I think it's just one bad cell...

[the8bitlegend]

I have yet to go through some of the smaller maintenance tasks with my recent MiEV purchase, like a new 12v aux battery, and I haven't given the car more than a full day on the charger to see if the cell will somehow reach the voltage of its comrades.
I did, however, buy an OBD MX+ and install OBDZero and CanIon on my Android, and was able to get what I hope is useful data, among a bunch of information I don't have a decent grasp on yet. I'm adding a couple of screenshots of the cells I'm working with.

In OBDZero, I can see that cell G of module 1 is sitting at 3.76v, and in CanIon, the same claim is made in regards to cell #7, which I'm guessing is the same one.

I was wondering if anyone on myimiev has been in a situation like this one, and what direction anyone might think I could start moving in terms of research and resourcing.

If I don't manage to somehow improve the range on this vehicle, I'll still be satisfied knowing I spent $1000 on something that can safely get me to work and back through winter this year while I save for something more substantial and reliable.

I appreciate this community's persistence. Thanks for reading!

 

[MickeyS70]

Hello and welcome to the forum

This issue is most likely caused by a CMU (cell management unit) problem i.e it’s not reading cell #7 voltage correctly.

To test this theory, take screen shots at various SoC levels and compare #7’s reported voltages to the others, if the difference remains constant, the LTC chip on CMU1 is probably faulty.

The traditional way to fix this problem is to take out and open up the HV pack and replace CMU1 (or just the chip on it) however there is an alternative:

https://myimiev.com/threads/cmu-can-‘spoofing’.5505/

Mickey

 

[the8bitlegend]

Hello and welcome to the forum

This issue is most likely caused by a CMU (cell management unit) problem i.e it’s not reading cell #7 voltage correctly.

To test this theory, take screen shots at various SoC levels and compare #7’s reported voltages to the others, if the difference remains constant, the LTC chip on CMU1 is probably faulty.

The traditional way to fix this problem is to take out and open up the HV pack and replace CMU1 (or just the chip on it) however there is an alternative:

https://myimiev.com/threads/cmu-can-‘spoofing’.5505/

Mickey

I took a trip for some errands today and got some shots at different SoCs. It looks like the difference between the regular and outlying voltages aren't perfectly linear, but I don't know if there would be any leniency there.
I did some reading in the linked thread, and the thread linked within it, and am wondering you mean to say that a potential fix for this being a CMU fault would involve using an Arduino (or equivalent) to hardware spoof the data. If that's the case, I do happen to have one handy. I do admit I'm still a little brain-scrambled after going through the discourse in those threads, and have a lot more to learn in order to keep up with all of the aspects that were brought up.
I am grateful that you reached out with so much insight, Mickey!

 

[MickeyS70]

I took a trip for some errands today and got some shots at different SoCs. It looks like the difference between the regular and outlying voltages aren't perfectly linear, but I don't know if there would be any leniency there.

The various screenshots don’t show much of a voltage difference in general, therefore maybe try to get the good ones down to around 3.7V and check on the outlier but so far it looks promising for a CMU fault.

…would involve using an Arduino (or equivalent) to hardware spoof the data. If that's the case, I do happen to have one handy.

Yes, in a nutshell the Arduino is intercepting all CAN communication from the CMUs, identifying and correcting the ‘bad’ cell data and then re-transmitting the information to the BMU.

I do admit I'm still a little brain-scrambled after going through the discourse in those threads, and have a lot more to learn in order to keep up with all of the aspects that were brought up.

It sounds a lot more complicated than it is…

 

[the8bitlegend]

so it looks like the difference varied about .06V between higher and lower average voltage.
it had been one full night and day of being left on a 120V plug when I did the drive for these screenshots, and it was showing an initial range estimate of 22 miles, which was the highest I've seen it read since buying the car. Other times I'd taken the car off the charge within a few hours of the charging light turning off, and gotten estimates between 14 and 19 miles.

Would it be worth the ~$150 investment to get rid of the potential faulty readings with the difference the car is claiming? Also, is it ever worth leaving the car plugged in to balance over an extended period of time, say, longer than the 20 ish hour period it got this last time?

I'm thinking it could be fun to leave the car plugged in for a few days or more, and just commuting with my electric moped..

 

[MickeyS70]

so it looks like the difference varied about .06V between higher and lower average voltage.
it had been one full night and day of being left on a 120V plug when I did the drive for these screenshots, and it was showing an initial range estimate of 22 miles, which was the highest I've seen it read since buying the car. Other times I'd taken the car off the charge within a few hours of the charging light turning off, and gotten estimates between 14 and 19 miles.

That’s promising, balancing might improve things significantly

Would it be worth the ~$150 investment to get rid of the potential faulty readings with the difference the car is claiming?

I’d say material cost for a CAN bridge are about half of your estimate, also bear in mind that cold weather will reduce current range significantly

Also, is it ever worth leaving the car plugged in to balance over an extended period of time, say, longer than the 20 ish hour period it got this last time?

Balancing times out after a while, so I suggest to fully charge it, then drive it for a bit (or turn on the heater) and plug it back in until the charging light goes off, then repeat a couple of times and see if the difference gets smaller.
Note: balancing is trying to discharge the 87 ‘high’ cells down to the lowest one, this may take dozens of cycles…

[edit] Your CanIon screen grab (post#1) shows a SoC of 53% when charging completed due to a cell reaching 4.1V. This may be another indication that cell #7’s low voltage occurred due to a faulty readout circuit.

Rather than just top balancing, a series of battery calibration routines may be more beneficial to increase your RR:

Basically you need to deep discharge the battery (below two bars as a guidance) and then AC charge in one go to full and leave it connected to do the balancing. Use OBDZero’s CAP 2 function to guide you through the process. (Note: ODBZero needs to be connected during the complete process to show correct results, but the car’s BMU will adjust capacity in parallel (very slowly though).

 

[martinwinlow]

Doesn’t look like a faulty CMU BMS PCB to me, TBH. You normally get a very distinctive corrupted display (with Canion at least, see pics in other threads involving a classic I-MiEV failed LTC chip scenario). I’d suspect a potentially sick cell and some use of the car charging and discharging will reveal the full type and extent of the issue.

 

[MickeyS70]

Doesn’t look like a faulty CMU BMS PCB to me, TBH. You normally get a very distinctive corrupted display (with Canion at least, see pics in other threads involving a classic I-MiEV failed LTC chip scenario). I’d suspect a potentially sick cell and some use of the car charging and discharging will reveal the full type and extent of the issue.

You’re referring to a ‘classic’ LTC chip failure where no cell data is reported on a particular CMU that causes a voltage spike on CanIon.

I have also seen CMU failures with less dramatic effects (e.g. a constant offset), but you’re right, too early to tell which way this swings…

 

[kiev]

Looks like 200 mV lag on cell #7 at the top, with close to 300mV at lower SOC.

Looks like a weak or worn out cell that needs to be replaced; a CAN bridge would only mask it and too risky to let that cell go too low without warning.

 

[MickeyS70]

Looks like 200 mV lag on cell #7 at the top, with close to 300mV at lower SOC.

Looks like a weak or worn out cell that needs to be replaced; a CAN bridge would only mask it and too risky to let that cell go too low without warning.

It might turn out that way, however just as an example, below turned out to be a CMU fault..


Maybe a quick way to verify is to do a load test, ODBZero keeps a log of cell voltages, so take her out for a spin and put the foot to the floor a few times, then upload the log file and compare cell #7 to the rest, a weak cell will drop even further…

 

[the8bitlegend]

I did the 'start all' function, thinking that'd catch worthwhile data. These are the three files that have date and time stamps that match the drive I just took.

 

[MickeyS70]

I did the 'start all' function, thinking that'd catch worthwhile data. These are the three files that have date and time stamps that match the drive I just took.

These PIDInt_xxx files contain all CAN messages transmitted (hex codes converted to integer) but there should be corresponding Cells_xxxx which contains cell voltages only, if you could post these also…

 

[the8bitlegend]

These PIDInt_xxx files contain all CAN messages transmitted (hex codes converted to integer) but there should be corresponding Cells_xxxx which contains cell voltages only, if you could post these also…

My fault.. I didn't take the time last night to properly identify the differences between the file categories.

 

[MickeyS70]

My fault.. I didn't take the time last night to properly identify the differences between the file categories.

Sound, will have a closer look tomorrow but if you want to have a go at it yourself compare module 1, cell G to the rest..

 

[MickeyS70]

Cell G certainly looks like it's closely tracking the others in CMU01, therefore CMU fault or cell failure?

 

[the8bitlegend]

Mickey, thanks for putting the data into graph format!
I was looking through the rows upon rows last night and had become fixated on the SoC of #7, thinking about how that reading was affecting my range estimate..

I was thinking it would be a good idea for me to:
-swap in a new aux battery
-check out and potentially correct the CMU
-source and swap in a new cell, or module, depending on availability

I keep having to remind myself to break this process up into smaller steps so I can avoid feeling overwhelmed.

I'm not the best with local research/resourcing, but I just started looking into nearby salvage yards, etc.

 

[kiev]

CMU fault or cell failure?

Nice graph.

Looks like a weak cell that got out of balance, and there is just no way to charge it and get it back.

Bleed balancing can only correct small differences by burning down the good cells to the weak cell's level. 20mV is a huge amount to bleed off of a LEV 50 A-hr cell at only 100 mA max, and that is not a constant current.

i had one like that and it progressively got worse. The "Full" charge level dropped from 16 bars down to 12-13, then later to 9-10, then 6-8 bars. i put in a claim and got the pack replaced under warranty--barely had enough range to get it to the shop.

If you were a diy sort and had more time than money, it might be possible to open the pack and charge just that one cell to get it back in line. But a lot of work for a probable temporary fix.

What causes just one cell to get out of balance like that--who knows?

 

[MickeyS70]

I was thinking it would be a good idea for me to:
-swap in a new aux battery

Always a good idea, just make sure you charge it overnight before replacing it

-check out and potentially correct the CMU
-source and swap in a new cell, or module, depending on availability

As @kiev mentioned, you need to remove and open the battery pack to correctly diagnose the problem. In order to be able to fix the issue on the spot, you need a ‘new’ CM01 and a LEV50 cell (with the same or larger capacity but the same voltage level as the rest of the pack)

I keep having to remind myself to break this process up into smaller steps so I can avoid feeling overwhelmed.

1. Check out what a battery removal entails, several excellent videos from @alviseven in thread below:
https://myimiev.com/threads/battery-pack-change-plans.5525/page-3

2. Determine your current pack’s capacity (use Canion or ODBZero)

3. Try and find suitable replacements

I'm not the best with local research/resourcing, but I just started looking into nearby salvage yards, etc.

Something to think about:
If you need (want?) to improve your range while you get organised, you could use a CAN bridge but only correct half the difference (100mV) to be on the safe side…

 

[martinwinlow]

It might turn out that way, however just as an example, below turned out to be a CMU fault..
View attachment 1234

Maybe a quick way to verify is to do a load test, ODBZero keeps a log of cell voltages, so take her out for a spin and put the foot to the floor a few times, then upload the log file and compare cell #7 to the rest, a weak cell will drop even further…

This plot really *does* look like a CMU (ie BMS) fault. What is the likelihood that 2 adjacent cells will have voltages one way over the norm Vs one way under the norm where the +ve and -ve difference is nigh-on identical?!

 

[martinwinlow]

Nice graph.

Looks like a weak cell that got out of balance, and there is just no way to charge it and get it back.

Bleed balancing can only correct small differences by burning down the good cells to the weak cell's level. 20mV is a huge amount to bleed off of a LEV 50 A-hr cell at only 100 mA max, and that is not a constant current.

i had one like that and it progressively got worse. The "Full" charge level dropped from 16 bars down to 12-13, then later to 9-10, then 6-8 bars. i put in a claim and got the pack replaced under warranty--barely had enough range to get it to the shop.

If you were a diy sort and had more time than money, it might be possible to open the pack and charge just that one cell to get it back in line. But a lot of work for a probable temporary fix.

What causes just one cell to get out of balance like that--who knows?

If you are going to go the trouble of opening the HV box (defo do-able, DIY, but *much easier if you have a proper vehicle lift), for the cost of a replacement cell, I'd just replace it. Just test the HV battery first without putting it fully back in the car to check the fix has worked. You can leave everything open and just connect the battery cables and the 3 signal connectors on the mid left side of the HV battery to check it's all working.