TCA6408A: Can't read correct register value after sending cmd

Hi Richard,

How are you reading from the I/O Expander? Do you have a scope trace of the SDA & SCL signals when reading from the device? 

Regards,

Tyler

Hi Tyler,

I didn’t use an oscilloscope to measure the I2C waveform; I simply used a multimeter to check whether the output levels of the IOs on the IO expander matched the H-L levels I had configured.

Any idea about my problems?

Thanks

This is I2C waveform for your reference, please help to check it. Thanks

Hi Richard,

The waveform looks correct. You are reading data from the configuration register at 0x03. It is 0x00 in data. After a power reset, it should be 0xFF. 

What does your power cycle waveform looking like? Can we confirm you are following the correct power sequencing? 

Finally, does an A-B-A swap test confirm that the issue follows the IC? Take a known working board, and replace the TCA6408A with the suspect unit. Does the board start failing? 

Regards,

Tyler

Hi Tyler,

Could you please help confirm whether the section in the middle (highlighted in the red box) is okay?

thanks.

Hi Richard,

We usually recommend a repeated start so that the MCU does not lose control of the read operation in circumstances of multi-master environments. 

A stop condition, followed by a start condition should still work. The command byte (pointer byte) is already set in the first half of the transaction. The second have is sending the device address and read bit of which data should be read from the selected command byte which is the configuration register at 0x03. 

Can we confirm on power cycling? 

You could try using a repeated start condition to see if that changes the outcome, I don't think this will change the outcome. 

Regards,

Tyler

Hi Tyler, thanks.

Can you teach me how to confirm on power cycling? actually I don't really know what is "power cycling" means.

Thanks

Hi Richard,

For sure. I am looking at figure 10-1 mostly. 

This is approximately the structure of what the VCC pin voltage should look like when resetting the power on the device. 

Are you resetting the TCA6408A through the /RESET pin or are you toggling the voltage on VCC for a power-reset? 

If you are resetting through the pin, can you scope the /RESET pin while you toggle? 

If you are resetting through the supply pin VCC, can you scope VCC pin while you power cycle? 

Regards,

Tyler

Thank you. I will proceed with the measurements next week.
By the way, may I ask—does that mean an incorrect power cycle or reset sequence could potentially cause the issue we’re seeing?

Hi Rich,

Given the datasheet default values for the configuration register, you should be reading 0xFF instead of 0x00. 

If you were to set the config register to 0x00 - all outputs, then conduct a /reset, you should see the register go back to 0xFF upon the read command. 

If an incorrect power cycle occurs on the VCC pin, you may think that you are resetting the device, but in reality, the device never reset because it had improper power-on-reset. 

This could also occur on the /RESET pin. It is possible that when you toggle /RESET, it does not go below a VIL. You would have to verify both of these tests on a scope to see how the voltage is behaving on the /RESET and VCC pins when toggling them. 

Regards,

Tyler

Hi Tyler,

May I ask — in my current circuit, both VCCP and VCCI are connected to the same 1.8 V rail, meaning their power-up timing will be exactly the same. Could this cause any potential issues?
If so, would it be possible to recover by toggling /RST?

On the other hand, if the /INT function is not required, can it be left floating, or does it still need a pull-up resistor?