting anymore... I will build these boards myself or raise hell until these things are out there in people's hands. This needed to happen 10 years ago. Judging from the parts list I put together for this board, you want to shoot for around 100-120 a piece. This cost will pay for having people assemble them efficiently with the right tools.

Please give rick sheets a call.

Bruce out.

rf[/quote]

Ask, and ye shall receive...

Thanks for the kind thoughts, but for me it has never really been about the money, although I certainly want to stop losing anymore. The problem is the time drain. The kits just have too many parts to try and do this myself. I am not trying to make a profit on these, just break even. I just can't aford to spend hours on end putting the kits together. I don't have a problem selling the boards at close to my cost, plus shipping, but selling the kits for cost, was just too much. What makes matters worse is that I have this consulting job that is consuming most of my free time of late, and will continue to do so for the next few months anyway. Sending the boards out is easy enough that I can have my wife and son help out while I'm off playing aerospace engineer again.

-- Gary[/quote]

 

[GGoodrum]

I'm not waiting anymore... I will build these boards myself or raise hell until these things are out there in people's hands. This needed to happen 10 years ago. Judging from the parts list I put together for this board, you want to shoot for around 100-120 a piece. This cost will pay for having people assemble them efficiently with the right tools.

Please give rick sheets a call.

Bruce out.

As I said, I'm not looking to get into the BMS making business right now. I may eventually do an SMT version, and have it made at any one of a number of shops that I've found near where I'm working right now, in Santa Ana, but that won't be anytime soon, I'm afraid.

-- Gary

 

[JB_EB]

Gary,

Just a thought. If you are not looking to make a profit out of this (and if Fechter agrees as original designer) why don't you get together with Knuckles and see if his Chinese connection can produce these BMS boards for us E-bike enthusiasts. I'm sure Knuckles would be able to arrange some compensation for you and Fechter as the original developers.
Or maybe the boards could be made in China and you could be the distributor? Many options to explore here.

Also, I think the proposed layout with the ability to cut-off the unused channels is a good idea. One further idea would be to put additional wire connection points on the PCB on each side of the cut line so additional boards could be daisy-chained onto the basic (20 channel?) PCB. Again, if there was a supplier set up, they could probably keep cut-off sections from lower channel orders and include the unused section for higher channel orders at very little extra cost.

It appears from the posts here that many people are really keen to get these BMS modules. It seems a pity to waste all the development and testing work that has been done by you and Fechter because you do not have the time to put together the kits. While your suggestion of selling the bare PCB is generous, the problem is that for individuals to buy all the parts in small quantities is probably going to increase the cost and difficulty considerably.

Gary - Please, please contact Knuckles to see if you guys can get together to to provide this fantastic product to us EV enthusiasts.

Regards,

JohnB

 

[commanda]

As far as physical layout of the pcb(s) goes, I'll tell you what I'm doing.

I have the common circuitry on one pcb. I then have multiple boards (4 of) with the per-cell circuitry, 5 cells per board. The boards all connect together.

My common circuitry board has somewhat more components than yours. I have 2 current sources, a time delay, and an R/S flip-flop so it shuts down when charging is finished. LVC drives a switched 12 volt output for the e-brake input on my EVT controller.

I'll start a seperate thread in due course when it's all working fully.

Amanda

 

[K-ray]

Hey Gary,

And of course, all who have been involved... kudos for your work! I'm sure we all hope
you continue to develope and produce products.
I REALLY like the tear off concept. I have no idea what kind of volumes are necessary for a
plalatable price, but I think production of 8 / 16 / 24 cell boards would all sell nicely.
I personally would be looking for 5 in a variety of these sizes. I also am hoping a 4 cell board
for stick packs happens, but I'm not sure about the desire for them. I guess one could clip an 8...
Cheers for your efforts

Kevin

 

[fechter]

Gary,

Just a thought. If you are not looking to make a profit out of this (and if Fechter agrees as original designer) why don't you get together with Knuckles and see if his Chinese connection can produce these BMS boards for us E-bike enthusiasts. I'm sure Knuckles would be able to arrange some compensation for you and Fechter as the original developers.
Or maybe the boards could be made in China and you could be the distributor? Many options to explore here.

I'd certainly like to see these as a fully built surface mount devices. I would like to see if Knuckles and Keywin can make them too, but I'd want to clear things with Gary first. Building the boards from parts by hand is way too time consuming.

Another approach to the modularity issue would be to build two versions of the board. One with the control circuit, and one without. Make them all 4 channel like Gary shows, but have them separate. You would daisy chain pieces together with 4 wires. Buy one "master" and as many add-on units as you need in multiples of 4 cells. I suppose some kind of 4 pin connector and jumpers would be cool to connect them.

 

[PJD]

If I may bring the topic back to the existing BMS, can anyone explain the effect I described in my post on Sep 01, 2008 1:02 pm (item #2).

Is this just an expected effect of the heating of the reference voltage resistors or other parts on the board?

Also, the shunt leds on my board operate with a very pronounced strobe-like flicker - maybe 30hz or so. Any concern about this condition?

Thanks.

 

[PJD]

Amanda,

I'm loooking forward to your thread.

While Gary and Richards board is a good start, we really need a timer so we can have a true plug in the evening and-go-to-bed type solution.

The 12 volt LVC high signal will be welcomed by all e-scooter/motorcycle users too - especially since a throttle pull-down won't work with EVT's.

 

[Bruce_Wayne]

As I said, I'm not looking to get into the BMS making business right now. I may eventually do an SMT version, and have it made at any one of a number of shops that I've found near where I'm working right now, in Santa Ana, but that won't be anytime soon, I'm afraid.

-- Gary

Have you considered putting this in the public domain so at least people have "something" to work with? The design could be improved upon as well. I already know several people willing to contribute to the design. There are people here who would get the smt version put together in a week if it was public domain material. What are your thoughts on an open archictecture? I apologize for being such a pest, but I know I'm by far not alone in wanting these things to get out.

 

[ggracer]

Gary,
This 24 cell option is EXACTLY what I've been looking for. I was more than happy to purchase 2 of the 16 cell and only stuff them for 12 cells each and link them together, but this solution would be perfect. I think you'll get more and more requests for 24 cell as more people upgrade to 72V solutions (I'm using a PERM 132 personally).

Many, many thanks for all your hard work, as soon as I can purchase either solution, you've got my business!!!!

Glen

 

[fechter]

If I may bring the topic back to the existing BMS, can anyone explain the effect I described in my post on Sep 01, 2008 1:02 pm (item #2).

Is this just an expected effect of the heating of the reference voltage resistors or other parts on the board?

Also, the shunt leds on my board operate with a very pronounced strobe-like flicker - maybe 30hz or so. Any concern about this condition?

Thanks.

The change in voltage set point with temperature is most likely due to the temperature coefficient of the orange LED, which is acting sort of like a zener diode in this application. While this is undesirable, the amount of drift is probably not enough to be a problem. Locating the heat producing parts away from the heat sensitive parts is always a good idea.

A fan wouldn't be a bad idea.

In future versions, I suppose there could be a way to compensate the voltage better.

The 30hz strobing is OK. With smaller cells, this is fast enough that you can't see it. It becomes a problem if the switching frequency gets too high. Under some conditions, I've seen it flash even slower, like 10hz, which won't hurt anything.

I'm looking at additional circuitry that will terminate the charge once all the shunts are full on. I don't think it will matter to the cells much, but will reduce heat generation once the pack is fully charged.

 

[PJD]

Richard,

Thinking aloud here..

I am wondering of I can achieve charger shutoff by setting my charger's "charging done" current setpoint to an appropriate value. If I am understanding things right: Icharger = Ilowest shunt+Ithru pack.

So, If I assume all shunts are on and each cells charging current is down to a practical "fully charged" value (about 250ma) than all shunts are nearly all on - say 400ma for the high current version. So, if I set my charger shutoff point for 650 ma, I should be able to get an automatic shutoff of charging. Am I correct?

The probelm is that the charger's current control itself is being "commandeered" by the BMS. One high cell's shunt will turn the BMS's regulation down a value so low the charger might shut off prematurely. Still, with some fiddling and assuming a resonably balanced pack, I should be able to find a value that will work. Ideas?

 

[fechter]

I think if you get the charger voltage matched to the string voltage, the current will drop off when all the cells get up to the shunt voltage. If your charger is adjustable, you can play with it. My guess is a typical SLA charger can be setup so it goes into float mode when charging is nearly complete. One possible down side to this is if the current tapers off enough to stop charging, the cell LEDs will will no longer be lit.

If you can adjust the current where the charger goes into float such that it is barely over the shunt current, then it will trigger float mode as soon as the first cell gets full. This may result in lack of balancing, which may be OK most of the time.

I think in actual practice you'll find nothing bad happens if the charge does not terminate (other than heat dissipation and some wasted electricity). The actual current through the cells will drop off to near zero and everything will be carried by the shunts. This will ensure that the cells are balanced as well. In a seriously mismatched pack, balancing could take quite a while.

 

[GGoodrum]

All the current has to go through all the cells, so as the cell voltages get to the crossover point, the shunts start also conducting. Since the shunts act as individual CV controls, the cell starts to reduce the amount of current they can let in. the shunts make up the difference so that at least 250/500mA can flow through the whole pack. If you had one cell that was out of balance, it could take longer for it to get full, so while the rest of the cells have very little, if any, current flowing back into the cell, its shunt is still bypassing 250/500mA on through to the next channel. The unbalanced cell can then have at least the 250/500mA of current that can all go back into the cell until it catches up. When it does, its shunt will end up bypassing all the current, just lke the rest.

Richard and I have been talking about adding logic to shut the charger current off completely, once all the shunts are on, but what I'm not sure about is whether or not a cell trying to "catch up" will be full enough when the LED first comes on. Actually, it is when the opto circuits conduct enough that the "ALL SHUNTS ON" logic gets tripped. The trick is going to be to make sure the shunt is fully on, as opposed to when it first starts to bypass current. In the case of the latter, the cell would still be absorbing current, so you wouldn't want it to cut off yet. When the shunt is fully on, however, not much is going into the cell.

-- Gary

 

[jeffkay]

Charge cutoff: Yeah, and the A123's need that last 15-30 minutes after reaching full charge...
Jeff K.

 

[PJD]

Gary,

I agree with your remark. We need current balance as well as voltage balance in all cells for the pack to be considred genuinely balanced. The amount of residual current for all cells to be considred full on isnt necesarily a tiny amount. I consider my 40 AH TS cells "fully charged" at the desired voltage when the charge current is down to 250-300 ma, as it takes theoretically infinite time for the current to appraoch zero.

One question:

Is it possible to adjust the "gain" of the BMS's current regulation? I don't think my shunts ever come full-on because the charge current is getting throttled back a bit too much - down to 400 to 500 ma. as I stated above, my cells are probably going to be passing at lest 200 ma unless they are charged a long time. This no doubt helps the heating situation, but lengthens the time for the lagging cells to catch up.

 

[fechter]

I tried to make the 'gain' variable, but was unable to come up with an easy way to do it. Keeping the parts count down was a primary consideration. If you need a higher current, then you'd use a lower value for the shunt resistor and put heat sinks on the shunt transistors.

The way it is now, the shunt slowly comes on as the voltage rises and when the voltage across the shunt resistor reaches around 3v, the opto starts to come on. A little more than 3v and the transistor can't turn on any more (full on). At this point on the highest cell, there will be full charger current going through the cell and the shunt. The shunt is only taking 500ma, so the rest is going into the cell. This starts the charge current throttling, which will taper down slowly as the cell charges. Eventually the charge current will drop to 500ma. By this time, the cell is going to be mostly charged and the cell current will be very low since the shunt is taking most of the current.

I guess the issue is from the time the first cell lights up how long does it take for the last (all) of them to light up.

If the cells were perfectly matched and balanced, the BMS would not really be doing anything until end of charge when all the cells were up to voltage. Once the pack is balanced, it should not require that much balancing on each cycle unless a cell has an internal drain or something. They all have the same amp hours drained from them each time during discharge. I guess we'll have to see in actual practice with your cells.

It may take an extended period of time to really balance everything at first, but I'd expect succeeding cycles to be much faster. If the cell manufacturing variations are such that a considerable amount of balancing is needed each time, you may need to settle for a longer charge time or less than a full charge each time. From what I can tell, holding the cells at their max voltage for an extended time doesn't really hurt them much.

 

[buzzforb]

Hello,
i am new to this string and need to ask a question that probably has already been asked. I apologgize ahead of time if this is true. I have read so much on these forums the last few weeks, i have admittingly gotten lazy. Once again sorry! Can your bms syystem be epanded to manage a larger ac motor pack setup for use in EV.

 

[ErikK]

I'd love to get the tear-off version in 20 or 24 cell configuration. But I'd be quite happy to just order a pair of the existing 16-cell bare boards and populate them myself. When will they be back in stock? I checked the website http://www.tppacks.com and it still lists complete kits, out of stock. What is the plan?

 

[PJD]

Richard,

Thanks for the reply. I described my first test of the BMS on "virgin' pack a few posts back.- some cells were still so low (3.45v) when the current was fully throttled back that I stopped the test, individually charged them, resumed charging, found I over did those cells a bit, individually discharged them a bit, tried again and got it to work fine.

On the first in-service use of the BMS, two cells were "laggards" - one at 3.60v the other at 3.55v when the shunts started coming on. They took about 15 and 25 minutes respectively to reach the 3.68 volt shunt point, and it took a total of 50 minutes from the first shunts coming on before all shunt LED's were on with equal brightness and minimal flickering - total charger current at this point being about 500 ma. I took this to indicate the cells were all passing roughly same low amount of current, and stopped charging at this point.

I sympathize with Erik K and others. Gary, when are you going to make the board- and part-list kits available?

 

[PJD]

buzzforb,

Based on my limited experience so far, the maximum practical pack capacity that these BMS's in their current configuration is about 50 AH. With large packs, any lagging cells can take an impractically long time to catch up because the charging current gets cut back to less than an amp fairly quickly after the first shunt is activated.

Addendum - If the pack is reasonably well balanced, the upper practical limit may be higher - 60 to 80 AH. But automotive sized packs are certainly too big for this design - although it could probably easily be scaled up to as big as needed using different parts and heat sinks.

 

[fechter]

Hello,
i am new to this string and need to ask a question that probably has already been asked. I apologgize ahead of time if this is true. I have read so much on these forums the last few weeks, i have admittingly gotten lazy. Once again sorry! Can your bms syystem be epanded to manage a larger ac motor pack setup for use in EV.

Yes, each 16 channel board can be daisy chained to another and you can use as many channels on each board as you want. The practical limit may be the charge control FET. The maximum voltage difference between the charger and a fully discharged pack cannot exceed 100v with the 4110 part.

PJD: Interesting. Those are 40Ah Thunderskys? If the cells were better matched it would be faster, but an hour for balancing does not seem too bad. From what I've heard, it may not be necessary to fully balance the pack each cycle. You might try running the pack after charging it until all the LEDs just come on and see if you notice any difference in capacity.

 

[GGoodrum]

Sorry for the delay, but since Richard and I are looking into adding a feature to cut the charge current when all the shunts are fully on, I've held off submitting the "tear-off" layout. One way or another, I will tun something in on Monday, which should get them back before the end of the week.

-- Gary

 

[PJD]

Richard,

On the second charging the lagging cells (different cells than before - good sign) were at 3.63 at start of shunting, and all led's were on in about 15 minutes. I still gave it a total of 50 minutes - the current regulation was at about 550 ma and still going down slowly at 50 minutes. It may a bit nit-picky and not be real significant, but my understanding is that if we want to follow the recommended CC-CV charging protocol we should aim for equal low current through each cell (and same voltage of course) as the criteria for being equally and fully charged (about 250-300 ma is low enough for the 40AH Thunderskys). So, I still think an adjustable time delay from the "all shunts on" condition would be the best solution for a shutoff feature.

The current did continue to decline through the 50 minutes of shunting, so in my case, I should be able to adjust the current setting on my Chinese scooter chargers to have them shutoff automatically. So, perhaps a shutoff based on a total charger current measurement?

On the heat management front, my board is mounted on the underside of a hardboard cover over an access hole in the bottom of the storage well under the seat. I drilled 3/8" (9mm) holes over the resistors - these seem to act like little flues and provide effective passive air circulation across the PC board, as nothing has burned out yet - and the ambient temperatures have been quite warm - around 80F.

I don't have a cycle analyst, and in the spirit of keeping things simple and cheap, don't have plans to get one, so I can't give quantitative capacity information.

 

[Tom_D]

Sorry for the delay, but since Richard and I are looking into adding a feature to cut the charge current when all the shunts are fully on, I've held off submitting the "tear-off" layout. One way or another, I will tun something in on Monday, which should get them back before the end of the week.

-- Gary

From Gary's post, page 19:

Anyway, if anybody is interested in this style of BMS board, please PM me, and let me know how many channels you plan to use.

PM sent. I am interested in (2) 24 channel tear-off boards when you get them.
--Tom_D