Laser Filament Monitor - test results

wilriker

So, I added some more results to my Spreadsheet and they line up with the previous results pretty good which means that for the same conditions, e.g. filament, housing and settings the sensor will perform quite stable.

Probably tonight or maybe tomorrow night I can switch the housing to one having a cone-shaped interior as @dc42 suggested - took me some tries to design it in a way to be actually printable. I will then do some more prints with the current filament to see if the redesigned housing performs better or worse before switching filament.

T3P3Tony

@wilriker thanks for the update. So from those results you could probably be comfortable having the filament sensor error outside 40%-120% which over 3mm mean it would under read by 1.8mm before having an error - that is short enough to save some prints, but not all, depending on how big the object is, layer height etc.

fma

Did someone try to reduce/increase the check length (default to 3mm)? Could it improve things?

T3P3Tony

@fma I did a lot of testing with 1mm earlier on, this increased the % range of min and max, which is to be expected as there is less difference to average over and some time maybe that 1mm is all during a fast retract or something. On the other hand if you have 20%-140% of 1mm then the maximum underextrusion before an alarm is only 0.8mm so worth testing and tuning for specific filaments, retract lengths etc.

fma

Ok, I see. Thanks.

wilriker

@t3p3tony Based on my current results and as stable as they are I would even be comfortable to set the limits to a narrower range of 65-105%. I will add this to my filament's config.g but still with S0 as long as I continue testing.

EDIT: Regarding check length: would it make sense to synchronize this somehow with retraction distance? On my MK8 direct extruder I use a rather unusual 3.5mm retraction length which is quite close to 3mm check length.

wilriker

I switched now to a cone shaped interior. First print with it might be tomorrow or the day after.

But when I removed the PCB from the housing I took a look at the sensor itself and it seems to have some kind of dirt on the shiny part. It is hart to spot or to explain. What would be the best way to clean it? My intuitive approach would be to deal with it like with my camera lenses that have sensitive coatings on it and therefore need careful cleaning with just cigarette paper (the softest paper available) and Isopropyl alcohol. But I am open to suggestions.

Mounted in its new housing it just happily blinks now... green - red - green - red - green - red...

EDIT: And yes I read and adhered to

Take great care not to damage [the sensor] or to touch it. Do not put the filament monitor PCB component-side down on any surface.

dc42

I used cotton buds and isopropyl alcohol to clean the image sensor. AFAIR this is close to what the manufacturer recommends.

wilriker

@dc42 Thanks, will do it like that then.

dc42

PS - the manufacturer says you must not squirt alcohol on the image sensor because it could get inside. So moisten the cotton bud with alcohol and then use it to gently wipe the sensor.

wilriker

@dc42 said in Laser Filament Monitor - test results:

PS - the manufacturer says you must not squirt alcohol on the image sensor because it could get inside. So moisten the cotton bud with alcohol and then use it to gently wipe the sensor.

Did it like that. Sensor seems to be clean now. I have not printed anything yet but the status gives a lower brightness than before. No idea if that is somehow related.

Also did a first print with the internally cone shaped housing and the results are worse than with the default square window and open top. Not much worse but measurable. See my Google Spreadsheet for details.

wilriker

And a second print with the cone-housing. Again slightly worse than the straight square open-window version.

I printed @T3P3Tony's housing with divergence today but when I tried to mount it I noticed that it does not fit. Has there been a rearrangement on the PCB at some point in time? Because the window for the sensor is much to far down. I took a photo with mounting holes aligned because they fit:
housing comparison

Left is Tony's design, right is my cone-shaped based on @dc42's version on Thingiverse.

wilriker

@wilriker said in Laser Filament Monitor - test results:

the straight square open-window version.

Can I get a patent on this term - or better word-spaghetti?

T3P3Tony

@wilriker that is very odd - maybe i uploaded the wrong version i need to check this when i get a chance!

wilriker

So, while porting @T3P3Tony's changes to get a diverging light path into my "fork" of @dc42's OpenSCAD file I went back to the original housing and now with the cleaned sensor this gave the best results so far. min/avg/max where about 2% above the average for all of these values.

I modified my Spreadsheet to include some color coding now. Green means greater or equal to the average in that column, red means below average.

The print that had this very good measurements was the new housing with diverging light path so I am going to switch housings again tonight.

wilriker

@t3p3tony said in Laser Filament Monitor - test results:

@wilriker that is very odd - maybe i uploaded the wrong version i need to check this when i get a chance!

I just checked the documentation page and the screenshots you exported from OpenSCAD on this page also show the sensor window being offset from the mounting holes.

wilriker

Last test with black PLA is done and entered into my Spreadsheet. This was done with the diverging housing and it was a tiny bit below the quality of the default housing.

I now loaded grey PLA and also recalibrated my steps/mm (they were to low). I did not do this with the black PLA but I think this would only introduce a relative shift because all tests with black PLA were performed with the same steps/mm. I still have the diverging housing mounted and will use it to set a baseline for the grey PLA.

wilriker

Added some results for the grey PLA now to my Spreadsheet. AVG and MAX values are somewhat consistent (among all results for grey PLA) though still wider spread than with black PLA. The MIN values however have an enormous range.

I will switch back to the straight housing tonight and see how this influences the results.

keyz182

Hey, sorry, work has been busy for me, so I've only now had a chance to get back to printing. I'm still finishing off a run of parts I need that are in Spannerhands Yellow translucent PETG, which is a far cry from the Black PLA requested! Here are the sensor readings for that anyway:

M591 D1
Duet3D laser filament monitor on endstop input 4, disabled, allowed movement 30% to 100%, check every 3.0mm, current position 9.0, brightness 82, shutter 13, measured minimum -22%, average 30%, maximum 92% over 8353.5mm

Once these are printed, I'll get the black PLA running

wilriker

@keyz182 You should print with whatever you just print. Only the sensor housing is supposed to printed in black PLA.