BQ78350: Considerations for Impedance track vs. CEDV guage for UAV application

Hello Nico,

This question has been assigned and a response is being worked on.

Thank you,
Alan

I also noticed over the weekend that for our design capacity of 11.25 Ah and pack voltage of 49V, our design energy of 55,125 cWh does not fit in the register that stores design energy. I went ahead and set it to the max allowed value of 32767 cWh. I am assuming I can just ignore this value and it will not affect the SoC, SoH, or other critical battery state parameters. I am also assuming that I can simply calculate the remaining energy on a seperate MCU using the pack voltage and remaining capacity values. Are these assumptions correct? Does this change your fuel gauge/AFE pair recommendation?

In general, CEDV is great for applications with constant discharge scenarios from cycle to cycle. The gauge learns Full Charge Capacity by measuring passed charge from a full charge condition to the time the battery voltage drops below a voltage threshold. It uses this learned FCC as a reference for SOC = 100* (FCC - coulomb count)/FCC.

Learning requires a substantial discharge (93% by default) and it's limited to temperatures between 10deg.C and 40deg.C. Applications that don't discharge that deeply or operate outside this temperature won't work well.

Also, CEDV doesn't pro-actively compensate for changes in conditions so if there's a low load discharge followed by a high load discharge, SOC can be significantly inaccurate.

Impedance Tracking uses a model of the cell to calculate FCC and Remaining Capacity. This makes it possible for the gauge to adjust FCC and SOC for changes in conditions, like load and temperature. It uses prediction instead of past measured passed charge for FCC so it's "forward" looking. This works well if the application allows the gauge to learn properties of the battery's cells. Like Qmax and cell resistance. This requires that cycles are compatible with the rules for Qmax measurements (not every cycle but often enough for the gauge to learn Qmax) and load profiles that allow the gauge to measure cell resistance (this works well for sustained discharges with a an averaged constant load - it does not work well if the discharges are short or very low load).

You can use current scaling for both CEDV and IT if you use a cell that's larger than the numerical limits of the algorithm allow. For example, if you have a 100000mAh cell, you can configure the algorithm for a 10000mAh cell and scale every current and capacity dependent parameter by a factor of 10.

Hi Dominik,

So it seems that CEDV is not the right for this application due to the following: the battery may/ may not be fully discharged between charge cycles, our operating temperatures will most likely fall outside of 10C-40C, and finally our discharge delta being 13A which may cause SoC inaccuracies. Please confirm that these statements are true and CEDV is not recommended for this application.

In that case, I believe Impedance Tracking is the way to go. I do have some follow-up questions though:

1. "Like Qmax and cell resistance. This requires that cycles are compatible with the rules for Qmax measurements" - Can you list the rules or link me to a document that explains them?

2."load profiles that allow the gauge to measure cell resistance (this works well for sustained discharges with a an averaged constant load - it does not work well if the discharges are short or very low load)" - Does the load profile I described in the original post adhere to this statement? What is considered a "short" discharge?

I am currently using BQ78350 paired with BQ7694000, can you recommend a similar combination that uses Impedance Tracking instead of CEDV, pairs with AFE chip with functions similar to BQ7694000, and is compatible with 14S1P cell configuration?

#1: The main rules are:

* Charge above the flat zone of OCV. This depends on the chemistry.
* Discharge below the flat zone of OCV.
* Let the cell relax until voltage is stable after discharge. This usually takes around 30 minutes for a load current that's less than C/25 but it depends on the chemistry.
* Minimum change in DOD is 0.37 (charge or discharge). This means that 37% of Qmax must pass (charge or discharge) between OCV measurements (or end of charge).
* Temperature must be between 10deg.C and 40deg.C when the gauge takes OCV measurements for Qmax updates. It can be outside these limits during active charge or discharge.

#2: A short discharge is anything less than 10 minutes. For example, if you have an application that stops discharging (current drops below C/25) after 10 minutes of a load that exceeds C/10, then Impedance Tracking will be able to measure cell resistance with reasonable accuracy (unless the load itself is very dynamic, like if it changes between C/10 and 1C without ever reaching a steady state during active discharge then the resulting resistance measurement will not be of high quality, which can have an impact on applications with an average high load (e.g >= 1C).

Our 4s Impedance Tracking gauges combine the AFE and gauge in one: For example bq40z50: www.ti.com/.../BQ40Z50

Thanks for the explanation in #1.

1. How often does the battery need to run a re-learning process?

2. Can we do this once at time of production in a controlled environment and then at set maintenance intervals?

3. Please confirm in my previous response that "CEDV is not the right for this application due to the following: the battery may/ may not be fully discharged between charge cycles, our operating temperatures will most likely fall outside of 10C-40C, and finally our discharge delta being 13A which may cause SoC inaccuracies."

4. Our typical load profile will be between C/2 to 1.6C. Changing between these values will take 5-15 seconds with steady states of roughly 30s in between. Then when the UAV lands there will be a C/2 to 0 discharge case. Since it never falls below C/10 or C/25 are we outside of the zones where poor impedance tracking occurs?

5. Our pack is 14S, the one you linked is 4S. Can you recommend a 14S IT gauge and accompanying AFE?

#1: This depends on the application, the cell chemistry and how much it stresses the battery. I can't give a concrete number as this doesn't depend on the gauge itself. The algorithm needs to be able to measure Qmax and cell resistance as the battery ages so this could be anything from once ever 7-8 cycles to once a month.

#2: Yes. The controlled environment is what TI considers a Learning Cycle. Several customers use maintenance learning cycles (e.g. rarely discharged battery backup applications).

#3: From the information that you provided, it appears that CEDV is not a good fit.

#4: This most likely will work with Impedance Tracking.

#5: We do not have a 14s Impedance Tracking gauge that can track 14 series cells individually. The only Impedance Tracking gauge that can handle 14s is the bq34z100, which is a lower-end solution without cell balancing. It has the AFE combined into one chip. www.ti.com/.../BQ34Z100

1. What makes the bq34z100 lower end?

2. Since this will run independently from the AFE, sticking with my current BQ7694000 AFE should be fine, correct?

3. I see that the bq34z100 takes in a scaled full pack voltage, is this all it needs as far as a voltage sample is concerned?