Quote
VDX
... we've made the same here in Germany (or EU) IP-range -- filed a patent and abandoned it shortly after, so no other party can "block" our open-source development with a similar claim

Quote
OutcastZeroOne
very cool to see how this design has progressed. Hope to see it continue to compact down, and hopefully not get stolen by anyone. Keep up the great work

Quote
oliof
@rq3 would a hollow shaft stepper like the OMC 17HS13-1504H be an option? ID of the shaft is 4mm, I'm thinking of a PTFE tube insert (OD 4mm/ID 2mm) to constrain the filament better ...

Quote
MarksAlot
To me, an ideal extruder would measure both diameter and travel, calculate volume dispensed, and use that to close the servo dispense loop.
Thermwood does similar with their pellet extruders.
They do a pressure condtrol loop with the screw, the melt then flows through a gear pump to meter/measure the flow into the nozzle, and nozzle velocity follows melt volume dispensed.
We could do similar. A strain gage for pressure control, and a filament encoder+diameter gage right at the cold end entrance. I know all have been done seperately, but not sure volume metering+ pressure control+ nozzle velocity following + predictive/adaptive/lookahead has been rolled together outside high end pro-level printers.

Quote
MarksAlot
I'm looking at something a little closer to a Uhing rolling ring drive,
I've done miniature versions of the Uhing on 1/8" piano wire as a linear actuator. Somewhat more than 20 lbs of thrust.
I'll have to support filament a little better if I can get it to work.

Quote
dekutree64
Ah, I'm glad I didn't bother with magnets yet then. I recommend using coarse thread screws if you're screwing into plastic. Fewer turns to take them in and out, and at least in my experience machine threads in plastic tend to wear out more quickly.

I have unloaded mine once, and with some gentle turning back and forth by hand I was able to get the lumpy end of the filament through the rollers without having to unscrew the mount. Probably would be better to do it while the filament is still warm so it puts less pressure on the rollers, but it will take some experimentation to find the right time window when it's soft but not sticky. Another option would be to retract part way and then reach in with some nippers and cut the filament between the rollers and heatsink. Then purge the remnant through the nozzle when loading the next filament.

My two-post mount is very nice in terms of usability, since it gives such a clear view and finger access. The long screws support the upward pressure from extrusion just fine, but a three-post mount or full cylinder like yours may be better for high lateral acceleration, to better resist whiplash in all directions. You only really need finger access from one side since it's so easy to turn, but loading and unloading is twice as fast when you can use both hands to spin it.

I look forward to the results with your NEMA 11. My guess is that it will be a bit underpowered. The NEMA 14 seems just right, with enough headroom that you can run cool without worrying about missing steps. Though with an aluminum carrier, a heat-tolerant mounting surface, and no bowden adapter epoxied to it like mine, the stepper should be fine running a little hot.

I still want to try the outrunner design I posted before as well, especially now that we know the anti-torque device is unnecessary. But I won't have a place to put the bowden adapter, so I may have to do it on my other printer. The design is a little mechanically iffy since it puts the extrusion force pulling upward on the roller screws, and pulling upward on the rotor so the shaft retainer clip will have to be shimmed just right so there's no wiggle room. But it should be about as lightweight as you can get.

Quote
MarksAlot
How do you go about getting sufficient grip should the filament change diameter?

Quote
dekutree64

Quote
MarksAlot
I've been working on a similar idea, kind of inside-out. Just have to machine a few more parts.

By inside out, do you mean using an outrunner motor, or something different? I look forward to seeing it. I just finished constructing my outrunner idea that I wrote about several posts back. Check it out! [reprap.org]

Quote
rq3
Thomas Sanladerer of YouTube fame just did a segment on the new E3D hot end, including measurements of just how much force it takes to squeeze molten plastic out of a nozzle. It's not as much as I thought, and the Schnekenstruder has a mechanical advantage of about 21:1.

Thanks for the heads-up, I hadn't seen that video yet. Brilliantly simple way to measure extrusion force. That integrated nozzle/heatbreak design does look great for eliminating the possibility of leaks between the heatbreak and nozzle, but I think the patent will ensure that regular threaded nozzles remain the standard.

What exactly do you mean by 21:1 mechanical advantage? I thought it would be a leadscrew type calculation. In the case of mine with 288 steps per mm, that gives a lead of 400/288 = 1.39mm per revolution, so 1Nm torque should give 2pi/.00139 = 4520N linear force. According to the Sanladerer video it only takes about 1kg to extrude, so that's only 2.2mNm. And the NEMA 11 is rated for 15mNm, so it probably is indeed enough, especially since you don't need microstepping.

Quote
MarksAlot
I've been working on a similar idea, kind of inside-out. Just have to machine a few more parts.

Quote
MarksAlot
I've been working on a similar idea, kind of inside-out. Just have to machine a few more parts.

Quote
dekutree64
Ah, I'm glad I didn't bother with magnets yet then. I recommend using coarse thread screws if you're screwing into plastic. Fewer turns to take them in and out, and at least in my experience machine threads in plastic tend to wear out more quickly.

I look forward to the results with your NEMA 11. My guess is that it will be a bit underpowered. The NEMA 14 seems just right, with enough headroom that you can run cool without worrying about missing steps. Though with an aluminum carrier, a heat-tolerant mounting surface, and no bowden adapter epoxied to it like mine, the stepper should be fine running a little hot.

I still want to try the outrunner design I posted before as well, especially now that we know the anti-torque device is unnecessary. But I won't have a place to put the bowden adapter, so I may have to do it on my other printer. The design is a little mechanically iffy since it puts the extrusion force pulling upward on the roller screws, and pulling upward on the rotor so the shaft retainer clip will have to be shimmed just right so there's no wiggle room. But it should be about as lightweight as you can get.

Quote
dekutree64
Finally got around to finishing my poor man's schnekenstruder, and it actually works! At least mostly. For some reason it doesn't seem to be able to retract. Even with jerk/acceleration set very low, motor current set much too high for continuous running, and retraction distance 0.5mm, it still just makes a suffering noise and fails to rotate backward. But if I command it to rotate forward and backward from the move axis menu, it does it just fine. Takes no effort to do it by hand either, so it must be some sort of software issue.

The motor is NEMA14 pancake. I epoxied a 3D printed pneumatic fitting adapter to it for reverse bowden setup. The tube should also serve the purpose of the anti-torque wheel, although from the other thread it sounds like it isn't necessary anyway. I'm still using the bowden heatsink, so there is about a 15mm space between the rollers and where the filament goes into the 2mm heatsink hole. But I don't think it's enough to matter for PLA, which is all I plan to use on this machine.

I reworked the carrier for 4mm bearings. The cutter is 4x8x3mm flanged, sharpened on both sides (to sharpen the difficult side of the flange, I used the same angled block as in my previous post, but with a diamond grit dremel wheel protruding off the top edge). The passive rollers are 4x9x4mm. I just used regular m4x10mm screws and split lock washers instead of the fancy shoulder screws and bellevilles. I did have to file a bevel around the edge of one of the lock washers since it was near impossible to keep it from jamming the smaller diameter cutter bearing. The other two didn't need any custom work, although it can take a couple tries to get them screwed down in the right spot to not jam the bearings.

I currently have it mounted to the printer with screws, but I may change to magnets. Haven't tried unloading it yet...

Thank you rq3 for sharing this brilliant extruder design!

EDIT: Got the retraction issue sorted. Apparently any speed above 5mm/s results in missed steps. Acceleration does not seem to affect it, so I've gone back to the default 10000 and jerk set to 10. Not sure why speed would be such an issue though. Steps per mm is 288 with 0.9 degree full steps, so 5mm/s is only 1440 steps per second. And that's 216RPM, so I wouldn't think torque dropoff would be the problem, especially since motor current doesn't seem to affect it either. I'm currently running a conservative 150mA. The motor is rated for 500mA, but even 250 gets uncomfortably hot after running for half an hour. 200 may be ok, will need to do another half hour test.

I definitely need to upgrade to ABEC 5 bearings. These cheap ones are pretty floppy, giving it some backlash. I can hear it every time it reverses direction, and after printing a retraction test the filament looks pretty chewed up due to not following the spiral groove back. I think the screws and lock washers are fine though. As far as I can tell, the inner race of each bearing is firmly held in place.

Quote
dlang
one bearing thick, but a largish bearing 10mm dia, x 4 mm tall is ~315mm^3

something 4mm dia and 25mm tall would be about the same volume (so similar mass, although bearings are not solid), but smaller diameter, so less angular momentum (i.e. easier to stop and reverse)

I am NOT saying that their design is better, just that it may not be as obviously worse as first glance would have it. I was just pointing out other similar ones, with the idea of cross pollinating ideas.

Your design looks easier to make the working end of.

I'm not thrilled about trying to drill through the center shaft of a stepper. I would just prefer to be able to use a BLDC with the solid pin replaced by a tube rather than trying to drill through the center shaft of a stepper (on many BLDC motors they are designed to be able to be pressed out and replaced)

Quote
dlang

Quote

It's interesting that all of these have rollers that appear to be 20 or 30mm tall. If you play with 3 pencils, it becomes quickly apparent that there can be only ONE central point of contact between 3 canted rollers, which is what makes the roller positions and spacing so critical.

Any roller taller than that one point of contact is just wasted space and mass.

The added length would give them room for the bearings and to get the bearings far enough away from the filament to hold them.

I could see tall skinny bearings being lighter than short fat ones (and lave less angular inertia)

you do get a little more than a single point of contact as the sharp edges of the rollers do bite in to the filament, but I agree it's not much more.

Quote

It's interesting that all of these have rollers that appear to be 20 or 30mm tall. If you play with 3 pencils, it becomes quickly apparent that there can be only ONE central point of contact between 3 canted rollers, which is what makes the roller positions and spacing so critical.

Any roller taller than that one point of contact is just wasted space and mass.

Quote
dlang
you mention a few posts back that you considered using a bldc, rather than re-wiring it, consider going with a smart controller like the odrive. It needs an encoder, but you don't need to go with a high resolution encoder like people normally do, as you note with 28 positions/rev you can get to 125 steps/mm, so a 7 ppr quad encoder setup (say with a copy photodiods) would be enough resolution, and the bldc will be very happy to spin faster, and to stop/reverse for pulling the filiment back.

a couple other youtube videos of folks doing similar things (but going with much smaller diameter, custom rollers) [www.youtube.com] [www.youtube.com]