How fast are board interupts or polling speeds.

sinned6915

Hi all -

I am trying to work out some homing speed and probing speed ideas. My notion is that there is an optimal speed for RRF to process the trigger signal for homing or probing..

Is there a signal speed or buffer speed reaction time for the polling and intterupt signals? Is it limited by RRF or the MCU?

I am trying to go through the source code and feel like I am on a hike and once i get a few feet into the woods, I just get hopelessly lost. I've gotten some notion of the speeds from DDARing and Steptimer.cpp but can't figureout where to look next.
It seems that you need to know how the next timer works and how many steps it emits per interrupt.

https://github.com/Duet3D/RepRapFirmware/blob/dev/src/Movement/StepTimer.cpp

Any pointers or ideas out there?

thanks in advance

sinneD

EDIT: Here are some plots of a series of tests I have run. I am trying to figure out fast I can probe and maintain accuracy. The vertical bars are for 1 stanrd of deviation. I am averaging anywhere from 0.0015 to 0.0024 for a deviation.

dc42

@sinned6915 the Z probe input is checked on every microstep (which limits the maximum step rate when homing ot probing). So if you use P8 or P9 in your M558 command then the response is very fast. If you use one of the analog modes (P1, P2 or P5) then the Z probe reading gets sampled by the ADC at intervals of about 1ms, then passed through a filter; so the response will be slower.

sinned6915

@dc42 I am using P8 for the probe currently.

So RRF is polling the status on every step?
What is the maximum step rate that might be reasonably practical? I don't want to approach the bleeding edge, I want to hold the speed back to get reliable results. Am I in the right place with StepTimer? => ~940kHz for Arm, which is still freaking fast.

thanks again

sinneD

dc42

@sinned6915 said in How fast are board interupts or polling speeds.:

@dc42 I am using P8 for the probe currently.

So RRF is polling the status on every step?

What is the maximum step rate that might be reasonably practical? I don't want to approach the bleeding edge, I want to hold the speed back to get reliable results. Am I in the right place with StepTimer? => ~940kHz for Arm, which is still freaking fast.

thanks again

sinneD

Yes
Do you mean the maximum step rate while executing a homing move? That depends on what microstepping you use, and which Duet it is.

A Former User

@sinned6915 said in How fast are board interupts or polling speeds.:

Hi all -
...

Hi,
very very impressive! What is your hardware-setup?
Best regards

sinned6915

@dc42 I am conclusing that I am not out-running the MCU in my tests.

@lbi I am trying to test/prove some microswitches. I am using a custom built printer with 3 Z-lead screws and doing sucessive single probes and logging the data.

Here for example is Panasonic switch. I think that if used as a probe, a probe height of 3.03mm would be sufficient. As an endstop, I can live with an error of 0.015mm.

sinned6915

Here is an example summary of my probing results-