Speak EV - Electric Car Forums banner

1 - 20 of 130 Posts

·
Registered
Joined
·
73 Posts
Discussion Starter #1 (Edited)
I am starting a thread specific to my problem as it evolved beyond the scope of the thread I originally started posting on (C Zero / ION Recalls).

So I have a 2012 left hand drive Citroen C-Zero here in south west France that has about twenty thousand kilometers on the clock and has recently started to go into tortoise mode maybe once or twice a week with daily driving. Normally when this happened, it would be possible to pull over, turn off the car and then start back up again. Then you could continue maybe without a problem for the rest of the day. It got worse such that even doing this reset would be a short lived solution as the problem would recur almost immediately.

I took the car to Citroen who kept it a few weeks and diagnosed a problem with CMU_08 and offered to replace the whole battery pack for €18k or give a €2k reduction on a new Citroen. The car was well out of warranty so there was no possibility to get any commercial offer better than this.

Needless to say, I did not go this route and decided to try and fix the car myself. I bought an OBDLink LX bluetooth adapter and used that with Canion to monitor the cell voltages. Normally, the voltages on all 88 cells were within 10mA of one another, but occasionally the voltages on cells 53 to 60 inclusive would shoot up to a few hundred volts (I have attached an image of this to this post). This of course must be a measurement issue and not real values.

strange Canion voltages.png


I had some further checks made on the car with the Citroen Diagbox software which confirmed the problem lay with CMU_08. Specifically, the error code was P1A7D which is an error in the voltage measurement of the cells in traction battery 08. If the code was cleared, it would recur soon after.

So the next step, now that I had enough diagnostic information, was to remove the battery from the car. I decided to do this on my garage floor using two ramps - one for each of the rear wheels, and a jack with blocks for the front of the car. I disconnected the auxillary battery and removed the service plug (I had to remove the front left seat for that). Also I had to access the cooling duct to the battery under the front right seat and so that also had to be removed.

I removed the two large covers under the car. There are loads of bolts and fixings for those. Then I had to disconnect the high voltage cables (I checked them with a multimeter first!) and unbolt the battery from the car underside. I used a pallet mover to receive the battery and I had only just enough clearance to wheel the battery out from under the car.
2019-04-08 19.18.18.jpg


I removed the cover of the battery the next day and surveyed the layout. I wrote on each group of cells what the CMU number was going from CMU01 at the negative end of the battery (where the service plug resides) to CMU12 at the positive end. I attach a photo of the open battery with my numbering. [!!!NOTE THIS NUMBERING IN THIS PHOTO HAS BEEN FOUND TO BE INCORRECT!!!]

2019-04-16 18.54.34.jpg


To summarise the CMU / cell numbering (thanks to DBMandrake for this info):

CMU_01 - 8 cells - 1 to 8
CMU_02 - 8 cells - 9 to 16
CMU_03 - 8 cells - 17 to 24
CMU_04 - 8 cells - 25 to 32
CMU_05 - 8 cells - 33 to 40
CMU_06 - 4 cells - 41 to 44
CMU_07 - 8 cells - 45 to 52
CMU_08 - 8 cells - 53 to 60
CMU_09 - 8 cells - 61 to 68
CMU_10 - 8 cells - 69 to 76
CMU_11 - 8 cells - 77 to 84
CMU_12 - 4 cells - 85 to 88

So I removed CMU_08 and took that apart to look at the PCB's.

2019-04-17 10.12.19.jpg


My current task is searching for the fault on the PCB's - likely a degraded capacitor or cracked solder joint. I attach a photo of the CMU_08 PCB. It is difficult to probe the components due to a conformal coating on the PCB (both sides). I am going to have to use a glass fibre brush to selectively remove this coating where I want to probe.

2019-04-17 11.24.01.jpg


So that is where I am now and I will post more updates to this thread. Thanks especially to DBMandrake and freddym who have helped me get this far.

I very much hope I can find the smoking gun on this PCB. I checked all the cell voltages with a multimeter and they were all ok and within 10mV of one another.
 

·
Registered
Joined
·
4,322 Posts
Watching with great interest. I'm impressed with the idea of using the (hydraulic lift ?) pallet mover as a trolley to get the battery out. (y) That gives hope to those of us without access to a two post hoist... hopefully it allows the precision of movement and height adjustment you need to line everything up to get it back in place again.

Regarding identifying the correct board and cell group for CMU_08 - in the other thread I suggested that it should be possible to figure out for certain whether the CMU's were numbered from the negative end (as I believe is the case) or the positive end, based on the asymmetric positioning in the series string of the two 4 cell modules - did this in fact turn out to be the case and we're now confident that CMU_01 is the most negative group ?

Regarding the actual fault - unlike most other reported faults of these CMU boards that seem to centre around a fault with the voltage measurement of a specific cell in the group, yours seems to lose contact with the CMU altogether and thus give completely erroneous readings for all cells in the group, so I'd probably start by looking at power lines/voltage regulators and Canbus Data lines.

I wonder if any circuit diagram is available online anywhere or whether you'll be working in the dark.. I assume you've had a look here ?

Mitsubishi i-MiEV: Service Manuals — MMC Manuals
 

·
Registered
Joined
·
4,322 Posts
Here's all I've been able to find in that service manual, just external connections for those boards, but that might be better than nothing:

[W/M]i-MiEV (HA3W)

Select MY 2012, 90-Circuit Diagrams, Main battery management system, the first diagram is the external Canbus connections going to the battery pack, as well as power.

The third one shows the canbus connections between CMU's and the outside of the pack unfortunately it doesn't give pin numbers on the actual CMU boards so you might have to trace that through the harness from where it enters the pack...

Not sure if that's of any use though, looks like the CMU boards are just a black box in the service manual, not surprisingly.

Excellent photos by the way - a great resource for anyone else who may have to disassemble a pack in the future.
 

·
Registered
Joined
·
4,963 Posts
Is this a known issue? Such that Citroen have a technical service note? Or has anybody seem a PCB with a later version number on it? I have generally found Citroen parts departments to be more helpful with technical information than the repair mechanics.

It might be worth checking the continuity of the CANBUS lines that are from the bus to the PCB and physically flexing them. Cell voltages going high sounds like a digital value being read as all 1's in a code block that has already been made up to include cells 53 to 60 or the PCB has dropped off the CANBUS completely and is out of touch.
 

·
Registered
Joined
·
4,322 Posts
Just looking at the picture of the group of 8 cells with the board still on it - is that crazed look on most of the chips and a lot of the board some kind of lacquer ?

That LTC6802-2 seems to be a standard chip, I think that's the one that @HEVRA was referring to being replaced on one that was fixed for slightly different symptoms and has a public datasheet:

https://www.analog.com/media/en/technical-documentation/data-sheets/68022fa.pdf

What would be interesting is to figure out what type of interface is used between that cell measuring chip and the Canbus controller chip. I suppose it's theoretically possible that a loss of communication between the two chips could cause completely erroneous readings for all cells.

I take it most of the other chips have any identifying numbers rubbed off ?
 

·
I'm not crazy, the attack has begun.
Joined
·
29,252 Posts
Looks like it has overheated to me.

Can you get a 2nd hand battery pack/crash/dismantling and take the risk just to swap over the board, hoping there is no special encoding on it?

18k for a teeny pack? What a joke. Was the 2k reduction only on ICE cars, perchance!?
 

·
Registered
Joined
·
4,322 Posts
The unique Canbus ID of each CMU board is precoded so a swap from another pack would only work if a board from the CMU_08 position was used otherwise there would be a Canbus ID clash with an existing board.

Nobody knows for certain wherever there is also VIN based coding to the car or whether any recoding of the board can be done.

I have a Diagbox dealer diagnostic tool and don’t recall seeing any programming options for the CMU units. (n)

A swap of CMU_08 from another similar year 88 cell car might work but it would be a case of being the first to try it and crossing your fingers that there is no VIN coding and firmware versions were sufficiently compatible...
 

·
Registered
Joined
·
4,963 Posts
The unique Canbus ID of each CMU board is precoded so a swap from another pack would only work if a board from the CMU_08 position was used otherwise there would be a Canbus ID clash with an existing board.

Nobody knows for certain wherever there is also VIN based coding to the car or whether any recoding of the board can be done.

I have a Diagbox dealer diagnostic tool and don’t recall seeing any programming options for the CMU units. (n)

A swap of CMU_08 from another similar year 88 cell car might work but it would be a case of being the first to try it and crossing your fingers that there is no VIN coding and firmware versions were sufficiently compatible...
Is @Mike Schooling in a position to help with a substitute PCB for CMU - 08 version 0.0.2?
 

·
Registered
Joined
·
3,869 Posts
Well done!

I seem to recall that an AEVA poster had a bunch of spare iMiev battery boards. Another source might be the Dutch company that sell a lot of used iMiev battery packs.

Did we also not have a recent UK HEVRA company posting wrt a similar repair.

Best of luck
 

·
Registered
Joined
·
4,322 Posts
I have no knowledge of this to help sorry but would the battery pack work if you just removed the cmu_08 cells?
No. Lack of valid data from CMU_08 will prevent the battery as a whole from being charged.
 

·
Registered
Joined
·
73 Posts
Discussion Starter #15 (Edited)
Regarding identifying the correct board and cell group for CMU_08 - in the other thread I suggested that it should be possible to figure out for certain whether the CMU's were numbered from the negative end (as I believe is the case) or the positive end, based on the asymmetric positioning in the series string of the two 4 cell modules - did this in fact turn out to be the case and we're now confident that CMU_01 is the most negative group ?
Thanks to the asymmetric positioning in the string of the two 4 cell modules, it is only possible that CMU_01 is at the negative end of the pack and CMU_12 the positive. Following the busbars from one battery to the next revealed how they are located in the pack.

Here is a photo that I have labelled with the CMU numbers. I am 99% sure this is correct, and also found info on another forum Battery only charging halfway - Page 2 - Mitsubishi I-Miev Forum that corroborates it. [NOTE: THIS NUMBERING SCHEME HAS BEEN FOUND TO BE INCORRECT AS DISCUSSED LATER IN THIS THREAD. DO NOT USE FOR REFERENCE!!!]

layout.PNG
 

·
Registered
Joined
·
73 Posts
Discussion Starter #17
Just looking at the picture of the group of 8 cells with the board still on it - is that crazed look on most of the chips and a lot of the board some kind of lacquer ?
Yes there is a conformal coating on both sides of the PCB which makes it difficult to probe the pads and devices.
 

·
Registered
Joined
·
4,963 Posts
Just looking at the picture of the group of 8 cells with the board still on it - is that crazed look on most of the chips and a lot of the board some kind of lacquer ?

That LTC6802-2 seems to be a standard chip, I think that's the one that @HEVRA was referring to being replaced on one that was fixed for slightly different symptoms and has a public datasheet:

https://www.analog.com/media/en/technical-documentation/data-sheets/68022fa.pdf

What would be interesting is to figure out what type of interface is used between that cell measuring chip and the Canbus controller chip. I suppose it's theoretically possible that a loss of communication between the two chips could cause completely erroneous readings for all cells.

I take it most of the other chips have any identifying numbers rubbed off ?
Having slept on it, one has rethought my earlier suggestion of checking the CANBUS where it is unique to this PCB. That would I think flag separate CANBUS fault codes. High values received for all the 8 cells in the 53-60 block points to the block framing bits being correct but the eight way multiplexed ADC chip may be faulty. I think the ADC is more likely to be multiplexed rather than reproduced on each cell. Likely to be a standard part with the ID rubbed off. So I am more inclined to consider it to be a fault on the PCB.

See:

Multicell Battery Stack Monitor | Analog Devices

Then AD7284 for data sheet and pin out as one possible. Chip is under $7 so looks suitable for automotive applications.
 

·
Registered
Joined
·
4,322 Posts
Having slept on it, one has rethought my earlier suggestion of checking the CANBUS where it is unique to this PCB. That would I think flag separate CANBUS fault codes. High values received for all the 8 cells in the 53-60 block points to the block framing bits being correct but the eight way multiplexed ADC chip may be faulty. I think the ADC is more likely to be multiplexed rather than reproduced on each cell. Likely to be a standard part with the ID rubbed off. So I am more inclined to consider it to be a fault on the PCB.
I'm leaning a bit that way too - one way to have told for sure would have been to try to interrogate CMU_08 in Diagbox - both checking the identification of the ECU (firmware version etc) and also looking at the standard parameters measurement, which show not just voltages but also temperatures.

If you can get the firmware version and board revision from the ECU and/or temperature readings and it's only the voltage readings that are whacky then that immediately rules out a the Canbus chips and bus.

Unfortunately now it's out of the car it's not possible to check this and I think Rupert was only borrowing access to Diagbox and doesn't have his own Diagbox system anyway ?
See:

Multicell Battery Stack Monitor | Analog Devices

Then AD7284 for data sheet and pin out as one possible. Chip is under $7 so looks suitable for automotive applications.
No need to look at other chips, the chip on the board that contains the ADC is the LTC6802-2 I already posted the data sheet for - it's a fully integrated chip that does everything in the one chip including measurement, balancing etc...

The MUX for the ADC is internal to the chip though, so unless the PCB fault affected a reference I'm not sure how it would affect the reading for all 8 cells. More likely the ADC in the chip is stuffed and therefore all 8 MUX'ed voltage readings will be wrong.
 

·
Registered
Joined
·
4,963 Posts
I'm leaning a bit that way too - one way to have told for sure would have been to try to interrogate CMU_08 in Diagbox - both checking the identification of the ECU (firmware version etc) and also looking at the standard parameters measurement, which show not just voltages but also temperatures.

If you can get the firmware version and board revision from the ECU and/or temperature readings and it's only the voltage readings that are whacky then that immediately rules out a the Canbus chips and bus.

Unfortunately now it's out of the car it's not possible to check this and I think Rupert was only borrowing access to Diagbox and doesn't have his own Diagbox system anyway ?

No need to look at other chips, the chip on the board that contains the ADC is the LTC6802-2 I already posted the data sheet for - it's a fully integrated chip that does everything in the one chip including measurement, balancing etc...

The MUX for the ADC is internal to the chip though, so unless the PCB fault affected a reference I'm not sure how it would affect the reading for all 8 cells. More likely the ADC in the chip is stuffed and therefore all 8 MUX'ed voltage readings will be wrong.
Thanks. I missed your earlier pointer to the correct chip. A quick look at the circuit diagram for it shows an external 100k resistor between the temperature reference pin in and the voltage reference pin in . That 100k resistor should be easily measurable either directly or by comparison with other boards.
 
1 - 20 of 130 Posts
Top