BQ34Z100-G1: Golden Image for new packs

Hi Kevin,

Can you upload the following and we will take a look: .gg, logs showing the behavior, and the GPCCHEM report.

THanks!

Hi Nick,

I have the requested files attached.

I only have one log file that kinda shows what I've described.

I don't have files from the packs in the field that are reporting the behavior.

The chem report is what was provided as the closest cell match.

Appreciate the feedback.

Thanks,

Kevin

0246.GPC_chemID_Data-report.zip

Hi Kevin,

In this case you may want to try doing two learning cycles back to back and then use that as the golden image. Let us know if this works.

The chemID report looks ok.

Thanks,

Thanks, Nick.

I'm in the process of running another learning cycle right now.

I will keep you posted on the results.

Best Regards,

Kevin

Hi Nick,

I've completed two learning cycles and so far the bq34z100-G1 has seemed not to have corrected itself.

The learned capacity is off.  It should be 5400mAhr, but it's only showing 4253mAhr. The True RC at least shows an improvement at 4865, but the Learned Status is still stuck on 4.  I'm using a battery analyzer to automate the cycle and it reports good capacity at near 5400mAhr.

Is there something else I'm missing?

Thanks,

Kevinbq34z100g1_new learning cycle.zip

Hi Kevin,

What kind of cells are these? The chemID you have programmed is a LiFePO4 is that correct?

Thanks,

Hi Nick,

Correct, they are LiFePO4 cells.  They are actually a BAK 18650 size cells rated at 1800mAhr. I have three in parallel so my FCC should be 5400mAHr.

It's interesting that best CHEMID to use from the report came back with 4103, which is a 26650 cell rated at 3350mAhr.

I'm attempting another learning cycle to see if it works this time as I'm not having luck getting the FCC to update properly either.

The True FCC will read correctly (close to 5400mAh) at times, but the FCC will update to something quite lower.

So, again hoping the two learning cycles back to back works this time.

Thanks,

Kevin

Hi Kevin,

There are two LiFePO4 specific functions ijn Pack Configuration B Register: DoDWT and LFPRelax

Have you tried these?

The GPCCHEM tool report shows that the match is ok, in general LiFePO4 chemistry does not perform well with our impedance track gauges. Bq34110 would be easier to get working.

Best regards,

Hi Nick,

Yes, both the DodWT and LFPRelax bits are set.

It's too late for us to change to the bq34110 and it's too bad it's pinout doesn't match up.

We used the bq34z100 because it states that works with LiFePO4, so that is frustrating to hear it doesn't work well with that Chemistry.

Since Impedance Tracking doesn't work well, would it makes sense to simply not Enable IT?

Is there any other settings we can look into to "dumb" down this IC to work better with LiFePO4?

Thanks,

Kevin

Hi Kevin,

We will still support this issue but yes, bq34110 is the better choice for LiFePO4. 

IT must be enabled or the gauge does not update anything (SOC, FCC etc.)

Someone from our algorithm team will be looking into this more.

Thanks,

Hi Nick,

Thanks for continuing to look into this.

I'm in process of performing the learning cycle again, and the Learned Status is still stuck on 04.

It should have now gone to 05.

This has been an issue before and is described in this post below.

However, there was never any resolve.

I appreciate the support and updates as the come along.

Thanks,

Kevin

Hi Nick,

I have some new cycle data logged. Please find it attached.

With this cycle, I didn't see the pack get to 100% too soon,or to 0% to soon, either.

However, the learned capacity is way off from what it should be.  I have three 1800mAhr cells parallel, so the capacity should be 5400mAhr.

I get this capacity with my battery analyzer, but the bq34z100-G1 reports well under that.  FCC reports 4914mAhr.

If I could get the fuel gauge to report the actually real world capacity (there isn't any reason it should not be reporting at or near 5400mAhr), then I think our problems would be less apparent.

My feeling that even though we get a cycle that behaves without getting to 100% too soon, when these get into the field, they get to 100% too soon because the gauge is reporting FCC incorrectly.  We get to 100% at 4914mAhr, but the pack keeps charging until we reach the actual rated capacity.  I've seen this happen during the learning cycle where the True RC goes above the True FCC.  However, the IC never updates to that proper value.  It jumps back to somewhere around 4900mAhr.

All we need is for this fuel gauge to work consistently.

We will be creating a new golden image with this latest cycle and then applying it to a fresh pack for testing.

To be clear, this is what I've read is the steps involved in creating a Golden Image.

1. Save the Data Memory file and open it with a text editor to change both cycle counts to "0", and set Learned Status back to "02"
2. Program the IC with the default bq34z100G1 srec file.
3. Load the proper CHEM ID.
4. Import the edited Data Memory file
5. Save it as a Golden Image srec.
6. Then, program that new srec to a new pack
7. Calibrate
8. Send the IT_Enable command.

Please let me know if that is correct.

I'm concerned that when we do this for new pack, the Remaining Capacity may also be off for a new pack.  The cells we are using are about 25% state of charge, but when we've programmed the golden image to new packs, calibrated and enabled IT, the fuel gauge reports near 50%.  This, too, could be a reason the packs get to 100% too fast on the first charge.

To summarize, we just want to have consistency with fuel gauge reporting the correct FCC and also to have a golden image be somewhat accurate when programmed to new packs so we do not have to perform a learning cycle on each and every one.

I appreciate your time looking into this matter and the support.

Thanks,

Kevinbq34z100_New_Log_data.zip

Hi Kevin,

How long ago did you measure the battery capacity on the battery analyzer? The capacity can change over the lifetime of the device which is what the gauge tracks

Can you provide the battery datasheet?

Thanks,

Hi Nick,

This analysis was performed just a couple of days ago.
I know there will be some changes in capacity, but I'm seeing big jumps in FCC during rest.

This pack was discharged and during a OCVTAKEN event, the FCC went from 4914mAh to 5114mAh.
That I could live with, as 5114 is at least near our rated capacity of 5400mAhr.  But then it would jump down to 4600mAh on the next OCVTAKEN, only to jump back to 4900mAr.

I feel like this is leading to errors during charge and us seeing packs get to 100% too soon if the FCC Is suddenly lower than what it should be.

At least that is what we are seeing when programming the golden image to new battery packs.
Is there something I should be looking at there?

We have SMOOTH enabled along with RELAX_JUMP_OK and RELAX_SMOOTH_OK.

Should we turn those off?

I have the datasheet for the cell attached.  It's rated at 1800mAhr and we have three in parallel.  We should be seeing 5400mAHr for a FCC, but even when testing on the bench at room temp we can't get that to report as FCC.

Appreciate the support.

Kevin18650-Fe-1800mAh.zip

Hi Kevin,

Let us look into this more, will get back to you next week.

Thanks,

Hi Kevin,

Thanks for letting us know.