BQ40Z50: OC bit gets set while charging, Calibration and other things

Hello Dominik,

1) For the cycles you do not need firmware to control the FETs so I would send the FET_EN command again to prevent the gauge from stopping the cycle.

2) Yes you are correct, SLUA903 is the best document for the learning cycle.

3) It is recommended to calibrate all the gauges because each system is a little different, the error could be from calibration, if this is a custom PCB then there could be some board offset error as well

4) Yes you can, as long as you don't exceed the SRN and SRP voltage range. A 1mOhm to 3mOhm resistor is recommended.

Sincerely,

Wyatt Keller

Hello Wyatt,

Thank you for your reply.

3) I calibrated our custom PCB with 2A and now all seems to report fine. I am guessing in production I do not have to calibrate every board individually because the errors (they arise from different impedances I assume?) should be similar for the same board - correct?

1) Great, I tried that and managed to do a full charge and discharge cycle with two cells in series by sending CHG_FET_TOGGLE and DSG_FET_TOGGLE.

I managed to create a CHEM ID match with max. deviation of 1.57% so that seems to be close enough.

I did _not_ manage to follow the exact charge and discharge cycle though. The problem is we are using cells that are UV protected and while discharging for 10h at some point the UV protection kicks in for one of the cells. So I could not have the 5h of OCV voltage in the end.

"Warning: Last OCVpoint has non-zero current I= -0.052 Please make sure you have a 5 hr relaxation period after the end of discharge."

So the last reported voltage would be 3253mV and 2979mV (6232mV together) with the second cell UV protection kicking in. (As a side note I was also surprised how uneven the voltage was between the two)
After that I do not manage to get any data out of the battery gauge. As far as I understand that is because the gauge gets it power from the VCC/BAT there is no voltage due to the UV protection.

Should I monitor the cell voltages and do not discharge to more than 3050mV (hysteresis of the protection) to keep the gauge alive?
What would you recommend doing here? 

Kind regards,

Dominik

Hello Dominik,

Yes you can average some of the values from your system and use those averaged values to program other boards.

Max deviation of 1.57% is good, you just want to keep it under 3% ideally.

Are you trying to perform a learning cycle with the chemistry ID you have programmed? If your packs have integrated protection that prevent a 90% DOD change you may not be able to get a successful learning cycle.

The gauge does get its power from VCC/BAT, so if your protection cuts those voltages then the gauge will shut down. Is the protection you're referring to enabled by the gauge or something else in the pack?

Sincerely,

Wyatt Keller

Hello Wyatt,

Thank you for your fast answer. 

I just programmed a CHEM ID and I am trying to run a learning cycle now.

The protection I am referring to are built into the cells. Or more accurately we get pouch cells with a protection circuit pcb (basically an IC and two FETs) soldered to the tabs with leads coming out. So I cannot change the protection other than physically removing it.

We do have other cells (for a different application), where the termination voltage is also at 3V, but the undervolt protection is at 2.3V. There should not be a problem with those cells I think.

I think I should be able to get 90% DOD by having the learning cycle between 4.2V and 3.05V, what do you think? As far as I know specially the last part of the voltage drop does not have much energy stored in it, so 90% of the energy should be out by then.
Or I could just use it as cell that works between 3.1 and 4.2V..

Sincerely,

Dominik

Hi Wyatt,

I cannot edit my previous post, so here is what I wanted to edit:

Do I need to reach Term Voltage for the Relax Time after the first discharge?

Because when I set it to 6100mV (2x 3050mV as discussed above) it is hard to reach, because one cell drops below 3050mV while the other one is still at 3200mV during discharging, so I do not reach the term voltage. (They were both balanced before the discharge.)

Regards,

Dominik

Hello Wyatt,

Thanks again for your quick answer!

So I managed to do a learning cycle stopping the discharge when one of the cells was close to 3030mV (while the other one was still up at 3200mV).
The gauge seems to have completed the R and Qmax updates and the update status is 06. I logged it all with the bqStudio as suggested.

I am now running another charge discharge cycle to get the status to 0E to enable cell balancing.

I am still confused about the creation of the golden image though:

SLUA903 mentions a Golden Image tab, which I do not have on my bqStudio.

1) So I assume I get the golden image by reading out the .srec file from the Firmware tab after manually setting the cycles to 0 and the Update Status to 2 in bqStudio?

2) What are the steps for production? Just flash the .srec and it is ready to go? Do I need to program CHEM ID again or is that included in the .srec file?

3) How would I do the averaging values from different systems part? Because the .srec is a hex file where cannot average for example the R values?

4) One last question, if I would have a single cell pack for another application and I would need a 3.3V termination voltage (e.g. it uses a buck converter and cannot boost the voltage and needs the 3.3V) - would I just set the termination voltage to 3.3V and nothing else? Would I do this before or after running the learning cycle?

Thanks again for your patience and your explanations,

Sincerely,

Dominik

Hello Wyatt,

The update status is on 0x0E, so everything seems to be fine. I should only turn sleep back on in the DA configuration.

I am assuming the averaging of calibration values are done before any learning, and then the learning data (Qmax and R Tables) are just not averaged.

Thanks for the extensive answers, I really appreciate it.

Sincerely,

Dominik

Hello Dominik,

No problem, glad you completed the learning cycle successfully!

Yes you are correct, the Qmax and Ra don't need to be averaged. Only one learning cycle is needed.

Sincerely,

Wyatt Keller