Fix multiple servos with STM32

[MangaValk]

Description

Fixes issue #16138

Depends on: stm32duino/Arduino_Core_STM32#822

Benefits

fixes multiple servos not working on stm32

[sjasonsmith]

Thanks for looking into this.
This is failing because of course it depends on your own customizations in the STM32 framework.

I think you can avoid these by adding a constructor and storing the information you need inside the libServo object.

I'm attaching some files where I experimented with this. I haven't tested these, so they might need some work. I believe they will achieve your needs without requiring any changes to the STM32 framework.
Servo.zip

[sjasonsmith]

I am confused by this. TIM6 is a basic timer, and TIM2 is a General-purpose timer.
This contradicts the comment in the line above. Do you know why this is working using the less functional timer?

[fpistm]

Hi all,
the comment is a legacy one. Since HardwareTimer library has been deployed, this comment is no more relevant.
Using TIM6 is a good option like it is not linked to a pin.

[sjasonsmith]

In that case we may need to check whether TIM2 is attached to any pins that might cause problems for the TONE_TIMER. Perhaps you are using some of the pins which attach to TIM2 as PWM outputs, and those are interfering with the timer operation.

(Sounds like the comment should be removed as well)

[MangaValk]

I'll give your change a try next weekend when I have time again :)

About the timers, I have no idea, I'm not a stm programmer. I just know c++ and mess around some till it works. (php programmer by day)

Using tim2 the servos stoped working when using the tool changer mid print.
With tim6, stolen from the speaker, they keep working.

[sjasonsmith]

I can help debug this, but I also won't be available to do so for a week. I don't have a tool changer, but I could probably reproduce just by tracing output on the servo lines. Right now it seems like we can't really explain why the change works, which concerns me.

Given the way the underlying servo code works, I don't really understand why one servo would work if the timer was insufficient. The manner in which the timer is used does not change by having more than one servo.

[MangaValk]

Thanks for looking into this.
This is failing because of course it depends on your own customizations in the STM32 framework.

I think you can avoid these by adding a constructor and storing the information you need inside the libServo object.

I'm attaching some files where I experimented with this. I haven't tested these, so they might need some work. I believe they will achieve your needs without requiring any changes to the STM32 framework.
Servo.zip

I had 5 minutes to check and your fix works!

[yopla31]

For your information, I had the same anomaly, resolved by applying this fix (copy Servo.cpp and Servo.h into directory stm32). No side effects observed until now.
Thanks for your work.

[yopla31]

Unfortunately I observed a side effect :
My X and Y axes are configured in "Sensorless Homing" with stall sensibility to 50 et 55 for X and Y, the Z axe homing with BL-Touch. If I run M502 and M500 then a G28, in 90% of the cases the probe does not deploy and the hotend crashes on the bed. Sometimes the homing of the Y forces like with an incorrect stall value or the homing of the X does not go as far as the stop.
The anomaly trigger seems to be the sequence M502 + M500, then G28.
Attached my configuration files.
SKRpro11_CFGv051.zip
I complete my observation:
After having to reconfigure the standard firmware (one said servo BL-Touch), I conducted a series of tests of this sequence, and the result is the same.