Sunday, 16 February 2020

Mercane 8.8Ah pack repair

This all started back at this post. However I think that this time I've nailed it, mainly because unlike last times I've identified a very significant physical factor: a rusted through connection bus strip across the postives of one parallel bundle of cells (yes, the one that was down on that last BMS test).


I found it (and as you can see it was buried beneath the paper where the BMS sat) while trying to identify which set of physical cells that wire loom (just to the right of the picture) connected to, as I'd been making my voltage measurements off that. Its hidden under the paper behind the BMS as seen in this picture:


(featuring a Reddit Approved Kitchen Utensil as a tool).

Anyway, the voltage measurements at the pack didn't marry up with what the BMS was seeing, so I dug in to see. When I peeled it back I saw a big scab of rust across the cells, which basically just crumbled away when I hit it with the vacuum cleaner.


So essentially this leaves the last three cells unconnected to the BMS and so only one cell was getting charged and supplying power.

I cleaned it up with the wire wheel on my Dremel tool ...


which showed me that I just needed to bridge that set of cells and I'll be right.

Now lacking a spot welder I pondered how to best do this, and decided that because it was the positive ends I could prepare that lone cap with some solder if I worked fast, because the top of that cap is not directly connected to the electrode. So if I worked the right way I could solder with minimal damage. The cells left and right of it I could solder onto the nickel (hah ... fucking nickel my arse) strip.

First however I gave them a balance charge with my iMax and then soldered two strands of copper ~1mm in diameter across those three cells.

I charged it till it was only pulling 0.1A and took it for a 5km ride to test it.  All went well, so over night I charged it again. This morning when I checked the pack it was at 54.6V, with 0.00Amps going in to it. I thought that was very encouraging and ideal at about 4.2V per cell (bundle).

So maiden voyage on the repaired pack was one of my standard test routes.


which is an excellent result and compares very well to what I got before I put in the new BMS and just balance charged the pack by hand (hoping that would fix the issue, but it didn't):


Comparing those two results in more specific details:


shows that the pack has a bit more pep in places but what is really significant is the voltages were way up when I got back, suggesting I'm going to have increased range (as well as I'm sure the cell balances will be better with the new BMS).



This is the voltage upon return of that trip (and an additional little 2km victory lap I did). Further as I found in my pull down and mentioned in that post, the balance charging is doing much better with the new BMS than the old one:


the blue columns are the result of a "full charge"with the old BMS, which clearly wasn't balancing anywhere near as consistently as the new one. Probably equally important is that my testing to BMS shutdown yesterday (which prompted this new pull down) gave this result:

V discharged
4.03
4.04
4.03
4.03
4.03
4.03
4.03
4.03
4.03
4.04
4.2
3.66
4.03
4.02
0.12
50.7

which:

  1. shows how bloody well balanced the cells were after initial "full charge" levels are shaved off by a few km riding and
  2. is what led me to to this pull down to investigate why the BMS shutdown was happening and why that cell was low.

Discussion and Conclusion

I think that this time I've got it nailed, and unless there are other cells in this condition the only reason I'll need to pull the pack down again is if the solder joints I put on need repairing (which will only happen if the solder melts because the copper heats up under power: meaning my calculations on the required diameter are wrong).

Its pretty clear that this pack has failed (dead pack walking) by catching it early I've minimised cell harm and saved money. because of improper assembly and cheaping out on materials. I'd speculate that in the Chinese Sweat Shop where this was assembled the worker involved dripped some sweat right there on that part of the bus connector and started this rust process right then and there, before the pack was even sealed. I've seen the impact of a single drop of sweat on a circuit board before (the salt in sweat just helps electrolysis and oxidation reactions go nuts). Had they used decent nickel this wouldn't have happened.

I don't believe that there are other cells with suspect (rusty) joints because I can see all those other joints and this one was the only one hidden by the paper under the BMS.

This highlights the importance of pack construction being at least as important as cell choice. For even if these were LG cells and even if said LG cells were better, it would still fail because the assembly of the pack was flawed.

I'll do some range testing and perhaps do another pull down to check again.

Basically for

  • my time, and 
  • 2c worth of solder and scrap copper wire


I've saved a $250 pack which I anticipate will go on to last another year (its already lasted nearly one).


I hope this is also helpful to someone else.

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.

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 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.

Saturday, 8 February 2020

detachable mudguard

I got a bit tired of applying and removing the duct tape for my "mark 1" mudguard (mainly because rain has been so sporadic but that's changed now) and so I decided to mung up one that's operable more permanently and conveniently.


basically its just a piece of plastic (from an empty 2L plastic milk bottle) which I cut to a shape, then put some duct tape over it to make it "black" and then affixed it with self adhesive velcro.

Loop side on the flap to keep it from getting clogged:


hook side under the guard


The easiest way to make sure its all "aligned" is to put the two parts of velcro onto the flap and put it in place.


... pressing firmly to make sure that the adhesive has taken, then just remove it.


Make sure the underside of the guard is cleaned first.

Cost breakdown:

  • milk bottle (fished out of the kitchen bin) = $0.00
  • velcro ($2.5 / 1.5 meters = 60c / meter, used 3cm) = about a cent
  • duct tape ($4 /  30 meters = $0.13 / mete, used about a quater of  a meter) = about 3 cents
and people want to buy this sort of thing from eBay and pay $15 for it.

hmmm

Tuesday, 4 February 2020

Mercane (single motor) new BMS

So I decided that I didn't like the way things were appearing on charging since I did my rebuild, so I pulled the battery again. The first thing I did was measure it.


so while no cells are down below 4V (which is good) one is close, and two are below 4.1V and no cell is at 4.2 , which at least one should be given it was charged till it "would take no more" over night and in repeated phases with a timer to boot.

So up on the ($20 folding) bench (in the kitchen, because I still haven't built my workshop!!) I pull the battery and reopen it after the last surgery.


and am able to plug the new BMS (red arrow) directly into the existing balance wiring harness (yeah), but will need to redo the input wires because the new BMS is a 3 wire BMS and the old is a two.


so this new BMS should be better. So it went in with some minor adjustment, where the charging, battery and power -ve each gets its own port.

Plugged in and working.


Seems good ... lets see


But alas it was too good to be true, on the way to work this morning:
Speed dropped, voltage dropped and then everything went dark.
So either:

  • my soldering failed
  • the BMS failed
  • something else

Either way, it was an Uber back to my accommodation, stuff it in my truck, and drive that to work.

Worth noting for anyone who thinks having only one scooter is a good solution.

More to follow