Enabling PWM prevents some GPIO pins from working

Official Hardware UNO Family UNO Q
1

I am attempting to use PWM on the Uno Q, and when I include the analogWrite() function in my sketch, it causes some (but not all) of the GPIO pins to stop working. I am not using the pinMode() function for the PWM pins due to this known issue.

Here is my test sketch:

Lines 35 and 36 have the analogWrite() functions for 2 PWM pins (D3 and D11), which I can comment out to see the resulting effect on the GPIO pin behavior.

When a GPIO pin stops working, the GPIO pin always shows a logic low (0 voltage) as read by a DMM attached to the pin. When the GPIO pin works, I see that the pin toggles from logic low to logic high and back again, as expected, using the above test sketch.

The PWM pins always work as expected, when they are either enabled or disabled, by either including or commenting out the analogWrite() functions, respectively.

If I comment out both of the analogWrite() functions, then the GPIO pins D2, D4 to D10, and D12 and D13 all work correctly and toggle low/high, as programmed in the sketch.

If I include the analogWrite() function on line 35 for pin D3, then the following GPIO pins stop working: D6 and D8

If I include the analogWrite() function on line 36 for pin D11, then the following GPIO pins stop working: D5, D12, D13

If I include both of the analogWrite() functions on lines 35 and 36, then the following GPIO pins stop working: D5, D6, D8, D12, D13 (basically a combination of the two).

I have a Google spreadsheet that summarizes the results here:

The first two columns of the spreadsheet show whether the PWM function is Active (uncommented) or Inactive (commented out) and the remaining columns show whether the GPIO pin works (0/1) or does not work (0).

In rows 9 and 10, I've included the STM32 Timer block and channel number, respectively, from the Zephyr DTS. It seems like there is a correlation between the enabling of PWM on a Timer block and the problematic GPIO pin that appears to share the same block (?).

Am I doing something wrong here? Is this a known issue? I could not find anything in the documentation or the forums about this behavior.

I ran into this problem when helping a friend get his Arduino Motor Shield to work with the Uno Q. The shield appears to use a fixed pin mapping, so changing pins here is not an option. My test sketch here models the pin assignments of this motor shield (direction, brake, pwm).

I am using App Lab version 0.7.0

image

2

Note: I have not tried your sketch, but there are several changes in the current sources trying to update a pins mode when you change which logical device/usage the pin will have.

That is when you do an analogWrite on a pin, it should switch that pin to the appropriate Alternate function to support PWM... if you then do pinMode(... it will switch it to either digital in or digital out...

The current sources (ArduinoCore-zephyr) was marked with a new tag on Friday, so I suspect they will do a board release very soon. So you might want to try it out when this release comes out.

Note: Your sheet is marked as being private.

3

Thanks for that suggestion. I updated the sketch to re-initialize pinMode() for a digital pin after every analogWrite() invocation for PWM, and that appears to cause all of the digital pins in the sketch to toggle correctly.

If you are talking about this PR, I definitely do not have those changes yet, I am still on tag 0.55.0.

For whatever reason, this forum web site shows the spreadsheet as private, but I've seen anonymous browsers in the sheet, and it was set up as world readable.

6

This particular problem has gone away with the release of 0.55.2 ArduinoCore-zephyr. My sketch now works as expected with no required workarounds.