GPIO + MCP23017 input/output numbering

[thermseekr]

Hello,

I have a design where I use two MCP23017 on MODE 2, one for inputs and the other for outputs. Everything works perfectly.

What happens is that i just found about a bug on the MCP23017 die that can disrupt I2C communications when pins GPA7 or GPB7 are used as inputs, so I need to disable them. I will move those inputs to GPIOs 34 and 35 on the ESP32. What I need to know is how the inputs will be numbered in that case, and how to adjust mcp23x.dat. Will the inputs on the ESP32 numbered as IN1 and IN2, and the inputs on the MCP as IN3 to IN16, or the other way around?

Here's my current mcp23x.dat:

{"NAME":"MCP23017 expanders","BASE":0,"GPIO":[128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,224,225,226,227,228,229,230,231,232,233,234,235,236,237,238,239]}

The docs mention using relative or absolute numbering on mcp23x.dat using "BASE":0 or "BASE":1, but I see nothing about mixing MCP with GPIO, so I thought better ask to be sure. If I keep relative ("BASE":0), can I adjust mcp23x.dat as below?

{"NAME":"MCP23017 expanders","BASE":0,"GPIO":[128,129,130,131,132,133,134,0,135,136,137,138,139,140,141,0,224,225,226,227,228,229,230,231,232,233,234,235,236,237,238,239]}

Thanks.
Tales

[sfromis]

Not sure what you mean by "can I adjust". You post a change simply omitting the two pins with the issue you found, and then move the numbers down to keep an unbroken sequence of 128-141 instead of 128-143. What would the possible problem be here?

When it comes to combining with native gpios, the usual way is that native pins gets the first numbers in the order, with others added after that. To be clear, not having worked with MCP23017, I'm not stating this as a "fact", only mentioning a more general approach used by Tasmota. Still, this seems to align with your question, and not "the other way around".

But just try it to confirm how it worked. You can do that without wiring anything, just configure a native gpio, and check if the expander ones moved.

[thermseekr]

Hello there

Not sure what you mean by "can I adjust".

Sorry, it was really not clear. What I wanted to ask, was how to build the template considering that the first two physical inputs will now be on the ESP instead of on the MCP. If I should set the first MCP input to 128 or to 130, or actually to 192 or to 194 as I'm using switches to GND with external pullups.

What I found upon testing:

If I set the MCP template to start on 194, status 10 will show 4 inputs that are not related to the MCP instead of only the two configured on the ESP. I know that because I have all switches configured as switchmode 5, so the ones that are on the MCP, being physically pulled up on the board I'm using for testing, will show as ON in status 10. In this case, I don't know what Switch3 and Switch4 are referring to. And, see that I have only 12 inputs with ON state, instead of the 14 that are configured on the template. One could conclude that the last two MCP inputs would be shifted to Input17 and Input18, but if they exist they are not showing up in the status 10 output.

Status 10 output for MCP template starting in 194
11:43:58.183 MQT: trafar/stat/fararmio066100/STATUS10 = {"StatusSNS":{"Time":"2026-02-10T11:43:58","Switch1":"OFF","Switch2":"OFF","Switch3":"OFF","Switch4":"OFF","Switch5":"ON","Switch6":"ON","Switch7":"ON","Switch8":"ON","Switch9":"ON","Switch10":"ON","Switch11":"ON","Switch12":"ON","Switch13":"ON","Switch14":"ON","Switch15":"ON","Switch16":"ON","AHT2X":{"Temperature":34.7,"Humidity":39.3,"DewPoint":18.8},"TempUnit":"C"}}

When instead I set the MCP template to start on 192, which is the code for input 1, the output of status 10 reflects the actual hardware, with Switch1 and Switch2 being on the ESP (I confirmed), and Switch3 being the first MCP input. Here, I have 14 inputs with ON state, being in line with the actual hardware.

Status 10 output for MCP template starting in 192
11:41:32.271 MQT: trafar/stat/fararmio066100/STATUS10 = {"StatusSNS":{"Time":"2026-02-10T11:41:32","Switch1":"OFF","Switch2":"OFF","Switch3":"ON","Switch4":"ON","Switch5":"ON","Switch6":"ON","Switch7":"ON","Switch8":"ON","Switch9":"ON","Switch10":"ON","Switch11":"ON","Switch12":"ON","Switch13":"ON","Switch14":"ON","Switch15":"ON","Switch16":"ON","AHT2X":{"Temperature":34.6,"Humidity":39.6,"DewPoint":18.8},"TempUnit":"C"}}

Conclusions, for when mixing I/Os between native GPIOs and I/O expanders. Take this with a bit of skepticism, as I did not have time to test thoroughly.

(1) As expected, the I/Os on the ESP will be numbered first.
(2) On the expander template, always start with the code for the first I/O, it will be numbered in sequence after the native GPIOs.

Regards,
Tales