Magnet Filament Monitor V3 - Calibration Issue

Diamondback

Hey,

just got my fancy new Duet filament monitors set up and I am somewhat stuck at the calibration phase.

My MFMs are located a fairly long way away from the extruders at the initial entry into my reverse bowden system (the entire way of the filament from drybox to direct extruder is a reverse bowden with 3mm ID)
I am well aware this means that I won't get ultimate accuracy from the sensors, but I'm not looking for that either (overall, the sensors are about 1m away from the extruders).
My main goal is to detect runouts/missing filament (switches ordered for that) and total blockages. I am not at all interested in detecting partial clogs or stuff like that, it's not something I ever had to worry about in practice.

With that out of the way, I need some help with the calibration.
In my config.g I have this line:

M591 D3 P3 C"100.io3.in" S0

When I run M591 D3 after that, I get the following log:

M591 D3
Duet3D rotating magnet filament monitor v4 on pin 100.io3.in, disabled, sensitivity 28.80mm/rev, allow 60% to 160%, check every 3.0mm, version 4, mag 128 agc 42, no calibration data

So far so good I guess, seems like the sensor is detected correctly and the gain control also seems ok?

I visually confirmed the behavior of the debug LED, without movement it just blinks red and green and with movement the green becomes more frequent until it's basically always lit. As far as I understand, this is the expected behavior. The magnetic assembly also seems to spin correctly when just looking at it.

When I then start a print, after a while it displays this instead:

M591 D3
Duet3D rotating magnet filament monitor v4 on pin 100.io3.in, disabled, sensitivity 28.80mm/rev, allow 60% to 160%, check every 3.0mm, version 4, mag 130 agc 42, measured sensitivity 21.04mm/rev, min 68% max 112% over 19.4mm

Still fine I guess?
What confuses me is that these values do not ever change, it's always exactly the same printout when it comes to measured sensitivity and the percentages (agc sometimes goes up or down by one)

I was under the impression that there would be a longer calibration phase over several meters of filament or whatever the print takes and not just "19.4mm". Where is that distance coming from? Can I change it?
I tried using the E variable in M591 to set it to something else than the default 3mm and that does not really change anything other than the reported "check every X":

M591 D3 P3 C"100.io3.in" S0 E100
M591 D3
Duet3D rotating magnet filament monitor v4 on pin 100.io3.in, disabled, sensitivity 28.80mm/rev, allow 60% to 160%, check every 100.0mm, version 4, mag 132 agc 42, measured sensitivity 21.04mm/rev, min 68% max 112% over 19.4mm

I'm sure I'm just being silly and missing some crucial step, but I can't figure it out...

(This behavior is the same across all my 4 sensors btw)

Hardware/Firmware: Duet 3 6HC as CAN master, Duet 3 Mini 5+ as CAN expansion, extruders AND filament monitors are connected to the Mini 5, 3.4.5 firmware on 6HC and 3.5b1+ on the Mini5 (custom build from @dc42 to fix a memory leak), standalone

Thank you

Diamondback

Can anyone help? I'd really like to calibrate the sensor over a longer distance than just a few millimeters...

T3P3Tony

@Diamondback the distance that the calibration is displayed for should be how much filament has been through the sensor during the current print since the last time you checked with M591. What happens if you run M591after printing for a longer time?

Diamondback

@T3P3Tony That's what I assumed it should be, but it always stays constant at the very low mm count.

T3P3Tony

@Diamondback I just saw you have it checking every 100mm. Try setting that to the default of 3mm

Diamondback

@T3P3Tony I did that already, the first two logs in the first post use the default 3mm (ie not specified in the M591 call)

T3P3Tony

@Diamondback ahh yes I see I have not tested the monitors as much with 3.5b1 as I have with 3.4.x are you seeing the same behaviour with both firmwares?

dc42

@Diamondback use the default 3mm unless you get a very large measured sensitivity range, in which case you could use 6mm. As you have the filament monitor a long way from the extruder drive, you are likely to need an even larger range than that.

When running a print, the measured sensitivity should be displayed after about 10mm of filament should be extruded. The length over which calibration has been done should gradually increase and the measured sensitivity range will widen at first and then settle down.

If you pause the print, calibration will restart from the beginning when you resume.

Diamondback

@T3P3Tony I can't run 3.4.5 on the board where the sensors are connected atm since it has a memory leak that was only fixed by @dc42 in a custom 3.5b1+ build for me.

All filament sensors are connected to a Mini 5 in Expansion mode, same board as the extruder motors.

Diamondback

@dc42 Ok, that's the behavior I expected from the documentation, but unfortunately it's not what I see.
After it first displays some form of calibration data, it just stops there and won't calibrate any further. The reported values then stay constant for the entire duration of the print.

If I use the reported values in my M591 call, the sensor appears to "work", as in, it does spawn filament error events when the runout switch triggers etc. However, the calibration distance of just a few mm of course leads to incredibly inaccurate results...

dc42

@Diamondback thanks. It may be that there is a bug in the support for filament monitors on main boards used as expansion boards. I will add this to my investigations list.

Diamondback

@dc42 Thank you

Diamondback

@dc42 Some further info, yesterday my filament didn't detach from the empty spool correctly and got caught during a testprint. The active filament monitor did not catch this, it did not generate any filament error at all even after a few minutes of continuing to print.

Its config:

M591 D2 P4 C"100.io2.in" S1 L23.2 R70:150 E5

I guess something is indeed not working?
(The switch part of the monitor works fine btw)

Phaedrux

Did you send M591 to see what it reported after the fail?

Diamondback

@Phaedrux Nope, unfortunately not, but I guess I can easily replicate this. I'll do that when I find some free time

Diamondback

@dc42 said in Magnet Filament Monitor V3 - Calibration Issue:

@Diamondback thanks. It may be that there is a bug in the support for filament monitors on main boards used as expansion boards. I will add this to my investigations list.

Any news on this?

Unsolved Magnet Filament Monitor V3 - Calibration Issue