Faulty charge circuit?

[Chriull]

Btw: I would strongly recommend to not buy and use such GW wheels with 3rd party battery packs! Bad and broken welds are dangerous and not acceptable!

 

10 hours ago, Mantraguy said:

However, when I de-soldered the tab at B7 (middle of the board), the tab fell onto the workbench - it had separated from the cells

 

10 hours ago, Mantraguy said:

They were the two that were no longer connected to their parallel partners between B6 and B7 at the top of the board (due to the broken tab

And the already some posts before linked teardown of another such 3rd party battery:

https://forum.electricunicycle.org/topic/18494-warning-gotway-nikola-100v-1800wh-battery-fire/?do=findComment&comment=322601

Which came from, as i just read again this post from Wheel Riders Store, too!

So they seem to have a cheap and unreliable source for batteries. And they sell without quality control such dangerous packages - they wheels battery pack from the link started to smoke without the wheel beeing turned on or charged!

Edited April 5, 2021 by Chriull

[Mantraguy]

4 hours ago, Chriull said:

One group dropped from 4.12V to 3.94V within some days?

That would be >~15%S(tate)o(f)C(harge) within a couple of days!

Sorry, that's not what I meant. The 3.94v cell group was lower than all the others (4.11-4.12) when I first opened the pack to check voltages. I charged it to 4.12 and it has been fine (balanced with other groups) since then. I'm not seeing any self discharge of any groups (probably because I ride almost daily so they don't have time to self-discharge lol)!

4 hours ago, Chriull said:

As you've seen with the 1.5A charger you can charge more (downto 0.5A). If you then let the pack rest a bit and charge again you'll get further a bit more... Until after some cycles the cells should be all at full!

Maybe some very small discharges (short rides) inbetween the full charges.

 Ok I'll try that. I find it odd that one charger drops to zero immediately but I can then use a different charger and it gradually decreases current the way you would expect. If it were the BMS cutting off charge current because of an over-voltage, wouldn't it do the same thing if a different charger was then connected? 

5 hours ago, Chriull said:

I'm not sure if i get tge right meaning of  this sentence: They did _solder_ a tab to a cell!? Or this means that the tabs, all welded to the cells are soldered to the pcb?

Yes, the tabs are welded to the cells and then soldered to the PCB (this is normal). I can't be sure but it would seem that one of the tabs broke where it was bent over the cutout in the PCB. Perhaps it just broke when I desoldered it, but it would be a big coincidence that the bad cells were in that same group. I suspect that this was the primary problem all along, although I did get ~1500km before any problems so it must have broken after some time (not broken from the factory). 

I'm still waiting for new heat shrink to arrive so I can properly close the packs, but until then I will check balancing at discharged and fully charged states after every 4-5 rides to make sure everything is good. Any irregular cells will be flagged for replacement once I get the 12 new cells from China. But for now, the wheel works perfectly and I haven't found any unbalanced groups yet. The only things I'm suspicious of now are the chargers - they don't all do the same thing like I would expect them to. 

[alcatraz]

Cool. I didn't know the 1500Wh teslas were dual 10s4p packs. How unusual. So it's got daisychained balance boards. That could be an issue.

The group that was off after your last test ride, is it another group that was fine before or is it the one you previously charged or replaced?

Were the bad cells corroded on the same side that the nickel strip fell off from? Or were they corroded in other places away from that loose nickel strip? What about visual damage/corrosion/loose nickel at and around the 3.94v group?

If you intend to heatshrink the packs and leave no way to monitor the voltages, then that's not so wise. All packs will show issues given a long enough time. Some earlier, some later. Try to figure out a way to probe it even after you're done.

The good news is that this tesla has the potential to run a whole lot farther than original teslas because you have battey service skills most don't.

Edited April 6, 2021 by alcatraz

[Mantraguy]

10 hours ago, alcatraz said:

Cool. I didn't know the 1500Wh teslas were dual 10s4p packs. How unusual. So it's got daisychained balance boards. That could be an issue.

Actually no - there's just one BMS that handles both packs. There are photos previously posted in this thread. 

10 hours ago, alcatraz said:

The group that was off after your last test ride, is it another group that was fine before or is it the one you previously charged or replaced?

I'm not sure - it may have been the 3.94v group but I'll have to consult my notes. I'll be keeping an eye on any groups that show any imbalance.

10 hours ago, alcatraz said:

Were the bad cells corroded on the same side that the nickel strip fell off from? Or were they corroded in other places away from that loose nickel strip? What about visual damage/corrosion/loose nickel at and around the 3.94v group?

The cells with the most corrosion were the two bad cells, and only at the positive terminals. No, the nickel strip that broke was on the other side of that group, so I don't think corrosion was a factor there (the broken tab was otherwise clean). A couple of cells near the bottom of each pack showed a small amount of rust so they had gotten wet on one of my rides it seems, but I had a close look at those cells and they seem ok now. They are not the ones that were out of balance so the small amount of corrosion seems not to have affected them, thankfully. 

10 hours ago, alcatraz said:

If you intend to heatshrink the packs and leave no way to monitor the voltages, then that's not so wise. All packs will show issues given a long enough time. Some earlier, some later. Try to figure out a way to probe it even after you're done.

I've been thinking about that. The pack that has the replaced cells has an 11 pin connector that goes to the BMS on the main pack, so I can check balancing by disconnecting it and testing voltage from one pin to the next. I don't even have to remove the pack from the wheel, so it's easy to do just by removing the side panel. 

The main pack isn't quite as easy; I have to remove it from the wheel and check voltage at the tabs for each cell group. I hoped to solder a second set of wires to each tab but there isn't a lot of extra space for more wires and terminals. I'll have to figure out a better plan. I could have little 'windows' in the heat shrink but that defeats some of the purpose of using the heat shrink in the first place. I'll work something out when the heat shrink finally arrives...

[alcatraz]

Aha. So the circuit board in the secondary pack just connects the cell groups together. 

It's an interesting construction to do two 4p10s. I might do something similar with a new pack build.

If the other pack charges to its full 42.0V then it should be ok.

For topping up that weak group you might want to build something like this. A high current USB-C 1s charger. It can output up to 3A which is what the balance leads can take at most. You might need to top it up like once every 6 months.

Edited April 7, 2021 by alcatraz

[Mantraguy]

2 hours ago, alcatraz said:

Aha. So the circuit board in the secondary pack just connects the cell groups together. 

It's an interesting construction to do two 4p10s. I might do something similar with a new pack build.

If the other pack charges to its full 42.0V then it should be ok.

For topping up that weak group you might want to build something like this. A high current USB-C 1s charger. It can output up to 3A which is what the balance leads can take at most. You might need to top it up like once every 6 months.

That's an interesting contraption! I have a few chargers from RC hobbies I've been using to charge individual cell groups; I just made a little adapter that will charge through the balance port (since there's no BMS on that pack, the balance port leads go directly to the cell groups). On the other pack I'd have to connect to the tabs but haven't had to do that. 

The forecast today is for rain here so I may spend some time checking the balance of all the groups. They're discharged now (~35%) so I'll check first, then charge and test again. I've probably done at least 8-10 rides (charges) since I last tested so I should get some indication of whether or not I have any weak groups. Fingers crossed!

[alcatraz]

A tip when topping up is to never do it on a full battery all the way to 4.2v. It takes forever.

You can top up in like 1/3 of the time if you're below 80%. Actually even lower is better to reach max current.  Put a timer on your phone and don't forget to check the voltage. After 1-2 top ups you will know about how long it takes.

It doesn't matter if it's +/- 0.05v anyway.

Edited April 8, 2021 by alcatraz

[Mantraguy]

When I replaced the two bad cells (Mar 29th), I checked all of the cell groups at 95% charged and they were all within 0.02v of each other. After 3 rides (down to about 30% remaining), I checked them again and all of the cell groups were within 0.02v of each other at the 60% discharged state (3.66v/cell). After charging, they were all between 4.17v and 4.20v except for one group, which was 4.11v. That is the same group that had 2 cells replaced, so it seems the other 2 cells in that group may be very slightly damaged as well. But they only drop by 0.02v at discharge (well, resting after discharge - there's no easy way to determine what the voltage is while under load) so hopefully they'll be ok.

Since then, I have done 7 more rides, usually down to about 20-35% remaining. I then took advantage of a rainy day to remove the side panel again and check all the cell groups both at discharged and fully charged states. This time, at discharge, all the groups were 3.67v +/- 0.02v (one as low as 3.65v - the same one I replaced 2 cells in, and a few up to 3.68v). For our purposes, I consider this fully balanced. After fully charging and allowing balance to complete, they all measured from 4.16 to 4.19v, except for that same group that is always very slightly lower - it was 4.11v. I still find it odd that it shows more imbalance fully charged than it does at discharge...

As luck would have it, my nice big foot-wide (28cm actually) heat shrink finally arrived that same day, and I wanted to wrap the battery packs up 'for good' but also wanted a way to check balancing of each group every so often. So, I soldered 20g wires to each cell group in pack #1 and made up an 11-pin connector that hangs into a recess in the shell. I can now check the voltage of each cell group in both packs simply by removing the right side panel and disconnecting the balance lead from pack 2 to the BMS. Here's a photo of the additional wires from each cell group and the test lead I made - this is pack #1 (the one with the BMS):

 

I didn't have an 11-pin connector, so I made do with two 3-pin and one 5-pin connectors, which are now secured with hot glue and heat shrink:

 

Each group can now be tested simply by checking the voltage between any two adjoining pins in the group, as shown here:

 

Here's a photo of the balance-checking port all finished with hot glue and heat shrink:

 

[Mantraguy]

I forgot to take a photo of the packs all wrapped up with their new blue heat-shrink, but here's a test-fit photo of Pack 1 with the new balance-checking lead, all snugly placed in the side of the wheel:

There isn't really any extra space on the side with the other connectors (top of photo), since that side of the pack houses the wires and connectors for the main motor wires, the connectors joining the two packs, the BMS leads from Pack 2 to the BMS on Pack 1, the main discharge lead to the control board, and the charge lead from Pack 1 to the front charge port. There was more available space on the other side (bottom of photo), near the connectors for the LED rings, speaker, and bluetooth control board, so that is where I decided to place the new balance-checking lead. I covered the end with electrical tape just to be safe, and tucked it securely unto that gap out of the way of other components. I can now check the cell group voltages with that connector for all of Pack 1, and the cell groups in pack 2 can be checked by disconnecting the 11-pin connector between the two packs (barely visible at the top of the photo) and testing with the same method. I don't have to disassemble anything else to check balancing now; I can have it all checked and reassembled in about 5-10 minutes.

[Mantraguy]

Some other interesting discoveries with respect to charging. I bought a YZ450 5A charger that is 'tuned' for 84.2V and puts out a steady 4.67A. As the voltage nears 84.0V, the current decreases as you'd expect, but only to about 2.34A (which is exactly half the current it puts out at CC - seems unlikely to be a coincidence). It then maintains 2.34A for a while and then cuts off completely to 0A. At that point the measured voltage of the pack decreases to just under 83v (usually somewhere between 82.5 and 83.0). If I then use my 'old' charger, which normally puts out a max current of around 2.7A (it constantly fluctuates between about 2.3 and 3.0A), I can see it slowly decreasing the current and balancing the pack, right down to 0.1A or so. I'll attach a graph of a recent charge below. Any ideas? Is the 'new' 5.0A charger defective in some way? Why wouldn't it decrease gradually all the way to zero the way the 'smaller' charger does?

[Tawpie]

I returned a 5A YZ450 because one of my packs set its charge stop early (82.4V) the first time I used it. The pack was defective and was replaced, but I lost faith in the charger. Sucked it up and paid 2x for eWheels’ version—at least I know they’ll stand behind it.

[Chriull]

7 hours ago, Mantraguy said:

It then maintains 2.34A for a while and then cuts off completely to 0A.

This is a very strange behaviour - staying fix at 2.34A. As this happens during the CV (dropping current) phase the forced constant current again from the charger causes the voltage to rise! Probably to high so the BMS cuts off!

I could not think of any valid reason why a charger should behave like this!?

Seems very likely to be some fault.

7 hours ago, Mantraguy said:

At that point the measured voltage of the pack decreases to just under 83v

This is normal behaviour - once the voltage drop over the internal resistance is gone once no charge current is flowing.

Secondly after "interrupted" charging processes the "charge distributes internally" leading to some voltage adoption processes with different time constants.