Fixing a low voltage cellpair in the HV battery without battery removal.

[nlspace]

Internal balancing diagram from datasheet:

 

[Frankleaf14]

You have to be careful about making assumption on L1/L2 charge counts.
The counter clicks up on click when the cord is plugged in and another when charging starts, so if charged immediately, it records one charge, if charged on the timer later after being plugged in it records another, so two recorded for one charge event, if someone unplugged the cord for any reason then plugged back in there would be 3 recorded for that one charge event.
The timer record is only a rough guide to charging events. recording only the charge contactor close events.

True, though I've seen a printout from a more sophisticated tool that shows a lot more detail about charging. I pulled it a while ago from another discussion about whether dealers etc can check if owners have been abusing the HV battery. I'll TRY to attach it to this post. It shows way more detail than leafspy, such as number of charges started at different starting SOC and peak temps reached during the charge. It also shows a number for AC (L1/L2) charges, which makes me think perhaps it can differentiate between contactor clicks and charges? Sounds like all those gory details are still held somewhere in the car's brain.

 

[Frankleaf14]

Looks like it is using the EMERGENCY Discharge Mode, defined in the MAX 17823 datasheet,

"Emergency Discharge Mode
The emergency discharge mode performs cell-balancing in a controlled manner so that the cells can be discharged to a safe level in the event of an emergency. The BALSWDCHG and DEVCFG2 registers provide control for this mode. A timeout value for the mode is configured by DISCHGTIME[7:0] as shown in Table 13.
The emergency discharge mode is activated by setting the EMGCYDCHG bit with DCHGTIME[7:0] ≠ 00h. In emergency discharge mode the following occurs:

1. 1) The CBTIMER[3:0] is cleared to prevent the cell-balancing watchdog from disabling the cell-balancing.
2. 2) Cell-balancing switches are controlled by BALSWDCHG, not BALSWEN.
3. 3) The discharge timer starts to countdown.
4. 4) The read-only counter DCHGCNTR[3:0] increments at a 2Hz rate with periodic roll-over at Fh. The host can read this counter periodically to confirm that the mode is active.
5. 5) The GPIO3 pin is driven high while the countdown is active.

The emergency discharge mode alternates between a 1-minute discharge cycle for odd cells and a 1-minute discharge cycle for even cells. There is a 62.5ms minimum off time at the end of each discharge cycle to ensure no overlap between even and odd discharge cycles. The duty-cycle of each discharge cycle may be configured by DCHGWIN[2:0] "

This is interesting. I haven't seen this kind of information before so thanks for posting. I've only briefly looked over it but a couple of things are already occurring to me. I wonder what the EV system would consider an emergency? Also, the cycle times, for example, for the odd/even cell flipping are different from my observations (a 34 second cycle)

 

[Frankleaf14]

Internal balancing diagram from datasheet:
View attachment 6187

Good information. It looks to me that when the transistor (FET?) turns ON, then there are TWO shunts (the two Rbalance ?) in series across the cellpair. Does this mean that there are 192 shunt resistors on that bms pcb and not 96, so the shunt resistance would be 200 Ohm and not 100 Ohm per cellpair?

Any thoughts or comments?

 

[cornbinder89]

True, though I've seen a printout from a more sophisticated tool that shows a lot more detail about charging. I pulled it a while ago from another discussion about whether dealers etc can check if owners have been abusing the HV battery. I'll TRY to attach it to this post. It shows way more detail than leafspy, such as number of charges started at different starting SOC and peak temps reached during the charge. It also shows a number for AC (L1/L2) charges, which makes me think perhaps it can differentiate between contactor clicks and charges? Sounds like all those gory details are still held somewhere in the car's brain.

never seen that before, Thanks! Wonder what they were using Nissan Consult?

 

[nlspace]

Looks like 200R in the balancing circuit, so 20mA on a 50% duty cycle (even/odd pairs cycling), probably a programmed event rather than Emergency.

 

[Frankleaf14]

Life was getting in the way yesterday, so the car got down to second battery warning at 10% car SOC with the 'flashing dashes' and low cellpair in the 3.4V region before I got to it. I noticed that the low cellpair delta had increased several mV from where it was where I usually recharge, just above first battery warning (about 19% car SOC). Makes sense, as down at 10% the low cellpair is now on the 'slide' at the righthand side of the lithium charge curve, while the rest of the pack is still on the flat 3.5 to 4V part of the curve. The improvement returned during the recharge. Phew! (Note to self: keep comparing apples to apples - this is a slow, gradual process - try not to fool yourself).

The improvement continues so the experiment continues...

 

[Lucky]

Shunts on 40kWh BMS are 100 ohm so current will be around 40mA. I don't know how would this emergency discharge the battery. When I balance the cells I use 10A (or 10,000mA) for charging and discharging.
You could easily charge a single cell in an assembled pack if you know what you are doing. Cells are insulated from the chasis and the power supply can be directly connected to the terminals. Problem could be in LH and RH stacks because of overlaping bus bars.

But there is one thing bugging the BMS and it can't be just noise. Perfectly balanced battery pack still shows more delta then there really is. I will attach two images. The 1st pack was balanced to 4mV max and the second was almost perfect at around 2mV.
Sorry I didn't zoom in when doing a capture.

 

[Frankleaf14]

never seen that before, Thanks! Wonder what they were using Nissan Consult?

Don't know. Another misty area for me. It's somewhere on the forum, so maybe a search? But you know how that goes ...rabbit holes...

 

[nlspace]

Shunts on 40kWh BMS are 100 ohm so current will be around 40mA.

That would be true if there were external transistors using a single shunt, but now we know that is not the case as shown in post #61.

Perfectly balanced battery pack still shows more delta then there really is.

The RED cells are shunting so the picture is dynamic, it is not a static voltage measurement. Each cell is being measured (it's own voltage) with respect to it's neighbor as the reference.

 

[Turbo21]

An interesting experiment, the only time my battery on 18' 40kwh leaf appears to be properly balanced is when i L1 indefinitely and let the car choose to stop the charger , all the cells turn "blue" on leaf spy then and it is something like an 8mv total difference , this seems to allow the battery longer to top off and gives the balancer a longer time to do its thing than during the course of the main regular charge. I charge to 100 every so often just to give the car the chance to "properly" balance to "Blue". Perhaps charging to 100 will speed up your plan by a few months? I suspect your rear pack will be puffy and blown out of shape and the physical internal volume of the cells will be decreased making it impossible to return to normal but its great you are trying something outside the box and not just throwing in the towel , at the least we can all learn from it. it will probably look much like the pack of Hiroshi from Japan with his 300000km Leaf 40Kw on its original pack , he attributes this to repeated high speed charging and the heat it creates

 

[Lucky]

That would be true if there were external transistors using a single shunt, but now we know that is not the case as shown in post #61.


The RED cells are shunting so the picture is dynamic, it is not a static voltage measurement. Each cell is being measured (it's own voltage) with respect to it's neighbor as the reference.

Yes, you are right. The current goes thru two resistors, so it is actually around 20mA. I missed that. But the PCB has 104 resistors so they are shared somehow. I must check the datasheet but then again, it doesn't make any difference. Thanks for your notice.

 

[Frankleaf14]

An interesting experiment, the only time my battery on 18' 40kwh leaf appears to be properly balanced is when i L1 indefinitely and let the car choose to stop the charger , all the cells turn "blue" on leaf spy then and it is something like an 8mv total difference , this seems to allow the battery longer to top off and gives the balancer a longer time to do its thing than during the course of the main regular charge. I charge to 100 every so often just to give the car the chance to "properly" balance to "Blue". Perhaps charging to 100 will speed up your plan by a few months? I suspect your rear pack will be puffy and blown out of shape and the physical internal volume of the cells will be decreased making it impossible to return to normal but its great you are trying something outside the box and not just throwing in the towel , at the least we can all learn from it. it will probably look much like the pack of Hiroshi from Japan with his 300000km Leaf 40Kw on its original pack , he attributes this to repeated high speed charging and the heat it creates

"Perhaps charging to 100 will speed up your plan by a few months?"

I really want to avoid any significant time at 100%. I put a lot of detail about SOC levels etc in post #12. There's more discussion of this on posts #33 and #38.

"I suspect your rear pack will be puffy and blown out of shape and the physical internal volume of the cells will be decreased making it impossible to return to normal"

I don't think that's my situation. I had a similar discussion with poster Lucky on posts #13, #23 and #24. He had a situation like the guy in your video with well over 1000 fast charges and many inflated rear-stack modules. I only have one low voltage cellpair #29 and the car only had 1 fast charge. Even the other cell pouches in the same module as #29 (#s 30. 31 and 32) are working well in leafspy - another hint to me, that there is likely no physical damage.

 

[Frankleaf14]

UPDATE ON CELLPAIR#29 DELTA-V

It's about 2 weeks since the last update on the delta-V on my car. So, on 14dec my low cellpair delta-V was 136mV and the displayed leafspy delta-V range was 154 -160mV. Now, on 28dec the low cellpair delta-V is 109mV and the leafspy delta-V range is now 128-134mV. All these measurements were taken at 18 or 19% car SOC, depending on when I got to it.

It's good to see that the improvement continues at about 2mV per day, comparing the corresponding values above. The car was OFF for about a day in total, as it had to be out of the garage for various reasons, plus when I use it, I obviously have it OFF and secured when I arrive at a destination and I'm away from the car.

My charge/discharge procedure is evolving as #29 improves. At the beginning of November, I was doing about 2 charges per day to leafspy 61%. That discharge is now running for over 24 hours, meaning a recharge would be needed in the 'wee small hours', so I have steadily reduced the charge point to mid 50s% meaning just one charge per day in the late evening.
....and so it goes....

 

[Frankleaf14]

MILESTONE?

This evening, 2Jan, the low cellpair delta-V dropped to double digits for the first time - 99mV, but double digits all the same! The leafspy-displayed delta-V is about 15mV higher, as it is the difference between the low and high cellpairs.

The reducing delta continues 'at pace' - about 2mV/day so I still expect it to be within 50mV of the cellpack average in my original timescale estimate but something is crossing my mind more and more.....diminishing returns? In other words as the delta-V reduces will the daily improvement reduce?
There is no sign of this happening yet, but it remains a possibility at some point.

Another thing I would like to do in the meantime, is observe leafspy when the battery is under heavy load, eg climbing a hill at speed. I've run leafspy several times during my drives but really can only view it briefly at stop signs or red traffic lights and the car is stationary at those times. I need to take one of my kids with me and they can read out the details to me or video the leafspy display. I'll try to organize that soon.

 

[ginetto]

MILESTONE?

This evening, 2Jan, the low cellpair delta-V dropped to double digits for the first time - 99mV, but double digits all the same! The leafspy-displayed delta-V is about 15mV higher, as it is the difference between the low and high cellpairs.

The reducing delta continues 'at pace' - about 2mV/day so I still expect it to be within 50mV of the cellpack average in my original timescale estimate but something is crossing my mind more and more.....diminishing returns? In other words as the delta-V reduces will the daily improvement reduce?
There is no sign of this happening yet, but it remains a possibility at some point.

Another thing I would like to do in the meantime, is observe leafspy when the battery is under heavy load, eg climbing a hill at speed. I've run leafspy several times during my drives but really can only view it briefly at stop signs or red traffic lights and the car is stationary at those times. I need to take one of my kids with me and they can read out the details to me or video the leafspy display. I'll try to organize that soon.

just install a screen recorder application in the phone and record during driving. I would suggest to put the screen in horizontal mode to show the A drained from the battery. I strongly suggest to activate leafspy logging that is really useful to make statistics during heavy loads. You can drain more turning off Eco mode too.
My rule of thumb is that if during a 150kW drain delta is higher of 150mV (referring to the weakest cell respect the medium of the others) then probably you 'll face a mileage dance during low SOC (SOC level depend on state of weak cell). Newest cells with higher capacity obviously tend to give a higher delta because do not drop voltage as the older and with higher internal resistance, so their delta shouldn't be counted.

my 2c

 

[hornstudio]

I had two fairly new 30kWh LEAF with one absolutely dead cell each when I bought them. Leafspy Pro looked horrible.

I drove them hard over the mountains and down again with one full charge nonstop, like 90 miles at once. Not using the brakes but changing D and B. I let them cool down and recharged to 100% the next day.

Dead cells woke up and could not be found again. I think racing and B mode is like little Chademo charges but allowing the cells to cool down.

 

[Frankleaf14]

NEW UPDATE ON CELLPAIR#29 DELTA-V

So, another 2 weeks have passed, since my last update on the delta-V on the car. On 28dec, the low cellpair delta-V was 109mV and the leafspy delta-V range was 128-134mV. On 11jan, the low cellpair delta-V is 84mV and the leafspy delta-V is 103 - 110mV. All these measurements again were taken at 18 or 19% car SOC, depending on when I got to it.

The car again was OFF for about a day during this period, for the same reasons I mentioned before. Taking that into account, the improvement (decrease) in the delta is continuing at about 2mV per day (25÷13), comparing the corresponding values above.

My charge/discharge procedure now requires only a charge to mid/high 40s% leafspy SOC, to keep me on a convenient 24hour cycle. The only reason for this must be the low cellpair #29 is holding more energy at these lower SOCs than before. I am also seeing that the gap between the car SOC and leafspy SOC is reducing as time passes - another good sign.

I'd like to give more detail on what I'm seeing during the charge cycle but will leave that for another post.
....and so it goes....

 

[cornbinder89]

I've been very interested in your progress. I'm waiting to see when you call it "good enough" and how long it stays in that range. Persistence may pay big dividends.

 

[nlspace]

Does the low cell pair catch up to the rest of the pack during the charging sessions, or is there still a delta? If a delta then is it also getting smaller over time?