STM I2C-3 not generating a clock and data signal on one of the I2Cs.

[CGrov.1]

I am running an STM32L476RG at 1.8V and have noticed that I2C-3 does not generate a clock or data signal. I have tried isolating the MCU from the rest of the circuit and found that there is still no signal at the MCU pins, so i know the circuit is not loading the chip. I have changed the MCU and still get an identical signal, and I have run the circuit with an external MCU at 3.3V with no problems. Because of all this I a confident the problem is with the MCU, although I2C-1 is working perfectly. When I look at the output of the SCL3 and SDA3 with a scope, i can see the voltae rise to about 0.9V and then slowly drop back down. In comparisson, the SCL1 and SDA1 lines of the working I2C-1 rises to about 2V, and then the clock and data signals start as the voltage level drops back down.

I am wondering if it could be a product of running it at the low voltage threshold? are there any other pins (pulled high or something?) I need to worry about when using the I2C3? I can't find anything in the datasheet about this.

[TDK]

I2C requires an external pullup as pins are driven in open-drain mode. Do you have one on the I2C3 lines?

> Because of all this I a confident the problem is with the MCU

You could be right, but I don't think you've provided enough detail to back up this claim.

[waclawek.jan]

Which pins? As TDK said, do you have external pullups?

Read out and check/compare to working/post content of I2C and relevant GPIO registers.

JW

[CGrov.1]

The I2C does have an external pullup. The reason I believe it is with the MCU is because I can use a separate dev board to drive the circuit with no issues, it is only when I am using the on-board MCU that I have issues. So I am fairly sure that the circuit downstream from the MCU is working correctly. The dev-board is running at 3.3V vs the on-board the MCU which is only running at 1.8V. However, i have tested the dev-board at 1.8V and do not get any issues. I am also able to detect a clock signal on the dev-board right at the MCU pins, on the on-board MCU I still get a weird pulse with no clock signal. I'm thinking that there is something wrong with my layout of the MCU.

[CGrov.1]

I am using pin 8 (PC0_SCL) and pin 9 (PC1_SDA), i have external pullups installed. I setup the initialisation using the CubeMX tool, and it works fine on my dev-board.

[TDK]

If you configure those as open drain output pins, does the voltage level rise to 1.8V?

[CGrov.1]

It only rises to about 1V on each of the pins. If I set PC0, I can see a small increase in voltage on PC1 and vice versa. There is no visible short on the MCU pins.

[TDK]

If it’s truly isolated from everything else except the pullup resistor, seems like a hardware issue. Maybe the protection diodes failed or some other ESD damage occurred.

[CGrov.1]

That's possible, I do use a strap and ESD safe clothing but you never know. The fact that it's appeared on two different MCUs though is surprising. I'll keep playing with it, thanks for your time!

[waclawek.jan]

Check power/ground pins (including VDDA/VSSA) first, physically, measuring continuity *directly* on the pins.

Said pullups, what do they measure on the "high" side?

JW