INA228: compatibility with ISO1640

Dejan Hrovatin1

Part Number: INA228
Other Parts Discussed in Thread: ISO1640, ISO1641, , INA233, BQ76942

Hello,

I'm working on a circuit based on STM mCU, with some I2C devices used. One INA233 and one INA228 device is placed on mCU's I2C side, wheras BQ76942 is used remotely. Its I2C lines are isolated from mCU's I2C lines with ISO1640 isolator. ISO1640 isolator is used instead of ISO1641, because ISO1640 supports also devices that perform clock stretching and BQ76942 performs clock stretching.
A problem is that Vol1 of ISO1640 is 570 mV (min.) and 710 mV (max.) which is higher than VIL(max.) of both - INA233 and INA228 that equals 0.4 V.
Since INA233 and INA228 would just listening what BQ76942 will replying (via ISO1640) to the mCU, could this lead to glitchy behaviour of INA233 and INA228 devices? mCU on the other side have VIL (max.) higher than Vol1 of ISO1640, so there will be no a problem with mCU.
Below is a schematic:

I must say that I already have some circuits which are simmilar - there are also INA233 used, but ISO1641 is used instead of ISO1640. However ISO1641's SDA line has also very high VOL on the side 1, which is higher than VIL of INA233, but it looks like that everything works without problems.
Can you please confirm that because INA233 and INA228 will just "listening" what BQ76942 IC will replying to mCU via ISO1640, this would not lead to glitchy behaviour of INA233 and INA228?

Many thanks,
Kind regards,
Dejan.

4 months ago

0 Clemens Ladisch 4 months ago

Guru 244390 points

The INA2xx's I²C inputs are designed to work with 1.8 V MCUs, so their switching thresholds are somewhere between 0.4 V and 1.2 V. There is no guarantee that the actual threshold is above 0.71 V.

When the bq76942 does clock stretching, the rising edge of SCL is detected too early from the INA2xx's point of view. This can happen only after the adress byte has been sent, so the INA2xxs should ignore it.

0 Dejan Hrovatin1 4 months ago in reply to Clemens Ladisch

Prodigy 220 points

Hello Clemens,

I'm thinking about another problem. Any zero generated by ISO1640 isolator would be higher than VIL(max) of INA233/INA228, because VOL(min) of ISO1640 is higher than VIL(max) of INA233/INA228. However, this would happen only when BQ76942 will replying to MCU, so INA233/INA228 would just listen to this signal. I'm worry that only just listening to "undefined" signal could lead to glitching behaviour, or this would not be a problem?

Many thanks, Kind regards, Dejan.

+1 Clemens Ladisch 4 months ago in reply to Dejan Hrovatin1

Guru 244390 points

During normal operation, the clock is generated only by the MCU. When the INA2xx is not being addressed, the SDA values do not matter.

0 Mohamed Hamida 4 months ago in reply to Clemens Ladisch

TI__Expert 5352 points

Thank you for the answer Clemens! For now I will close this thread, if there are any more questions feel free to open it back up.

Best,

Mohamed

0 Dejan Hrovatin1 4 months ago in reply to Clemens Ladisch

Prodigy 220 points

This means that I2C buffer is not needed here, which is great.

Thanks for support!

0 Mohamed Hamida 4 months ago in reply to Dejan Hrovatin1

TI__Expert 5352 points

Thank you Clemens!

Amplifiers

Amplifiers forum

INA228: compatibility with ISO1640