RE: Stepper Motor Control with Raspberry Pico PIO and DRV8711 driver- Part 6: autonomous PIO with speed control

DAB — Sat, 12 Apr 2025 19:31:49 GMT

Nice update Jan.

You are making great progress.

RE: Stepper Motor Control with Raspberry Pico PIO and DRV8711 driver- Part 6: autonomous PIO with speed control

Jan Cumps — Sat, 12 Apr 2025 19:07:49 GMT

I added an interrupt option. The PIO will raise an interrupt when it has completed the number of steps requested.

It's optional, but useful if you want to do other activities when the motor has done some activities - and you don't ant to rely on calculations/estimations of time passed.

[mention:fc48193383514ad18b3997352ebcdb02:f7d226abd59f475c9d224a79e3f0ec07]

RE: Stepper Motor Control with Raspberry Pico PIO and DRV8711 driver- Part 6: autonomous PIO with speed control

Jan Cumps — Thu, 10 Apr 2025 21:01:51 GMT

It's not possible to debug a PIO, but there's a nice emulator: https://rp2040pio-docs.readthedocs.io/en/latest/index.html

That one allows you to step and watch registers, FIFOs, and simulate in and out (either data or pins)

Here, it's loaded with my stepper design, running on WSL ubuntu:

RE: Stepper Motor Control with Raspberry Pico PIO and DRV8711 driver- Part 6: autonomous PIO with speed control

ahsrabrifat — Thu, 10 Apr 2025 15:30:14 GMT

This is the first time I have come across PIO codes. Are these based on assembly language?

RE: Stepper Motor Control with Raspberry Pico PIO and DRV8711 driver- Part 6: autonomous PIO with speed control

Jan Cumps — Wed, 09 Apr 2025 13:47:36 GMT

> The granularity is still rough. With these slow signals (850 Hz as fastest that my motor can handle in full-step mode), each delay adds 2 extra ticks.
> One tick is added to each half of the pulse. It's a lot. Making it more granular is something for a next exercise ...

I have a solution for this. A compromise for the developer, but works well. No changes to the PIO state machine.

In my first approach, I brough down the clock to a reasonable speed for a stepper motor. By default, it churns out pulses at several MHz.
By dividing the clock, it dropped to a range that makes sense for steppers.
But that has as side effect, that each extra clock tick I steal to make the motor go slower, is rather big.

What I've done now, is make the clock divider programmable by the end user. That way they can choose how fine the delay mechanism acts.
At the cost of having to set a delay that's in the 1000s. Each delay step will add 2 clockticks of the frequency the PIO runs at to each pulse's period.

So if you set a high clock (say: keep PICO clock, or divide it by 2), you have to set high delays, but they are very fine and give precise control.

If you set that divider to a higher value, you may not have to use delay at all, or lower values of delay. But the granularity at each delay step becomes rough.

In the end, there is no efficiency gain for one vs the other. It's a choice.

Here is an example function that steps with delay:

[embed:dc8ab71f-3b98-42d9-b0f6-e21e02a0f8e2:3378b21b-f8fc-4f7d-8da8-dfaadfc4965f:type=c_cpp&text=void%20demo_with_delay%28const%20std%3A%3Aspan%3Ccommand%3E%20%26%20cmd%2C%20uint32_t%20delay%29%20%7B%0D%0A%20%20%20%20pio_pwm_set_delay%28piostep%2C%20sm%2C%20delay%29%3B%0D%0A%20%20%20%20for%28auto%20c%20%3A%20cmd%29%20%7B%0D%0A%20%20%20%20%20%20%20%20%2F%2F%20printf%28%22Steps%20%3D%20%25d%5Cn%22%2C%20c.steps%29%3B%0D%0A%20%20%20%20%20%20%20%20pio_stepper_set_steps%28piostep%2C%20sm%2C%20c.steps%2C%20c.reverse%29%3B%0D%0A%20%20%20%20%7D%0D%0A%7D]

and code that uses it:

[embed:dc8ab71f-3b98-42d9-b0f6-e21e02a0f8e2:c182fe11-e09e-477b-9858-9165d5bffae2:type=c_cpp&text=%20%20%20%20std%3A%3Aarray%3Ccommand%2C%206%3E%20cmd%7B%7B%0D%0A%20%20%20%20%20%20%20%20%7B200%20%2A%20microstep_multiplier%2C%20true%7D%2C%20%0D%0A%20%20%20%20%20%20%20%20%7B200%20%2A%20microstep_multiplier%2C%20false%7D%2C%0D%0A%20%20%20%20%20%20%20%20%7B200%20%2A%20microstep_multiplier%2C%20false%7D%2C%0D%0A%20%20%20%20%20%20%20%20%7B400%20%2A%20microstep_multiplier%2C%20false%7D%2C%0D%0A%20%20%20%20%20%20%20%20%7B250%20%2A%20microstep_multiplier%2C%20true%7D%2C%0D%0A%20%20%20%20%20%20%20%20%7B350%20%2A%20microstep_multiplier%2C%20true%7D%7D%0D%0A%20%20%20%20%7D%3B%0D%0A%0D%0A%20%20%20%20%7B%0D%0A%20%20%20%20%20%20%20%20wakeup_drv8711%20w%3B%20%2F%2F%20wake%20up%20the%20stepper%20driver%0D%0A%20%20%20%20%20%20%20%20sleep_ms%281%29%3B%20%2F%2F%20see%20datasheet%0D%0A%0D%0A%20%20%20%20%20%20%20%20demo_with_delay%28cmd%2C%204300%29%3B%0D%0A%20%20%20%20%20%20%20%20sleep_ms%282000%29%3B%20%2F%2F%20give%20enough%20time%20to%20complete%20the%20action%0D%0A%20%20%20%20%20%20%20%20demo_with_delay%28cmd%2C%207000%29%3B%0D%0A%20%20%20%20%20%20%20%20sleep_ms%283000%29%3B%20%2F%2F%20give%20enough%20time%20to%20complete%20the%20action%0D%0A%20%20%20%20%20%20%20%20demo_with_delay%28cmd%2C%209000%29%3B%0D%0A%20%20%20%20%20%20%20%20sleep_ms%284500%29%3B%20%2F%2F%20give%20enough%20time%20to%20complete%20the%20action%0D%0A%20%20%20%20%7D]

I divided the PIO clock by 2 for this example:

[embed:dc8ab71f-3b98-42d9-b0f6-e21e02a0f8e2:b8b12751-bd72-473f-bf8c-ea6003e40728:type=c_cpp&text=const%20float%20clock_divider%20%3D%202.0%3B%20%2F%2F%20works%20well%20for%208%20microsteps%0A%0A%2F%2F%20...%0A%20%20%20%20uint%20offset%20%3D%20pio_add_program%28piostep%2C%20%26stepper_program%29%3B%0A%20%20%20%20stepper_program_init%28piostep%2C%20sm%2C%20offset%2C%20dir%2C%20clock_divider%29%3B%0A%20%20%20%20pio_sm_set_enabled%28piostep%2C%20sm%2C%20true%29%3B%0A%2F%2F%20...]