Saturday 15 February 2020

difference in BMS outcomes

This is sort of a part 3 in the ongoing issue of (what I believe to be) the BMS caused issues in the battery pack of my 2019 (supposedly, but I somehow think 2018 fluffed up) Mercane Wide Wheel single motor 500W scooter. For other parts see here and here.

What happened

I was riding to work and then suddenly the battery voltage just plummeted and I had to take an Uber the rest of the way. The failure happened on a flat just a km or so from work.

So I've pulled the pack apart a few times now (of course after the first pulldown it was only sealed back up with duct tape, so no cutting required after then) so that I could measure cell situations. This has been prompted by an early "flat battery" situation (at about half the range I normally got) before. Investigations ensued.

Firstly this is the situation of state of charge after a full charge (defined by observation of no further amps going into the battery monitoring with my 150A charge monitor tool).


V 1st represents the first full charge still using the BMS that came in the pack, the other V readings represent state of charge with the new BMS and all readings are taken with the pack having had an hour or so to "settle".

What stands out to me is:

  1. just how badly the original BMS behaved, with the lowest cell after charging being 3.86 (or just over nominal) and most cells not getting a proper saturation charge.
  2. just how much better the new BMS has managed to balance these cells after the punishment of the last 9 months of 3 or 4 times a week discharge <-> charge cycle
  3. how tolerant the cells are of mild abuse (given by the old BMS which probably dragged at least one cell bundle below an ideal low level.

Analysis and Speculation

I have been unable to find much on the BMS that it came with but what I've found implies it was an older design (initially made for 3 cells and then simply scaled) which is a passive system (not active, and reading this supports that view and why its failed) that I anticipate used the following simple triggers:
  • shut down pack at a pre-determined whole of pack voltage (which seemed to be 42.5V based on my experience, which would be a safe 3.2V per cell if all cells were equal (less likely in a bigger pack)
  • shut off charging when the highest cell got to above 4.22V (even if bleeding hadn't raised the others much)
So pretty clearly this new BMS (which is active looking at the video below) is doing a much better job.


But the over discharge seems to have done some damage to a couple of the cells in parallel bundles in at least one bundle because even with 49V showing on the display the pack goes into shutdown. When I get it home (often 30 min later) I can see this:

V discharged diff
3.94 0.26
3.9 0.3
3.94 0.26
3.96 0.24
3.91 0.29
3.92 0.28
3.94 0.26
3.91 0.29
3.76 0.44
3.96 0.24
4.06 0.14
3.47 0.73
3.92 0.28
3.89 0.31
0.14 0.14
50.7

where the bottom rows are average and standard deviation of the column.

Of course with the new Active BMS it will trigger a pack shut down to protect the lowest cell at 2.8V and not release the pack until that cell recovers at 3.0V, from the seller specs (assuming they're correct):

Type of batteries: lithium cobalt oxide / manganese lithium / ternary materials
Temperature protection: 55/65/75
Rated Discharge current: 20A
Equilibrium voltage: 4.18v
Equilibrium: ≤ 30MA
Overcharge protection voltage 4.25V
Overcharge recovery voltage: 4.1V
Over-discharge protection voltage: 2.8V
Over discharge recovery voltage: 3.0V
Protection Current consumption: ≤30μA
Short circuit protection current: 60A
Short circuit protection time: 500MS
Fine workmanship and durable.
Applicable: 48V13 string lithium battery protection board.

Meaning that when those low cells get down to that level (probably well before the others) the pack will shut down to protect them.

I'm now going to see if I can "groom" those two bundles up a bit with my iMax charger but if I can't it's going to mean more serious pack surgery to slice out those two bundles and replace one or more cells in that parallel bundle.

PostScriptum:

Further surgery and exploration after a failure to be able to cover even 1/4 of previous ranges revealed this:


which essentially presented two different halves to the BMS and the Discharge. This has clearly been coming for a while, and I believe fits my (and many other peoples) experience of a hesitation when hitting bumps.

Essentially only one cell in that bundle was carrying the load.

I'll repair this and report.
Lets see how this goes.

No comments: