Genereting a PWM signal in Port 0 pin using GPIOTE + TIMERS + DPP

[Pablo-Focus]

Hi,

I would like to generate a PWM signal specifically on the P0.03 pin of the nRF54L15 using GPIOTE + DPPI + TIMERS.

As far as I’m concerned, Port 0 in nRF54L15 does not have a dedicated PWM peripheral for its pins, but it seems there is a workaround using the peripherals mentioned above.

TIMERS are supposed to be in the PERI PD. I believe I need to initialize two timers: one to handle the “on-time” or duty-cycle duration, and the other to define the PWM period. I also believe I need to connect them through PPIB22 so they can publish events to the LP PD GPIOTE30, which should toggle the P0.03 pin to emulate a PWM signal.

I would appreciate someone could guide me to implement this since I'm not being able to find examples to start with it.

I've seen the GPPI one to one example, yet it does not build for nRF54L15 as it is.

I'm using the nRF Connect SDK 2.9.0.

[AHaug]

Hi,

Just letting you know that your query has been picked up and we will look into it. I expect to have some more substantial questions/follow up comments early next week than this.

Thanks in advance for your patience.

Kind regards,
Andreas

[Simonr]

Hi Pablo

Is there a specific reason you need to generate the PWM signal on the P0 port? I assume you would like to do this only on the low power domain to have lower current consumption, but if you start a TIMER in the peripheral domain that would defeat the purpose since that would wake up the peripheral power domain anyways. 

This does sound possible without me having looked too close into it, but I don't see a reason why since you would have to use the Peripheral domain either way. Can you try to explain your use case here a bit more?

Best regards,

Simon

[Pablo-Focus]

I have a custom board with an LED connected to P0.03, and I want to control its brightness using a PWM signal.

I understand that using GPIOTE + DPPI + TIMERS allows the PWM system to minimize CPU usage, and I would like to control the LED that way instead of using a thread (which I already implemented) that relies on the CPU. 

[Susheel Nuguru]

Hi Pablo, 

I have tried similar thing before and I think you should be able generate PWM on P0.03 using two PERI TIMERs feeding LP GPIOTE over PPIB so the CPU stays idle. TIMERs live in PERI; P0.03 + LP GPIOTE live in LP. While PWM runs, PERI must stay on. If runtime PM turns it off, disable runtime PM for those TIMER devices 

CONFIG_NRFX_GPIOTE=y
CONFIG_NRFX_TIMER0=y 
CONFIG_NRFX_TIMER1=y 
CONFIG_NRFX_DPPI=y
CONFIG_CLOCK_CONTROL=y
CONFIG_PM_DEVICE=y 

I do not have a code snippet for it but the logic would be using DPPI/PPIB to fan out two timer compare events to a single GPIOTE task that toggles the pin, so you synthesize PWM edges entirely in hardware. One timer marks the duty edge, the other marks the period wrap; both events hit the same toggle task, so the pin flips high at duty compare and low at period compare. Because the GPIOTE task is subscribed to those published timer events, no CPU intervention is needed after setup