Magnetic filament monitor - cannot dial in the configuration

[cherringshaw]

Good morning Duet collective,

I have read the other threads and have seen similar issues, but nothing that has resulted in consistency with the monitor.

I am having some calibration issues for the rotating magnetic sensor. My setup: New-ish Railcore printer, BMG extruder, v1.7a magnetic sensor board mounted right before it, black FDM self-printed case sanded down to achieve AGC of ~80-81. There is an approx 1m cable from the sensor back to a Duet Wifi v1.02 board, running firmware 2.05.1 per a previous recommendation.

My issue is inconsistent triggering of the stop actions.

I started with this somewhat generic but tuned config:

M591 D0
Duet3D magnetic filament monitor on input 3, enabled, sensitivity 25.20mm/rev, allow 70% to 130%, check every 3.0mm, version 3, mag 130 agc 81, measured sensitivity 25.10mm/rev, min 96% max 105% over 446.9mm

In this configuration I get no more than 20-30 minutes of runtime before it bombs out with "Too little extrusion", "Too much extrusion", or "Sensor not reporting"...? When extruder gears are turning, the status light is almost solid green. I've tried a variety of PLA filaments from light to dark, same result. I hit "Resume" and the print keeps going fine.

The prints themselves comes out nearly flawless, like most things printed on the Railcore. I've tested with and without the sensor enabled on smaller items....they look identical coming off the plate, so I do not think the filament is under/overflowing at any time during the print, at least to my eye.

Based on suggestions in other threads, I changed slightly to this new configuration:

M591 D0
Duet3D magnetic filament monitor on input 3, enabled, sensitivity 25.10mm/rev, allow 50% to 150%, check every 5.0mm, version 3, mag 130 agc 80, measured sensitivity 25.10mm/rev, min 91% max 107% over 22108.7mm

This went better, and on a 32 hour (scheduled) print it only stopped three times...but one was overnight so I lost several hours in standby until I could hit "Resume", which is a bummer.

Last night I hit "M591 D0 P0" during a long print to disable the sensor overnight, then re-enabled it in the morning and it ran fine (one more stop) before finishing a few minutes ago....but disabling it kinda defeats the purpose.

Just curious if this is more of a configuration / tuning issue, or is there something maybe with the wiring (too long?) that could cause an intermittent bad signal? Is the error calculation somehow cumulative, so eventually the small errors build up into a larger one that hits the threshold?

Overall happy with the unit and it seems like it's doing it's job very well....I just need to dial it in to avoid these false alarms on long, overnight prints.

Thanks!
Chris

[engikeneer]

@cherringshaw It could be a few things:

• Noise/interference from motors. 1m of cable should be okay, but if it's 1m of unshielded cable running right next to your unshielded extruder motor cable, you might get problems. Particularly if you're getting "Sensor not reporting" errors, this could be because some of the data has been corrupted by noise? I'd try separating the MFM cable from the motor cables as much as possible first, and maybe then shielded cable if you can't separate them.
• Similar to this I'd check your crimps. If it is usually working, this is very unlikely your issue, but nevertheless might be worth checking
• You can probably up your measurement length. I have an MFM feeding a BMG extruder and measure the filament distance from sensor wheel to extruder hob to be ~50mm. For detecting filament runout, you just need your measurement length to be less than that (maybe say half so 25mm?). For jams, admittedly lower is better, but there might be a tradeoff with false alarms. I'd guess somewhere in between might be okay? Having said all that, I use 5mm and rarely have any problems now my mahine is set up.
• Make sure your filament path is well constrained between your extruder and the MFM, and that you've got the collet clips fitted properly. If the filament can flex, you will get some variation between what goes through the extruder and what the sensor detects.

Does your setup have a long bowden tube with high pressure advance? I have ~0.7m bowden and need pressure advance set at ~0.8. This gives good prints, but means the extruder jerks back quite hard at every corner. I find that I get "too little extrusion" errors if I set my measurement length too low or pressure advance too high, particularly in areas where there is narrow zig-zag infil. Whether this is real under extrusion from the extruder grinding the filament from constant kicking/retracting/unretracting at high speed, whether it is an error from what the MFM sees (e.g. maybe the filament flexes, or everything vibrates too much and there's some play), or whether it is an issue with the sensor not correctly reading/processing so many fast, jerky moves, I am not sure...

For your question on the error calculation, I do not beleive it is cumulative. As far as I know, it measures each 5mm chunk separately too see what the measured filament movement is and whether it is within the tolerance bands.

Next time you do get it stop a print, send M591 D0 to see what the actual measurement was that caused the pause and post it back - might give a bit more clue as to whether is is a data/corruption issue, or a mechanical/setup issue

[cherringshaw]

Really appreciate the thoughtful, detailed response, thank you! I've not had many issues with my various Duet products, but when I do this forum has always been incredibly helpful.

I will check the cable crimps for sure, great idea (I created the cable from the Duet wiring kit). The cable is not shielded, and it is zip-tied into the wire bundle, so your point about noise from the extruder motor cable is very plausible; I will separate it from the bundle on the next run and see if that eliminates the false alarms. Perhaps I should even try a shielded, pre-made cable to eliminate both possibilities.

Sounds like my setup is similar to yours...I created a custom mount to place the MFM just before the BMG extruder....eyeballing it it's probably ~50mm or less.

I do have a long bowden tube to get outside of a normally enclosed case, but it is almost entirely pre-extruder. Collet clips are in-place, and I did carefully cut (and re-cut, sigh) the short tubes between the MFM and extruder, and extruder to hot-end, so there is no obvious flex when the extruder is pushing. Here is my setup:

Thanks again for all the great tips, I will check all of these as well as increase the measurement distance a little further, and see if I can eliminate the false alarms.