[dougingraham]

Everyone thinks about doing their BMS this way at first. The big problem is that you are powering the PIC and LED's from the cell it is monitoring and every cell has a little different load on it. This will imbalance the pack.

[PStechPaul]

Originally Posted by dougingraham

Everyone thinks about doing their BMS this way at first. The big problem is that you are powering the PIC and LED's from the cell it is monitoring and every cell has a little different load on it. This will imbalance the pack.

I admit that I do not have any direct experience with this, but I am used to dealing with precision instrumentation and measurement technology, and one of the main principles is the effect of the measurement on the system being measured. While in theory a current draw of 5 mA will affect the voltage of a cell, and there might be tiny differences in the current draw based on tolerances of components, there is no way that it will cause any perceptible imbalance on the cells of the pack.

In fact, if the battery monitor circuit draws more current from a cell with a higher level of charge, it would tend to balance the cells. I don't know the circuitry of a commercial BMS, but I would venture to guess that its current draw would be on the same order as the circuit I showed. And since 2.5mA is 0.01% of the one hour discharge rate of a small 25 A-H cell, it is probably well below the internal self-discharge current. In any case, it would take 10,000 hours, or over one year, to deplete the cell if it were just stored and unused and not on a float charge.

Moreover, the circuit I would use for a commercial version (or even for my own purposes) would use a microcontroller such as the PIC10F320, which has an operating current of just 25uA at 1.8V, and can also operate in a standby mode with just 500nA for the watchdog timer which can awaken it every 10 seconds or so to perform its measurement and display functions, such as flashing a status LED, and then resume sleep. If the status LED flashes for 0.25 seconds every 10 seconds, and uses 10 mA, the average current is 0.25 mA, which would deplete the cell after about 11 years of unused storage, at which time it's probably dead anyway.

I don't want to start an argument, but I'd just like to hear an explanation as to why the ideas I presented would cause problems. Thanks!

[dougingraham]

Originally Posted by PStechPaul

And since 2.5mA is 0.01% of the one hour discharge rate of a small 25 A-H cell, it is probably well below the internal self-discharge current. In any case, it would take 10,000 hours, or over one year, to deplete the cell if it were just stored and unused and not on a float charge.

The problem isn't so much the 25 microamps the PIC draws. And it isn't so much the parasitic draw bringing down the batterys after a few months. The problem is the variation in the LED's and the resistor dividers. If the board on one cell draws 3ma and the one on the next cell over draws 2ma that is a difference in state of charge of 0.72 AH in a one month period. This may be a severe example but if it was 1/10th of that it would be an amp hour per year of drift. Since I expect these batteries will last more than 10 years and they dont seem to go out of balance on their own without some outside agent causing the imbalance you have to ask how much imbalance is allowable? Well if you are bottom balancing in order to avoid bricking cells on an overdischarge event a 1% imbalance might be too much. In a top balanced situation a 1% variation is certainly too much.

I haven't worked the numbers but I am guessing that you could get away with 0.1% resisters in the dividers. Since the LED's will only be lit when there is some sort of event so at that point I expect you could rebalance the cell that has had the LED turned on. And this could be on for days since you are not bringing back the error to a central display point and you will get tired of popping the hood or trunk to look at the LED's after a few weeks of seeing nothing.

My own view is that it it were free I would have a millivolt accurate reading on every cell and a temp sensor that could do a 0.1 degree reading on that cell with no load on the individual cells. This would be displayed on LCD screen on the dashboard. It would be cool stuff right? Actually no. Engineers and scientists are the only ones who want to see this stuff. And I know that I would watch it for a few weeks and then wonder why I thought it was a good idea to spend all that time and effort on something that is useless to the normal operation of a car. Something that inherently imbalances the pack over time, maybe not enough to cause a problem in days or months but in years. Differences in microamps matter over a 10 year period.

[PStechPaul]

Thanks for the detailed explanation. I found more information on BMS ICs such as the http://cds.linear.com/docs/Datasheet/68021fa.pdf, and I see that it has provisions for discharging individual cells to balance the charge, and a means of daisy-chaining to achieve communication with a controller for display purposes. It's a bit costly, at about $10/1 and $5/100, but since it can monitor 12 cells at a time the per-cell cost is minimal.

I agree that this is a better solution than the individual simple cell monitors, although I think it would still be useful and I doubt there is any way that it could affect the balance of charge as you claim. But since the cells are generally concealed in the vehicle, such a system would not be useful, although it would be highly desirable for cells that are in storage outside the car. It would also be a great help for batteries that are used only for backup purposes or for occasional starting of an ICE. And it may be ideal for my own intended use for 12V lead-acid batteries used in electric tractors, where there may be only 3 or 4 batteries and they are generally exposed for inspection.

[dougingraham]

Originally Posted by PStechPaul

Thanks for the detailed explanation. I found more information on BMS ICs such as the http://cds.linear.com/docs/Datasheet/68021fa.pdf, and I see that it has provisions for discharging individual cells to balance the charge, and a means of daisy-chaining to achieve communication with a controller for display purposes. It's a bit costly, at about $10/1 and $5/100, but since it can monitor 12 cells at a time the per-cell cost is minimal.

I agree that this is a better solution than the individual simple cell monitors, although I think it would still be useful and I doubt there is any way that it could affect the balance of charge as you claim. But since the cells are generally concealed in the vehicle, such a system would not be useful, although it would be highly desirable for cells that are in storage outside the car. It would also be a great help for batteries that are used only for backup purposes or for occasional starting of an ICE. And it may be ideal for my own intended use for 12V lead-acid batteries used in electric tractors, where there may be only 3 or 4 batteries and they are generally exposed for inspection.

The only reason for the discharger on the cells is to fix the inbalance caused by the device itself (at least in the case of the LiFePo4 cells).

There are documented cases of LiFePo4 cells in storage sealed in their original boxes for over 4 years that when opened are still all at the original 3.30 volts that the ones put in service 4 years before were. There is no self discharge of these cells. Now they might have lost some capacity just sitting there in boxes but the state of charge didn't change. I dont know of anyone who performed a capacity test and then put the cells back to 3.30 volts and stored for many years and then tested capacity again. I've only done 6 months and could not see any degradation. If there was any it was below my measurement threshold. I also didn't see any degradation in a cell stored fully charged for 6 months. I did see degradation of about 1/2% of a cell stored for 6 months at a resting voltage of 2.7 volts. So don't store them discharged for any length of time.

Since LA batteries have high self discharge characteristics there could be some merit to a monitoring system. I am not certain it would help the life of the batteries but it would let you know when one went bad. And since they seem to last only a couple of years if you take really good care of them this seems like it would get used.

Note: My comments apply only to LiFePo4 type cells. This includes ones with a pinch of Manganese and those with a pinch of Yttrium in the mix. Other varieties of lithium cells almost certainly behave differently.