We were excited to learn that someone had started working with force sensors on filament extruders, especially after we posted about a recent development in filament thickness sensors.
[airtripper] primarily uses a Bowden extruder, and wanted to be a little more scientific in his 3D printing efforts. So he purchased a force sensor off eBay and modified his extruder design to fit it. Once installed he could see exactly how different temperatures, retraction rates, speed, etc. resulted in different forces on the extruder. He used this information to tune his printer just a bit better.
More interesting, [airtripper] used his new sensor to validate the powers of various extruder gears. These are the gears that actually transfer the driving force of the stepper to the filament itself. He tested some of the common drive gears, and proved that the Mk8 gear slipped the least and provided the most constant force. We love to see this kind of science in the 3D printing community — let’s see if someone can replicate his findings.
20 thoughts on “Experiments With A Bowden Extruder Filament Force Sensor”
It’s an old post, nobody has duplicated or improved on it yet.
You should be able to do the same with a direct drive extruder, by putting the sensor between the hotend and the extruder. It would even double as a contact probe for z levelling.
And the mk8 gear has one fault: one set screw that only engages 1.5mm of material. This could be rectified by redesigning it slightly to have the portion where the setscrew is thicker, it shouldn’t interfere with most designs, and the setscrew already sticks out.
It can also be used to calibrate temperature for most filaments, though viscosity varies it can help find the sweet spot. And it can detect clogs. Combine it with a filament diameter sensor and your on your way to making a plug and print machine.
This. Filament force sensor + z probe in the same. I’m going to try this out with some cheap load cell.
We’ve done the tests at Ultimaker. But didn’t publish the results. However, the basic result wat the same. A mk8, if proper cut, has the most grip.
What makes most difference (with PLA) is how sharp the teeth are. Dull teeth is less force till it starts to slip.
However, we did do 1 major discovery that let us back to the UM style teeth:
And that’s a straight cut tooth, while having more grip, loses this grip when it gets dirty and filled with grind material. And needs cleaning then. The UM hobbed style bolt however, cleans itself and keeps grip even when it grinds.
Attaching a few toothbrush bristles removes the plastic debris from the teeth as the extruder spins:
(Also note the double-side grip there; seems to be quite effective.)
First, with a correctly set up driving gear and hotend you can print away a whole spool without a single debris accumulating between the teeth.
Your double-side grip is less efective then using a simple ball bearing, simple because it is idle, only one of the gears are driven by a motor.
Your toothbrush is really just a patch for a problem which should not exist in the first place.
A force sensor would be useful to detect if there is filament coming out at the other end of the hotend,
aka. a safe measure. All the other listed usecase are workarounds…
The outer gears engage, so it is not idle.
But yeah, I can believe that it is possible to make a perfect setup that never grinds. I’ve not yet achieved that, so I’m happy to have some useful workarounds in the meantime. As far as I know, there is no foolproof way to achieve such a perfect setup, it is still more of a trial-and-error.
Absolutely. I have printed over 5KM with a straight cut 8mm drive gear with a bearing idler without a single incident. These people typically introduce more problems trying to solve it the wrong way, especially when there really wasnt an issue to begin with and all that was needed was little adjustment.
Tip to all: You do not need nearly as much pressure on your idler as you are currently using , back it up a bit. If you DO need that much pressure , then the problem lay in your filament path after the extruder. Fix that.
If both gears are powered (connected for example) then a double grip is more effective.
A force sensor is a more expensive solution then self-cleaning. And having your feeder properly calibrated depends on the filament.
Self cleaning just improves reliability for us (as Ultimaker).
Note that you get grinding and dust generation when you put too much or too little tension on your idler. And your idler has a different perfect tension depending on PLA, ABS or other materials.
I’m using the flint wheel of a lighter, it has even better characteristics than the MK8: it smaller (more torque), it self-cleaning (more or less same pattern as the UM), it’s hard as hell (high carbonated steel). The only inconvenient is that it can be tricky to install on a 5 mm stepper shaft. There is actually possibilities: either a coil (typically from the BIC brand and others solid wheels that require further machining like from the Clipper).
If you’re interested in more details, check on the Saintflint (http://www.thingiverse.com/thing:979113 or https://www.youmagine.com/designs/saintflint-extruder).
Additionally, it was fitted on a double action extruder that is as simple as it is efficient with a not-so-frequently-seen-design.
I built one of these yesterday. The design is super clever. I loved the m4 nuts for stopping the tubes, the pivoting on the nema screws, and the main threaded tightener. All in all it’s an awesome design.
Havent had a chance to see how it performs over time, but it did very well on my test print. No slipping or grinding, and i was printing pretty quick too.
I made one of these also. The double contact makes it extremely unlikely to slip. When I first saw it, it was unclear to me why it was so much better, after watching the video and seeing it work, this appears to be one of the best extruder designs out there and it’s very compact. A lot of the 3D printer parts are incremental improvements over previous designs.
For now I need a compact bowden extruder with a lot of power, and this Saintflint (http://www.thingiverse.com/thing:979113 or https://www.youmagine.com/designs/saintflint-extruder) fits the bill just right. I’m using the mk7 version.
Later, when I build a bigger printer, I’d like to try the Libro-Struder (http://www.thingiverse.com/thing:1212687). I’m intrigued that it gripped the filament so well that it broke when he tried to pull it out of the extruder.
Both are very different than the other designs I’ve seen, and appear to work well enough for me to think “why didn’t I think of that.”
This could be useful to compensate for variations in filament thickness. I would imagine it could be adapted to use a combination of filament speed, stepper motor current and extruder temperature to calculate filament volumes.
There’s a good reason the Airtripper is the go-to Bowden for most printed extruders, and it’s at least partially because of this kind of verification
That is far from true. The Airtripper is ONE of the go-to, non-geared extruders , but just about any geared extruder will be the number one go-to for bowden. The Airtripper really doesnt offer anything over other non-geared extruders other than printability. Mechanically , it is essentially identical to the majority of other non-geared designs. Its a fine non-geared extruder however. For bowden though , you will get along much better with a geared extruder. A compact design such as the Infill3D ones are quite nice and much more capable than a non-geared.
I was always wondering why there is no extruder which uses a screw in parallel to the filament, like a worm gear thing or maybe not 100% parallel maybe with an angle 10-15 to get pass the stepper motor. The screw’s teeth and the filament should have a lot of contact.
It might twist the filament around too much.
There’s these two: http://www.bondtech.se/ (my favorite atm) and this crazy belted thing: https://0x7d.com/2014/07/improved-ultimaker-2-material-extruder/ .
Older extruders used this idea, but they were larger, heavier, more complex, and slower. It’s a little hard to find older information, but here’s a nophead post circa 2009: http://hydraraptor.blogspot.com/2009/04/gm17-stepper-hack.html and you can see the screw drive towards the bottom.
There is if you look. http://www.thingiverse.com/thing:209136
Interesting discussions. I found this whilst looking for force data. I am working on a modification to the Gunstruder, in the hope that it will have near zero filament damage. As part of this I have been working on my own design of a Bowden filament force sensor. I have published details on thingyverse. https://www.thingiverse.com/thing:2429390
I hope this will be useful to the 3D printing community.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)