Daniel828 said:
Hi Gary. thanks for responding.
After the first pack (in the pic) melted I built a new on. I used cells I bought on ebay. I checked the voltage on them. Most were at 3.3 or better but two were below 3 volts. About 2.7-2.9. The same thing happened with this pack, it melted. Was the two lower voltage cells the culprit? I thought they would self balance if you charge them a little higher?
The other dilema. I like to use three 8 cells per pack, packs. I charge them in parallel, then connect them in series for roughly an 80 volt working pack. The question is, should I parallel the cells first then in series. So, I use them as 2 cells soldered in parallel, then 8 pairs of those 2. For a total of 16 cells per pack (8s2p).Or, should I have (6) 8 cells in series per pack then hook two in parallel, then the three pairs in series to get 80 volts. Gosh I hope that makes sense.
With out going thru the Care and feeding thread, does your BMS provide a per cell LVC capability? If not what could I do?
Thanks
--Dan
This is a good thread because it shows others exactly what
not to do with a123 cells, or any other LiFePO4-based "homemade"/DIY pack. First of all, you need to make sure you start with healthy, fairly matched cells. Cells that down around 2.7-2.9V are not "well within range". You need to first
individually charge each cell to about 3.7V (set the Mastech to 3.70V and about 2A...), wait about 1/2 hour, and then check the voltages. Healthy cells will retain a surface charge, and the cells will be somewhere in the range of 3.55-3.70V. Cells that are "stressed", meaning they have been previously damaged due to over discharging them, will have lost about 10-15% of their capacity and althought they will appear to take a full charge, up to 3.70V, after a few minutes the surface charge will bleed off by itself and the voltage will drop to about 3.38-3.45V. Do this test with each of your remaining cells, and with any of your replacements. Only build packs with the ones that pass the surface charge test, that is the ones that will hold the voltage up to something above 3.55-3.60V.
Running a single string of these only gives you 2.3Ah of capacity. On an ebike, that will not last but a few minutes/miles. The problem is that because a123 cells are so strong, you don't get any indication that the cells are about to dump until it is too late. It can feel just as strong 10 seconds before you are killing cells, which you apparently have done repeatedly, as it does at the beginning of the capacity. I think I remember telling you, or maybe someone else, that you were making a big mistake with this setup, in another thread. You have componded the problem by not even bothering to start with fully charged, matched cells.
If you are trying to run a 24s2p setup, using 8s2p "sub-packs", it would definitely be much better to wire the cells in parallel first, and then the pairs in series. You definitely need to have a 24-channel LVC function, so that each parallel pair of cells can be monitored (once connected in parallel, the cells will self-equalize to the same level, so they can be treated as one, larger cell...). The way the LVC function, which is part of the BMS, works is to trip an opto-coupled output if any one cell drops below 2.1V, under load, or otherwise. These opto outputs are all connected in parallel and then you use this "ganged" output to switch on the brake input on your controller. Think of it as 24 individual pushbutton switches, all connected in parallel to the brake input. Any one circuit trips, and the controller immediately cuts the throttle. Once the load is removed, the voltage will recover up above the cutoff, and so the throttle is re-enabled. If you don't let up on the throttle, the cutout will occur again. It has been my experience with healthy packs that when the cutout first occurs, usually under full throttle, you have about 5-10% of the capacity left. If I back off the throttle, I find I can usually go another couple of miles until the cutouts happen with even the slightest throttle. At that point, if I charge the pack, it will take almost exatly the full rated capacity back in. BTW, a fully drained pack will still have a "resting" voltage that is well above 3.0V per cell, more like 3.10V.
The problem is charging. I'm assuming you still want to use the Mastech to charge the three 8s2p sub-packs in parallel, using the Mastech. I'll have to think about how one of the new 24-channel "tear-away" BMS boards might be configured to allow parallel charging of three sub-packs. Actually, I think this might work. What you would do is break the 24-channel BMS board, shown below, into the 8-channel sections.
The left-most section would act as a "master". The other two boards would have to have the six opto "bus" lines connected together with those on the main board, but the connections between the 8th and 9th cells, and between the 16th and 17th cells would be left cut. Instead, the bottom of the 2nd 8s2p pack would need to be connected to the top of the 1st power resistor on the 2nd board (the 9th...), and the bottom of the third sub-pack, to the 1st power resistor on the 3rd board (the 17th...). On the next version of the board, I will add extra connection holes, next to the existing ones, to make this easier. Anyway, this should allow the 8s2p sub-packs to be connected in parallel, for charging, and in series, for discharging, and still get the full benefits of the BMS, including the LVC function.
-- Gary