Magnetic monitor too high agc value

jamkas

Hi,

I'm using two magnetic filament monitors on my dual extrusion setting. They are placed right before the extruder. But, since i bought them I had some false under extrusion alerts (on both of them). Now since I started using pressure advance, they record false values quite often, sometimes every few minutes. So, to print reliably I had to turn them off.
I think the culprit might be the agc value which is always 128 (so the maximum as far as I understand) on both of the sensors. The mag is always 127. I tried sanding one of them to get the magnet closer to the sensor but agc still is 128. I don't dare to sand it even more as I'm afraid it might start to grind against the sensor.

Is it possible, that the wiring might be troublesome? The wires are about 1m long and run in the same canal with some stepper motor wires. Does the sensor send an analog signal which might be damaged because of some interferences or does it transmit a digital signal which shouldn't be affected? (sorry, I'm not an electrical engineer and therefore a bit unsure in that matter)

The voltage on the sensor is 3.28V so it's probably not a power issue.

the setup is as following (even with this ridiculous R20:180 I'm still getting false errors):

M591 D0 P3 C"e0_stop" S1 R20:180 E4.0 A0 L25.41   ;Filament sensor left
M591 D1 P3 C"e1_stop" S1 R20:180 E4.0 A0 L25.56   ;Filament sensor right

Does anyone have an idea how I can improve the sensor's stability?
Thank you very much in advance!

jamkas

I now tried a shorter (20cm) wire by placing the sensor near the mainboard. AGC is still 128. Then I sanded the lid even more. Now the magnet very slightly touches the sensor chip and left a tiny circular scratch. Still AGC is 128.

While tinkering I realised, that the idler slides to the side when I insert the filament and only the idler's very edge slightly reaches the filament. This led to too little pressure and the filament started slipping. I fixed this problem by modifying the lid so that the idler is more constrained (MFM_enc_lid_modidied.step ). With this modification I didn't have any false alerts within an hour of printing (min 92% max 113% currently). Maybe this solves my problem. Still it does not solve why the AGC value is so far away from what it should and I think I still need to fix that!

dc42

@jamkas it sounds to me that the magnet is too weak, because with the magnet touching the sensor you should get an AGC below 90. Has the AGC always been very high, or has it crept up over time?

The wiring does not affect the AGC value.

We are working on some new idler assemblies with stronger magnets so that the separation between the magnet and the sensor is less critical, however this does require a change to the housing.

jamkas

@dc42 Yes, the AGC has always been 128 on both of my sensors as far as I remember (I bought them in september 2020 but had them disabled for a long time because of the issues). It's strange how both of them are affected the same. The magnets are not demagnetized, they do attract objects but I don't know if they are weaker then they should be. I'm afraid I don't have any tools to measure a magnetic field.
Could it be some software related issue? I never did any firmware update of the sensor (if this would be possible). The Duet 2 firmware is up to date (version 3.3)

A version with stronger magnets would be great if that improves the reliability! I'd gladly buy an upgrade for mine if it makes them reliable. Printing a new housing is the least problem for me as I have a small SLA printer.

My fix of the idler's side movement didn't change too much. At my current print, after 1 hour of printing, the measured values are "min 37% max 117%", so still really far apart. To be fair, I have 0.8 pressure advance (long bowden setup), so it is jittering quite a lot...

dc42

@jamkas we will supply you with new magnet/idler assemblies when we have them. I will try to get an expected date for when we have them available. Currently we just have a few prototypes.

T3P3Tony

@jamkas please drop me an email, tony@duet3d.com. We will sort it out.

jamkas

Thank you very much, I really appreciate this! @T3P3Tony I sent you the email