New 16-cell Battery Management System (BMS)

[GGoodrum]

Sorry, been busy.

I'm aware of the Firefox problem with my website. This is actually an old problem, that was fixed at one point, but I recently updated the shoping cart software and I guess I re-introduced the problem. I'll take a look at this again, when I get a bit more time.

Re: the BMS/CMS, I think Bob has everything working now. Unfortunately, I have found that although the design is sound, I just can't see anyway that this is going to be a commercially viable product for me, primarily due to the labor incolved. Frankly, I just don't have the time required for any commercial involvement on my part. Bob, who works much faster than me ( ), is going to take over 100% of the sales and support for these. You can PM or email him for a quote.

What I'm doing for my own a123 and LiFeBatt packs is to use these with the LVC-only boards, and then I'm just using two ThunderPower TP-210V auto-balancers once in awhile, to keep the packs reasonably balanced. Then, I just bulk charge them with my Soneil and Zivan chargers. I made a special harness that connects the two TP-210Vs into the pack, overlapping one cell. Here's an example of my 48V LiFeBatt pack being charged with a Soneil 4808 4A charger (with a WattsUp power meter...), and with the dual TP-210Vs:

Since I've had several requests for this "quick and simple" charging/balancing solution, I went ahead and added them to my website. You can find the here: http://www.tppacks.com/products.asp?cat=26.

-- Gary

 

[rf]

[...]
Unfortunately, I have found that although the design is sound, I just can't see anyway that this is going to be a commercially viable product for me, primarily due to the labor involved.
[...]

Thanks for the great effort, Gary.

Is the problem the need for per-cell alignment or the shear number of parts? (Just trying to think of ways of solving this problem in the future, perhaps with a microcontroller.)

Richard

 

[GGoodrum]

Thanks for the great effort, Gary.

Is the problem the need for per-cell alignment or the shear number of parts? (Just trying to think of ways of solving this problem in the future, perhaps with a microcontroller.)

Richard

A bit of both, actually. It does have more parts than most controllers I've seen, which takes time, but it also requires custom heatsinks that I was hand-making. It is also not the easiest to adjust and test. I didn't have the right setup here for that, but Bob does. The design could be converted to use SMT parts, and the assembly farmed out, which is the ultimate solution, but not unless the demand is there, which we just don't know right now.

While doing all the work on the BMS/CMS, I did a lot of testing on the LiFeBatt and a123-based packs I have here and found some interesting information about how they seem to work. With the LiFeBatts, the cells will bleed off the surface charge on their own, after a charge, until they are down to about 3.340-3.345V. It only takes about an hour, or so, for this to happen. The a123 cells, however, will hold this surface charge, pretty much indefinitely, but as soon as you apply even a small load for a few seconds, the surface charge will bleed off, and the cells will settle down to somewhere around 3.45-3.50V.

I then started doing a few more tests, to compare individual cell charging, with bulk charging, Using my setup with a bunch in VoltPhreaks 2A CC/CV chargers, I charged a 12-cell LiFeeBatt pack and then waited 24 hours. Here's where the cells ended up at:

• 3.345
• 3.345
• 3.346
• 3.345
• 3.351
• 3.347
• 3.349
• 3.347
• 3.349
• 3.347
• 3.347
• 3.348

I then discharged the pack a bit, and measured again:

• 3.324
• 3.322
• 3.324
• 3.223
• 3.324
• 3.323
• 3.323
• 3.326
• 3.322
• 3.322
• 3.323

Next, I simply bulk charged the pack, using a Soneil 3610 SRF 4.5A charger with a cutoff set at about 43V, or around 3.60V per cell. The numbers were very similar:

• 3.341
• 3.340
• 3.341
• 3.340
• 3.341
• 3.340
• 3.339
• 3.342
• 3.339
• 3.341
• 3.339
• 3.341

What this tells me is that for healthy LiFeBatt cells, it doesn't make a huge difference whether they are bulk charged, or individually charged. The differences between cell voltages are similar. I then decided to see if using RC-type auto-balancers would make any difference, so first I discharged the pack about 3Ah:

• 3.287
• 3.285
• 3.286
• 3.284
• 3.283
• 3.284
• 3.286
• 3.284
• 3.283
• 3.284
• 3.285
• 3.286

I then bulk charged the pack, but with the Soneil turned up to 45V, which is about 3.75V per cell:

• 3.430
• 3.419
• 3.421
• 3.416
• 3.413
• 3.412
• 3.417
• 3.412
• 3.400
• 3.403
• 3.411
• 3.444

Finally I cycled the pack again and this time I used the two TP-210V auto-balancers, which I left connected until all the LEDs went off, which took about an hour. Waiting overnight I got the following readings:

• 3.346
• 3.345
• 3.346
• 3.345
• 3.346
• 3.346
• 3.346
• 3.346
• 3.345
• 3.345
• 3.345
• 3.346

Myconclusion, so far, is that bulk charging, with occasional balancing, is going to work fine, at least for LiFeBatt-based packs. This also holds true for most of my a123 packs, but several of these have stressed cells, that have lost 5-10% of their capacity. So for these, I will continue to individually charge each cell block, using the VoltPhreaks chargers.

-- Gary

 

[PJD]

Gary,

Thanks for the info. From the testing I've done so far, my Thunderskys behave similarly to the LiFeBats but a bit more variability in the voltages.

So, one alternative to the lifecycle BMS for my two scooters would be an LVC on each of them, plus one pair of the TP-201V's, kept off-board at home for balancing every few cycles, with the proper connectors for convenience.

Question: When do the TP201's operate? Whenever connected to the battery pack? Via an on-off switch? In response to a user-supplied signal? Also what are the dimensions of the the TP201's (and the LVC). Maybe you can add this info the your website. Thanks!

 

[PJD]

but it also requires custom heatsinks that I was hand-making...

With regard to the heat sinking, I noticed the heat sink options would be better if the darlingtons had been mounted on the PC board with their mounting surfaces facing outward, so they could be attached to larger finned extruded heats sinks or the sides of an extruded case. Curious why this wasn't done?

 

[GGoodrum]

Gary,

Thanks for the info. From the testing I've done so far, my Thunderskys behave similarly to the LiFeBats but a bit more variability in the voltages.

So, one alternative to the lifecycle BMS for my two scooters would be an LVC on each of them, plus one pair of the TP-201V's, kept off-board at home for balancing every few cycles, with the proper connectors for convenience.

Question: When do the TP201's operate? Whenever connected to the battery pack? Via an on-off switch? In response to a user-supplied signal? Also what are the dimensions of the the TP201's (and the LVC). Maybe you can add this info the your website. Thanks!

The TP-210Vs start a cycle as soon as they are connected to the pack. The first thing they do is count how many cells there are, and to make sure they are connected properly. It will flash all the LEDs for the ones where it finds a healthy cell. If an error is detected, the orange status light will flash, indicating the type of error found. If okay, it moves on to start balancing the cells. The LEDs for each cell that is being discharged come on, so you can tell what is going on by simply watching them. For cells that are reasonably close, the LEDs seem to come on in a fairly random order. Once all the cells are within .001V of each other, the balancer shuts off for a period, but if it detects a new imbalance, it will start a new cycle. There is also a "Start" button that can be used to initiate a new cycle at any time.

The best way I've found, with the liFeBatts, is to hook the balancers up while charging. This allows the balancers a bit more "room" to work. I've also taken a pack that was fully charged using only the bulk charger, and had similar variations as always. I then hooked the charger back up, with the balancers connected, and it basically just went through the CV portion of the bulk charging cycle again. I let it go down to where the current was down to about 200mA, and then I disconnected the charger. Once again, it took about an hour for all the LEDs to go out on both balancers. When I checked later, the cells were all within .002V, from the highest to the lowest.

The harness, which is shown below, plugs into both balancers, overlapping one cell so that all will balance together, and then has a matching multi-pin plug that mates up with the connector on the LVC board.

The basic LVC option comes with a connection harness that can be wired into the cell junctions. In this basic configuration this connection harness is plugged into the board-mounted connector. An option is provided that adds another wired connector, so that one of them can be used with external balancers, or individual cell chargers. This is what I would recommend for your setup. You could also wire the cells directly up to the basic LVC board, but it is easier to make mistakes. I prefer to wire up all the cell junctions into a connector so that I can verify the wiring before plugging in the LVC board. Here is what the board looks like, with the second wired connector:

In any case, you are right, all you would need is one LVC board for each pack, and then the dual TP-210V setup, with the harness.

-- Gary

 

[GGoodrum]

but it also requires custom heatsinks that I was hand-making...

With regard to the heat sinking, I noticed the heat sink options would be better if the darlingtons had been mounted on the PC board with their mounting surfaces facing outward, so they could be attached to larger finned extruded heats sinks or the sides of an extruded case. Curious why this wasn't done?

When we starte with just the "CMS" concept, we were actually going to mount the TIP105s facing down, bolted to a solid block of aluminum. Then we decided to try and combine the CMS and LVC functions together, and still make it fit in the same footprint as the LVC board. This lead us to re-orienting the TIP105s so that they could be mounted to the sides of a smaller block of aluminum. This becam impractical to do, with the LEDs for each channel, so then we went to a configuration that used two bars, one for each row of TIP105s, and then added a plate that bolted to the top of the bars. When we tested this configuration, we found it could manage to dissipate about 1-1.5A of current through the shunts, if all are in full bypass. This was about what we expected because doing the math, it would take about 1/2-pound of aluminum to dissipate 2A at about 50V.

the problem is that I couldn't figure out how to make this work in a LiFeBatt-based pack, without a LOT of heat, so I looked at just trying to use the bars. This limited the max current to about 1/2A, which is pretty low. That's when I decided that I really didn't want that much heat in the closed confines of my G10-based packs, and running it at only 1/2A kinda defeats the whole advantage of using the shunts. By going back to the original CMS concept, and move the shunt-based regulators to a separate board. By putting the CMS board into an extruded aluminum box, I was able to get the current up to about 1A, but still no higher. It really needs a bigger heatsink, and a bigger box. One nice feature about the latest version is that there are thermistors use so that the current is automatically reduced in order to keep the temps down to around 150F.

At this point, I started adding up all the costs, along with all the money I had spent to this point, and decided I just didn't have the resources for any further experimentations in order to optimise the configuration. I also decided that because of my own labor that would have to go into the heatsink/box/harness/connector end of things, I just couldn't do it in anything close to an economical fashion. Bob, on the otherhand, is still willing to build/test the boards, and I think can also have heatsinks made, for those that want them, plus I think he's still willing to try and do kits, if somebody wants to save some money, and who is reasonably handy with a soldering iron.

As I said, after I've done all these tests, I'm just not convinced anymore that it makes all that much difference, trying to effectively individually charge each cell/block of cells. In fact, Simple bulk charging, along with occasional balancing, seems to work just fine.

-- Gary

 

[rf]

Please pardon my french, but I find the return to `balancing' somewhat horrifying.

Balancing was originally devised for a very different application with batteries of a very different chemistry.

Balancing is appropriate for packs of small numbers of cells where the weight and cost of a BMS is prohibitive -- and the occasional loss of a cell or two is acceptable. LiFePO4 cells have a very flat discharge curve, so relying on voltage to gauge `balance' is problematic. Worse, balancing means we're sacrificing charge for `balance' -- and I can't imagine why we would do that!

Balancing through discharge buys us nothing if we assume we will always operate with an LVC. Nothing. We always want the maximum charge we can safely achieve for each cell. We never want to waste energy. Balancing wastes energy for no apparent reason.

Richard

 

[GGoodrum]

Please pardon my french, but I find the return to `balancing' somewhat horrifying.

Balancing was originally devised for a very different application with batteries of a very different chemistry.

Balancing is appropriate for packs of small numbers of cells where the weight and cost of a BMS is prohibitive -- and the occasional loss of a cell or two is acceptable. LiFePO4 cells have a very flat discharge curve, so relying on voltage to gauge `balance' is problematic. Worse, balancing means we're sacrificing charge for `balance' -- and I can't imagine why we would do that!

Balancing through discharge buys us nothing if we assume we will always operate with an LVC. Nothing. We always want the maximum charge we can safely achieve for each cell. We never want to waste energy. Balancing wastes energy for no apparent reason.

Richard

That's what I though as well, but my tests have convinced me otherwise. If you look at the numbers I posted, you can see that there's no loss in capacity at all. After a full charge, the cells will bleed off the surface charge, and end up at around the same value. Using the auto balancers just means the cells will be closer to the same number, but the total pack voltage is about the same, a little over 40V in each case. In fact, you could argue that you might get slightly more capacity before the LVC kicks in, because otherwise the low cell would trip it before it would if it were closer to what the others are at. The reality is it is probably not a big difference, in either case.

I would agree with you if you have cells that have significantly less capacity, balancing down to the lowest common denominator doesn't gain you anything, but with healthier cells, it is better to have them balanced. As I said, I have some a123-based packs with blocks of "stressed" cells, that have 5-10% less capacity. For these packs, some form of individual cell charging is still the best option. For everything else, I'm finding simple bulk charging, with occasoional balancing, does just fine.

-- Gary

 

[rf]

[...]
That's what I though as well, but my tests have convinced me otherwise. If you look at the numbers I posted, you can see that there's no loss in capacity at all. After a full charge, the cells will bleed off the surface charge, and end up at around the same value. Using the auto balancers just means the cells will be closer to the same number, but the total pack voltage is about the same, a little over 40V in each case. In fact, you could argue that you might get slightly more capacity before the LVC kicks in, because otherwise the low cell would trip it before it would if it were closer to what the others are at. The reality is it is probably not a big difference, in either case.

I sorta/kinda see your point. But it is so fleeting -- and subjective.

What we want is: amp-hours.

With bleed-off-balancing we're sacrificing amp-hours from all our cells but the weakest. Using voltage as a gauge -- a flimsy viewpoint at best.

Bleed-off-balancing is akin to an obsessive-compulsive straightening the fringe on the rug each time he passes. The difference being his efforts may have tangible benefits, since things at least look better.

Burn off that `excess' energy for the sake of `balance' and you use up portions of the finite life of all the stronger cells -- eventually going beyond just making them look weaker for the moment. Actually wearing them out.

I'm sorry all the time and funds you spent on the CMS didn't pay off. A loss for all of us, but especially for you. But balancing isn't the answer.

Richard

 

[Jeremy Harris]

Thanks for all the hard work you've put in Gary, and for making the results available here for all of us to see and learn from.

It's certainly helped me resolve the charging dilemma on my electric motorcycle conversion. I wanted to have a charger that would actively balance the cells, like your shunt system, but also really want the option to be able to charge away from home, which means having a charger built in to the bike. I have a 350 watt switched mode power supply that is small enough and light enough to fit in the bike, but finding room for a box with a big heatsink, fan etc for the shunt balancing system was going to be a problem.

I think I'll risk going for a constant current/constant voltage charge solution built in to the bike, coupled with your LVC hooked up to cut the controller off. I'll make a separate shunt balancer box, with a big fan cooled heatsink, which I can just plug in at home every now and again to equalise all the cells.

I just need to make up a PC board for the shunt balancer, which sounds like a job for next weekend. I've spent most of today milling thick alloy plate for the motor mount, so a bit of electronic work will be a bit like therapy...............

Jeremy

 

[Deepkimchi]

I'm going for the 36V pack, charger and balancers, with an LVC board for my existing pack (from Gary).

With the family and a bunch of stuff going on, don't have time to build my own shunt balancer (or the know how).

Lastly, makes a big difference with who you are buying from.

Gary tried to talk me into just getting the battery pack and LVC boards, he would adapt the connectors to work with my Flintstone charger. He was trying to save me money. But I wanted an extra charger anyway..

That says a lot to me. :wink:

DK

 

[EMF]

Gary, sorry to hear that the project was not going to be profitable for you. In light of that, I think then you made the right decision, however hard it was to make. My dad used to say, that a man can work himself to death, or a man can starve himself to death and that's alright. But he is a fool to do both at the same time. :wink:

 

[PJD]

But meanwhile, Bob M. is is absent from this board, and hasn't answered an e-mail I sent him a few days ago either.

So, at this point, with warm weather here, my LiFePO4 battery investment sitting on the workbench, gasoline at $3.50 a gallon, and an existing "silicone" battery pack fading fast, I find myself wondering about the ultimate availability of the the Lifecycle BMS. I wish Bob could provide a firm time frame - even if it is that it is not going to be available for 6 months. It seem my entire journey through this world of alternate EV transportation so far has been about open-ended waiting, and un-returned phone calls and e-mails.

I don't men to sound like an "irate consumer" about this, but that is how I feel right now.

 

[GGoodrum]

I sorta/kinda see your point. But it is so fleeting -- and subjective.

What we want is: amp-hours.

With bleed-off-balancing we're sacrificing amp-hours from all our cells but the weakest. Using voltage as a gauge -- a flimsy viewpoint at best.

Bleed-off-balancing is akin to an obsessive-compulsive straightening the fringe on the rug each time he passes. The difference being his efforts may have tangible benefits, since things at least look better.

Burn off that `excess' energy for the sake of `balance' and you use up portions of the finite life of all the stronger cells -- eventually going beyond just making them look weaker for the moment. Actually wearing them out.

I'm sorry all the time and funds you spent on the CMS didn't pay off. A loss for all of us, but especially for you. But balancing isn't the answer.

Richard

It has long been my experience that voltage levels track very close with caapacity, both with a123 cells, and now with LiFeBatt cells.

Again, I would agree that balancing doesn't get you anything if you have cells with significant (>4%...) differences in capacity. For healthy cells, with similar capacities, balancing doesn't hurt, and it can actually help. If you don't believe me, do your own tests.

-- Gary

 

[GGoodrum]

Gary, sorry to hear that the project was not going to be profitable for you. In light of that, I think then you made the right decision, however hard it was to make. My dad used to say, that a man can work himself to death, or a man can starve himself to death and that's alright. But he is a fool to do both at the same time. :wink:

Your dad was/is a wise man..

 

[ngocthach1130]

What is the estimated time of arrival for the CMS board. How about pricing. How much would the CMS board cost.

 

[Deepkimchi]

I think if y'all read Gary's last post on page 17, you'll find out IT (CMS) AIN"T COMING.

That's why I've ordered the Soneil charger and TP balancer set up from Gary.

DK

 

[Jozzer]

Gary, you say the cells settle back to even after a bulk charge, but what worries me more is that the first cell to complete will be exposed to higher than spec voltages without individual cell protection...do you have figures for voltages in the last few mins of charging?

 

[miro13car]

I also sent Bob Macree e-mail few days ago when BMS kit will be available, I wanted to buy LifeBatt cells but with only LVC that's not enough for me , I need balancing I am going to buy LifeBatt battery and charger. I don't care LifeBatt batteries were designed for automotive, it is my role to adopt to my bike.
MC

 

[PJD]

Bob's absense and non-response to e-mails has got my conspiracy-instincts going.

Wonder if Bob has encountered legal troubles with the BMS and Don Harmon/LiFebatt, and is, let's say, "under the advice of an attorney".

 

[EMF]

Bob's absense and non-response to e-mails has got my conspiracy-instincts going.

Wonder if Bob has encountered legal troubles with the BMS and Don Harmon/LiFebatt, and is, let's say, "under the advice of an attorney".

Although it is not like Bob McCree to not be around here much, or answer email...perhaps he is on a vacation.

As far as the conspiracy thingy goes, I find that most unlikely - at least the part about procuring an attorney. Bob and Gary were making these items out of the goodness of their hearts, in an efffort to help the ebike community. It's my understanding that there was not much (if any ) profit to this endeavor. So, it seems Bob would not want to go into a negative cash flow situation by hiring a lawyer. It would make more sense to simply post that he has abandoned the project.Which, he has not. But, he did say he had a falling out with LifeBatt and was no longer dealing with them.

 

[GGoodrum]

Gary, you say the cells settle back to even after a bulk charge, but what worries me more is that the first cell to complete will be exposed to higher than spec voltages without individual cell protection...do you have figures for voltages in the last few mins of charging?

All LiFePO4 cells I've tried can all take as much as 4.0V without issue. Both a123 and LiFeBatt have done their own internal tests, and haven't seen any long-term reductions in cell life if higher than "normal cutoff voltages are used. One a123 engineer told me that although they haven't really seen any problems, he still said that since it really doesn't gain you anything over having a cutoff down around 3.7V, it probably is better to not use such a high cutoff. Same with LiFeBatt. Their original VMS design used a brute force cutoff of 4.0V, but without using a CC/CV profile. Now, they use about 3.7V. The VoltPhreaks individual cell chargers actually use 3.80-3.85V as a cutoff. The point is that the cells have a fairly wide tolerance for where the CC/CV cutoff is set to.

I really don't see how this is going to be an issue. If the charger limits the voltage to a total value that is 3.65V x the number of cells, even if you had one cell that had somehow managed to end up with a voltage significantly lower than the rest, like say .5V, which is a huge imbalance, the rest of the cells would only go up to about 3.7V before the charger starts limiting the voltage. By balancing every so often, you just simply wouldn't have this case, as the balancers would not try and balance a pack that has such an imbalance. It will kick this out as an error.

-- Gary

 

[GGoodrum]

Bob's absense and non-response to e-mails has got my conspiracy-instincts going.

Wonder if Bob has encountered legal troubles with the BMS and Don Harmon/LiFebatt, and is, let's say, "under the advice of an attorney".

Ha! Good one...

I can assure you that neither Bob, or myself, are having any legal issues with Don and/or LiFeBatt. Bob and Don had some "differences", which caused Bob to just not want to be involved with LiFeBatt anymore, but they parted company on good terms I have no issues with Don, and I get along with him just fine. Michelle can be difficult, but I ususally only deal with Don.

LiFeBatt is not going to be directly involved in producing ebike-specific packs, mainly due to an agreement they have with Phostech, but they are still allowing me to do LiFeBatt-based packs, for those here and elsewhere who want them, as long as I don't use their name/logo on the labels.

As for Bob, I think he just decided, now that it is warming up a bit, to take his dog and go fishing. He told me he aoccasionally does this when he starts to get stressed out. I'll check today, but last time I heard from him, he said he had a bunch of controller repairs to do this last week, so he was going to spend most of last week doing that.

-- Gary

 

[EMF]

As for Bob, I think he just decided, now that it is warming up a bit, to take his dog and go fishing. He told me he aoccasionally does this when he starts to get stressed out.
-- Gary

Ha! What a great way to relax. 8) I don't have a dog, since I have to work all day...but it sure is nice to go down by the river and do a little fly fishing for rainbow and cutthroat now and again.