Tuning stall detection

justus2342

Hi,

I would like to use stall detection to detect skipped steps on my big Delta Printer (Modified Tevo Little Monster). I use:

Board: Duet2 Ethernet (Standalone)
Firmware: RRF3.4.4
Tower Motors: 3x Nema17 0.9° "17HM19-2004S1" (Product page incl. specs)

In a first step I want to determine the right values for F, H and S. Currently I use

M915 P0:1:2 S3 R1

That gives me lots and lots of false positives. Thus I have to tune my arguments, starting with H. In the docs there is a formula to calculate this. It says:

Hmin = full_steps_per_rev * rated_current * actual_current/(sqrt(2) * pi * rated_holding_torque)

Looking into my config.g and the stepper motor's data sheet I use:

full_steps_per_rev = 400
rated_current (A)= 2
actual_current (A) = 1.5
rated_holding_torque (Nm) = 0.46

That gives me 587.163 mm/s as lower speed limit. That seems very wrong.

I might have made a mistake, but I do not see it. Could someone please explain what is wrong or why I have such a high minimum speed value?

Thanks in Advance.

Phaedrux

@justus2342 said in Tuning stall detection:

Nevertheless, I would like to have at least a reasonable value for the H-parameter.

I'd say H is less critical.

Hnnn (optional) Minimum motor full steps per second for stall detection to be considered reliable, default 200 (try 400 for 0.9deg motors)

Try 400 and tune sensitivity.

jay_s_uk

@justus2342 FYI the general consensus is stall detection isn't accurate enough for delta printers

justus2342

@jay_s_uk We are not talking about sensorless homing, right? What could be a better option to detect skipped steps on a delta printer?

jay_s_uk

@justus2342 oh, i read that wrong. ignore me then

Phaedrux

@justus2342 said in Tuning stall detection:

rated_holding_torque (Nm) = 0.46

Does your motor have a spec sheet that shows a torque curve? You may find the rated holding torque to be quite a bit lower.

@justus2342 said in Tuning stall detection:

actual_current (A) = 1.5

When doing stall detection homing the current is usually lowered, so that helps with the speed as well. During a print though, you're not going to be using lowered current.

Have you tried adjusting the sensitivity yet?If you're getting lots of false positives that's where I would start.

justus2342

@Phaedrux said in Tuning stall detection:

@justus2342 said in Tuning stall detection:

rated_holding_torque (Nm) = 0.46

Does your motor have a spec sheet that shows a torque curve? You may find the rated holding torque to be quite a bit lower.

Yes, here you go:
https://www.omc-stepperonline.com/download/17HM19-2004S1_Torque_Curve.pdf

It seems indeed a little lower, but since the formula has it in the denominator, smaller values do even raise my minimum speed. That's the wrong direction.

@Phaedrux said in Tuning stall detection:

Have you tried adjusting the sensitivity yet?If you're getting lots of false positives that's where I would start.

Yes, I played around with it a little. It seems the truth lies somewhere between 3 and 10. I will have to figure it out by trial and error. (It would be really nice to see the trigger value in the warning log-message by the way. )

Nevertheless, I would like to have at least a reasonable value for the H-parameter.

Phaedrux

@justus2342 said in Tuning stall detection:

Nevertheless, I would like to have at least a reasonable value for the H-parameter.

I'd say H is less critical.

Hnnn (optional) Minimum motor full steps per second for stall detection to be considered reliable, default 200 (try 400 for 0.9deg motors)

Try 400 and tune sensitivity.

justus2342

@Phaedrux Thank you for citing the docs to me. No irony here. I absolutely overlooked that the value for the H parameter is in stepper steps and not in mm/s.

Thus there is no problem and my value is way less off than i thought.

I consider this problem solved now.