Duff/weak cell?

[hohisilver]

I'm beginning to suspect the BMS/wiring even more now as the problem is still there (although not as extreme as before). Under load, cell 5 drops to <3.00V (even though it was the one that triggered a charger stop on Saturday when I last charged it). Any clues where the problem might lie? I'm thinking of initially swapping the PIC chip for another on the same board to eliminate the possibility of bad programming (although I'm not convinced this is the problem as this has gradually "crept in" over time). Could a dying comparator exhibit this effect?

I'm going to check the actual voltage with my meter next time I charge (bit difficult to test it under load ) and see if it is a misreported voltage. I'm also going to check that the connections are secure on the actual BMS board. I really hope I find something as although it wasn't as difficult as I'd imagined, removing the pack again is not desirable

[GregsGarage]

First step is to check the bms voltage reading accuracy with a multi-meter (or even 2). That will tell you if you need to be looking at the bms or the cell.

[hohisilver]

Well I think I can rule out the BMS - I tested the voltage at rest and charging (as I said, pretty difficult to test under load ) last night and it pretty much matches to within 20mV (which is obviously the resolution of the BMS anyway). I also swapped the PIC with its neighbour to rule out any issues there. So it's either a high resistance in the wire to the BMS or I killed my new cells when I had the "fireworks" when connecting it back up

What I could *really* do with is a cell tester - how's yours coming along Greg?

[GregsGarage]

Just for comparison, on my pack the difference between highest and lowest cell voltage under load is around 150mv at 0.5C, 200mv at 1C and 300mv at 2C. These differences increase as the pack nears empty and can be as much as 500mv. My hipower cells are rated for max 3C discharge for short periods). I am hoping that when I finally get the pack better balanced these will come down, but I never let the lowest cell go below 2.3 volts under load (and that has only happened a couple of times in 2,500 miles), mostly I can keep it above 2.5 volts during acceleration and usually above 2.8 volts while cruising.

What I could *really* do with is a cell tester - how's yours coming along Greg?

I am hoping to start soldering some components next month, a few priority jobs to do first. The project is here. Although the site hasn't been updated, Lee has made some pcbs (and revised the schematic) and I have one. To the best of my knowledge none have been built yet. A bit of history is in this thread

[hohisilver]

Yikes! That looks a little more complicated that I expected

However, I had a sudden brainwave last night on what's happened. I realised that my reconnection "fireworks" only involved one cell in my new cell pair. I reckon that took it out so that pair is now running on half capacity. Given that I've proved I had a loose connection which was probably causing dodgy readings I have a certain amount of confidence that the original cells are ok, so it looks like "pack out, replace one cell" time.

[GregsGarage]

Yikes! That looks a little more complicated that I expected

Lee is a analog guy, so lots of bits. I was hoping to replicate it with a micro, either a pic or an arduino. It would have been a simpler interface, but the software probably requires some sort of PID controller. I put it on hold while building my Matiz. In the meantime Lee has got some pcbs made and since I have all the components sitting on a breadboard, thought I may as well solder them to a pcb and see if it works.

Yes, a duff cell would explain your readings. If your fireworks discharged the cell below 2 volts it probably ruined it.

[hohisilver]

Swapped that cell last night, and although it still dips it's not as bad as it was, so I may have to live with it. It still charges quickly so maybe if I can get a better balance in the pack over time it won't be such a problem. I hope

[hohisilver]

As an update, I'm still getting strange behaviour on that cell. The scooter is rideable now but that cell still dips to 3V or so under acceleration, and still causes the charge to cut off prematurely. It's really annoying because I can never get the cells balanced now, I'm lucky if I get a "fully-charged" pack voltage of 84V these days because of that one cell.

I even tried temporarily increasing the balancing load on just that cell this evening but it still cut the charger way before it should have. I suppose the next thing may be to try the other old cell in there (and maybe even swap that cell pair with its neighbour to rule out a BMS misreading issue), but I'm running out of options and to be honest I haven't got the time or energy at the moment

[GregsGarage]

Ian, it still sounds like you still may only be using 1 of the 2 paralleled cells for some reason. I had the same issue with a slightly loose connection on one of my paralleled cell groups. With parallel connected cells, the only way to insure they share current equally is to make sure the connections are as perfect as possible. Any differences that could cause poor connections at one of the cells will force all the current through the other cell. Another possible cause could be differences in internal resistances between cells. I could see where 2 cells are connected in parallel, if their internal resistances are very different, only the cell with the lower internal resistance will do any (or most) of the work. Maybe, for best results, only cells from the same manufacturing batch should be paralleled? Or if from different batches, some individual cell testing to pair up cells with similar characteristics? I would swap that pair of cells with another pair that is working o.k. and see if the problem moves with the cells. I had thought maybe you could mix up the cells, but I think it is better to keep them paired.

[hohisilver]

I had thought about possible variation in batches, the new cells I bought are visibly different from the old ones, the negative end is silver in colour instead of black. Given that there may not have been anything wrong with the original cells I'm now thinking of putting the other old one back in and swapping that pair with its neighbour, as you say to see if the problem moves with it.