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

My Nissan Leaf Forum

Help Support My Nissan Leaf Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
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.
So today, I decided to charge-up the car to 100% to see what I can do now. I haven't done that for quite some time and the charging experience was different from what I recall. First, all balancing stopped around 80% SOC (all cellpairs blue) and from about 85% the charging throttled back from about 6kW to 1kW then eventually to about 400W, so that took a while.

Long story short I, got about 81 miles before first battery warning and decided to call it a day as it was after 3+ hours of driving and I needed to go, plus it was bedtime! I arrived home with 84.2 miles. There was a lot of driving up hills and about 10% freeway at 60+ mph.

Those numbers are way better than anything previous to this experiment but it seems there is still some improvement possible. I glimpsed some quite high deltas when pushing hard on the freeway (250+mV), so I think any further reduction on the delta will be helpful.

Yes the car is probably now good enough for my needs, but I am so curious to see how far I can take this. There's no great effort involved and I calculate a cost of about $35 in the electricity necessary for a 90 day test.
 
Last edited:
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?
You are correct. I'm going to soon post my observations from a very recent charging session. I think it will be interesting reading
 
So today, I decided to charge-up the car to 100% to see what I can do now. I haven't done that for quite some time and the charging experience was different from what I recall. First, all balancing stopped around 80% SOC (all cellpairs blue) and from about 85% the charging throttled back from about 6kW to 1kW then eventually to about 400W, so that took a while.

Long story short I, got about 81 miles before first battery warning and decided to call it a day as it was after 3+ hours of driving and I needed to go, plus it was bedtime! I arrived home with 84.2 miles. There was a lot of driving up hills and about 10% freeway at 60+ mph.

Those numbers are way better than anything previous to this experiment but it seems there is still some improvement possible. I glimpsed some quite high deltas when pushing hard on the freeway (250+mV), so I think any further reduction on the delta will be helpful.

Yes the car is probably now good enough for my needs, but I am so curious to see how far I can take this. There's no great effort involved and I calculate a cost of about $35 in the electricity necessary for a 90 day test.
Seems like you are close to the finish line on this project if not past it , charging to 100 has answered the question i had about this situation , i think you probably have a puffy cell bank or two , you could drop out the pack and identify the few that need changing , buy second hand good condition packs and swap them and transform the car most likely. the car will increase in value by much more than it will cost if you, find some fantastic aftermarket techs to assist you.
 
Last edited:
OBSERVATIONS FROM A RECENT CHARGE CYCLE

Since starting this experiment, I've been noting the delta-Vs after charging is complete, though I have not reported those numbers on this thread to avoid confusion. What happens during charging is that both the low cellpair and high cellpair voltages immediately jump around 20+ millivolts, in the case of the high cellpair, going from the teens to around 50 millivolts. The low cellpair#29 delta-V, therefore appears artificially lowered during the first few minutes of charging. Not only that but the low cellpair delta-V continues to reduce slowly during the rest of the charge (To me, this may be an indication that the cellpair#29 is lacking capacity to some degree). As soon as I disconnect the charge connector, this artificially low delta-V immediately starts to increase and gradually moves back to the 'real', larger delta-Vs that I report in the thread. However, this 'after-charge' delta-V has also become a useful marker of the improvement of the low cellpair#29. Recently, this after-charge delta-V has become so low that the low cellpair is almost level with some of the lower cellpairs in the pack. Presently, my charging routine is to charge to the mid/high 40s% spy SOC but this past week I did a test going higher to see the effect on the low cellpair #29.The result was that low cellpair voltage CAUGHT UP with the next lowest cellpair voltage at spy SOC 53% (first time I've seen this happen). By 55%, it was higher than 4 other cellpairs (the low cellpair delta-V was now showing as 33mV...but no longer caused by the 'low' cellpair #29). By 58%, it was higher than 10 cellpairs. By 60%, it was higher than 12 cellpairs and finally by 70% it was higher than 16 other cellpairs.

My thoughts? First, as I've said, what happens by the end of the charge is an artificial situation and cellpair #29 voltage immediately starts falling faster than the other cellpairs, BUT SO FAR HAS ALWAYS ENDED-UP WITH A SMALLER DELTA-V AT THE END OF THE DISCHARGE (...the 2mV per day...). Looking ahead, given that the minimum low cellpair delta-V appears to be about 33mV (see above paragraph), the most improvement I can make from here is about another 50mV (that is, my recently reported low of 84mV minus 33mV). I can't see the bms balancing doing any better than that, and indeed it may taper-off before I even get to that point....we'll see.
 
OBSERVATIONS FROM A RECENT CHARGE CYCLE

Since starting this experiment, I've been noting the delta-Vs after charging is complete, though I have not reported those numbers on this thread to avoid confusion. What happens during charging is that both the low cellpair and high cellpair voltages immediately jump around 20+ millivolts, in the case of the high cellpair, going from the teens to around 50 millivolts. The low cellpair#29 delta-V, therefore appears artificially lowered during the first few minutes of charging. Not only that but the low cellpair delta-V continues to reduce slowly during the rest of the charge (To me, this may be an indication that the cellpair#29 is lacking capacity to some degree). As soon as I disconnect the charge connector, this artificially low delta-V immediately starts to increase and gradually moves back to the 'real', larger delta-Vs that I report in the thread. However, this 'after-charge' delta-V has also become a useful marker of the improvement of the low cellpair#29. Recently, this after-charge delta-V has become so low that the low cellpair is almost level with some of the lower cellpairs in the pack. Presently, my charging routine is to charge to the mid/high 40s% spy SOC but this past week I did a test going higher to see the effect on the low cellpair #29.The result was that low cellpair voltage CAUGHT UP with the next lowest cellpair voltage at spy SOC 53% (first time I've seen this happen). By 55%, it was higher than 4 other cellpairs (the low cellpair delta-V was now showing as 33mV...but no longer caused by the 'low' cellpair #29). By 58%, it was higher than 10 cellpairs. By 60%, it was higher than 12 cellpairs and finally by 70% it was higher than 16 other cellpairs.

My thoughts? First, as I've said, what happens by the end of the charge is an artificial situation and cellpair #29 voltage immediately starts falling faster than the other cellpairs, BUT SO FAR HAS ALWAYS ENDED-UP WITH A SMALLER DELTA-V AT THE END OF THE DISCHARGE (...the 2mV per day...). Looking ahead, given that the minimum low cellpair delta-V appears to be about 33mV (see above paragraph), the most improvement I can make from here is about another 50mV (that is, my recently reported low of 84mV minus 33mV). I can't see the bms balancing doing any better than that, and indeed it may taper-off before I even get to that point....we'll see.
with interest to be constructive and happy that you are reporting all your experiments, probably it is my limitation, but to me is really hard to follow you though without having graphs or leafspy screenshot that visually show your though. IMHO there is also a lack of test under load. Just installing a video recording app in your phone could allow record leafspy variations during charge or discharge (important to show A drained or in input to the battery).
I hope all these tests can be paired with leafspy screenshot or videos.
regards
 
with interest to be constructive and happy that you are reporting all your experiments, probably it is my limitation, but to me is really hard to follow you though without having graphs or leafspy screenshot that visually show your though. IMHO there is also a lack of test under load. Just installing a video recording app in your phone could allow record leafspy variations during charge or discharge (important to show A drained or in input to the battery).
I hope all these tests can be paired with leafspy screenshot or videos.
regards
Ginetto. I meant to reply to your post#76 on the recorders. I tried several of them but they were hopeless because of advertising bloat. I now have one which might work ok. It puts an overlay on the leafspy, though the controls are really tiny - not great while driving. I plan to try it out soon on my 'test hill', once I'm a bit further down the learning curve with it. Though it's really just going to confirm what I said in my previous post#84....high (250+mV) delta when under heavy load, then it falls back as the load lightens - that's a classic symptom of highish internal resistance. My battery total internal resistance is about 2.5 times that of a new, perfect battery (Hx 38 and change). Now, for a gracefully aging battery that's not a disaster - but my battery is not graceful! I suspect cellpair #29 has got more than it's fair share of that increased resistance.

BTW you also mentioned in your post about running at 150kW. I think that's the 62kWh battery? My battery is spec'd a little over 100kW, I believe.

I take your point about more pictures - worth more than 1000 words etc. I'll put up some 'before and after' screenshots. I took some screenshots just after I got the car and will try to locate them on the phone. Also, like I said I will try out my screen recorder soon. Thanks for all your input.
 
Here's a video of the 'before' times, just after I got the car. It shows the problem I was up against. Video starts at the bottom of a 2 mile hill climb. (I tried to get in position with 80% but it had already dropped to high 70s%. So from 100% SOC, I already had 32.7 miles on the trip gauge). Video finishes with 34.5 miles and low battery warning.

Apologies in advance for the quality - it was my kid's phone.
 

Attachments

  • 1000010410.mp4
    38.4 MB
Last edited:
Here's a video of the 'before' times, just after I got the car. It shows the problem I was up against. Video starts at the bottom of a 2 mile hill climb. (I tried to get in position with 80% but it had already dropped to high 70s%. So from 100% SOC, I already had 32 miles on the trip gauge). Video finishes with 34.5 miles and low battery warning.

Apologies in advance for the quality - it was my kid's phone.
I mean video of leafspy not fot eh car dashboard :)
 
BEFORE AND AFTER LEAFSPY SCREENSHOTS

The first diagram shows a screenshot of the leafspy cellpair bargraph shortly after I got the car. I have several others but this one had a total delta-V 20 or 30mV less than the others, so about the best result from that time. The 'odo' reading is actually miles not km. Another oddity is SOH should have shown 83%, not 0%

The second diagram is from today, 15jan. Trying to have an 'apples to apples' comparison, I used a 300mV scale. It corresponds to a car SOC of 19%, just before I started a charge cycle.

So there is a significant improvement in the low cellpair#29 delta-V. It appears that maybe another 50 or 60mV reduction might be possible.

...fingers crossed I selected the correct files...
 

Attachments

  • Screenshot_20240624_114650.png
    Screenshot_20240624_114650.png
    147 KB
  • Screenshot_20250115_111319.png
    Screenshot_20250115_111319.png
    147.8 KB
BEFORE AND AFTER LEAFSPY SCREENSHOTS

The first diagram shows a screenshot of the leafspy cellpair bargraph shortly after I got the car. I have several others but this one had a total delta-V 20 or 30mV less than the others, so about the best result from that time. The 'odo' reading is actually miles not km. Another oddity is SOH should have shown 83%, not 0%

The second diagram is from today, 15jan. Trying to have an 'apples to apples' comparison, I used a 300mV scale. It corresponds to a car SOC of 19%, just before I started a charge cycle.

So there is a significant improvement in the low cellpair#29 delta-V. It appears that maybe another 50 or 60mV reduction might be possible.

...fingers crossed I selected the correct files...
In my experience it is not so significative a single screenshot. voltage drop is much more important than a single weak cell. again a video (of the leafspy scree in Horizontal mode) during heavy load is much more descriptive. And if you activate leafspy log much much better (btw finding logs in the phone is hard for normal users)
 
i.e. Use Landscape mode with the phone rather than portrait. Then If you monitor/graph total pack voltage and current while someone else drives and gives a few wide open throttle bursts, it will show a great story.
 
ginetto and nlspace. I think the heaviest load I could apply, would be the hill climb I did in the video in post#87. That trip took me from 80% car SOC to first battery warning in 2 miles. Besides the screen recording, it would make sense to video the car SOC display to compare with the 'before' video - leafspy is useful but at the end of the day it's what the car/bms decides (for example, my recent 84 mile trip in post#81 - leafspy said close to 40% SOC remaining but the car said low battery, because of cellpair #29).

I definitely will need another person to help me. My kids are now fulltime in work/college but I'll figure something out soon. Meantime, I've never used the leafspy logging function. Is that something I start and stop for the trip? Is it in Settings or explained in the HELP somewhere?

Another final question, besides showing the gyrations of cellpair#29 what else might this test usefully tell me?
 
Another final question, besides showing the gyrations of cellpair#29 what else might this test usefully tell me?
usually a weak cell could hide more less weak cells (here a log could really help). The histogram (because scaled due to weak cell) could not show other voltage drop in other cells but could help to have an idea.
Leafspy logs are activated in leafspy settings and are saved in the mobile intrnal memory: Android/data/com.Turbo3.Leaf_Spy_Pro/files/LOG_FILES/ (you need an app to navigate folder)
probably your sons could help you.
 
FUTURE EXPECTATIONS

It's been occasionally discussed on the thread about what might happen to any improved delta-V after this experiment. How long might it take to lose the improvement? What might be required to maintain it? I think I now have a little insight into this....

Recently, the low cellpair.delta-V moved in the wrong direction. It was coincident with the long trip I made in post #81. So the day previous to the trip, delta-V was at 84mV at end of discharge (18% car SOC) and the day after it was at 86mV. I did 84.2 miles that day which equates to about 31k miles annually. Now the 6 year historical average for our second car is around 4.5k miles and I've been roughly averaging that mileage during this experiment, so there appears to be no problem with this low mileage. However, by simple extension of this result, I could expect to lose all the improvement in the delta-V in about 3 months, if I were averaging a daily mileage of about 84 miles. OTOH if I can use proportion on this result for my own expected mileage (12 miles/day average) that would be a loss of about 0.28mV per day which I expect I could overcome, simply by slow overnight L1 charging, so maximizing the time the car is being balanced on a daily basis.

BTW I've arranged for some help with the hillcllimb test on 23jan, so expect some posts soon after. At the very least, it will make for an interesting comparison with the 'before' video of post#87.
 
Last edited:
HILL-CLIMB PART 2 - THE 'AFTER' VIDEO

So I repeated the hill-climb of post#87. I tried to maintain an 'apples to apples' comparison but there are uncontrollable variables such as other traffic and stop lights (there are 4 stop lights during the trip). I tried to start the test at 79% car SOC but it clicked up to 80% just as I rolled down to the start, though this immediately went back to 79% in the first few yards of the trip. Also, in the 'before' video I was stopped by only one light but in this test, I stopped at 2 lights, so had to regain momentum twice and so use more energy than the first hill-climb. SPOILER: the next paragraph contains my analysis of the trip, so maybe watch the video first!

First, I have edited the video as about half the time, the car was stopped, waiting for 2 light changes. In the 'before' video, the car range estimate had dropped in half in the first few seconds and the low battery warning turned on, a few seconds before the top of the hill. In the 'after' video the range starts at 117 miles and drops to 111 miles by the end. The car SOC went from 79% at the start and fell to 75% at the end. This behavior is very consistent with a normal battery whereas the car was essentially unusable by the end of the 'before' video. To me, that is a big improvement which is due only to the 200+mV reduction of the delta-V of cellpair#29. That reduction helps for 2 reasons.The first reason is the cellpair holds more energy when it's 200mV higher. The second reason is the 'voltage dives' under heavy load take longer to reach down to low battery warning.

BTW, the results of the leafspy screen recording were fascinating. I'm preparing a post with the recording and my thoughts on what I see there.
 

Attachments

  • VID-20250125-WA0005.mp4
    31.6 MB
DETAILS ON THE HILLCLIMB

The post I have made for the leafspy recording is rather long, so here is a little more information about the hillclimb route which takes me from the lowest to highest point in my town in the shortest distance. I made 2 stops because of stop lights, so that divides the hillclimb into 3 sections. The first section is about a half- mile and the steepest part. The second section is the least steep and a little over a half-mile. The third section is about 0.8 mile and the second steepest section. I removed most of the stopped periods from the hillclimb video at post#95 as it was essentially dead-time and so reduced it to 3min 17sec, but kept these periods in the leafspy recording as it contains some variability that may be of interest. So the screen recording will be 5min 8sec.

I'll post the leafspy recording later today.
 
DETAILS ON THE HILLCLIMB

The post I have made for the leafspy recording is rather long, so here is a little more information about the hillclimb route which takes me from the lowest to highest point in my town in the shortest distance. I made 2 stops because of stop lights, so that divides the hillclimb into 3 sections. The first section is about a half- mile and the steepest part. The second section is the least steep and a little over a half-mile. The third section is about 0.8 mile and the second steepest section. I removed most of the stopped periods from the hillclimb video at post#95 as it was essentially dead-time and so reduced it to 3min 17sec, but kept these periods in the leafspy recording as it contains some variability that may be of interest. So the screen recording will be 5min 8sec.

I'll post the leafspy recording later today.
where is the leafspy video? or the leafspy log?
 
LEAFSPY RECORDING OF HILLCLIMB 2

First section (0 to 1min 40sec)
I have to admit to some surprise on what I saw on the recording, especially at 22 seconds (the highest load, at 177amps). No surprise with #29 doing its dive, but #53 doing an even bigger dive to just 10mV above #29 and a couple of smaller but notable dives by #78 and #96. So using the cell voltages at 0 seconds as a no-load reference, the dives were #29 did 324mV, #53 did 400mV and the cellpack average dived 190mV. A little later, say at 50 seconds, when the car has stopped, #59 has returned to a middling position within the 94 'good' cellpairs.

Several thoughts occur to me about this behavior. At first glance, it looks as though there is not much point in trying to improve the delta-V of #29 by any more than 10mV from here because #53 would then become the defining cellpair for a low battery warning. Next, I'm puzzling about #59's behavior. It appears to have adequate capacity given its relative voltage position among the good cellpairs but is suffering from excessive internal resistance, as shown by the large voltage dive. IMHO I think it's quite unusual for a lithium cell to have good capacity, yet high internal resistance, which leads me to believe there might be an issue with the connection of #59 to the busbar, such as an under-torqued cellpair terminal bolt or even an over-torqued bolt which might damage the busbar conductor - after all, it only needs an increase of 1 MILLIOHM over the desired resistance to cause a 177mV volt-drop at 177 amps. WHATEVER THE REASON, MY EXPERIMENT CAN DO NOTHING TO IMPROVE THIS BEHAVIOR, AS THE BMS SEES NO DELTA-V ON #59 TO BALANCE.

Second section (1min 40 sec to 4min)
The second section is a shallower version of the first section due to it being less steep and I was traffic limited to about 45 mph. This time #59 got no closer than about 40mV above #29 at a lower load of 154A (1min 48sec), though #29 continued to show a smaller volt-drop (no-load - load volts) than #53 of 310mV and 350mV, respectively.

Third section (4min to 5min 8sec)
During about the last quarter mile of the third section, I took the opportunity to speed-up due to the absence of traffic. The speed limit was 45 mph and all I'll say is I went well beyond my comfort zone (the speed pointer disappears behind the display hood during the last 30 or so seconds of the video!)...still no range 'dance' nor low battery warning. I think the low point for #29 is 3.273V and its delta- V peaked at 320mV (at 4min 34sec). This was the only time that the volt-drop of #29 exceeded that of #53 (590mV versus 480mV)

Conclusions: Despite the surprise behavior of cellpair #53, I think #29 remains the main limiter of further improvement because #53 only got close to #29 at the beginning of the climb. In the near future, there may come a point where #53 will be the lowest cellpair, with further reduction in the delta of #29 but I think this will only happen at high SOC and high load, so it won't have much, if any, effect on the car's range or SOC.

I continue with reducing the delta-V of #29 and in the meantime, may also do a leafspy recording at a lower SOC (say 50% SOC), just to confirm that #29 is still the main block to further improvement.
 

Attachments

  • VID-20250128noaud-WA0002.mp4
    17.9 MB
Last edited:
LEAFSPY RECORDING OF HILLCLIMB 2

First section (0 to 1min 40sec)
I have to admit to some surprise on what I saw on the recording, especially at 22 seconds (the highest load, at 177amps). No surprise with #29 doing its dive, but #53 doing an even bigger dive to just 10mV above #29 and a couple of smaller but notable dives by #78 and #96. So using the cell voltages at 0 seconds as a no-load reference, the dives were #29 did 324mV, #53 did 400mV and the cellpack average dived 190mV. A little later, say at 50 seconds, when the car has stopped, #59 has returned to a middling position within the 94 'good' cellpairs.

Several thoughts occur to me about this behavior. At first glance, it looks as though there is not much point in trying to improve the delta-V of #29 by any more than 10mV from here because #53 would then become the defining cellpair for a low battery warning. Next, I'm puzzling about #59's behavior. It appears to have adequate capacity given its relative voltage position among the good cellpairs but is suffering from excessive internal resistance, as shown by the large voltage dive. IMHO I think it's quite unusual for a lithium cell to have good capacity, yet high internal resistance, which leads me to believe there might be an issue with the connection of #59 to the busbar, such as an under-torqued cellpair terminal bolt or even an over-torqued bolt which might damage the busbar conductor - after all, it only needs an increase of 1 MILLIOHM over the desired resistance to cause a 177mV volt-drop at 177 amps. WHATEVER THE REASON, MY EXPERIMENT CAN DO NOTHING TO IMPROVE THIS BEHAVIOR, AS THE BMS SEES NO DELTA-V ON #59 TO BALANCE.

Second section (1min 40 sec to 4min)
The second section is a shallower version of the first section due to it being less steep and I was traffic limited to about 45 mph. This time #59 got no closer than about 40mV above #29 at a lower load of 154A (1min 48sec), though #29 continued to show a smaller volt-drop (no-load - load volts) than #53 of 310mV and 350mV, respectively.

Third section (4min to 5min 8sec)
During about the last quarter mile of the third section, I took the opportunity to speed-up due to the absence of traffic. The speed limit was 45 mph and all I'll say is I went well beyond my comfort zone (the speed pointer disappears behind the display hood during the last 30 or so seconds of the video!)...still no range 'dance' nor low battery warning. I think the low point for #29 is 3.273V and its delta- V peaked at 320mV (at 3min 48sec). This was the only time that the volt-drop of #29 exceeded that of #53 (590mV versus 480mV)

Conclusions: Despite the surprise behavior of cellpair #53, I think #29 remains the main limiter of further improvement because #53 only got close to #29 at the beginning of the climb. In the near future, there may come a point where #53 will be the lowest cellpair, with further reduction in the delta of #29 but I think this will only happen at high SOC and high load, so it won't have much, if any, effect on the car's range or SOC.

I continue with reducing the delta-V of #29 and in the meantime, may also do a leafspy recording at a lower SOC (say 50% SOC), just to confirm that #29 is still the main block to further improvement.
need to change the modules where this 4 weak cells are present. if you read the leafspy log and plot the delta of each cell respect the previous one (or next one) (I mean delta respect cell it'self and compared respect the delta of the neighbour cells) you have to find all cells that delta is higher or near 150mv during 150kW drain. that you have at least to change. this is my super simplified rule of thumb
 
Last edited:
Back
Top