FastPWM timer re-enable

Question

FastPWM timer re-enable

gke opened this issue 9 years ago · comments

Should the timer re-enable in pwmWriteSyncPwm be delayed until all channels for that timer, or better all timer channels are set?

dongie · Answer 1 · Mon May 18 2015 11:40:21 GMT+0800 (China Standard Time)

I don't know, what do you think?

Prof. Greg Egan · Answer 2 · Mon May 18 2015 12:27:59 GMT+0800 (China Standard Time)

Not absolutely sure but the issue relates to Fiendie's branch where I think the one-shot code is currently. I see anomalies in my logic analyser trace and these seem to go away when the re-enable is done outside the motor setpoint update loop that calls pwmWriteSyncPwm. I just re-enable all of the PWM timers at once.

dongie · Answer 3 · Mon May 18 2015 12:29:16 GMT+0800 (China Standard Time)

pretty sure cumshot stuff is in main branch now, but I'd be surprised if
anyone uses or cares about it.

On Mon, May 18, 2015 at 1:28 PM, gke notifications@github.com wrote:

Not absolutely sure but the issue relates to Fiendie's branch where I
think the one-shot code is currently. I see anomalies in my logic analyser
trace and these seem to go away when the the re-enable is done outside the
motor setpoint update loop that calls pwmWriteSyncPwm. I just re-enable all
of the PWM timers at once.

—
Reply to this email directly or view it on GitHub
#272 (comment)
.

Prof. Greg Egan · Answer 4 · Mon May 18 2015 12:35:55 GMT+0800 (China Standard Time)

I am definitely NOT a oneshot believer, trust me, just curious. All seems like solid gold wiring on sound system bull. My UAVX code just runs async and at around 500Hz it is good enough ;). I think I do believe in spin up and spin down symmetry though.

dongie · Answer 5 · Mon May 18 2015 13:26:03 GMT+0800 (China Standard Time)

I'm pretty sure without external triggering there's no way to guarantee
separate timers to start at same time.

But I've not looked into it for all the reasons you've already stated.

On Mon, May 18, 2015 at 1:35 PM, gke notifications@github.com wrote:

I am definitely NOT a oneshot believer, trust me, just curious. All seems
like solid gold wiring on sound system bull. My UAVX code just runs async
and at around 500Hz it is good enough ;). I think I do believe in spin up
and spin down symmetry though.

—
Reply to this email directly or view it on GitHub
#272 (comment)
.

Andreas Tacke · Answer 6 · Mon May 18 2015 16:07:37 GMT+0800 (China Standard Time)

@gke, the stuff was merged long ago. There is some real data about the "oneshot125" stuff. The gist of it is that the "oneshot" part (SyncPWM) actually yields some improvements but the "125" part, namely FastPWM, is questionable at best. Which is why we split it up into two features in the first place. SyncPWM also works with any ESC. A little more than "anomalies" would also be pretty helpful to narrow down if there really is a problem :)