Speak EV - Electric Car Forums banner

81 - 100 of 130 Posts

·
Registered
Joined
·
3,746 Posts
The problem of the CMU reporting the voltage as abnormal remains, as does a report of all cells except cell B having low voltages, despite the fact that the voltages look normal and similar to all the other cells in the battery.

However, the fluctuating low voltage value on cell H had gone, so maybe that was caused by the LTC chip, or a connection to it from the PCB.
I know you've probably tried this, but did you try clearing the fault codes in the BMS then checking for codes again ? For that type of fault they probably won't clear themselves after the fault is repaired, you will have to manually clear the codes.

Although you only noticed one cell visibly dropping to an abnormally low voltage keep in mind the communication rate to the diagnostic tool for the voltage readings is very low - only one or two updates a second would be typical.

So the other cell readings could have been dropping for milliseconds - long enough to set faults by tripping a low voltage threshold in the chip, but not long enough to be caught by the real time display of cell voltages due to the low effective sampling rate you have viewing the cell voltages via the diagnostic tool.
 

·
Registered
Joined
·
73 Posts
Discussion Starter #82
I know you've probably tried this, but did you try clearing the fault codes in the BMS then checking for codes again ? For that type of fault they probably won't clear themselves after the fault is repaired, you will have to manually clear the codes.

Although you only noticed one cell visibly dropping to an abnormally low voltage keep in mind the communication rate to the diagnostic tool for the voltage readings is very low - only one or two updates a second would be typical.

So the other cell readings could have been dropping for milliseconds - long enough to set faults by tripping a low voltage threshold in the chip, but not long enough to be caught by the real time display of cell voltages due to the low effective sampling rate you have viewing the cell voltages via the diagnostic tool.
I checked the BMS and there were no fault codes since my last clearout.

Yes, I agree that the sampling rate of the display value may not catch actual variation in cell voltage. I will try and measure the real values with an oscilloscope.
 

·
Registered
Joined
·
3,746 Posts
I checked the BMS and there were no fault codes since my last clearout.
Yet it still claims "abnormal low voltage" for 7 of the cells in the module ? o_O

Back when you were locating the CMU modules with Diagbox by unplugging them, was that before or after you did the most recent board swap / removal ?

If it was before the swap, can you verify again that if you unplug CMU_02 and CMU_08 one at a time that the correct CMU's stop responding in Diagbox ?

We want to be 100% sure that there are no address conflicts between the boards and that the identity of the modules is where they should be in the off chance that the address is not in the EEPROM.

I wonder whether the CMU has some non volatile memory (the EEPROM ?!) which stores the abnormal cell voltage status ? By transferring the EEPROM could you have transferred stored fault codes as well ?

I presume when you did the LTC chip swap that the board was removed for a good few hours from the module thus powering it down and hopefully clearing any fault status in ram ?

Clutching at straws a little here now I admit...
 

·
Registered
Joined
·
73 Posts
Discussion Starter #84
Here is a possible test: if you zoom in on your picture of the replacement CMU board you will see that there are test pads on the cmu board near the LTC chip. Now you could use a voltmeter to measure the cell voltages that the LTC chip thinks it is reading.

You can check the return path first (since cell H was fluctuating) for any stray voltage, measure between the module negative terminal and the Test Pad for pin 26 of the LTC, there should be no voltage difference.

Then for cell H measure between the TP of pin 26, to the TP for pin 24. For G, between TP24 and TP for pin 22. For F, TP22 to TP20, ...etc repeating along the test points in the same pattern of even pins down to cell A between TP12 and TP10.

If not too much trouble, you could repeat this measurement set for another module, for comparison of results or to illustrate that a measurement error really exists for that CMU board. Hope something points to the root cause and it's easily fixed...
Thank you Kiev for the suggestion of that test and your description of the methodology was perfect. I was able to check everything above and found no anomalies at all. 0V delta between pin 26 and the module negative terminal, and then 3.75V between each pair of pads corresponding to each cell, with no variation observed using a high sampling rate voltmeter.

I will stick with the plan of replacing the PCB with another - a brute force way of eliminating the problem.
 

·
Registered
Joined
·
73 Posts
Discussion Starter #85
Yet it still claims "abnormal low voltage" for 7 of the cells in the module ? o_O

Back when you were locating the CMU modules with Diagbox by unplugging them, was that before or after you did the most recent board swap / removal ?

If it was before the swap, can you verify again that if you unplug CMU_02 and CMU_08 one at a time that the correct CMU's stop responding in Diagbox ?

We want to be 100% sure that there are no address conflicts between the boards and that the identity of the modules is where they should be in the off chance that the address is not in the EEPROM.

I wonder whether the CMU has some non volatile memory (the EEPROM ?!) which stores the abnormal cell voltage status ? By transferring the EEPROM could you have transferred stored fault codes as well ?

I presume when you did the LTC chip swap that the board was removed for a good few hours from the module thus powering it down and hopefully clearing any fault status in ram ?

Clutching at straws a little here now I admit...
I have confidence that CMU_08 is the one that is the one I am working on and the one I see listed as CMU_08 in Diagbox because when I made the diagnostic yesterday of cell temperatures, CMU_08 was the only one showing 20°C whilst all the others 14°C - because I had just brought CMU_08 home from the warm laboratory whilst all the rest had been sitting in the cold garage.

When I have replaced the PCB on the real CMU_08 and perform the diagnostics, I could always try unplugging CMU_02 as I do it to see if that makes any difference (if the problem persists).

When I made the LTC chip swap, the board had been unplugged for at least 24h.

What are the chances of the problem persisting anyway, if I replace everything apart from the cells, casing and EEPROM of CMU_08. This has to work! (famous last words) :LOL:
 

·
Registered
Joined
·
3,746 Posts
So just to confirm, it still complains about abnormal low voltage of 7 of the cells in the status section of the CMU despite the actual voltage readings being OK ? :(

I hope that EEPROM isn't being used to save that fault...
 

·
Registered
Joined
·
73 Posts
Discussion Starter #87
So just to confirm, it still complains about abnormal low voltage of 7 of the cells in the status section of the CMU despite the actual voltage readings being OK ? :(
Exactly, so it is not the connections from the battery cells to the LTC chip or the LTC chip itself that is at fault, but something upstream of that (and hopefully downstream of the BMU!).
 

·
Registered
Joined
·
3,746 Posts
Exactly, so it is not the connections from the battery cells to the LTC chip or the LTC chip itself that is at fault, but something upstream of that (and hopefully downstream of the BMU!).
I wonder why it's no longer setting a fault code in the BMU though ? Maybe it would set a fault again if the car was put into READY or a charge was attempted ? (Obviously with the pack reinstalled)

I guess we don't actually know for sure whether the car is still refusing to charge without trying it - which would be a lot of work getting the pack hooked up again. :(

There's always the possibility that it would in fact charge now with the fluctuating voltage fault solved and the BMU clear of any stored fault codes because ultimately it is the BMU that makes the decision whether charging is allowed or not - the CMU's really just report the status of the cells (voltage and temperature) and perform balancing but don't really do anything else.

I wonder if @kiev has any further ideas on which way to go next ?

I don't suppose you still have your 30 volt supply handy that you were originally going to use to charge the cells in that module up by 100mV each ?

Is it worth connecting that across that 8 cell module and charging it slightly while simultaneously monitoring the voltage and cell normal/abnormal status with Diagbox to see if you can get the abnormal cell status to shift at all ?

Even just commencing a charge at say 10 amps (about 0.25C) should raise the voltage of the cells by probably 50mV or more immediately due to the compliance of the cells - if the status flicks from abnormal to normal then you know you might be onto something and should continue to charge them up 100mV.

I wonder if there is some deliberate hysteresis in the setting of the abnormal/normal status - in other words you have to go very low (below 2.3 volts I think @kiev said ?) to set the abnormal low cell voltage status, but maybe it needs to go higher than 3.775 volts to clear that status again ? Again, clutching at straws...
 

·
Registered
Joined
·
73 Posts
Discussion Starter #89
I don't suppose you still have your 30 volt supply handy that you were originally going to use to charge the cells in that module up by 100mV each ?

Is it worth connecting that across that 8 cell module and charging it slightly while simultaneously monitoring the voltage and cell normal/abnormal status with Diagbox to see if you can get the abnormal cell status to shift at all ?

Even just commencing a charge at say 10 amps (about 0.25C) should raise the voltage of the cells by probably 50mV or more immediately due to the compliance of the cells - if the status flicks from abnormal to normal then you know you might be onto something and should continue to charge them up 100mV.

I wonder if there is some deliberate hysteresis in the setting of the abnormal/normal status - in other words you have to go very low (below 2.3 volts I think @kiev said ?) to set the abnormal low cell voltage status, but maybe it needs to go higher than 3.775 volts to clear that status again ? Again, clutching at straws...
This experiment sounded worth a try, so I set up to charge the 8 cells in series with my 30V power supply. I applied 3A constant current (maximum possible with this PSU) whilst monitoring the voltage across one of the cells as seen here:

2019-05-22 17.10.10.jpg


I periodically disabled the power supply to check the cell voltage until I had increased it by around 10mV, which took about 10 minutes of charging. I didn't have time to do more than that so could not reach 100mV. Then I took the module back home and installed it back in the battery.

Diagbox showed the following voltage values:

cmu_08 tension.PNG


Compared to how it was before I charged (and before I replaced the LTC chip which resolved the low value for cell F):

cmu_08 tension low.PNG


The abnormal status remained for CMU_08 voltage with undervoltage reported for all cells except cell B:

etat2.PNG


BMU status is now:

bmu etat.PNG


So it is reporting a cell that I charged as having the highest voltage in the battery, but still reporting undervoltage for those cells at CMU level.
 

·
Registered
Joined
·
523 Posts
Hi is it possible that you have 2 failed LTC chips? Seems like you've resolved one but 1 remains. With my LTC failure, undervoltage was reported for all cells in the module, this was because the voltage reported by the LTC to the BMU (under the back seat) was bouncing up and down from 0 to hundreds of volts, so all cells were "undervoltage". (tho actually all fine, just reported wrong by the LTC).


 

Attachments

·
Registered
Joined
·
73 Posts
Discussion Starter #92 (Edited)
Hi is it possible that you have 2 failed LTC chips? Seems like you've resolved one but 1 remains. With my LTC failure, undervoltage was reported for all cells in the module, this was because the voltage reported by the LTC to the BMU (under the back seat) was bouncing up and down from 0 to hundreds of volts, so all cells were "undervoltage". (tho actually all fine, just reported wrong by the LTC).
Maybe I have two failed LTC chips. Did you ever get Diagbox measurements of the cell voltage values from the CMU? If so, were those appearing normal or not?
 

·
Registered
Joined
·
8 Posts
@Rupert,
Can you read any DTC from the EV-ECU related to the BMU?

For example, P1A30 in the EVECU points to a low cell voltage DTC reported by the BMU, e.g. P1A59 in the BMU points to CMU08 as the location.

Also P1A33 in the EVECU points to an abnormal cell sensor DTC reported by the BMU, e.g. P1A7D in the BMU points to CMU08 as the location.
 

·
Registered
Joined
·
3,746 Posts
I've also had another thought about this situation.

What if the cell abnormal low voltage and cell abnormal high voltage status flags reported by the CMU are designed to be a permanent record of any incident of extreme low or high cell voltage that might constitute abuse of the cells, (such as deep discharge) for the purposes of warranty etc ?

This information may be written to the on board EEPROM for permanent storage (?) so transferring the EEPROM to a new board may have transferred these flagged "incidents" of abnormal low voltage as well.

And as far as I can see there is no facility to clear fault codes or fault status in the CMU units - they don't have the regular "read fault codes" and "clear fault codes " operations that you find on other ECU's like the BMU. They are more like status codes than fault codes as there is no P number.

It's possible that now that the fault with the LTC is repaired this abnormal low cell voltage status is simply an informational warning as a permanent record in the pack for any future technician to see that at some point in time cells were subjected to extremely low discharge. (Although in reality this was due to a faulty voltage measuring chip not actual discharge of the cells)

As long as the BMU and EV ECU no longer report any faults after the codes have been cleared I wonder whether the problem has been fixed and we are just chasing our tails reading a status flag (warranty void ?) that is just an informational record of a past problem which can't be cleared but won't actually stop the car charging or driving.

I hesitate to suggest it because I know it's a lot of physical work and hassle to refit the battery even partially and connect it fully to the car including HV, but I wonder if we're now at the point where connecting the battery to the car fully and trying to charge it normally is worth trying ? (but don't completely bolt everything up so you can get it out again easily)
 

·
Registered
Joined
·
523 Posts
Maybe I have two failed LTC chips. Did you ever get Diagbox measurements of the cell voltage values from the CMU? If so, were those appearing normal or not?
Hi i was using iCarSoft i909, this showed the voltages on CMU10 jumping up and down constantly, for me all other CMUs reported the correct voltage so my issue was limited to 1 CMU only. Once I swapped the LTC chip on CMU10 and refitted everything, it went to ready and charged.

It is a pain getting the battery pack in and out, but to test it you dont have to put the lid on or bolt the pack in, just refit the modules, bus bars and balance connectors, then jack it up roughly into place and connect the multi plugs and orange cables (drivers side only, you dont need the quick charge ones on the left).

Once you test and go to remove you need to double check for voltage across the orange wires on drivers side with a multimeter, as it may have charged the capacitor in the MCU via precharge resistor, it takes time for that to bleed away so if you turn the key, turn off, undo that cover there could be 360v on those cables, even tho main pack is not engaged. Apologies if i'm insulting your intelligence in these matters, i'm not aware of your background/experience level.

Thanks.
 

·
Registered
Joined
·
73 Posts
Discussion Starter #96 (Edited)
I've also had another thought about this situation.

What if the cell abnormal low voltage and cell abnormal high voltage status flags reported by the CMU are designed to be a permanent record of any incident of extreme low or high cell voltage that might constitute abuse of the cells, (such as deep discharge) for the purposes of warranty etc ?

This information may be written to the on board EEPROM for permanent storage (?) so transferring the EEPROM to a new board may have transferred these flagged "incidents" of abnormal low voltage as well.

And as far as I can see there is no facility to clear fault codes or fault status in the CMU units - they don't have the regular "read fault codes" and "clear fault codes " operations that you find on other ECU's like the BMU. They are more like status codes than fault codes as there is no P number.

It's possible that now that the fault with the LTC is repaired this abnormal low cell voltage status is simply an informational warning as a permanent record in the pack for any future technician to see that at some point in time cells were subjected to extremely low discharge. (Although in reality this was due to a faulty voltage measuring chip not actual discharge of the cells)

As long as the BMU and EV ECU no longer report any faults after the codes have been cleared I wonder whether the problem has been fixed and we are just chasing our tails reading a status flag (warranty void ?) that is just an informational record of a past problem which can't be cleared but won't actually stop the car charging or driving.

I hesitate to suggest it because I know it's a lot of physical work and hassle to refit the battery even partially and connect it fully to the car including HV, but I wonder if we're now at the point where connecting the battery to the car fully and trying to charge it normally is worth trying ? (but don't completely bolt everything up so you can get it out again easily)
@DBMandrake you hit the nail on the head. Since I last posted, I have done a few things so will go through that chronologically.

So the last thing I had done was to swap out the LTC chip on CMU_08 for the one that came from the PCB of CMU_02 (that I had replaced with a PCB from a second hand seller). This did not reset the flags seen in Diagbox for abnormal or low voltage in this CMU, although it did correct a real low voltage measurement problem on cell H.

I sourced a replacement PCB for CMU_08. I wanted to replace everything in the CMU to test if the "abnormal" voltage flags would be eliminated or not. So I moved again the EEPROM and all CMU_08 was left with that was unchanged were the cells, the mechanical supports and the EEPROM. Diagbox showed that the flags remained so I conclude the flags are either in the EEPROM (most likely), or stored in the BMU or some other component upstream of the CMU.

Before I moved the EEPROM, I wanted to take the opportunity to read it using the parameter that @kiev had suggested of ADW=9. I tried this and it worked - I can read all 512 words now. This modification also means that the values I read previously have absolutely no similarity to what I am reading now. I wanted to compare directly with another CMU EEPROM and so removed CMU_07 as it was the most convenient. Below are the data tables I read from CMU_07 and CMU_08 respectively:

CMU_07 EEPROM contents 512 word.PNG


CMU_08 EEPROM contents 512 word.PNG


I cannot even be sure that the data in these tables is valid or not - I need to see the data from other people who have read these EEPROM's for comparison. Anyway, I concluded that I could not sufficiently understand the numbers in order to eliminate the flags set in the EEPROM of CMU_08 and not in CMU_07.

Unfortunately, in my haste to put the CMU_07 PCB back into its place on the cells, I put it (for just a short time) the wrong way around - rotated 180°. Diagbox showed that some cell readings were then 0V and the undervoltage flag against those cells became set. The abnormal voltage flag did not become set though. So I gave myself extra work to now replace this PCB as I had damaged it. So in fact I moved the EEPROM of CMU_07 onto the old PCB of CMU_08 that had the LTC chip that had come from CMU_02. This worked fine.

So at this point, I had the battery hooked up to the car only via the comms ports and no HV connections (as the battery was next to the car, still open). I followed the suggestions above to make a test of putting the battery back on the car in a loose fashion just to test it. I did this using the threaded bar method detailed previously. This worked well and I had the battery back up under the car in around an hour. I connected to Diagbox and cleared all the error codes that were there from having the battery partially connected previously (things like battery cooling fan not connected). Then I tried charging, and it charged! Around 5 hours later I was back up to 100%. This screenshot was taken after I had played with it a few minutes to check I could get into Ready mode etc so is slightly less than 100%:

Capture1.PNG


So then I removed the battery again, fixed on the covers properly and put it back on again (this took another couple of hours and my arms ached from turning the nuts to lower and raise the battery).

Then I retested once more before doing all the finishing work of putting the covers back on the underside, seats back in, carpet well fitted etc. Last thing was getting it off the blocks, but that was quite easy with the jack and some oak beam sections.

I found my OBDLink tool in the car when reassembling it so was able to do a test drive. Here is a screenshot:

pict_2019-05-31_07-21-19.png


So all looked good.

The only problem I have is that the airbag light is illuminated so maybe I need to reset something for that. I will look on the forum to see if I can find a solution.

So, conclusions of this repair:

1.) The fault was caused by the LTC chip on CMU_08
2.) Lack of clarity of where CMU_08 is located made me a lot more work as I worked on the wrong one (CMU_02)
3.) If a CMU doesn't work and the cells are healthy, then just buy a second hand CMU PCB (it doesn't matter what CMU number it was in the original car) and then transfer the EEPROM to the new PCB. Put it back in and will likely work. These PCB's second hand cost around fifty euros so worth a try. This is easier than replacing the LTC chip in its own right.
4.) Do not destroy the original EEPROM! I have not yet been able to write to a replacement EEPROM and that is a project for another day as I have one lying on my lab bench. Once we have the capability to read from the old EEPROM and write to the replacement, then I will rest much more easy because the EEPROM's that I have already moved may be killed if I solder them again and then my battery may not ever work again :eek:
5.) The threaded bar method for battery removal and remounting is much easier than the jacks method and much safer for the battery.
6.) Diagbox was essential for this repair given the need to reset many codes and perform in depth diagnostics.
7.) Abnormal voltage flags seen in a CMU with Diagbox do not seem to prevent normal operation of the car and do not generate regular fault codes. They also seem not to be resettable, even if the original cause of them has been fixed.
8.) Don't attempt the repair unless you know how to do it safely and take ESD precautions when handling the PCB's. Also it is useful if you have soldering experience and equipment.
9.) ChipQuik solder and flux is really great for removing SMD components such as the EEPROM or LTC chip and then resoldering them back onto a new PCB with minimal thermal stress to the component.
10.) With hindsight, this repair could have been done for €70 versus €18,000 Citroen wanted to do it themselves. I spent around €200 as I bought two replacement PCB's and also the OBDLink tool which could be useful in the future. The ChipQuik was another €20.
11.) Removing the battery, spinning it around 180° and connecting it to the canbus system (with no HV connection) was great as it was possible to make diagnostics on the open battery and test most things out before putting the battery back under the car.

Thank you to everyone on this forum who has contributed. This place is rich in knowledge and ideas. :D
 

·
Registered
Joined
·
2,637 Posts
Well done and congratulations.

Fantastic learning for anyone removing a battery pack from an iMiev clone and attempting repair.

As you say, EEPROM replacement seems a real concern.

Amazed that the ChipQuik solder worked so well. Did you use any aids like magnification viewer, pcb holder, desolderer.
 

·
Registered
Joined
·
523 Posts
Nice one, glad you got it working. From experience and reading on the forums, the 2 main electrical "failures" on these cars are the LTC chip (CMU10 seems to be prone but also happens on others) and the charger/dc-dc converter unit popping fuses/capacitors.

Of course there are battery degradation issues but if you can keep the car going and live with the range, these are great little vehicles. I just sold my old petrol car !

Cheers.
 

·
Registered
Joined
·
73 Posts
Discussion Starter #99
Well done and congratulations.

Fantastic learning for anyone removing a battery pack from an iMiev clone and attempting repair.

As you say, EEPROM replacement seems a real concern.

Amazed that the ChipQuik solder worked so well. Did you use any aids like magnification viewer, pcb holder, desolderer.
I used a binocular microscope at work and the PCB was sitting on a grounded ESD mat. The microscope was essential for seeing the necessary detail as I was soldering / desoldering the fine pitch components and also for peeling off the protective coating of the PCB.
 

·
Registered
Joined
·
3,746 Posts
Outstanding result! :D

Although there were a few mishaps and dead ends along the way that could have caused some people to give up you persevered and got a successful result in the end, and you are probably one of only a handful of people that have successfully done this component level repair of a CMU now. (y)

Those mishaps whilst no doubt frustrating and time consuming have provided additional insight into how the system works and how repairs can and can't be affected, so in that sense it wasn't such a bad thing as it will no doubt help others in the future.

By the way I'm surprised that it's possible to fit the board backwards (certainly something to watch out for!) and also that you were able to get any dialogue with Diagbox, as the board actually powers itself from the cells...

So just to clarify and confirm for the record:

1) The board that in the end has gone into the CMU_08 position to solve the original fault was a second hand replacement board that you just fitted the original EEPROM to without any other changes ?

2) The board that you had to fit to CMU_07 to fix the reverse polarity oopsie was the original faulty CMU_08 board with the LTC chip replaced and the EEPROM swapped from the reverse polarity damaged CMU_07 ?

If so that seems to validate both swapping of the LTC chip, and swapping of the entire board but installing the original EEPROM as valid repair mechanisms.

One thing you haven't said is whether the new CMU_08 board with the original CMU_08 EEPROM in it still shows 7 of the 8 cells as having suffered abnormal low voltage ? Clearly the car charges now, so it would be very useful to know if the car is happy to charge despite the abnormal low voltage status of certain cells, provided that the BMU does not have any related fault codes set, or whether those abnormal low voltage flags that plagued you for a while have now somehow been cleared. (But if so, how)
 
81 - 100 of 130 Posts
Top