UPDATE 14-04-22: Now released here: https://github.com/KerbalFoundries/Duet-Magnetic-Filament-Monitor-Housings

Hello!

I'm fairly new to actively posting here, but been using Duet/RepRap Firmware for a while now and recently got tempted to buy a filament monitor after having one or two issues with long prints getting ruined. And because it's a cool little gadget. Who can resist a cool gadget, right?

To cut a long story short, I had a few issues. Some of those (sensorError) were due to loose crimps and something that got tweaked in one of the 3.4 release candidates that dc42 was kind enough to clue me in on. The other issues was "tooLittleMovement". After a while I figured out that I had a few issues with the printed case. In fairness, I know it's supposed to be resin printed really, but I don't have a resin printer. I do have some 3D CAD skills and a love of tinkering, so I set about first recreating the existing design, then tweaking it. Editing STL files is not my idea of fun, so I redrew it in OnShape.

I came up with a few design goals and tweaks:

• Keep the filament path clear of anything but the PTFE inserts, hob and idler.

• Align the PTFE so the filament is lightly pressed against the hob even before the idler comes into play

• Add slightly more clearance for the hob and idler in the filament region

• Add an indent to keep the idler spring in the correct position

• Achieve a tight fit for the plastic collets holding the PTFE

The housing I came up with worked a treat, I'm happy to say.

Pic for comparison:

I felt an extra mounting lug couldn't hurt either.

Fast forward a little bit... I ended up spending some time on the forum and chimed in on a few threads where people had similar problems and shared my design a few times. It's a funny thing, though: Once I start sharing out stuff I create, it's hard not to want it to be a good as it can be.... I've been down this rabbit hole before making a KSP mod a few years back, but let's not get sidetracked! It's an enjoyable thing getting involved in an open source project/community, so why not?

I printed and tested many prototypes:

Given that this needs to be printable on any printer, the sane thing to do is print it on the worst machine possible. In my case, it's an ANET A8. I can hear what you're thinking... "You came in that thing?!? You're braver than I thought!" In my defence, it was my first printer, has been modified and runs RRF (though on a Fly CDY V3 - I hope I'll be forgiven, I couldn't justify a Duet for this one). However, it's got a horrid build bed and no bed compensation. Realistically a worst-case scenario for most people playing with Duet, I suspect.

Anyway, the point I'll get to is that I found it really had to hold the tolerance between the bottom of the lid and the stop of the standoff, which all other things being equal is what defines the distance between the sensor and the magnet. A well set up 3D printer can hold surprisingly good tolerances, with the bottom layer being one of the exceptions in a lot of cases. The solution that came to me was to make the standoffs part of the body, not the lid. The critical tolerances (bottom bearing flange seat, top deck of body, top of standoffs) are all then defined by the layer height tolerance without a first layer in the stack:

MK1:

At this point, I should add that I have the MK1 rotating assembly. The MK2 has a shorter hobbed section and fatter magnet. Moving the deck of the body cuts into the reliefs for the PTFE collets, creating a gap where the collets can push the lid up. This didn't seem ideal, so I moved them out to some small extensions. This has the added bonus of giving more length within the housing to hold the PTFE nice and stable.

I made a few "quality of life" adjustments, making the bearing seats and PTFE holes slightly tapered for easy starting without resorting to files or suchlike.

I had a quick think about the idler carrier. I'd designed with slightly more room for a beefier carrier, so came up with something that still allows the halves to be identical, but a little more surface area for the glue joint, and the joint running perpendicular to the layer lines for strength:

The result:

A quick view of how the standoffs work:

And the strain relief works nicely too:

All printed PLA, 0.1mm layer height, 3 walls, 0.4mm nozzle, 0.4mm line width on a cr**tacular printer
and snapped together with nothing more than a light deburr. M591 D0 from its first print:

Duet3D rotating magnet filament monitor v3 on pin e0stop, disabled, sensitivity 26.67mm/rev, allow 70% to 130%, check printing moves every 6.0mm, version 3, mag 131 agc 86, measured sensitivity 26.62mm/rev, min 95% max 104% over 5150.3mm

That's probably accurate. That print was 0.6mm line width, 0.3mm layer height and fairly fast. Just happened to be what came up for printing when the latest prototype was ready. Typically, I'd expect and get 98%-101% on a more standard 0.4mm width .1 or .2mm layer print. I'm happy enough with the mag and agc right there straight out of the box. Yes, it's disabled for the moment. I tend to log M591 for a few prints before sending new designs live in case it starts doing weird stuff and pausing prints that it shouldn't.

I've got some testing to do to make sure this really is as solid as I hope and have been in touch with one of the team about getting hold of a V2 rotating assembly to make sure that version is working nicely too.
I'll start sharing links in a couple of days when I'm sure it's all behaving. I hope I'm not treading one anyone's toes - I figure this is all in the spirit of open source

Thanks for reading!

I'm happy with the V2 assembly, so I've uploaded everything to a GitHub repo:

https://github.com/KerbalFoundries/Duet-Magnetic-Filament-Monitor-Housings

You'll find both V1 and V2.

Happy printing

@phaedrux Thanks! It's really nice contributing to a project again, even in a small way. I'd forgotten how much I enjoy it.

Good news is the new V1 housing is performing perfectly. I'll test it printed with my other printer next, and thanks to the Duet team a MK2 magnet assembly is on its way for me to test too.

@charliedrums

I've just had a similar issue myself which turned out to be noise. I've been developing the design of the housing and during testing had moved the sensor wire closer to the extruder wiring to neaten things up. It always stopped during large circles, which if you think about it could make sense: the frequency at which X and Y are driven continuously varies when printing a circle. At some point while printing that circle, if you're going to get RF coupling between the stepper wiring and the sensor wires, there's a small range of frequency at which it's most susceptible that the drive will run through. This isn't necessarily true when printing things that aren't arcs, and on arcs that big may occur for a significant amount of time.

Just a theory, and couldn't say with certainty, of course, but it's the exact same issue that I've been musing about for a few days. Forgive me if I've missed it, but are you using screened cable? Might be worth a try as a cheap and easy diagnostic step. 4 core is easy enough to get hold of. Moving the sensor wire out of the way fixed the problem in my case, but it's a pain not being able to bundle it all up with the extruder wiring, so I'm going to experiment with screened cable myself.

Anyway, thought I'd throw that into the discussion. Hopefully it's helpful!

@dc42 Quick update:

I redesigned the housing and re-printed, paying particular attention to having the filament only in contact with the PTFE, having the PTFE securely mounted closest to the hob and passing the filament over the hob such that the idler has very little work to do. I believe RC2 fixed the "sensor error" issue and the housing update seems to have cleared the frequent spurious movement readings I was still getting.

I'm now consistently getting:

Duet3D rotating magnet filament monitor v3 on pin e0stop, enabled, sensitivity 26.67mm/rev, allow 70% to 130%, check printing moves every 6.0mm, version 3, mag 131 agc 69, measured sensitivity 26.69mm/rev, min 98% max 102% over 13106.6mm

Not had an unwelcome print pause in days now.

I could probably refine the design a little more. I think a full case with strain relief for the cable(s) would be a definite plus.

Housings for the V1 rotating assembly:

MFM Housing - Lo-Fi Lid - mag V1.stl
MFM Housing - Lo-Fi Base - mag V1.stl
MFM Idler - Idler Mount V1.stl

I've tested these printed from both my printers with different filaments. All successful right off the print bed with no sanding or shimming and very little variation in Mag or AGC.

The only super critical print setting is to use 0.1mm layer height. I really cannot stress this enough. Moving down to 0.05mm is fine, as is any other layer height the divides neatly into 0.1. Anything else is likely to land you with some "odd" dimensions that are really quite important. Otherwise, just do what you'd normally do for accurate functional prints. In Cura, the "Horizontal Expansion" setting can be quite handy to compensate for any deviation of your printer - I'm sure other slicers have similar settings.

I've designed for 0.4mm line width, so best to stick to that. 10-20% infill of any type that ties the walls to the top/bottom will work fine. I used lines an gyroid while testing. You should get 3 x whole lines in the 1.2mm "walls" of the lid.

The lid might need a light deburr if you get some elephants foot on the first layer.

I opted for using flanged RC servo screws to mount the board, though you can easily drill and tap to
M2.5 for the board mounts. All the supplied hardware is used as standard otherwise.

Push the PTFE tube into the housings so it's close to, but not interfering with the hobbed section. This runs in further than the official housing and is best done before the rotating mag assembly is installed so you can be sure where it ends up. Running the point of a knife or deburring tool inside the entry to the outgoing PTFE completely eliminates any snag when loading filament.

The bottom bearing is best pressed in with the head of an M5 cap head screw if this is required. This puts the force purely on the outer race, though not much should be needed all other things being equal. As explained in the original post the seats have a slight taper which seems to account nicely for variations in prints and gives a snug fit without any fussing. The bearings snap nicely into place on all my prints now with no play or fettling needed.

The idler carrier halves will need gluing together around the bearing as per the original and need no special attention. I'd advise a tiny blob of glue where the stubs meet inside the bearing, but careful with the bearing itself, obviously.

I'll move on and test V2 now In the meantime, please do treat this as a development prototype and of course send some feedback if you try it! I had a stupid moment this afternoon where I forgot to add the idler bearing altogether and was annoyed I got one or two under extrusion warnings. That was until I figured out what I'd done!

@TheDragonLord

Hi, just found your post - sorry for the delay and thanks! I don't use Fusion so STEP is the best I can do:

MFM Housing - Lo-Fi Lid - mag V2.step

MFM Housing - Lo-Fi Base - mag V2.step

@bubblehead you've specified M950 SO when creating servo channel 0 on duex.pwm5, but then used M280 P7. I think you should be using M280 PO. Give it a go?

@jrcl m203 specifies in mm/min, rather than mm/sec. E6000 ought to set it straight.

First impressions are that it's much better! 100m of filament in and it's been rock solid so far.

@Phaedrux

Going back over my notes, I originally had extruder max speed at 3600, then subsequently fiddled with it. I've just rebuilt my config entirely and all seems well.

I do still have an issue with the MFM reporting too little movement (<40%) with pressure advance on, though I'm homing in on whether it's slipping itself or whether there is actually some under extrusion going on. My feeling so far is it's struggling to react to rapid moves, but only some slow motion footage will be able to confirm that. I can't hear any skipping and the filament feels like its movement matches the extruder. My prints usually run around the 80-110% mark and I've had the MFM set 70% - 120% for as long as I can remember.

@Phaedrux I think I probably muddied the water by posting the config after I'd been messing with it! You're right, of course.

I'm going to strip it right back and start again - I'll keep the thread updated.

@Phaedrux said in Pressure advance sounds like a machine gun:

M572 D0 S0.05
M566 E1500
M203 E6000
M201 E3000

Thanks for the reply. It was a starting point suggestion from the documentation based on the bowden length. My Pulley ratio is 3.75:1, so I have to be a little careful about speeds and accelerations, as I'm pretty sure those settings all specify conditions at the filament after the gearing, not the motor? I'm running 6mm, 40mm/s retract currently.

Hello!

I thought I'd have a play with pressure advance. I have a printer with a belt reduction extruder and .5m bowden, so it seemed high-time to investigate.

2 problems: firstly, the extruder sounds like a small machine gun half the time. Sounds like a motor missing steps.

Secondly, it fails the print all the time as the MFM calls out too little movement (which I believe is probably correct). Particularly for small features like infill.

I'm running M572 D0 S0.4

I can smooth it out by setting extruder instantaneous speed changes super low (50), but it's slowed things down hugely as your might expect. The extruder is bombproof otherwise - I can't remember the last time I had a movement warning from the MFM.

What, if anything, am I doing wrong? Apologies for the hiatus - usual life stuff, I'm afraid!

Thanks in advance

; Configuration file for Duet WiFi (firmware version 3.3)
; executed by the firmware on start-up
;
; generated by RepRapFirmware Configuration Tool v3.3.10 on Mon Jan 10 2022 00:22:29 GMT+0000 (Greenwich Mean Time)

; General preferences
G90                                            ; send absolute coordinates...
M83                                            ; ...but relative extruder moves
M550 P"Ender 5 Plus"                           ; set printer name

; Network
M552 S1                                      ; enable network
M586 P0 S1                                     ; enable HTTP
M586 P1 S0                                     ; disable FTP
M586 P2 S0                                     ; disable Telnet

; Drives
M569 P0 S1                                     ; physical drive 0 goes forwards
M569 P1 S1                                     ; physical drive 1 goes forwards
M569 P2 S0                                     ; physical drive 2 goes backwards
M569 P3 S0                                     ; physical drive 3 goes forwards
M584 X0 Y1 Z2 E3                               ; set drive mapping
M350 X16 Y16 Z16 E16 I1                        ; configure microstepping with interpolation
M92 X80.00 Y80.00 Z800.00 E1000.00              ; set steps per mm
M566 X1000.00 Y1000.00 Z100.00 E50.00        ; set maximum instantaneous speed changes (mm/min)
M203 X12000.00 Y12000.00 Z400.00 E1000.00        ; set maximum speeds (mm/min)
M201 X1200.00 Y1200.00 Z200.00 E1000.00       ; set accelerations (mm/s^2)
M906 X800 Y800 Z800 E800 I30                  ; set motor currents (mA) and motor idle factor in per cent
M84 S30                                        ; Set idle timeout

; Axis Limits
M208 X0 Y0 Z0 S1                               ; set axis minima
M208 X350 Y360 Z400 S0                         ; set axis maxima

; Endstops
M574 X1 S1 P"xstop"                            ; configure switch-type (e.g. microswitch) endstop for low end on X via pin xstop
M574 Y1 S1 P"ystop"                            ; configure switch-type (e.g. microswitch) endstop for low end on Y via pin ystop
M574 Z1 S2                                     ; configure Z-probe endstop for low end on Z

; Filament Monitor
M591 D0 P3 C"e0_stop" S1 R70:130 L26.67 E6.0; Duet3D rotating magnet sensor for extruder drive 0 is connected to E0 endstop input, enabled, sensitivity 24.8mm.rev, 70% to 130% tolerance, 3mm detection length
M591 D0 ; display filament sensor parameters for extruder drive 0

; Z-Probe
M950 S0 C"exp.heater3"                         ; create servo pin 0 for BLTouch
M558 P9 C"^zprobe.in" H1 F120 T9000            ; set Z probe type to bltouch and the dive height + speeds
G31 P500 X45 Y0 Z2.399                         ; set Z probe trigger value, offset and trigger height
M557 X45:345 Y25:325 S100                      ; define mesh grid

; Heaters
M308 S0 P"bedtemp" Y"thermistor" A"Bed" T100000 B4138 ; configure sensor 0 as thermistor on pin bedtemp
M950 H0 C"bedheat" T0                          ; create bed heater output on bedheat and map it to sensor 0
M307 H0 B0 S1 P19.56 I0.71 D134.26             ; disable bang-bang mode for the bed heater and set PWM limit
M140 H0                                        ; map heated bed to heater 0
M143 H0 S100                                   ; set temperature limit for heater 0 to 100C
M308 S1 P"e0temp" Y"thermistor" A"Nozzle" T100000 B4138  ; configure sensor 1 as thermistor on pin e0temp
M950 H1 C"e0heat" T1                           ; create nozzle heater output on e0heat and map it to sensor 1
M307 H1 R2.624 K0.576:0.000 D6.17 E1.35 S1.00 B0 V24.2 ;
;M307 H1 B0 R2.451 C156.9 D6.32 S1.00           ; disable bang-bang mode for heater and set PWM limit
M143 H1 S280                                   ; set temperature limit for heater 1 to 280C

;Sensors
M308 S2 P"exp.thermistor6" Y"thermistor" A"Heat Sink" T100000 B3950 

; Fans
M950 F0 C"fan0" Q25                           ; create fan 0 on pin fan0 and set its frequency
M106 P0 C"Part" S0 H-1                         ; set fan 0 name and value. Thermostatic control is turned off
M950 F1 C"fan1" Q500                           ; create fan 1 on pin fan1 and set its frequency
M106 P1 C"Tool" S255 H1 T65                      ; set fan 1 name and value. Thermostatic control is turned on

; Tools
M563 P0 S"Extruder 0" D0 H1 F0                 ; define tool 0
G10 P0 X0 Y0 Z0                                ; set tool 0 axis offsets
G10 P0 R0 S0                                   ; set initial tool 0 active and standby temperatures to 0C

@medicusdkfz

Apologies for the slow reply! Glad it's working out for you

@c310 mag is a little low, AGC is a little high but it should be workable. Try my redesigned housing from the sticky thread in this forum, it'll probably solve the filament skipping past. It'll print fine on an FDM printer so long as the instructions are followed.

The wiring is sensitive to interference no matter how good the continuity might be. How is it routed? If it's close to stepper wiring it's caused communication errors for me. Screened cable seems wise - USB cables are a good cheap source that everybody has. Don't forget to ground the shield only at the board end!

@c310 In my experience, this is typically a wiring or interference issue. Either way, communication has been lost with the MFM during the print. How do you have the wires routed?

That minimum movement figure looks suspiciously low too. That's a separate issue and means that the filament is either slipping in the extruder, or more likely not moving the rotating assembly in the MFM. What housing are you using?

@TheDragonLord

Hi, just found your post - sorry for the delay and thanks! I don't use Fusion so STEP is the best I can do:

MFM Housing - Lo-Fi Lid - mag V2.step

MFM Housing - Lo-Fi Base - mag V2.step

@jens55

Just to throw my 2P worth in.... I run glass on ali and had trouble with the glass taking on the plates shape and having inconsistent flatness. Both solved by dabbing heat transfer compound onto the ali plate in a grid then placing the glass on top. Better heat transfer is a bonus (no waiting for the glass to heat) but the real kicker is the HTC is viscous enough to let the glass sit on it without running away, so the glass can retain it's own (very good) flatness without following the plate. It also stops the glass easily sliding laterally, should the nozzle contact a blob or suchlike. I still run bed comp but I'm well under 0.1mm and so consistent it's almost pointless to remap. I use it to tell me if one of the two Z screws has dropped really, though I should stop being lazy and set them up independent and get the machine to sort.
Removal is easy enough, just pull a single strand of wire through like a cheese wire to break the HTC. Clean and reapply. If you get a hump in the middle, just heat up and put some weight on. Leave it half hour and the glass naturally want to go back to flat, squishing the HTC where it needs to.
I'm not using anything expensive: stock Ender 5 Plus plate and glass bed. Also pulled the same trick on my much Frankensteined Anet A8, the bed of which is truly atrocious but perfectly serviceable with HTC and borosilicate glass slapped on top. It doesn't even have a probe or bed compensation! At the end of the day, ali is terrible at dimensional stability when heated. It's also full of stresses whether billet or cast. If it's not really well annealed to normalise it, those stresses will appear every time heat is applied. Tooling plates should be better than any old cheap slab, of course.
Hope someone finds it helpful, thought I'd throw into the discussion.

@thooijer-0 said in Redesigned the MFM housing:

Thank you for the design , the original i printed skipped.
I noticed without connecting it to the duet.
It will probably work when printed with very tight parameters or with a resin printer.
Your design works good for me, have printed a few objects and it works fine.
Tested the runout function and it is really great! Thanks!

Thanks! I'm really glad it's working out.