BQ78350: BQ78350 Production Failure. Partial SMBus and no VEN pulses.

Part Number: BQ78350
Other Parts Discussed in Thread: BQSTUDIO

Good afternoon,

I am analyzing some BQ78350 gauges that failed in production.

I have one unit that won't communicate with BQ Studio.

It also does not show the pulses on VEN that are expected.

Chip text is:

BQ78350

2CKg4

A95K

Can you confirm the chip hardware version?

Interestingly it will communicate on SMBus RSOC and ASOC commands with:

RSOC = 0x0303

ASOC = 0x2CFE

Otherwise all other commands fail, and we cannot update the firmware.

Any thoughts on a failure mode?

Thanks.

  • How are you attempting to update the FW? With bqStudio?

  • Hi Dominik, yes, with BQStudio.

    Our procedure is to power up the BMS with a simulated cell stack (0.1% 1K resistors) on a test jig and program/test the PCBA with BQStudio before we solder the BMS to a pack.

    This usually works fine, however we have had a few failures. Most boards power up OK, and talk to BQStudio.

    I'm trying to figure out if the chip was bad before we received it, or if we have done something wrong (ESD, or procedure based).

    Thanks,

  • It looks like the gauge is in ROM mode, where it executes a small maintenance code instead of the gauging FW.

    Please read a word from smb command 0x0D and post the result.

  • 0x0d returns 0x0303

    These are the 3 commands that appear to return data:

    Average Current will read 0 briefly, then 4.

  • Can you please try the following on a good unit:

    1. Unseal
    2. In Advanced Comm, issue the following commands:
        write block to 0x44: 0x00 0x0F
        write word to 0x00: 0x0F00
    3. Read word smb command 0x0D and verify if the gauge behaves like a bad unit.

  • OK, here's a unit that passed PCB level test, and was successfully programmed with R2 firmware:

    It looks normal, now trying the change:

    Writing block and write word look like this:

    The write word has an Error status.

    Read Word of 0x0d shows 0x0303

    Device now has the same problems as the other units, with only these registers showing:

    So, yes, this created the same issue...

  • What this means is that the gauge is indeed executing the small maintenance code in ROM instead of the firmware in flash.

    bqStudio will, in most cases, be able to re-flash the firmware in this situation. What happens, exactly, if you try re-programming the .srec to a device in this mode?

    Also, does this resolve with a power cycle?

    There are several reasons why the gauge may enter this mode (the 0x00 0x0f sequence is one, others are corrupted FW or a HW controlled sequence to force it into this mode). The most likely is that the previous attempt to program the FW was interrupted.

  • Dominik,

    Awesome, thank you...

    I re-flashed a "bad" unit and it came up OK... of course BQ Studio barfed this at me:

    But restarting BQStudio found a working gauge.

    This is a big step forward for us, I suspect the PC on the production line is... not healthy... and might be crashing, or the operator has opened multiple instances of BQStudio or something...

    One last question (I can open as a new post if you prefer):

    With this chip text:

    BQ78350

    2CKg4

    A95K

    Is this a BQ78350 or BQ78350R1?

    I know the only hardware difference is something about sleep modes, but I need to know what HW version our vendor supplied. We flash with R2.

    Thanks!

  • bqStudio's dashboard automatically refreshes (it issues commands periodically in the background). I recommend turning this off by clicking on the green background "auto refresh" line on top of the dashboard if you have a gauge that's not running the gauging FW as this can cause errors with auto-refresh if the gauge won't respond as expected.

    I'll check the date code and get back.