automatic calibration?

[destroyer2012 0]

I've noticed that with the bowden setup and the high speed of ultimaker calibration is an absolute must. If the feedstock filament changes in diameter even a tiny bit your print quality will be massively affected. Trying to push too much filament to fill a small space results in back pressure in the bowden tube, and blobbing at the best to tube breaking and filament grinding at the worst. I've found that my feedstock is not nearly uniform enough for touch-free printing, and with a changing filament diameter the "bite" of drive gear into the filament changes so I have to recalibrate steps/mm and re-measure filament diameter fairly often.

So the question is: Why not measure the filament diameter and feed rate at the entrance, by attaching like a pedometer-type setup with some method to measure the filament's diameter too? That way you can automatically calibrate before every print (and possibly during?).

Has anyone considered this?

[mr_seeker 0]

It can be done, but you should consider looking at tolerances. The less tolerance your feedstock has, the less you need to measure. Measuring every time will only make the machine slower because you have to measure and calculate the feedstock rate before every move. Doing a good measurement and using filament with a low tolerance will help a lot more.

[destroyer2012 0]

That would be great but many websites don't list tolerances. For example makerbot and our illustrious ultimaker. Besides, looking towards a future of recycled plastic, wouldn't it be convenient to automate the calibration process? maybe we wouldn't be able to print as fast but wouldn't you want to go for weeks or months at a time without touching the machine besides removing prints and reloading filament?

Low tolerance feedstock seems like it would be expensive. Wouldn't you want to make cheaper products at a minimal time cost? You wouldn't have to calibrate before every move, only every once in a while, and maybe adjust as you go. I'm envisioning it more like, calibrate before every print, not necessarily during.

Not quite ready to throw out this idea just because it might make my printer a bit slower.

[alaris2 1]

Maplin sell a cheap but good digital caliper. I used it for measuring the filament that came with my new Ultimaker.

used to be £10 but I think they charge £20 now.

I bought 3 and was going to make a 3-axis CNC machine but never got around to it. why is this relevant? because this particular caliper has accessible pins and a described waveform for reading the values via an Arduino or similar - you could mount it and have it reading the filament size and feeding it to a CPU.

having just paid lots of money for my new Ultimaker and finding the silver filament supplied with it varies from 2.79 to 2.92mm and makes some atrocious prints, I'd be all for automatic calibration..

gonna have to throw this filament away I fear ;(

[destroyer2012 0]

having just paid lots of money for my new Ultimaker and finding the silver filament supplied with it varies from 2.79 to 2.92mm and makes some atrocious prints, I'd be all for automatic calibration..

gonna have to throw this filament away I fear ;(

I have the exact same problem with the white PLA I bought from UM. I didn't know what was happening at first and it got jammed in my bowden tube something fierce (3.09 mm diameter!) My silver doesn't seem too bad, could you share your batch number so I can check mine?

Yes mainly I was thinking about using those encoder wheels you find in inkjet printers (I have a bucketload of those), combined with some way of measuring the circumference of the filament (more dynamic range). Now I just need to figure out how to get data from the wheels; shouldn't be TOO hard I hope.

[jelleatprotospace 0]

Measuring thickness variation and doing something useful with it seems very hard to me. An encoder wheel on the input to the hot end would be a good idea, but at the same time measuring diameter and presumably compensating for off-round filament just seems like a bridge too far.

But why not try to make the filament you have now and make it conform to a better spec? Apparently you can limit the thickness with a tool I only know as 'pulling-iron' an iron plate with smooth slightly conical holes that you pull your filament trough. The extra thickness is converted to a little extra length and the diameter is limited to the hole used. Easy to do by hand for smaller lengths, but it should not be too hard to make a machine for.

[destroyer2012 0]

Apparently you can limit the thickness with a tool I only know as 'pulling-iron' an iron plate with smooth slightly conical holes that you pull your filament trough. The extra thickness is converted to a little extra length and the diameter is limited to the hole used. Easy to do by hand for smaller lengths, but it should not be too hard to make a machine for.

Cool. Sounds better than my idea if it works. I couldn't find anything online by that name, maybe you can post a picture? Of course this could have applications in filament recycling, which is why I think it's worth pursuing.

[alaris2 1]

other than 'Ultimaker PLA' there's no batch code or anything on my silver. it came with the machine which shipped here a couple of weeks ago.

I just noticed it says <=220 degrees on it too. which is strange, since I can't even make it come out the extruder below 230 degrees!

does the gcode take care of the extruder rate and filament thickness? if so, modifying it on the fly does sound tricky..

if you could just get away with changing the extruder motor speed with a single command you might be able to compensate?

[ddurant 0]

Measuring thickness variation and doing something useful with it seems very hard to me..

I agree.

Rob Guiseburt, the guy who coined the term "Volumetric 5D", has been talking about that for a while now. He seems to think it's not THAT hard. Not sure I buy that but Rob's a very clever guy.

Personally, I'm not sure it's worth it. For the most part, we're printing at resolutions where small differences in filament diameter are in the noise - unless you're printing at 100% infill, over-volume just gets smushed into the voids and under-volume ends up making lines that don't exactly join but are pretty close. Filaments that have big diameter variations tend to cause other problems (mostly at the cold end) and don't live long. People give up, the seller gets a bad name, etc..

What I've been saying for a while, diverging from Rob, is that I'd like to see things move more towards "true" volumetric 5d where the slicer specifies (via "G1 E" or a new "G1 V" or some such) a raw volume value to the firmware in mm^3. For filament-based extruders, like we all have, this doesn't get us a whole lot - it lets you change feedstock diameter and (maybe) rescale a print without reslicing but that's it. What it does allow is non-filament extruders - PLA pellets cost like a 1000 times less than PLA filament....