[retepsnikrep]

No updates but I am still fiddling and have just got a Lithium CR-Z (car) for research purposes.
~40mv down I would not worry for now unless it approaches the Lto low limits.

Just monitor and when it gets like 100/200mv + apart maybe charge with something like a wall wart while monitoring the voltage.

 

[jime]

I'm also willing to try charging that single cell, but I'm not sure how I'd do it.

I don't think 40 mV is enough to worry about yet. If you get your pack down close to 144V(2V/cell) then carefully watch that cell.

There is a cheap and dirty way to charge a cell, AND LIKELY DANGEROUS. You can use red and black jumper wires with alligator clips to jump a high cell to a low cell. Since the two voltages differ little, there isn't much current. The two cells will equalize fairly quickly. Two extremely important cautions: first you must remove the strap on that pack, and second, you MUST NOT jump two cells in the same sub-pack since this creates a high current short. I've done it, but it is very tricky and prone to error. If you get it wrong, you can damage a inter-cell bus strap, and of course a loose jumper wire is bound to create mayhem.

My first thought is get a spare low voltage wall wort, maybe one that outputs 5V, and alligator clip it to that cell, and wait until it rises. I'm assuming this is probably not safe or intelligent for some reason or another, so I'm asking here first -- A: Can I do that and B: if I did would it help me balance that cell?

This is a better solution if the wall wort has current limiting. Otherwise, it will probably destroy itself, but it would be an interesting experiment, and convenient for those who want to do a bit of balancing. A good quality constant current power supply would be better of course, but I don't have a specific one in mind - maybe mudder will weigh in since he knows lots about that sort of thing.

I intend to use my PL8 since it has balancing capability, but an adequate PL8 setup is very expensive, so not a practical solution for those who don't already have it

 

[minor4326]

How peculiar that some people have voltage issues. My pack voltage has remained pretty steady and my individual cells have changed very little, even delta hi-low has remained the same. I've had SOC issues but incredibly no voltage ones.

 

[Natalya]

Last night I plugged in the grid charger. I've got a low current one, 350mA, and I sometimes charge the car for about 12 hours up to about 170V over night, in order to use the car as a light PHEV. It started at 156V, and this morning it's now around 168V.

I checked that cell on the OBDIIC&C this morning (while the pack is still being charged). It went from being the lowest cell by about 40mv to now being the highest cell by 40mv. So maybe it's bad and it's not going to be able to be helped by balancing?

I do recognize y'all are saying that's not a big imbalance, but I've not seen the Ltd be this high before. My LTO install is about 1½ years old now.

Long-term I want to install the G3 batteries when mudder finishes the project, but I was hoping to remove my LTO setup and sell it to another Insighter when I made the switch to the G3 batteries.


EDIT
Another thought -- that cell is the one right before the breaker and the mega fuse -- is it possible its voltage swing may have something to do with that? I have noticed in two cars (mine, and in the 60 cell LTO car @Atheos built) that the very first sub-pack had the highest cell voltages. If cell position in the series matters, maybe I could "rotate" them by changing where the cables are hooked up?

 

[*sean*]

This is a better solution if the wall wort has current limiting. Otherwise, it will probably destroy itself, but it would be an interesting experiment, and convenient for those who want to do a bit of balancing. A good quality constant current power supply would be better of course, but I don't have a specific one in mind - maybe mudder will weigh in since he knows lots about that sort of thing.

Glad Jim chimed in here. All I could think was, one misplaced tool/wire and you are welding. I would not connect a wall wart for the reasons Jim mentioned and also because some "may" be tested for short circuit protection, none are probably tested connected to a cell capable of a hundred plus amps. You might look to how Honda brings out the voltage measurement taps for the G1 NiMH cells to the BCM for ideas. I believe they put current limiting resistors, or is it PTC devices, in series before the wires even leave the pack?

I sometimes charge the car for about 12 hours up to about 170V over night

What do you use to prevent overcharging during a period of inattention? It should stop automatically when any one cell reaches a set point.

It went from being the lowest cell by about 40mv to now being the highest cell by 40mv.

This sounds like a cell with degraded capacity; a cell both first to bottom, and first to top. If it is on the end of the pack, do you smell that lithium battery cell? Some of Fit batteries are damaged because they were stacked and the battery post was pressed into the battery, puncturing the case. Note that their eBay photos showed them stacked on pallets 4 or 5 deep. I have one pack with a damaged cell on the end. Is not really possible to look for damage because of how it is assembled.

It could also be degradation from an overcharge event if that cell won the race to the top in some past overcharge event.

I do recognize y'all are saying that's not a big imbalance

.04 V when the working range is about 0.5 V is about 8%. I wonder if a low current overcharge event could do this. I recall a 20% hit in the literature.

Another thought -- that cell is the one right before the breaker and the mega fuse -- is it possible its voltage swing may have something to do with that?

Under a high current situation, yes, it matters where things are connected. Here the currents don't seem excessive. Just fixed a problem like this today where I was measuring voltages on the same wires that connected cells to a load. Each time a load was applied on one cell, the voltage measured at its neighbors would jump.

 

[jime]

Yes, the way things are hooked together does seem to matter some. For example, when I balance cells 8 at a time with the PL8 I can see the effect on "measured" internal resistance caused by the strap bus. The IR of the cell containing the bus can be .001 ohm higher.

It wouldn't seem like these intra cells resistances would matter in the same way that I see it on the PL8 because each BMS board has 13 taps iirc.

The issue of contact resistance may not be getting adequately addressed in any of out conversions. Since LTO IR is so very low, I've been wondering if we should be making special efforts to lower contact resistance, like polishing the contacts and using some sort of contact grease like NoOX??

If there is contact resistance variation, the voltage differences are only going to show when there is current. The voltages should read differently when there is no charge current.

As Sean says, it's 8%, but that is with a tiny current. Drive the car and observe this cell under heavy assist. If it is well behaved under heavy assist, then I don't think you have anything to worry about. Statistically, there will always be small differences

 

[Natalya]

Under assist it reads as the lowest cell. Under charge it reads as the highest cell. Any kind of numbers I should be looking for?

So, and maybe y'all already answered this, I asked in previous post... like if i tried to balance it somehow (and I still obviously haven't worked out the particulars of that) would there even be a point, going by how its acting? Is there a possibility of helping the cell out in any way -- or is it a slow tick tick tick until death?

@*sean*
- Could I have overcharged it by accident because i didn't have a way to see when any particular cell got high?
Yeah maybe? Here's the thing though -- I don't have a way to turn off a grid charger based on individual cell voltages. I simply don't have that available. A year and a half ago I tried to build something to turn it off when the whole pack got to a certain voltage, but even that proved like beyond me. I bet I could try again and maybe get it, but I'm not an EE, and I've got a lot of other things going on too.

What I can say is I charge with a 350mA charger, and I only hook it up over night if pack voltage is 159 or less when I get home. How high would the whole pack have to get to damage things? Like past 180V right? I understand maybe this cell is weak or whatever, but I've never charged above about 172 total pack voltage. I guess I'm saying like if it was overcharge-able then like it must have been like that on its own or something, maybe compressed or something as y'all mentioned. I can take a look.

If I did try to go the wall-wort method I could use like a resistor to lower the current. I did find a 3V 300ma wall wort.

So uhh, possible intervention options as far as I know:
0: Do nothing
1: Attempt to balance somehow
2: Rearrange the cables to get this cell somewhere else in the series
3: Bypass it (should be easy since it's at the end of its module)

 

[jime]

I think you will be ok for now IF it doesn't drop below 2.0V when you use heavy assist. I thought you had Peter's cell level measuring conversion. If so, I would just watch that on a high current assist and see what happens. You may just now be noticing something you have had all along.

 

[*sean*]

A zoom call would be the perfect place to talk about these things.

 

[retepsnikrep]

I tried sending a valid packet sequence to a single CR-Z board on the bench and deviating the voltage on one or more channels of my voltage fooler, but it did not seem to enable any balancing, so that may have to come from the MCM via the 501 packet..

I now have a full OEM Lithium CR-Z Car and will be pulling my spare lithium pack to induce some cell deviation in a working bench setup to see how the MCM responds.

I also captured the 501 data from my working car on a long journey and it did not change from the bench test data..
The cells are all well in balance so I think for now I have to assume no balancing is being commanded by the MCM.

 

[retepsnikrep]

Trying to force BMS balancing activity in a Lithium pack on the bench.

I've charged two cells in the block shown to nearly ~3.9V when the pack is at 3.75V.
No change in id 501 CAN data at present or any sign of balancing activity.

I'm going to leave it powered up for 24hrs and see if the voltage of those boosted cells comes down at all.
If they do then it must be balancing, perhaps to the block average or a pack voltage in that constant 501 data..

Interestingly the series balance resistors on a CR-Z BMS board are not the same for each channel! (Weird)

The first and last channels 1 & 10 are 43R x 2 = 86R, the middle 8 channels are 40R x 2 = 80R.

Anyway if we ignore the FET losses and imagine balancing at 4V / 80R = 50ma x 4V = 0.2W.

So a ~1ah overcharge will take ~20hrs to burn off.

 

[retepsnikrep]

OK I'm not getting any balancing behaviour at present so what might be preventing it from happening.
What does this bench setup not have that a car setup does.

Let's forget all the ECM engine stuff as that's unlikely to affect the IMA.
So what about the IMA, what am I missing? Have you spotted it?

The only thing external to the IPU/IMA assembly apart from the IMA motor itself is ........... The IMA cooling fan....

It does make some sense that if the MCM cannot detect the fan operating when commanded it might shut down any potential heat generating functions like balancing.
So I'm going to attach an IMA fan to my bench setup and see what that does. Is it being commanded to run for instance?

Edit.. No change with IMA fan added..

Maybe it just won't go into balance mode with IMA errors present.
(Errors are present due to no IMA motor position sensor fitted)

That's a pain as it means I would have to try and influence the cells in my fully working lithium car.
I'm not ripping all the trim out of that at the moment. I hate doing that...
So for now we might be at an impasse and I will have to manually drain my two cells back down to match the others.

Edit 2

I manually discharged the two high cells to bring the pack back into balance and observed some interesting behaviour with two HDS parameters.

1) Useable capacity %.

This slowly increased from about 43% to 52% as the pack came back into balance and the deviation/difference between the highest and lowest cell decreased.
So it appears cell voltage deviation is one of the parameters used to calculate the useable capacity. The closer all the cells are together the higher the useable capacity.

2) SOC %. This decreased from ~73% to 60% as I brought the two high cells back down to match the other 38 cells.

So the cell with the highest voltage is used as one of the parameters to calculate the current pack SOC.
This would help prevent overcharging as the high cell would reach full first and the car will stop regen when the SOC reaches 79% anyway..

The total pack resting voltage is likely also used as one of the variables in the SOC mix.

 

[eprom]

Is it possible that the discharge automatically activates passing a certain voltage like a zener diode would ?
As far as I know lithium batteries can support up to 4.2V. Maybe you stopped cranking up the tension just before it start?
And If the discharge starts automatically, it might have an ID that would serve to indicate which cell is in balancing.

 

[Natalya]

A couple thoughts...

1: Maybe its worthwhile to see if it does anything to balance a low cell?
2: Maybe the CR-Z only balances cells during assist and regen?

 

[retepsnikrep]

A couple thoughts...

1: Maybe its worthwhile to see if it does anything to balance a low cell?
2: Maybe the CR-Z only balances cells during assist and regen?

Thanks for ideas, all are welcome.

1) To balance the pack if it detected one low cell it would have to burn off power in all the other 39 higher cells?
That just does not make much sense to me. (I will probably try it though.)

2) I have logged quite a bit of ID501 data when driving, there is no evidence of any changes at the moment during assist/regen. But of course the pack in the car is in balance anyway.

 

[retepsnikrep]

OK here is some exciting 501 data I captured yesterday. (Only just looked at it..)

Starting at row 526 until row 688 there is a change in the alternating data.

This coincided with an empty pack event and assist cut followed by forced regen.
I was climbing a steep hill in high gear with the throttle mashed.

So in effect a Lithium pack negative recal.

This is the first change in the 501 data I have seen...
Maybe it does balance at the bottom???

But what does the data tell us???

I'll try discharging a cell in my bench pack and see happens..

I wonder if it is the same for LTO?

 

[retepsnikrep]

I tried discharging a single cell from the average of 3.75V to 3V and nothing happened to the 501 data.

No evidence of balancing. Maybe it has to go lower than 3V momentarily under load to trigger.

Maybe It just won't trigger on the bench, and has to be in a working car with no errors and assist/regen etc.

The SOC fell slowly from about 61% to 20% at about 3.2V and stayed put.
The useable capacity ticked down from 52% to 10%..

I'll need to study that capture from above..

I charged it back up again to 3.75V and the SOC went back up to 61% and the useable went back up to 52%.

So the SOC is clearly very much voltage related.

 

[Mario]

Having recently worked on a very rudimentary balancing algorithm for a li-ion pack, I now know a little more about balancing li-ion cells than I used to.
Check this TI application note (SLVA115): https://www.ti.com/lit/an/slva155/slva155.pdf?ts=1614653397973

In short, balancing should be done at high SOC, during charging, and not at low SOC. At lower SOCs, the ESR (resistance) of the pack is higher. A charge or discharge current causes the cell voltage to deviate relative to its true open-circuit voltage. Differences in individual cells can cause cells of identical SOC to show a different voltage under charge or discharge. Trying to balance using this information can actually unbalance cells.

Based on this, I would assume cell balancing on the CR-Z only takes place at high SOCs and during charging. The TI appnote describes a simple algorithm that doesn't even do any balancing until a cell is above 3.9V, and only with a >40mV difference between cells (trying to balance cells to be within less than ~25mV can also cause imbalance).

High SOC is not always attainable in a scenario such as a hybrid battery pack, and charge/discharge current isn't deterministic. Maybe the CR-Z will apply a background charge to top up the pack specifically for purposes of balancing. Maybe they do balancing while the car is idle, I'm not sure. The algorithm in the appnote is very basic and surely not the only correct way to balance li-ion cells.

I think ultimately, the best way to try to capture balancing behavior (and potentially associated CAN messages) is to unbalance a cell in an actual car and drive it until something happens. You should be able to easily determine if a cell is balancing by measuring the voltage across its balancing resistor. It should be near 0V with no balancing and near the cell voltage when it is occurring. I'd recommend capturing the voltage across a balance resistor plus the CAN messages with your Saleae, HOWEVER, make sure to use the ground-referenced cell! And don't hook the ground lead up to the balance resistor, just one of the input leads. I can draw a diagram if this isn't very clear. I'm also not sure if you have the Saleae with analog input capability. You could maybe make a simple circuit to turn it into a logic-level high/low signal.

Also, there may not be an associated CAN message as the pack might handle its own balancing. This would likely require a CAN message to tell the pack the current in/out of it, though. I find it more likely that a central BMS talks to all the packs and decides when each individual cell should have its balancing circuit turned on or not.

 

[retepsnikrep]

All good info thanks.. I'm not prepared to rip my great condition working lithium car apart at the moment.
It cost me £6000 and you have to remove loads of easily scratchable interior trim to get the battery pack out.
It's not a 5 minute job like the G1 Insight, so it is just not happening for now..

So......

The 4 x 10 cell CR-Z BMS boards use the LTC 6803G-2 12 cell BMS chips.

Bizarrely each 10 cell board has two LTC chips, and it looks like one maybe dedicated to the voltage readings and the other to balancing..

https://www.analog.com/media/en/technical-documentation/data-sheets/680324fa.pdf

I've been perusing the data sheet trying to compare the CAN id 501 data in my post 96 above with some of the LTC balance commands.

I'm assuming for this comparison that the unknown processor on the BMS boards is just the SPI to CAN interface and hopefully it's not altering the LTC data. Or if it is perhaps it's in an obvious way i haven't spotted.

Perhaps someone more familiar with the LTC chip command structure might comment on the 501 data stream.

 

[Natalya]

Hey @retepsnikrep any updates on this?