Laser filament monitor
-
When will either sensor be available. I'm definitely adding a filament sensor to the printer I am building now and rather buy from the makers of the duet than some no-name AliExpress company.
-
I'm sorry, we can't give a date yet. I expect to have a few prototype laser filament sensors by the end of this week, which should put us on track to decide which version to manufacture at the end of this month.
-
First Laser Filament Monitor prototype.
-
I'd buy one. Is there already an estimate for the price?
-
Would love to betatest one
-
looks promising
-
Would look better in Blue.
-
The production boards will be blue, these are prototypes so I accepted the default PCB options for fastest turnaround and lowest cost.
-
Laser filament monitor assembled:
Installed and working on delta printer:
During a test print using black filament (probably one of the worst cases), the average detected motion was 7% low, and the tolerance was quite wide at +44% - 32%. I know from earlier tests that the optical sensor has difficulty tracking fast moves when using black filament. To improve accuracy, we may need to compare actual vs. commanded motion only during printing moves, not during retraction and re-prime moves. Would this be acceptable?
-
Is there an LED illuminating the filament for the sensor?
-
Hi David, I'd say yes it would be acceptable to just monitor extrusion, it's a lack of (or excess of) extrusion that counts. This also simplifies which presumably improves accuracy.
-
Yeah I would agree
-
-
Oh, interesting. Is there a possibility that focus or intensity needs to be adjusted? Perhaps even for each different filament colour?
Perhaps the laser monitor could read a pattern on the circumference of a hobbed wheel that is idling against the filament – or perhaps even just the grain of the metal of the hobbed wheel would track better than the surface of filament.
-
I tried a similar print with retraction disabled, although I still had pressure advance set at 0.2 so I could still see frequent backward movement of the filament. The calibration tolerance reduced to +27 -26%. That's much more manageable, not only lower but more symmetrical too.
If the sensor works at least as well with all other filament colours and materials including transparent, and ignoring retractions and re-prime movements works is feasible and works, I think we will be able to go ahead with this design.
-
@bot:
Oh, interesting. Is there a possibility that focus or intensity needs to be adjusted? Perhaps even for each different filament colour?
There is no focus or power adjustment, although the reflected intensity can be read. We might be able to use that to increase the tolerance automatically when the reflected intensity is low.
@bot:
Perhaps the laser monitor could read a pattern on the circumference of a hobbed wheel that is idling against the filament – or perhaps even just the grain of the metal of the hobbed wheel would track better than the surface of filament.
If we did that, we would lose the benefit of mechanical simplicity of this design over the previous rotating magnet design.
-
@bot:
Perhaps the laser monitor could read a pattern on the circumference of a hobbed wheel that is idling against the filament – or perhaps even just the grain of the metal of the hobbed wheel would track better than the surface of filament.
If we did that, we would lose the benefit of mechanical simplicity of this design over the previous rotating magnet design.
Yes, that's true.
How was the accuracy of the magnet based sensor? I would be willing to forgo a bit of simplicity if the accuracy was dramatically improved.
-
I thought the basic idea was sound, the hobbled wheel gripped the filament and turned reliably, it was the variable nature of the 3d printed casings holding it all together, which allowed the magnet attached to the hobbed shaft, to move away from the sensor chip that let it down, and I've a pretty accurate printer to make them with. An injection moulded or machined casing would have made a big difference and dropping monitoring of retractions, but it was still a clunky solution compared to a tiny pcb which could easily be built into an extruder if it works better.
-
I found the accuracy of the magnet based sensor was around 20% after calibration. It is limited by the need to use a low data rate to the Duet (1000bps) because of the filtering capacitors in the Duet endstop circuits.
-
@bot:
Is there an LED illuminating the filament for the sensor?
Not an LED, a tiny diode laser.
I took @bot's question to be a suggestion that perhaps the addition of a tiny LED to the PCB might supply additional light that would (perhaps) assist with the sensing.
(Out of my league here - but if by chance that is a useful suggestion - just wanted to call it out).
