bq27510-g3 learning cycle

Kat,

It appears we have a gap in our tools for bq27510-G3 that we are working on.

EVSW, as you may know, is a legacy tool.  For some older products, it is the only tool available.  It works well enough for bq27510-G3, except for bqEasy which is hardcoded to look for some parameters that don't exist, which is why you get those error messages. 

I would recommend that you instead use GaugeStudio for now.  It is itself a transitional tool that is being replaced by our latest GUI that should eventually support all of our latest products. 

The latest tool, and one which we will be focusing on exclusively going forward is called Battery Management Studio (aka bqStudio).  All new products are being launched with bqStudio as the only tool necessary.  Currently the installer only supports a handful of our latest products, but as we test with each existing older product, we will add support over time.  As of the time I'm posting this, bqStudio doesn't support bq27510-G3, but it will be added in the near future.

In the meantime, GaugeStudio is the latest tool that does support bq27510-G3.  However, GaugeStudio doesn't have a full equivalent of bqEasy.  It does have a Chemistry plugin to program the chemID.  It does have a plugin called Golden Files, but it simply reads out the flash contents as-is and does not change any flags or copy any learned data to other profiles.  This is different from the button in bqEasy, which resets everything and sets flags and other learned parameters to be ready for production.

So...

Temporarily this is what I recommend you do to manually "goldenize" your file after a learning cycle:

After you get a successful learning cycle where Update Status is 02 (for system-side gauges like bq27510 or bq27520) or 06 (for pack-side gauges like bq2754x or bq2774x), you will need to copy the learned data to a few other locations and change some flags.

1. The system-side gauges have four Qmax locations.  Find the one that updated from your learning cycle (probably Qmax 0) and copy it to the other three locations (Qmax 1 as well as two more Qmax locations within the OCV Tables).  Pack-side gauges only have one Qmax, so just leave it as the learned value.
2. Change Update Status values.
   1. For system-side gauges, change all four Update Status values to 02.
   2. For pack-side gauges, change Update Status from 06 to 02 if you plan to send the IT_ENABLE command in the pack production line or leave it at 06 if you want to leave IT enabled and skip sending the IT_ENABLE command in production.
3. Update Ra tables
   
   1. System-side gauges:
      
      1. There are six Ra tables (Def0, Def1, Pack0, Pack1, Pack0 Rax, and Pack1 Rax).
      2. Both Pack0 Ra or Pack0 Rax should have been updating during the learning cycle.  Find the one with the most recent update (Ra Status and Ra Flag will be 00 / 55) and copy the values of Base, Gain, and all points in the table to the other five tables.  Change all Ra Status and Ra Flag values to 00 / 55, except leave the two Rax tables at FF / FF.
   2. Pack-side gauges:
      
      1. There are only two Ra tables.  Find the one with Ra Flag = 0055 and copy those values to the other table.
4. Change any cycle counters or reset counters to 0.
5. Update any Manufacturer Info Block data to your desired values for version control or other purposes.
6. Update any other parameters you want, for example
   1. Change calibration values to the average values taken from samples of your actual production boards.
   2. Set Avg I Last Run and Avg P Last Run to typical system load values so that the gauge will initialize using those loads for predictions.
   3. Set SOH Load I to either a typical system load or a maximum system load.

That looks complicated when typed out, but it's basically just reviewing all of your parameters before exporting the golden file.  Let me know if you have any questions.

dMax,

If I am to manually "goldenize" the file, am I to use bqEasy in  bq EVSW?

With bq EVSW, I have the overall notion of the process you had given except for the cycle counters or reset counters. Where are these counters located?

Thanks in advance,

Kat

Hi Kat,

Just use the DataFlash tab to update the parameters manually.  Click on the one you want to change, type the value, then hit the Enter key to write it to flash.  Then you can read out the SENC file using I2C Pro once you have modified any values necessary.

Some products have reset counter and cycle counters, and some don't.  If you see any in dataflash then change them to 0.

dMax, 

Thanks, we will schedule another characterization and follow your instructions. We'll keep you posted.

Thanks,

Kat

dMax,

We started the characterization yesterday. I had attached here 3 screenshots (DataRAM, DataFlash.Gas Gauging and DataFlash.Configuration). Let me first explain the process we did.

1. Using bq27510-g2, we programmed to bq27510.g3 FW (bq27510-G3_4_00.senc downloaded from TI)in I2C Pro.
2. Verified Data Flash.Configuration.Data.Device Name = bq27510-g3
3. proceeded with bqeasy
4. Entered Project name and configuration data.
5. Used averaged Calibration data (same data as average calibrated g2 data)
6. Chem ID selected manually (chem id is 1134)
7. Proceeded with the Learning Cycle
8. Used and Electronic Load to Automate Discharge and wait for 5 hours
9. Manually initialize Impedance Track Algorithm
10. Manually Charged the battery and wait for 2 hours
11. After step 10, we selected "Charge/Wait Done" button. Pop up showed stating FC bit is not set.(provided in the screenshot)
12. We browsed thru the DataFlash and the Device Name is "þ" instead of bq27510-g3 (provided in the screenshot)
13. The setup is currently held on a "pause". (PC, FG, EV2300 are still interconnected and bqEVSW is still running)

Our concern here is that we had placed a halt midway through the learning cycle due to FC bit not setting. In addition, the SoC had registered to be -30343% and lastly, the device name had changed to þ.

What could be our best and fastest step in resolving this issue? We are still hoping to continue the characterization and learning cycle. Is there a way to bypass the error by manually encoding register values which would allow us to proceed with the characterization?

Thanks in advance,
Kat

4505.screenshots.pptx

Restart EVSW and you should be able to continue. 

To add more background:

Your screenshot shows that your EVSW still expects to be talking to a bq27510-G2 (v1.23).  After you upgraded your IC to bq27510-G3 (v4.00), you need to restart EVSW so it loads the correct register and dataflash mapping for bq27510-G3.  What your screenshots appear to be showing is that you have a register mis-match where values from one register are being reported under a different name.  You will notice lots of crazy values in DataRAM and DataFlash.

In the top title bar of the EVSW window you can see what product and FW version the tool was loaded to talk to.  In the very bottom of the window (when you are on the DataRAM screen, at least) you can see what product and FW version was read from the gauge at the moment the EVSW was started.  They should both match, unless you intentionally chose to load a different FW version at startup.  If you don't have a gauge plugged in when you start EVSW, it will pop up a window to let you choose which product/FW you want to load EVSW for.  In that case you would probably see zeros at the bottom of the window since there was nothing to talk to at the launch of EVSW.  Even if you subsequently connected a gauge to talk to, it will not update the info at the bottom.

The bottom line is that neither the top title bar nor the bottom info will update dynamically.  Anytime you change FW versions it is a good idea to restart EVSW.  Some of our newer tools will try to dynamically update when you change products/FW, but it's always safest to restart the tool to be sure everything is loaded freshly for the corresponding gauge.

Now, for your current situation, you should be able to simply restart EVSW and continue your learning cycle from where you left off.  The gauge doesn't care what it was talking to.  The algorithm inside the gauge is still progressing through the conditions for learning.  You may have trouble with the bqEASY wizard if you were going step by step, but the learning cycle tab in there is simply a forced checklist that you can follow yourself.  You can skip around in bqEASY if necessary and if you generally know what you are doing. 

If you have any trouble generating your golden file, you can upload your log file and gg file here and we can help to complete the process.