Difficulties producing a credible golden pack (bq20z80a-v110). Qmax updates but update status remains at 04

Craig,

Update Status should change to 05, if a Qmax update occurred. It is odd that MaxErr changed to 5% as it should have changed to 3% with a Qmax update. It will normally change to 5% when the Ra-table update occurs before the Qmax update. Can you provide a log file and gg files for me to review?
Tom

Tom:

Thanks for weighing in. We will try to produce a log file starting with a virgin circuit board that includes all steps as we go through the steps in bqEASY trying to create an engineering perfect pack.

In the meantime

1) One bit of new information from the pack that exhibited the symptoms described above. I manually changed  the update status from 04 to 00 using the Data Flash screen and generated a .gg. file containing the best learned info we had been able to generate to date. I was operating under the incorrect belief  that changing the update status to 00 turned off impedance tracking and so I set out to turn it back.on after creating the .gg. Instead of reversing the process and manually changing the update status, instead I used the means to turn on impedance track I was most familiar with. I entered  0041 into manufacturer access to reset the bq20z80a-v110, and then entered 0021 to turn on impedance track. The result was that all values became the expected values!  Update status changed from 00 to 06 and max error changed from 5% to 1%.

2) Regarding "update status should have changed to 05" if a Qmax update occurred - From slua364: "The data flash constant DF.Update Status increments by 1 when the first Qmax update takes place (e.g., from 4 to 5 if no resistance updates were made, or from 5 to 6 if a resistance update was made). SBS.Max Error( ) becomes 5% in the first case and 1% in the second case." I am unclear when Qmax and resistance updates are expected during a 1st learning cycle. In the symptomatic pack, the Ra values changed, but the Rax values remained at the default values even after many cycles.

Following the steps in bqEASY on a virgin circuit board, what is the expected behavior of the update status, maximum error, Qmax and Ra tables at the end of the 1st discharge, and at the end of the first cycle i.e. 1st discharge followed by first charge from near empty?

My sense in all this is that the bq20z80a-v110 is not functioning properly and the reason probably has something we did or didn't do correctly trying to follow bqEASY to produce our engineering perfect pack and golden image files. 

Craig

Hi Craig,

how are you getting on with this? I usually point people to the learning cycle outlined at top of page 2 of http://www.ti.com/lit/an/slua334b/slua334b.pdf?keyMatch=slua334b&tisearch=Search-EN

for "When no close QMAX value is known" as this seems most reliable.

In this process you first discharge battery to VTERM and then let it rest 5 hours before beginning the 'real' learning cycles, i.e. you want to start learning from a discharged state. You then turn on IT ( by sending 21) and start the learning process by charging to full ( Note for this first cycle you must charge >90% of design capacity for QMAX update).  So QMAX should update on the first charge phase.  Resistance table updates  only take place during discharge and so consequently after you charged to full you when you then discharge you should get the resistance table updates and max error goes to 1%. So everything happens in the expected order

Based on what was described above I wonder if in your first problematic attempt to learn you started by enabling IT and doing a discharge which might be why it seemed the resistance tables updated first.

Note the >90% passed charge on first cycle is important. On subsequent cycles for QMAX update you only have to pass 37% passed charge but must rest outside DQ zones. The other important detail is the discharge current during resistance table update , you want >C/10 rate.

Mike 

MIke:

According to TI documentation, it is not correct that the passed charge during the 1st learning cycle must be greater than 90% of QMAX. Passed charge must be greater than 90% of the value entered for design capacity. The distinction is essential since there is a good chance passed charge will not exceed 90% of QMAX under recommended learning cycle conditions while the User can assure that passed charge >90% of design capacity. Please verify the actual  criterion. 

QMAX is a value derived from a table of open circuit voltages generated by TI under standardized conditions for the particular cell (Boston Power Sonata 5300). According to TI documentation, one of these standard conditions that assures that passed charge during learning will be less than QMAX is that QMAX is the capacity at a discharge rate of C/20 while the TI recommended discharge rate during learning is C/5. Other undocumented standard conditions for the Qmax that could cause passed charge during learning to be further reduced relative to QMAX includes end-of-charge voltage (4.1 vs 4.2 v), end-of-discharge voltage (2.5 vs 2.75 v), and taper current (C/20 or greater).

Above you state that the value VTERM affects whether or not the 90% criterion is met. TI documentation makes it clear that VTERM affects the values displayed for full charge and remaining capacity when full and the gas gauge display. TI documentation is not clear whether the "passed charge" required to meet 1st learning cycle requirements is actual total passed charge or is equivalent to "full charge capacity."but if the latter one would think the documentation would say so. Please provide actual  criterion. 

Note 1: In our testing, we made sure the reported full charge capacity during learning cycles exceeded 90% of design capacity. Since passed charge is at least equal to full charge this requirement was met.

Note 2: The initial values for QMAX that were entered were the values that QMAX's settled at in a battery pack that was cycled multiple times i.e. close to true.

Craig

PS - The document you referenced, slua334b, includes references to chemistry "senc" files. Am I correct that chemistry senc files have been replaced with some other form of chemistry file and that the recommended order of events in this document is obsolete?

Hello Craig,

not sure where the confusion is coming from. Who said 90% of QMAX? . In my previous post I clearly said

" Note for this first cycle you must charge >90% of design capacity for QMAX update"

which is stated n section 2.3 of slua364.

Not sure I get the point of the second paragraph.  Not saying your incorrect , just what are you trying to say ?

In your third paragraph I think you are mininterpreting what I said about VTerm, I was just making the point to be sure you are fully discharged to Vterm before you start to  charge or else you may not be able to meet the 90% passed charge requirement. Maybe we should have a call, I'll drop you an email with my phone number. I thought the original issue was incorrect update staus but your concern now seems to be QMAX update accuracy?

regards the PS ,  Once upon a time  when there were not many CHEM IDs I am told we used to have individual default senc files for each CHEM ID preloaded with that chemistry data. Now we have a single default senc file that I believe comes with CHEM ID 0100 loaded and if you have a different CHEM ID you just use bqEasy to load that chemistry data to your senc file. So certainly some individual steps described in the document may have changed a little but much of the overall guidance i.e. the learning cycle is still valid and useful as a reference.

 Mike

Hi Craig,

As I have demonstrated while testing with your battery pack, I can get the update status to change to 06 under these conditions: Charge to 4.2V/cell; discharge to 2.75V (cell data sheet end of discharge); discharge at a rate of ~C/10.

After the update status changes to 06, modify the data flash to represent your application charge voltage and terminate voltage and any other data flash parameter changes, then cycle the pack between full and empty a couple of times using your application charger and representative load. This allows the gauge to learn your pack under your application conditions.

After complete, use this to generate your golden file. Are you saying that you have tried this and you have not been successful? Feel free to contact me via email if you would prefer.

- Mark

Mark

So far I have not been able to reproduce your results. I just completed 2 learning cycles at the higher end-of-charging voltage, lower discharge rate and lower end-of-discharge voltage  you suggested. Once again I witnessed updates of QMAX and the Ra table but no change in update status from 04 to 06 after either the 1st or 2nd discharge. I will continue a few more cycles, but am not hopeful. Note: bqEASY will not generate a golden image file with update status at 04.

But it is noteworthy that you confirmed my observation that update status does not behave as described in TI documentation. On your 1st learning cycle you witnessed QMAX updates with no change in update status. 

I need to know the real firmware requirements for update status to change from 04 to 06 during learning cycles in the bq20z80a-v110. You hypothesized that the  "passed charge" requirement on the 1st cycle is actually greater than the TI documentation claims, hence your suggestion we use a higher and more typical end-of-charge voltage, lower discharge rate and lower end-of-discharge voltage. It seemed to work for you, but after 1 try I was not able to repeat your finding. There are too many possibilities to try to work this out by experimentation. A definitive answer resides in the firmware code for this product.

Also I need to know how tolerant this technology is of  some extra milli-ohms in the means used to connect cells to the circuit board (wires).

Craig

Hi Craig,

sorry to hear your still having issues.

update status starts at 0 and gets set to 04 by the action of you sending the IT enable  command. Then the part auto increments it to 06 during the learning cycles, at which point you export it as a golden file and manually edit the 06 back to 02. Then after you load the golden file to a new pack and send that pack an IT enable the update stautus will set back to 06.ie impedance track is disabled on the new pack until you send the IT enable.

Do you have any gg and log files you can share with me from the initial configuration before you sent the IT enable , then after the first charge cycle, next discharge etc . We really need an EVSW  log file of these charge/discahrge operations and accompanying gg files to look at. The EVSW will auto export gg files at regular intervals for diagnostic use and that would be really usefull to do here so we can work out what is happening.

I have just been doing learning cycles plus a  few extra charge/discharge cycles on a 20z95 and everything I tested works on that as I would expect. I am not aware that the z80 should do anything different during learning but I will repeat some learning cycles with a 20z80 board to compare.

Mike