2
$\begingroup$

I have a Flex3Drive remote-direct-drive extruder I bought for my Ender 3 (with the original extruder motor), and I'm pretty happy with it except for abysmal retraction performance due to the 40:1 reduction, which requires the motor to turn about 480° to retract just 1 mm. This is minimally workable for PLA and rigid materials where I can get by with just 1 mm of retraction, but it takes just as long to retract 1 mm as the original bowden extruder did to retract 6 mm, defeating a lot of the appeal of a direct drive. For TPU I haven't been able to make it workable at all. By the time I retract 3 mm or so, which seems necessary to avoid stringing, the hotend has already melted and/or deposited a blob on the print. Failure to print TPU also defeats much of the point of a direct drive.

I'm running retraction at 8 mm/s, 500 mm/s² acceleration, which is already higher than the manufacturer of the extruder recommends, and about the fastest I've been able to get it to work. Based on this calculator and 8.9 mH motor inductance, that seems roughly expected. I have TMC 2209 steppers and tried playing with current but it doesn't seem to make much difference.

Otherwise, I love this extruder, and want to make it work. Is there a way I can salvage it by changing out the motor for something that can do higher RPMs and accelerate just as fast or faster?

$\endgroup$
2
$\begingroup$

In general I would use https://github.com/rcarlyle/StepperSim which takes into account more parameters.

You can play with voltage and current to see which combination gives you the best results for your motor.

Or you can change to a TMC driver with higher voltage (35-50 V) to keep torque at much higher speeds and push the current motor more.

Since the torque you require is likely not so high, you can increase the speed of your stepper motor with 3D printed herringbone gears, for example 4:1. They don't need to be super accurate, backlash is totally fine considering the ridiculous 40:1 reduction.

$\endgroup$
4
  • $\begingroup$ Yeah, adding a gear on the motor end of the flex shaft seems like it should be doable, just replacing the existing 3D printed linkage there, but I suppose I'll have to do at least some custom design. Do you know any good designs I might look at as an example or to customize? $\endgroup$ – R.. GitHub STOP HELPING ICE Mar 18 at 0:22
  • $\begingroup$ Maybe this, but with the shafe configuration for the input and output reversed? thingiverse.com/thing:3641458 The main adaptation seems like it'd be making the output shaft of the motor fit what was intended to be the output of the gearbox. The new output can just be merged directly with the existing 3D printed part that goes to the flex shaft. $\endgroup$ – R.. GitHub STOP HELPING ICE Mar 18 at 0:28
  • 1
    $\begingroup$ That's a good option indeed, it only extends the length of the motor, which is usually not an issue $\endgroup$ – FarO Mar 18 at 9:26
  • $\begingroup$ I've accepted your answer since the idea for how I solved this basically came from what you wrote, and added my own answer with some info on what I did (which I might expand on later). $\endgroup$ – R.. GitHub STOP HELPING ICE Mar 28 at 20:23
2
$\begingroup$

While the following doesn't conclusively prove changing the motor would work, I think the answer is yes! And I achieved the desired result via a different method, so I consider the problem solved.

Instead of changing the motor, I designed a 1:4 step-up planetary gearbox to put on top of the extruder motor, yielding a 10:1 net reduction instead of the original 40:1. This works, even without much attention to part strength or backlash in the gearbox (thanks to the subsequent 40:1 reduction), yielding the desired sub-50ms retractions with PLA. This seems to establish that the rotational rate or acceleration of the flex shaft is not a limiting factor.

One thing I noticed after adding my gearbox is that, even after scaling up the extruder speed and acceleration to 4x the values I was using without the gearbox, the stepper motor did not make the sharp "chirp"/"squeal" sound on retractions that I got before. This seemed odd, because it should be the same speed/acceleration for the output shaft and slightly higher load than before. So I think there may be something other than just exceeding the stepper's capabilities going on with the original configuration.

The manufacturer's recommended (provided STL files) coupling of the motor to the flex shaft is this long rigid coupler and guide cap:

coupler motor cap

That looks like a suspiciously long lever on the motor output shaft. And while the cap retains the flex cable sheath fairly rigidly, it can't be perfectly aligned with the motor shaft, and there's room for the flex shaft to move slightly side to side inside the sheath. So, my guess is that spinning the flex shaft at high speeds with it directly coupled to the motor like this put some serious lateral forces on the motor output shaft, interfering with the motor's performance and likely leading to long-term damage if not corrected.

The gearbox I added avoids extending the motor shaft at all (the planetary gear carrier slides fully over the D-shaft) and the small amount of backlash in the gears themselves allows the flex shaft positioning to be imprecise with (apparently) no serious ill effects. And if it does cause unwanted wear, well these are printed parts that are easily replaceable.

As for the 40:1 reduction in the extruder itself, after working with all of this to get it printing well, I don't think it's necessarily a wrong design. The flex shaft can handle the speeds needed just fine, and the 40:1 reduction keeps torque from the extruder gear off of it while extruding. The Zesty Nimble (which some call a clone of this design) used 30:1 instead of 40:1, and has dropped to 20:1 in their latest iteration, probably for reasons related to my troubles with the 40:1. But I think the gearbox at the motor side is actually a rather nice solution that preserves all of the potentially good properties of the high reduction and none of the bad ones. It does make the setup more complicated to manufacture and install though.

$\endgroup$
5
  • $\begingroup$ I wonder how they picked that huge reduction. $\endgroup$ – FarO Mar 29 at 8:55
  • $\begingroup$ @FarO: I wouldn't assume it's a bad choice for the reduction after the flex shaft. Being high reduction should prevent wear on the shaft and error due to twisting of the shaft. But failing to step up before it seems bad. I also have some other ideas I'll add in a bit. $\endgroup$ – R.. GitHub STOP HELPING ICE Mar 29 at 12:00
  • $\begingroup$ @FarO: I added a lot more now. $\endgroup$ – R.. GitHub STOP HELPING ICE Mar 29 at 13:32
  • 1
    $\begingroup$ That long lever on the carriage side is a weak point: youtube.com/watch?v=qvY_dChxj0E $\endgroup$ – FarO Mar 29 at 15:31
  • $\begingroup$ @FarO: Thanks for the link. I'm not too worried about that because I can justprint a new cap if needed. The Orbiter does look very attractive (and addresses the slow retraction issue) though. $\endgroup$ – R.. GitHub STOP HELPING ICE Mar 29 at 19:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.