Quad Cell Distributed Digital BMS design discussion

[John in CR]

Gordo and amberwolf,
Some of these concerns can be addressed with a BMS that balances cells by capacity, not voltage (e.g., as can be done with TI's bq78PL114). It's actually kind of creepy watching it work.

Depending on the initial imbalance, it might take a couple of cycles to balance the cells out but after that the BMS should have no problem keeping up with any capacity imbalance that forms due to differences between the the cells.

Won't a capacity balancing system just reduce the whole pack down to the level of the worst cell?

How about eliminating the need for a BMS altogether by coming up with a highly efficient DC/DC converter that can handle 3V input and output the voltage and current we need, so we can just connect all of our cells in parallel?

 

[Alan B]

How about eliminating the need for a BMS altogether by coming up with a highly efficient DC/DC converter that can handle 3V input and output the voltage and current we need, so we can just connect all of our cells in parallel?

This could be done by either winding the motor for 3V or by upconverting the voltage. Either way the weight and loss would be higher than a higher voltage setup. Dealing with current in the range of 500-1000A would be difficult and the copper required would be quite bulky and heavy. It is possible but probably not what we would want to carry on a bicycle.

 

[BatteryMooch]

Gordo and amberwolf,
Some of these concerns can be addressed with a BMS that balances cells by capacity, not voltage (e.g., as can be done with TI's bq78PL114). It's actually kind of creepy watching it work.

Depending on the initial imbalance, it might take a couple of cycles to balance the cells out but after that the BMS should have no problem keeping up with any capacity imbalance that forms due to differences between the the cells.

Won't a capacity balancing system just reduce the whole pack down to the level of the worst cell?

How about eliminating the need for a BMS altogether by coming up with a highly efficient DC/DC converter that can handle 3V input and output the voltage and current we need, so we can just connect all of our cells in parallel?

The cap.-balancing BMS will balance all the cells to an average capacity since the highest cap. cells will give charge to the lowest cap. ones. The loss of capacity from the highest cap. cells doesn't make a difference since the pack would have been limited originally to the lowest performing cell since that one would reach LVC first (typically, depends on the differing IR of the cells too). You end up ahead.

Designing 95% efficient (or even better) converters is easy but isolated ones are expensive and you'd still need a BMS since charging the cells individually doesn't address internal resistance (IR) or capacity imbalance between the cells. Each cell will charge to a different level and will discharge differently due to their different capacities and IR.

A chip like the bq78PL114 can compensate for IR, capacity, and temperature-induced (some cells hotter than others) differences between cells.

[Edit] Ahh...you meant to run the bike from individual cells with a DC-DC converter on each? Oops, I thought you were referring to charging only. See Alan B.'s response above.

 

[John in CR]

[Edit] Ahh...you meant to run the bike from individual cells with a DC-DC converter on each? Oops, I thought you were referring to charging only. See Alan B.'s response above.

No, just one DC/DC converter with all the cells in parallel. Due to the very high current at 3V+, you'd want to mount the up-converter directly to bus bars on the pack. How do we build a 95%+ efficient DC/DC converter that accepts 2.5-4.2V in and outputs 99V/100A, because I definitely want several. Since a big power single cell charger may be hard to find, it's probably better if the converter goes both ways so it can charge the big 1s pack as well.

 

[Alan B]

[Edit] Ahh...you meant to run the bike from individual cells with a DC-DC converter on each? Oops, I thought you were referring to charging only. See Alan B.'s response above.

No, just one DC/DC converter with all the cells in parallel. Due to the very high current at 3V+, you'd want to mount the up-converter directly to bus bars on the pack. How do we build a 95%+ efficient DC/DC converter that accepts 2.5-4.2V in and outputs 99V/100A, because I definitely want several. Since a big power single cell charger may be hard to find, it's probably better if the converter goes both ways so it can charge the big 1s pack as well.

I think this is good material for a new thread, it is not a quad series sell BMS design so it is out of scope here.

Thanks,

 

[Jeremy Harris]

How about eliminating the need for a BMS altogether by coming up with a highly efficient DC/DC converter that can handle 3V input and output the voltage and current we need, so we can just connect all of our cells in parallel?

I think this was first suggested a year or two ago by Tiberius. If you can get a switch mode boost converter that will do the job, yet end up giving you an overall efficiency that's no worse than a conventional high voltage battery and management system setup, then I think the idea is a really good one.

The best (in terms of efficiency) approach would probably be to combine the controller and switch mode boost converter into the same unit. The power stages would be best designed around a boost converter topology, with variable output voltage/current, rather than up-converting the battery voltage to some arbitrarily high value only to PWM it back down to what you need (at anything other than full throttle). It'd need some hefty inductors, but in principle I don't see why it need be significantly more lossy than a conventional controller. The advantages in getting rid of high DC voltages and removing the need for any sort of clever BMS would be well worth the effort involved in looking at this, I'd have thought.

The I²R losses could be managed easily enough, just integrate the controller to the battery pack and have short, massive, buss bars connecting it up. I'd need to do some thinking about the possible snags with the switching stage devices, but in principle I don't think it's an insurmountable problem, probably just an expensive one..............

Another approach would be the one you suggested, John in CR, but use a bank of ultracaps to meet the high peak current demand, charged by a DC DC converter sized to meet the average demand, or perhaps a bit above it.

Sorry for going off topic.

Jeremy

 

[Alan B]

Engineering realities cause boost converters to be lower in efficiency than buck converters. Efficiency falls as the ratio of supply to delivery voltage increases. These characteristics are well known and documented in the engineering trade.

Back to topic.

 

[Jeremy Harris]

Sorry for being a complete ignoramus, daring to suggest something novel and taking your thread off-topic. I guess I need educating some more by clever guys like you "in the engineering trade", rather than rely on my own humble qualifications and experience.................

 

[Tiberius]

How about eliminating the need for a BMS altogether by coming up with a highly efficient DC/DC converter that can handle 3V input and output the voltage and current we need, so we can just connect all of our cells in parallel?

I think this was first suggested a year or two ago by Tiberius. If you can get a switch mode boost converter that will do the job, yet end up giving you an overall efficiency that's no worse than a conventional high voltage battery and management system setup, then I think the idea is a really good one.
....
Sorry for going off topic.

Jeremy

It is certainly something I've been looking at. I built some DC-DC step up converters that would parallel up 36 V batteries to an output of 43 V.

I was thinking it might work better starting not with single cells but with packs of 4, nominal 12 V. Each pack is stepped up to the output voltage, say 60 V, and you then stack up more packs and step up converters to get the output current and Ah you require.

Suppose each pack/converter was 50 A in at 12 V and 10 A out at 60 V. Those aren't unreasonable figures for a DC-DC converter. If you had 5 of these it would look like a 60 V, 50 A pack, the same as if you connected all the cells in series. The difference is that
a) the voltage would stay constant as the packs ran down
b) the Ah is not limited by the weakest cell
c) if one cell fails completely you only lose 20% of the capacity

That leaves out the useful fact that the output voltage is controllable. So, as Jeremy says, that's another way of doing throttle control. Just run the controller at 100% PWM and vary the supply voltage.

Of course, in the example above, an intelligent 4 cell BMS would be needed. Sorry, seem to have strayed on-topic.

Nick

 

[Alan B]

Sorry for being a complete ignoramus, daring to suggest something novel and taking your thread off-topic. I guess I need educating some more by clever guys like you "in the engineering trade", rather than rely on my own humble qualifications and experience.................

My apologies Jeremy, not trying to give you a hard time, just trying to gently stay on topic.

The up converter topic deserves its own thread as it is a valid discussion (and I have considered it). Folks might wish to review online design guides for switching power supplies, this is a well documented area of expertise, and the vendors selling components provide various tools for design simulation and calculations. At the end of the day the extra loss, extra weight and increased component count caused me to not consider the boost converter further for a bicycle project. But that's just my understanding.

 

[Alan B]

See new thread:

http://endless-sphere.com/forums/viewtopic.php?f=14&t=25706