Migrating from PPI to DPPI

RE: Migrating from PPI to DPPI

Yaxit — Mon, 03 Mar 2025 14:10:15 GMT

Sorry for the long wait. I finally had the change to try this approach with the EGU and it works well. I will close the ticket, thanks for the help!

RE: Migrating from PPI to DPPI

Yaxit — Fri, 14 Feb 2025 13:31:40 GMT

I did not know of this peripheral, but it seems the layer of indirection that I need, I'll have a try and come back to you to confirm!

Thanks!

RE: Migrating from PPI to DPPI

Yaxit — Fri, 14 Feb 2025 13:30:44 GMT

Hi, I have not tried this yet but I too had the impression that it would not help. The EGU solution seems interesting though, I'll try that first.

Since the behaviour of groups is not very clear, it might be still useful to try it out and then update the documentation?

RE: Migrating from PPI to DPPI

Emil Lenngren — Fri, 14 Feb 2025 13:24:25 GMT

I think you should be able to use an EGU peripheral to form additional "edges" between channels (up to 16 per EGU instance). From what I see, DPPI groups as Maria suggests are more meant as a way to be able to enable/disable channels also from DPPI events, but not a way to give flexibility to event/task connectivity.

Example:

DPPI channel 0: e1 -> t1 & EGU.TASKS[0]

DPPI channel 1: e2 & EGU.EVENTS[4] -> t2 & EGU.TASKS[1]

DPPI channel 2: e3 -> t3 & EGU.TASKS[2]

DPPI channel 3: e4 & EGU.EVENTS[5] -> t4 & EGU.TASKS[3]

DPPI channel 4: EGU.EVENTS[0] & EGU.EVENTS[2] -> t5

DPPI channel 5: EGU.EVENTS[1] & EGU.EVENTS[3] -> t6

DPPI channel 6: e5 -> TIMER3->CLEAR & EGU.TASKS[4] & EGU.TASKS[5]

Then also configure a SHORTS from your TIMER4 CC0 to TIMER4 CLEAR.

See the following illustrature. Circles are DPPI channels and the horizontal lines are EGU task/events pairs.

RE: Migrating from PPI to DPPI

Yaxit — Fri, 14 Feb 2025 12:14:27 GMT

For now in software. Later I would like to stop and clear tim3 on the e5 (tim4 cc0).
However again, the issue is mainly with the GPIOTE tasks. Clearing the timer would just add a few other tasks to the event e5.

RE: Migrating from PPI to DPPI

Emil Lenngren — Fri, 14 Feb 2025 12:07:26 GMT

At what point are you clearing (restarting) the timer instances? By ppi or from cpu?

RE: Migrating from PPI to DPPI

Yaxit — Fri, 14 Feb 2025 12:05:27 GMT

- e1 to e4 are TIM3 CC0-3
- e5 is TIM4 CC0

- t1, t3, t5 are GPIOTE set
- t2, t4, t6 are GPIOTE clear

RE: Migrating from PPI to DPPI

Emil Lenngren — Fri, 14 Feb 2025 11:43:23 GMT

Exactly which events are you listening to? Which event is which cc and do you have any shorts from a cc to clear (which in that case)?

RE: Migrating from PPI to DPPI

Maria Gilje — Fri, 14 Feb 2025 11:34:58 GMT

Hi Andrea,

[quote user="Yaxit"]The problem I have here is that some tasks should be triggered by multiple DPPI channels (e.g. t6).[/quote]

The only possible option I see is to use the many-to-many connection and set up channel groups for each event-tasks group. I am not sure if that will allow you to have one event publish to several channels or have one task subscribe to several channels, though. It is not clear from the documentation and I have not found anyone internally who's tried this yet.

You may have already tried out the channel groups?

Best regards,

Maria

RE: Migrating from PPI to DPPI

Yaxit — Thu, 13 Feb 2025 13:01:15 GMT

Thanks. As far as I can see, this use case is different from the example you linked. I'm familiar with how to connect multiple events and multiple tasks to the same DPPI channel. The problem I have here is that some tasks should be triggered by multiple DPPI channels (e.g. t6).

This was achieved in PPI with forks, but I cannot reproduce it with DPPI. Looking forward to your solution!

Best,

Andrea

RE: Migrating from PPI to DPPI

Maria Gilje — Tue, 11 Feb 2025 15:33:50 GMT

Hello,

Thank you for your patience so far.

This reply from a few years ago seems promising: RE: Using the DPPI on the nRF5340 to enable the ADC and GPIO toggling based on a hardware timer

I will look into it more as well. I'm not experienced with (D)PPI yet, so I appreciate your patience while I figure things out.

Best regards,

Maria