BQ27621-G1: POWERMIN and i2c clock stretching details

Andreas Urke

Part Number: BQ27621-G1

We are currently evaluating the BQ27621 for our low-power application and have some questions we could not figure out via the doc.:

1. There is a clear relationship between the different modes (Normal, sleep, etc.), how the gauge operates, and the current draw by the gauge. However, how does the POWERMIN fit: How does it impact the different modes when set? And what current draw reductions can we expect in the different modes?

2. The clock stretching on all i2c traffic when in SLEEP and HIBERNATE is worrying. With e.g. bq27441 this is specified in the datasheet (Section 9.5.4.4) as <100 us, do you have this number for bq27621? How much is added to this stretch when POWERMIN is enabled (if any)?

Regards

over 2 years ago

0 Wyatt Keller over 2 years ago

TI__Guru** 102355 points

Hello Andreas,

I will take a look into this tomorrow.

Sincerely,

Wyatt Keller

+1 Dominik Hartl11 over 2 years ago in reply to Wyatt Keller

TI__Guru 51346 points

#1: The gauge has several internal oscillators. Some can be turned off in sleep mode and it's possible to control this to some extent. One control is POWERMIN. If this is set to 1, the gauge will turn off the HFO in sleep mode, which saves about 40uA in sleep mode (note that the typical normal operating current in the d/s is too low. It should be around 100uA, which is not a steady number but it will fluctuate during normal operation).

Normal mode: around 100uA
Sleep mode, POWERMIN = 0: typ. 60uA
Sleep mode, POWERMIN = 1: typ. 20uA

The main impact, besides the reduction in current, is that the gauge won't be ready immediately when the host tries to communicate because the HFO is off, hence #2:

#2: The bq27441 uses the same I2C core and oscillator design so the same timing applies to the clock stretching for the bq27621. When the HFO is off during sleep (POWERMIN = 1), the gauge will clock stretch if it detects communication on the I2C because it needs to spin up the HFO before it can respond.

0 Andreas Urke over 2 years ago in reply to Dominik Hartl11

Expert 1460 points

Thanks, that answers all my questions. A quick follow-up: Would the built-in profiles in the BQ27621 match a battery which has two cells in parallel? Should we expect any additional inaccuracy in that case?

+1 Dominik Hartl11 over 2 years ago in reply to Andreas Urke

TI__Guru 51346 points

This will work and effectively cause the gauge to estimate twice the current for the same change in voltage.

If your load is low (so that the voltage barely changes when you apply the load), then the gauge will mostly use OCV measurements to track the state of the (two) cells so the more parallel cells you use, the more OCV measurements will dominate (for the same system load).

0 Andreas Urke over 2 years ago in reply to Dominik Hartl11

Expert 1460 points

Related to this topic of BQ27621 and I2C:

Does the BQ27621 suffer from the same issue as BQ27421 and BQ27441 where it in rare cases erroneously pull down the i2c bus, as discussed in https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/838088/bq27441-g1-scl-line-stuck-low and https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/574055/bq27421-g1-scl-line-gets-stuck-at-low-state ?

+1 Dominik Hartl11 over 2 years ago in reply to Andreas Urke

TI__Guru 51346 points

None of these are regular scenarios and this gauge (and gauges with similar cores) is (are) used successfully in extremely large volumes.

That said, the gauge is effectively a small system on chip with CPU, analog front end, communications interface and not a simple device like an ADC so it's possible that electrical conditions (glitches, ESD events, excessive noise) can cause it to stop working if this affects important internal circuitry (e.g. startup circuit that controls how the different internal components turn on relative to each other, LDO stability, various internal oscillator stability).

It's important that the circuit design won't violate any of the operating conditions that are stated in the datasheet.

Power management

Power management forum

BQ27621-G1: POWERMIN and i2c clock stretching details