Random trigger height, bending stainless steel bed

redpanda

What you can see in this video is the Smart Effector v3.0 triggering "on time" only on the very first probe, while every consecutive probe is pushing the hotend in the bed with great force as the SE trigger is "late". I've ran the probing sequence more than 50 times and observed the same behaviour without exception. Probe recovery time is 400ms as suggested (higher values tested with same behaviour) and there is an additional 2s delay before every subsequent probe!! The build plate is 4mm of stainless steel, drivers are running on 48v.

I've been trying to make this work for about a year now and this is what I discovered just now.

As I'm running Klipper on RPi and SKR 1.3 my first thought was that something in that chain caused additional delay in registering the trigger event, so the first thing I did was move the SE Out signal from Z- pins on SKR (that has 10k pull up and 1k pull down resistors) directly to the RPi pins, to minimise the probability of such a delay. As that didn't change anything, I decided to shoot a slow motion video and try to see if the probe actually triggers before bending the bed, and that is not the case. Meaning that the problem is in the smart effector itself.

PLS OUR LORD DC42 I NEED HELP

https://youtu.be/3w_I38wAM3c

infiniteloop

@redpanda

PLS OUR LORD DC42 I NEED HELP

Don’t want to hurt your religious feelings, but when running Klipper, it looks like you’re in the wrong temple: this forum is dedicated to RRF.

redpanda

@infiniteloop That's what I did firstly, but then i shot slow motion video from which you can clearly see that smart effector is triggering "late" based on the on-board led. Watch the video, everything will be clear then

oliof

@redpanda Several things that you could check before assuming the smart effector is defective:

make sure that the smart effector is not clamped too hard by the screwtop hotend. Earlier versions had a metal nut and a washer, the new plastic nut on yours should alleviate this to an extent.
adjust sensitivity (see https://docs.duet3d.com/en/Duet3D_hardware/Accessories/Smart_Effector#programming-the-sensitivity -- I dont know whether klipper has support for this, both RepRapFirmware and Marlin do).
The SKR1.3 is capable of running the LPC/STM port of RRF (albeit being limited to v3.3 of RepRapFirmware), so you might be able to check whether the same late trigger happens using RRF.

I tried to order my suggestions in increasing effort and involvement, maybe one of these help

redpanda

@oliof thank you for your answer !!

it seems alright, tensions wise
sensitivity is 28, on 50 it is scary how hard second and every consecutive probe hits the bed
i'm aware as I used to run RRF with skr and esp8266 and moved to klipper, I'll be installing RRF today again to stop this blame-shifting EDIT: Moved to klipper cuz rrf does not support TMC5160 on LPC boards

alt text

droftarts

@redpanda What speed is the second probing move at? Smart Effector requires a decent speed to trigger promptly, usually 1200mm/min or 20mm/s. Slower is not better.

See https://docs.duet3d.com/Duet3D_hardware/Accessories/Smart_Effector#commissioning

Ian

redpanda

@droftarts Thank you for taking your time to answer !!

Speed is 20mm/s, I also tried it with 30mm/s with the same effect.
I took it apart, completely...it seems like nothing can be mounted on any of the "mounting" holes if one wants an accurate reading. without bending anything. It's still a bit less sensitive on every probe with the exception of the first one but it's not nearly as bad as depicted in the video.

Just to be clear, if it's not visible in the video, I went ahead and configured led/fan power to shut off before start of the probing moves. Idk, that's all

You can mark this as solved...

apak

in the video it looks like the fan shroud hits the bed before the nozzle. Also it goes way to slow to probe...

Make sure what part hits the bed before. If not the nozzle, fix it.
2.- Probe at 1200 mm/min at least