@AndyE3D
Hi, thank you for sharing this alternative method for calibrating the non-linear extrusion with the MFM based on quadratic regression. In any case, your macro looks quite clean and well thought out.

It seems that the macro collects the errors only once, without reevaluating the applied coefficients. Based on my experience, I can say that this approach is insufficient to achieve good values that result in an error of less than 1% across the entire speed range. That’s why I follow this process with my macro:

The minimum extrusion distance for the measurements should be at least twice as long as the melting zone. First, I collect the errors. I discard speeds with errors greater than 10%, as corrections would require at least a 20% adjustment, which is not feasible due to slip and grinding. I calculate the coefficients using least squares. I apply the coefficients. I test whether the errors are within acceptable limits. If the errors are within limits, I finish. If not, I adjust the speed values based on the corrected errors (e.g., an error of 2.5% at 2 mm/sec). I recollect errors with the adjusted speeds (e.g., 2.05 mm/sec). I recalculate the coefficients. I apply the coefficients again. I test if the errors are within limits. If the errors are within limits, I finish. If not, I drop the fastest speed and repeat the process.

Regarding your concerns about die swell or foaming, I can clarify that this is simply an extrusion multiplier in the slicer or filament profile when the non-linear extrusion is calculated using that method.

I am excited to see where the development of the MFM will lead.

Btw. you can find the latest macro of my MFM calibration at https://github.com/Meltingplot/BigPrint/blob/duet-3.5.3-sbc/duet-config/macros/meltingplot/filament-extrusion-calibration

@infiniteloop said in FilamentMonitor Short and fast errors:

Does the OP run a SBC configuration? Was not aware of that.

HTTP is enabled by default for standalone, according to https://docs.duet3d.com/User_manual/Reference/Gcodes#m586-configure-network-protocols, though 'M586 P0 S1' is usually added in anyway by the configuration tool. I didn't get the sense from the config.g that this is an SBC setup, or there wouldn't be any network codes.

Ian

@Phaedrux will do 👍

It seem to be a TPU thing. WHen loading Ninjaflex, it seem to just not grip the filament enough, leading to these unreasonable values.

Well, I guess I can live with that..

@greggles said in Filament monitor on Duet 0.6:

I am guessing low signal means tied to ground

Yes, you may have to experiment with that, as it depends on how the switch is wired. Ideally the microswitch will be 'Normally Closed', ie pin is grounded when filament is present, and only go 'high' (or rather floating) when filament runs out, as it is less susceptible to interference then. If you end up with a switch on a 3-wire PCB (one wire will go to 3.3V), you should still be able to connect it; see https://docs.duet3d.com/en/User_manual/Connecting_hardware/Sensors_endstops

The Duet 0.6 is less tolerant to wiring errors than newer boards, so check your wiring at least twice before turning it on. You can test it as a normal endstop switch first with M574, before setting it up as a filament runout sensor.

I also have a Fisher (I used to work for RepRapPro), and it still comes out to test firmware updates and print odd bits and bobs when other printers are broken or busy. It currently has a Duet 2 Maestro in it, but I have plans to update it to a Duet 3 Mini 5+ board. Also got rid of the original extruder, has a Bondtech BMG clone on it now, much more reliable! Also carbon rod arms. I need to redo the side panels though, as the acrylic cracked in a couple of places, and broken around one corner. I've got a pile of 3mm Dibond to make them, though I'll have to get my CNC router running again before I can!

Let us know how you get on.

Ian

@gallaghersart yes you can do that, and the filament monitors will tell you individually which extruders are working. If using Duet 3 boards then each filament monitor must be connected to the board that drives the corresponding extruder motor.

@Aurimas said in Filament error on extruder 0: sensorError:

why would it run sometimes for hours and hours on the same printer and then throw an error every 15min?

🤷 If I knew that, I'd just tell you what to fix!

Ian

@Grant_fm you have to extrude a minimum amount of filament (usually 10mm) with associated XY movement before calibrated values are available.

@Simon-4

I thought because of statistics which the sensor sends back with M591 D0 there has to be an internal memory.

In fact, there is - well, kind of. Apart from the RAM which the DUETs use to maintain, calculate and organise data during operation, some specific settings can be written to the SD card for permanent storage. These then can be recalled when the system starts up again. See the GCodes M500: Store parameters and M501: Read stored parameters for details.

The problem with this mechanism is that the settings are typically defined in config.g - which is a good way to understand how your printer is set-up - but are then altered by the contents of a second file (config-override.g) which may introduce some level of confusion. So, many of the Duet users prefer to edit the config.g instead of relying on this mechanism, just to keep all settings in a single place.

@Fvanp said in Duet magnetic filament sensor filament path:

I'm only having one problem when changing the filament it is quite hard to push the filament through the magnetic filament sensor.

Using an FDM printer, the assembly of idler, bearing and spring may end up to be bit ”sticky”. Sometimes it helps to file corners and edges of the two idler parts so they operate more smoothly.

@Acid I used this as the basis to get my first filament-error.g action running. Thanks!

I'm confused by the slot numbers of saved position vs. the resume position. We save in slot 0 using G60 S0 and resume using slot 1 with G1 R1. Is there a difference in 0/1 based slot counting? Typo, but works for other reasons? Not complaining since it works, but want to learn.

; Step 1 > Pause > move to a spot center-front for easy access G60 S0 ; Save position to slot 0 ... ; Step 6 > Resume the print G1 R1 X0 Y0 Z5 F36000 ; go to 5mm above position of the last print move G1 R1 X0 Y0 Z0 ; go back to the last print move M24 ; resume print

@Kwad3d yes, so that makes sense it looks like the magnet is very close to the sensor

@T3P3Tony said in issue with 1LC+3.5.0RC2 and magnetic filament sensor V3:

When you say you are powering up and down to change the filament monitors, is that just the toolboard or the whole system?

the whole system
yes - no changes in the loom or anything else

@jay_s_uk I am away at the minute, but will have a look and report back. Many thanks

@DocTrucker thanks you

@India Make sure you have updated DWC to the same version as the firmware.

Ian

@claustro Glad to hear of the progress. The mechanics around the MFM are indeed, well, ”delicate”.

Regarding the piece of PTFE tube below the sensor, I am having problems using this because it makes the filament load in the titan quite hard.

As you have a fixed distance between sensor and extruder anyway, there is no need to bridge it with a PTFE tube. To ensure better guidance of the filament through the MFM, however, collet and PTFT help a lot. With your arrangement, it will be sufficient to provide an extremely short piece of PTFE - just cut it directly at the neck of the collet on the MFM housing.

I hope this gives you enough headroom to fiddle the filament into the titan. I bear with you, as I have a titan as well. In my case, the MFM is attached to the titan's lever by a very short piece of PTFE, but free pending otherwise.

Good luck,
mb