Fillament-thickness sensor

fma

You need to measure in 2 orthogonal directions, as the filament may have the correct section, but not being a circle... In this case, measuring in one direction only will give you a wrong indication.

dc42

It's been done before, using linear CCD sensing elements. As @fma says, you need to measure the diameter in at least 2 directions.

However, good quality filament has fairly closely controlled diameter these days, so I'm not convinced that monitoring filament diameter is worthwhile. For example, rigidink's 1.75mm PLA is controlled ot +/-0.03mm, which equates to +/-1.7% in diameter or about +/-3.4% in area. Prusament is +/-0.02mm diameter, so even less. Ian (@deckingman) did some tests that indicated that even quite large changes in extrusion rate didn't have much effect on the print.

deckingman

@dc42 said in Fillament-thickness sensor:

Ian (@deckingman) did some tests that indicated that even quite large changes in extrusion rate didn't have much effect on the print.

Actually Michael Hackney did it much more scientifically and more in depth that I did - http://www.sublimelayers.com/2017/12/musing-on-under-extrusion-prepare-to.html

Definitely worth a read but if you are short of time I hope Michael won't mind if I post this extract - quote:

"The bottom line is, FFF 3D printing is surprisingly robust to non-trivial under-extrusion in the range up to 10% under-extruded, and possibly higher depending on your requirements."

IMO, over extrusion is very noticeable but I agree with Michael regarding under extrusion. So much so that I simply run 95% extrusion multiplier as my default setting and very rarely change it.

klcjr89

I am working on a sensor that measures in more than 2 axes and should eliminate all under and over extrusion issues for good. It even has the benefit of allowing a user to use random diameter filament without having to worry about issues.

Even good quality filament is far from consistent. It may be '1.75mm', but has an egg shaped cross section or non consistent cross section across the length of the spool.

This also doubles as a filament run out sensor.

thwe

Manufacturers of extrusion products (including filamtent) use for example sensors from sikora, which operate on the principle as shown in the following image

alt text
The product creates a shadow on the CCD line sensor. In this case, the number of dark pixels on the line sensor is equivalent to the diameter.

Decisive for the measuring accuracy is the ccd sensor

klcjr89

All of the parts have arrived for the sensor. If I can get everything designed and assembled, it will be a game changer!

devleon

@klcjr89 care to share some more details about how you're going to build it and what parts?

klcjr89

@devleon The parts were very expensive, which always seems to be the case with any prototype project I undertake haha.

The heart of it is a transducer, or known as LVDT in the industry.

nikscha

@klcjr89 sweet! Keep us updated!

devleon

@klcjr89 Holy ***, LVDTs are expensive, you weren't kidding. Definitely nothing for me if the sensor costs more than my 3D printer.

klcjr89

@devleon I got it new old stock, so wasn't too bad If the idea works, the manufacturer has volume price discounts!

DocTrucker

Work with one and rotate the assembly around the filament using a spare stepper drive, or perhaps target that as stage 2 of the testing. Read up on the laser filament monitor tests and get some of the filaments that caused large variations or problems with them.

How are you planning to verify your system? I've found the experimental noise on the weight of small test parts to be significant.

Solving under and over extrusion issues is a great aim but filament diameter is one part of the puzzle which other projects are also working on such as the magnetic filament monitors towted potential capability to automatically calibrate non linear extrusion.

I'm tackling calibration process issues myself this week and have decided to test setting my extrusion multiplier to 1 then adjusting the filament diameter to correct wall size, adjust extrusion multiplier over a sequence of solid 20*20mm test coupons, and finally tackle offset and scaling issues. This is before taking pressure advance, retraction settings, and extrusion correction in any detail.

Setting the temperature has been far too vague for my liking so far and mucks up other calibrations when I've had to change it later on.

Edit: I'll know I've got things nailed when a causal observer can't accurately tell what machine a test part has come from. I feel a coke/pepsi style test with some dissinteredted family members coming up!