BQ34Z100: Four BQ34Z100-Z1 gauges are in parallel. Problem with getting valid RC.

Hi, Steven.

I have added diagram of the system. It contains 4 modules, that include 4 bq34z100 (1 gauge per module).

1: Q: How the I2C of these bq34z100 connected, are the data and clock signals simply ORed with same pull up resistor?

A: I2C pins (SDA, SCL) are connected to I2C-MUX via pull-up resistors.

2: Q: How does the host MCU identify each battery pack?

A: Host MCU switches between every battery via I2C-MUX. So, MCU communicates with bq34z100 in series. No address conflict.
This approach was tested with another gauge and all was fine.

3: Q: Is there any circuit to manage the load allocation for the 4 battery packs connected in parallel?
A: As load I use resistor with constant value (couple of Ohms).

4: Q: Is there any protection circuit for each module?
A: Protection circuit helps against overcurrent, over- and udervoltage cases. When such event takes place, output from battery is cutted off.

Q: How the circuit manage the FETs if other packs outputs are higher than its self?
A: Unfortunately, I don't now much about protection circuitry. Before adding bq34z100 to module all batteries were connected
to equalize them in voltage.
Then was made couple of charge-discharge cycles.

Hi, Steven

1: Could the sense resistors for current measurement of bq34z100 are bypassed somehow?

Current data from every bq34z100 seemed good. Sum of four current from four modules fitted well with Ohm law (load equals to 3 Ohms).
Voltage data, temperature seemed also good.

But with capacity I have a trouble.

2: can you provide a log file with bqstudio when discharging for further inspection?

I have attached charge / discharge log files.

Also, I have an additional question.
During tests with two paralleled modules that has BQ34110 device each (two BQ34110 in sum), RC values from every module were good.
But with BQ34Z100 I have met the problem.

Could it be related with different algos?

"The key to the bq34z100-G1 device's high-accuracy gas gauging prediction is Texas Instrument's proprietary Impedance Track algorithm.", bq34z100 datasheet, p.10.
"The bq34110 gas gauge uses Compensated End-of-Discharge Voltage (CEDV) technology to accurately predict the battery capacity and other operational characteristics of the battery.", bq34110 datasheet, p. 9.

Does TI provide any reference design with couple of gas gauges in one schematic?
Is it approved solution by TI experts?

Can I use some ICs to prevent such a problem with RC in advance. For example, using OR-ings controllers (e.g. LM5050) or eFuse Power MUXes (e.g. TPS25944) to connect modules in parallell?

Thanks much for your help!

8054.bq34z100_data.zip

Hi, Steven

For debugging, I have experimented with just one module standalone and made a couple charge-discharge cycles.
Remaining capacity has no meaningful values. It really changed, not as increment value, but as boolean.
Current, voltage and Q max passed Q in good agreement with test measurements.

Concerning RUP_DIS. Yes, bq34z100 was power cycled. Module was in deep discharge state and protection circuitry cutted off the battery.
Now I test module without power cycle.
For data viewing and logging I use MSP430.

I have attached pics below:
- REGs before module charging.
- charging process.
- charging process.
- REGs after charging.

Hi, Steven

This is nice advice!
The problem was really with SRP and SRN pins. I have switched over them and data from module look like truth now.
I have attached pic, where single module charge process is shown.

Also, I have tied two modules together and start to test. At first view, they work good, no problem as I met before.
I plan to make additional tests and give feedback.

Thanks much for help!