New battery build: one parallel group behaves differently

[Jim Subzero]

I just built a 14s16p 40 Ah [edited] battery. When I put it together all cells (18650) were very close in voltage. I randomly tested a sample for capacity and found them close. I found one dead cell, and it could not be rehabilitated. A few cells were a couple of tenths of a volt below the storage voltage, but cycling them and testing their capacity showed them to be the same as the others.

With the wiring and packaging complete I put it on the charger. The BMS used has a bluetooth radio enabling me to watch the voltage of each parallel group. Group 10 charged faster and reached 4.2V well before the others. Group 4 lagged a bit behind the others. The BMS started balancing. I saw G10 briefly touch 4.25V than fall back to 4.2V a few times during this initial charge.

The next day I mounted the battery on the bike and went for a ride. The bike behaved well, a bit friskier than with its 48V battery. However, G10 instead of being high was a half volt lower than the rest. This is despite the BMS balancing, or so it says. Recharging does not change G10's lagging voltage; it stays roughly ½ volt lower than the rest.

What is wrong? Should I worry?

By the way, I need to mention a crucial fact. Each cell has its own pcb to protect it from over voltage, under voltage, and over discharge rate.

I speculate that perhaps one cell in G10 is bad somehow, but I have no idea what could produce the observed over and under voltage problem for that group. I am new to this.

 

[amberwolf]

You said G10 reached full first, but all started at the same voltage. This means G10 has less capacity than other cells, and higher internal resistance.

That means either one or more of it's cells are not properly connected to the rest (or it's protection has disconnected it or failed), or one or more of them are damaged or defective.

The same is true of other groups that don't track with the rest.


Additionally...if each cell has it's own protection, this interferes with the BMS's ability to monitor the groups, as each cell can cut itself out of the group, making voltage of the group rise and fall as charging current flows into less or more cells, or the reverse with discharging.

Whether that affects the actual ability of the pack to charge or discharge, you'd ahve to experiment with, but it could.

Generally, cells with built in protection are intended for single-cell use. A BMS is generally intended to be used with unprotected cells, so that it can itself protect them.

You may be able to use them together, depending on the application, but there are potential issues. They might not ever happen, hardly ever will, but it's usually a good idea to think about worst-case situations in designing a system, even if you don't end up worrying about the problem in teh actual design.

If one of hte protections kicks on and disconnects a cell under load, the load will increase proportionally on all the other cells. If all the cells in the group then disconnect themselves because of the overload, there's no current flowing thru the cells...but depending on the BMS electronics design, there might be current attempting to flow around the group thru the BMS sense/balance circuits, and that could damage or destroy them in the moment before the BMS reacts to loss of voltage in the group or the controller reacts to teh voltage drop by shutting off. Whether this is actually a problem depends on the BMS design, and would probably have to be tested experimentally to see what happens, if you want to know before encountering the problem.


Another potential issue is these protections built into cells are often pretty low-current units, and may even have high resistance in them, mulitplying the cell resistance many times, potentially creating a lot of heat in a pack made of them under loads typical of ebikes/etc.


A side note: you call it a " 14s16p 40 Wh battery.", but I suspect you mean Ah rather than Wh.

It is important to keep track fo the terminology, as many of these terms mean completely different things, and if you accidentally use the wrong number in a calculation somewhere, you can end up with undesirable performance or worse.

 

[Jim Subzero]

Correct about Wh vs Ah. My mistake. The pack's Wh capacity is about 2kWh.

I did check the resistance added by the pcb. With pcb, 130 milliOhms, without pcb, 90 milliOhms. [edited]

Why did I use protected cells? They were cheap, unused, and had wires attached, eliminating the need to spot weld.

It is true that all those pcbs add to the things to go wrong. And, it's possible that the frequent small voltage variations up and down by a millivolt or two of the parallel groups is due to the pcbs. I'm not clear why they would do anything when well within their "safe range". Those variations could also be artefacts of the BMS's measuring.

The motor controller limits current draw to 30A. A 16p battery asks for no more than 30/16 A from each cell. The cell's pcb limits current to 5A. It should be okay even if a cell or two are shut down in a 16p group. However, the voltage of the group should be little affected. My soldering, of course, is impeccable [imagine an emoticon here]. So far, the battery, which has two thermisters reporting to the BMS, shows temperatures close to ambient.

 

[amberwolf]

As long as the protection boards don't cause you problems, then it should work ok.

But there is still the problem of the group(s) with charging / balance issues, whcih is probably a bad cell or cells if you are certain the interconnects are fine.

 

[amberwolf]

FWIW, that is a lot of extra resistance. (just the extra resistance of the protection is higher than my whole 14s2p 2kwh pack's resistance). It may not matter for your application, but food for thought:

A bit of math (not my strong suit):
If you have 16 cells in parallel, then 130mohm (milliohm, not microohm, with cells like these) per cell becomes 130/16 = 8.125mohm per group. 14 series groups is then 113.75mohm, not including pack interconnects/etc.

30A thru a 113.75mohm resistance gives 3.4125v of sag. That's 102.375w of heat generated in the pack (a little less than half a watt per cell), 70.875w of that is in the cells; 31.5w of it is in the protection boards. Seems like a lot to me, but if it's acceptable to your usage scenario, and it all works...then it doesnt' matter.

Without the PCBs, it'd be 78.75mohm pack resistance, so at 30A it'd only be 2,3625v of sag. That's 70.875W of heat directly generated in the cells.

The voltage sag also makes a difference in how much power you actually get to the motor.

Actual resistance of the cells varies with state of charge, temperature, etc., so those numbers will all vary some in actual usage.

 

[Jim Subzero]

Your math looks good, but I should note that the motor hardly ever draws the full 30A. More typically under load I see about 6A or so, and that explains why the battery's temperature stays close to ambient.

I will take the battery (and the bike!) on a longer ride today. If anything interesting or unexpected happens, I'll report back. Sometime soon, if all is well, I will ride up Wolf Creek Pass, elevation 10864'. That will test whether heat is a problem. WCP is the local hill near me.

Normally, I would not choose protected cells for a battery, but cost and convenience won the contest. I was not eager to buy a spot welder. The cells were $1.13 each, $252 total. The Satiator charger for them cost more. I had been considering fusing the cells individually for the planned battery, but a protected cell takes care of that.

 

[Jim Subzero]

Photos of the battery build:

https://drive.google.com/drive/folders/1UZzZusqShNDQ-09lFpVJ_9Cr8hvwulhp?usp=sharing

 

[Jim Subzero]

my whole 14s2p 2kwh pack

What kind of cell are you using to get 2kWh from 28 cells? They must be big!

 

[Jim Subzero]

Just back from Road Test 2. Thirteen of the fourteen cell groups stayed close to 4V +/- 15mV; however, G10 apparently stayed at 3.4V. I say apparently because the BMS shut off power after a few miles claiming G10 was under 2.8V. I watched the BMS app, and G10 appeared steady at 3.4V. The app display updates about once a second. Waiting a couple of minutes, I could continue riding for a short distance, and then lose power, with G10 still looking like it's at 3.4V. Luckily, I had a fully charged 14.5Ah Reention Dorado clipped into the down tube. Flipping the battery selector switch got me out of trouble.

Clearly, G10 has a problem. Likely, it's just one cell, less likely two, and so on. Finding and replacing the bad cell(s) may be too horrible to consider. Less horrible would be to disconnect G10 and bypass it.

The BMS balancing wires from G10 on would have to be moved up one group each, and then the BMS would have to be converted from a 14s to a 13s, if that is possible.

Or, . . ., or, remove all the cell holders from the bottom of the cells; clip out all sixteen cells of G10; cycle test them all for capacity and internal resistance; replace any not nearly identical with nearly identical cells; solder them back in; install all new cell holders on the bottom side of the battery; shrink wrap and tape things; replace battery in its box; hope for the best.

I welcome suggestions.

 

[amberwolf]

my whole 14s2p 2kwh pack

What kind of cell are you using to get 2kWh from 28 cells? They must be big!

They don't seem big to me, but then...most everthing I do is on the "go big or go home" side. EIG NMC C020, 20Ah each. You can get them new over in Jimbob01's thread in the for sale section, though mine are old and used from a different source. I think his would cost about $170usd to make a 14s1p 20Ah pack, not including shipping and busbars. I forget what the busbars cost but I think it would bring it up to around $200. Shipping would be the killer from UK to USA.

They bolt together with busbars, so easy to build, and easy to reconfigure or repair if necessary, Mine were well-used when I got them a while back (several years for the first set, and a few for the second), so they're nowhere near new capabilities, but they still work well with good capacity and low Ri and not a lot of sag under load. New, they can handle 5C continuous, 10C for 10 seconds. (so 100A all the time, or 200A peak). They're not magic but they're a lot better than even the best 18650 cells I know of, overall.

It's on the SB Cruiser trike, which is over 500lbs with me on it, and can carry 350lb+ grocery loads or a St Bernard or two in the back (and pull a trailer with more, or a piano, piles of dogfood, etc). Presently has about 2kw per side in the rear, dual hubmotors, with about 80A+ peaks at startup, and around 18-20A cruising at 20MPH. (it's not very aerodynamic). The 14s2p pack is around 35lbs.

 

[amberwolf]

Clearly, G10 has a problem. Likely, it's just one cell, less likely two, and so on. Finding and replacing the bad cell(s) may be too horrible to consider. Less horrible would be to disconnect G10 and bypass it.

If it's only one cell, it would just about have to have an internal short to cause that kind of problem, by basically shunting all the charge current for that group thru it, instead of charging the other cells in the group. So it should be easy enough to find.

One way to locate a cell with this type of problem is temperature--it'll be enough warmer than the rest during charge that it should be "visible" to a point-and-measure IR thermometer (FLIR even easier, but less commonly available).

Cells near it will also warm up over time but not as fast as it will. Same is true of other problematic cells, in either discharge or charge, depending on the fault.

It might not be much of a difference...but they should read about the same (with core cells a bit warmer than edge cells, the longer the pack charges or discharges at higher currents).

 

[Jim Subzero]

And I think my bike big at 80 lb. The new battery weighs 31 lb in its polycarbonate box. That too seems big to me.

I'm steeling myself for plan two, remove G10 and rehabilitate it. Bypassing G10, while possible, seems cowardly. One must confront the problem.

Big, bolt-together batteries sound good now.

UPDATE, before posting the above: Just got your message about checking for the bad cell with an IR thermometer (I've got one). That sounds reasonable, but I don't think it will save a whole lot of trouble compared to the assurance I'd get from testing all the cells in G10. I don't want to have a repeat problem.