CMU reporting battery cell temp of 57 deg [solved]

[MickeyS70]

Hi there
My 2012 I-Miev doesn’t charge more than 3 bars (no 10 year battery warranty where I live). I have possibly 2 faulty cells #85 @ 4.095V and #86 @ 3.695V while the remaining are all around 3.89V.

So far everything points to faulty cells and I was planning to replace them tomorrow, however I only discovered today that temperature sensors 64,65 &66 report 57 deg Celsius, compared to 7deg on the rest.

I gather both issues seem to originate in one of the 4 cell packs, most likely caused by a faulty CMU?

Anyone come across this, if yes, how did ye fix it?

[kiev]

Howdy Mickey,

i would guess that the CMU board has a defect, such as power supply, loose connection or some other issue that is causing false readings. The car thinks that one cell is nearly full so it won't allow charging; the 3 errant temp readings are 3 different sensors on the CMU board--not likely for all 3 to fail at the same time.

when you open the pack be sure to check the voltages with a voltmeter to compare with the CMU values. i'm puzzled by how a cell could be higher than all the rest.

[MickeyS70]

Howdy Mickey,

i would guess that the CMU board has a defect, such as power supply, loose connection or some other issue that is causing false readings. The car thinks that one cell is nearly full so it won't allow charging; the 3 errant temp readings are 3 different sensors on the CMU board--not likely for all 3 to fail at the same time.

when you open the pack be sure to check the voltages with a voltmeter to compare with the CMU values. i'm puzzled by how a cell could be higher than all the rest.

You were spot on, opened up the pack this morning all cells (bright yellow) measure 3.88V in both 4 packs, and after messing about with CMU 12 ODBZero now reports an amazing RR of 410km (250 miles) with a battery voltage of 6.5kV and hot (205 deg Celsius) 4 pack…

Swapped the position with CMU6; no change in reporting so problem seems on CMU 12 PCB. Can’t remember exactly where I read about a similar problem, but the solution was to replace LTC6802G-2.

Will order a new one tomorrow and look out for a phone repair shop that might swap it for me, anything else I should consider (was thinking about ordering spare CMUs but they seem hard to come by at the moment) as there was another chip mentioned?

[kiev]

There is a thread about the circuits on the CMU boards, take a look here for a photo of the board with the parts identified;
http://myimiev.com/forum/viewtopic.php?p=42633#p42633

The LTC chip is a battery management chip, read cell voltages and turns on the balancer bleed-off transistors, etc.

[MickeyS70]

Looking at the link and other discussions I seem to have a couple of options:

A. Replace CMU12 with another CMU12
B. Replace CMU12 with a CMU6 and re-ID
C. Replace CMU12 with any other CMU, re-ID and disconnect slave board
D. Replace LTC6802 on faulty CMU12

In reality currently only C & D are viable (don’t seem to be any 4 cell CMUs available presently). Option C is what I plan to do next, unless someone can confirm D is viable (ID chip is easier to swap/ reprogram than LTC):

The PCB itself is 8-cell (G8D-795M-8), however some components are missing (e.g only 4 balance resistors). The sticker on the board (where the connector would be on an 8 cell) defines it as G8D-795M-4. Is there is a way to tell the CMU only to look for 4 cells or (crazy idea alert) could I use the slave board in parallel e.g measure the same cells again?

I have ordered one to do some testing…

[kiev]

eeprom info on another thread

http://myimiev.com/forum/viewtopic.php?t=4302

[MickeyS70]

Thanks for the info, I recon that the EEPROM data identifies if a CMU monitors 4 or 8 cells. It should therefore be possibly to take any 8 cell CMU and either swap the EEPROM or clone the data from a 4 cell to make it work. Should have all the necessary materials to test my theory next week…

[MickeyS70]

Received my programmer ahead of schedule, was able to read data of both CMU12 and CMU8 and proceeded to re-ID CMU8

Replaced the board and started ODBZero: temperature values are now reported correctly, however cell voltages were still off (same 329.78V as with original CMU12?)

DTCs P1A5D & P1A81 both relating to CMU12 low cell voltage which was confirmed 2.1V for IDs 85-88 (all measured a healthy 3.88V though).

Looks like cloning the eeprom isn't enough, some sort of HW mod is also required. Slight differences have already been documented, see link http://myimiev.com/forum/viewtopic.php?t=4302.

Connecting the slave board and running 'sense' wires back to the cells did not change any reported voltage. But during the course of swapping boards and messing about I lost CAN communication with CMU12 again (U11AB)?! Looks more and more like a CAN/BMU problem at this stage:

1. car only charging to 3 bars, CMU12: 2 cells reporting high/low voltage, battery temp 57deg
2. after dropping battery pack and troubleshooting actions: CAN communication lost with CMU12 (329.78V 205deg C)
3. new 8 cell CMU with cloned 4 cell EEprom reports correct battery temp (13 deg C) but low cell values (P1A5D & P1A81)
4. after troubleshooting actions: CAN communication lost with CMU12 (U11AB, 329.78V 205deg C)

Planning do main battery CAN termination resistance check next and swap position with CMU6 but if that doesn't reveal anything new then I'd say I'm dealing with a BMU issue??

[MickeyS70]

Finally managed to solve this problem by replacing all the SMD fuses and LTC chip on the original CMU12.

Be warned:
Do not connect the 12V aux battery & CAN connector if the CMU is not connected to all the battery cells: it can blow the on-board CMU fuse(s) and permanently damage the LTC chip!

If it wasn’t for that mishap I’m pretty certain that an 8 cell CMU could work in a 4 pack after changing/programming the ID chip and some simple HW changes on the PCB. Unfortunately I only had one new LTC chip at my disposal