BQ40Z50-R2: BQ40z50-R2 on Custom PCB blows discharge FET

Hello Jeff,

I don't think the files were attached correctly, could you also attach the .gg file for your setup before and after your test (and a picture of the bqStudio bit dashboard after the test may also be helpful)? Can you describe the test you are performing when the FET becomes damaged?

Sincerely,

Wyatt Keller

Hi Wyatt,

Here's the schematic.

22000-06120_ControllerBMS_DNP_RevA_2021-10-13.PDF

For some reason, I can't get the forum to attach the srec file.

Let me see if I can get a .gg file and screenshot.

Hello Jeff,

That's okay, just sharing the .gg file before and after would be very helpful, thanks for sharing the schematic. Also the test procedure you performed that caused this to happen.

Sincerely,

Wyatt Keller

Hi Wyatt,

Attached are the .gg files from before a test (fresh FET installed) and after a test (FET did not blow on this test).

During the test, I noticed the output voltage seen by the dummy load was <6V but I was expecting something much closer to 11.2-ish.  Most of the drop seems to be in the discharge FET and the discharge FET gets hot very quickly (I am unsure if the FET is leaving the linear operating region and going into saturation; I'll check on this tomorrow).  I also checked the voltage between the pack output and connector going to the dummy load (which is the voltage across Q8 in the schematic on the page with the charger) and I'm getting around 20mV.

Here is the test process I am using: I put 3x INR18650-30Q cells into the battery holder (stack voltage is about 11.2 or 11.3V) on the PCB and connect the EV2400.  I startup bqStudio and confirm communication with the BMS.  Then I have a dummy load connected to J7.  The dummy load is set for 1A or 2A for the purpose of this test (I think I blew the FET with a 1A load last time).  I enable the load and after a minute or thereabouts, I usually get smoke from the discharge FET.  If I run the test for a short period of time, say 30 to 45 seconds, the discharge FET appears to not blow--hence the remark about the discharge FET staying in the linear region instead of going into saturation).

FET_OK_18oct2021.gg.csvAfter_test_18oct2021.gg.csv

Hello Jeff,

It sounds like the gauge may not be driving the FET or there's some issue with the connection from the gauge to the FET. If there is some drain on the pin the gauge charge pump may not be able to continuously drive it and cause it to be in linear region.

What is the Vgs of the FETs during your test? The .gg files were the exact same, I believe that's probably because the FET was not damaged. If there is current flowing when the gauge wants the FET to be off the gauge should indicate a failure, does anything change on the bqStudio dashboard during the test? is XDSG and XCHG activated?

Sincerely,

Wyatt Keller

Hi Wyatt,

Vgs is about 1.35V during the test and 6.07V from the negative terminal of the first cell to the gate.  The CHG FET gets about 11V for Vgs and about 21V volts from the negative terminal of the first cell to the CHG gate.

The node with discharge FET's gate has the reverse polarity protection circuit (a 2N7002K and a 10K to GND/Pack- on the gate of the 2N7002) and the 10M pull down from the discharge FET's gate to source (Pack+).

I took a log using bqStudio during yesterday's testing.  I've attached that, in case it helps.  I wasn't really paying attention to my laptop screen during the testing yesterday.

I have intentionally not taken the discharge FET to failure on this run as I am running low on replacement FETs.  I can take the FET to failure if that will help debug the circuit

.FET_test - Copy.log

Hello Jeff,

Can you remove any circuitry on the DSG node that could be a load on the DSG FET charge pump? It seems like the gauge wants the FET to be on, but something is preventing the correct Vgs from being applied.

Sincerely,

Wyatt Keller

Hi Wyatt,

I removed the 2N7002 and the associated resistor from the circuit and nothing changed behavior-wise.  It is worth noting that the 2N7002 circuit is copied from the eval board (I started the design of this board with the posted Altium files for the EVM and added the charger and power supply blocks), so I was not expecting a change in behavior. I attached a log file from bqStudio of the test. Pulling R4 (10M resistor from gate to source) seems like it won't do anything, but I'll give that a shot.  I can also pull the FET that links the BMS to the charger/rest of the board and see if that changes anything (although I am similarly doubtful that will change anything).

Is it possible I am fighting something related to hot air reworking the 40z50 onto this board? 

The assembly house put a 40z555 on the board instead of the 40z50 (see https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/1032366/bq40z50-differences-between-bq40z50-and-bq40z555?tisearch=e2e-sitesearch&keymatch=bq40z555#), so we stole a 40z50 from a spare eval board to get the BMS section of the board to start up.

FE_test2.log

update: I removed the R4 and the FET connecting the charger to the BMS.  Vgs for the discharge FET dropped by about a volt, so I don't think removing R4 helped.

Image shows what I pulled off.

Log file from the test:

FET_test3.log

Right now we only have the one board I've been working with.  I'll build up a second board today and see if the problem persists.

admin-ish question: I am having issues with the other two TI parts on the board (bq24113 and TPS55288), should I make separate questions for each of those issues?

Hi Wyatt,

I got a second board working today and it looks like the problem with the 40z50 was a one-off issue.