Hey! This thing is still a Work in Progress. Files, instructions, and other stuff might change!

Remote & Direct Drive Extruder with dual drive gear-set V2

by lorinczroby Dec 16, 2019
Download All Files

Thing Apps Enabled

Please Login to Comment

Comments deleted.

Great design!
I'm working on nema8 wormgear extruder and I took great inspiration from your project (see attached image). I'm using a 1:36 gear ratio to overcome the very little torque of the nema8 but I'm still testing.

Can I ask you how you design the gear set? I'm having a lot of issues parametrizing that thing.

Hi, Thanks,

I also hit the issue of proper worm gear design, to get the shape correct is no easy task. I'm also not very good at good wormeager design, I've searched many variants from other but when you take a close look most of them a not really functional, sectional analysis reveals the gears intersects at some points. What I did I found 2 good sources of worm-gear designs done by professionals who offers the cad model as well. I took those models and scaled/cutt/shaped to my needs:

that's incredibly useful! Thanks! Have you tried 3d printing the gears with a resin printer? That helped me a lot getting consistent results (but ultimately I can always see a little slant pattern especially on 1-wall extrusions)

I have no access to a good resin printer, I just printed them out of PETG. I cut the worm in half and glued them together afterwards. This gave me very good accuracy for it. I'm wondering if the resin printed parts are strong enough, I tried printing out of IGUS I180 but all the gears gave up pretty quickly. With well lubricated PETG never had problems.
I'm trying now to design another version with the big 12mm diameter dual bondtech style gear version, hope this will further increase the extrusion accuracy since there are more tooths touching the filament and do not bite in the filament so much as the small gears.

I'm printing with some "technical" resin and so far it works very well but I don't have long term stats yet. It's ABS like toughness and very smooth. I can print you some gears for testing if you want (on me)!

Hi Robert, thanks for the cool design!
I'm still in Upgrade-It-All phase with my ender3 :)
Wanted to build your extruder, but decided to wait for nema14 version info/test results
Did you see the old design of using Cycloidal gearbox https://reprap.org/wiki/CycloidalExtruderDrive ?

Very interesting gearbox, thanks for the hint, I'm looking for a solution for my next extruder version in which I would like to use the big 8mm gearset from trianglelab. From my estimation there I would need about 30:1 gear ratio. Although the cycloidal gearbox seems to be heavier than a worm gear based one.
The version with the NEMA17 works better than I expected. The printing accuracy is much better, the waves on the surface are much reduced, I will upload some comparison pictures. On the downside the 50g nema 14 stepper is not that powerful so the maximum extrusion speed I got is about 5mm/s on a 0.4mm volcano nozzle. This is much less to what the volcano can do, it's more or less what a standard V6 can do. Anyhow it is not bad this is equivalent with 150mm/s printing @0.4mm width & 0.2mm height

HI, great design. I go to test it on my Tevo Tornado. But ... did you seen this https://flex3drive.com/g5/ , advertise from their website: "40:1 gearing + 4mm diameter shafts ensure bullet proof performance"

Thanks you, I have seen from them the old version. This seems to be much improved :). What I have experienced with the high gear ratio is that the pankake nema 17 extruder motor cannot keep up with acceleration / unless you limit it a loot this will slow down your printer very much especially with small prints. I had acceptable result using a nema 17 pancake plus the 10:1 gear ratio...
Now I'm also playing with a higher gear ratio 20:1 and 30:1 but direct extrusion mode using a small nema 14 stepper 50g, will publish the test results soon, looks thi combination works better than the nema 17 + the flexible driveshaft + 10:1 gear ratio.

Seems I need put my testing for stop now and wait you about 14 nema. What I mean, is that I need solution to my rework/redesign my Anycubic Kossel Pulley to Anycubic Predator and Tevo Little Monster mix. It is 90% done (on fusion 360) and wait for extruder solution. Yes the cost of materials (starting from zero, I mean you not going rebuild existing delta) right now is around $150 and I have same amount room to not hit Predator price.

You have to decide which option you would like to build.
Of course I can confirm the print quality of the direct extrusion using the nema14 stepper is better compared to the remote extruder version. This is somewhat expected since the elasticity effect of the flexible driveshaft is removed. The added additional weight of 50g for the stepper do not make a big difference.
Nevertheless I plan to further improve the design, like using 8mm drive-gears and normal gearing instead of worm gear so I could use the back pressure in the filament for a better retraction performance, so for me its a living project. I will upload some comparison photos as soon as I finish the calibration of the extruder with the nema14.

Very cool design and open source! I've been looking at trying my hand at some of these designs too, but I believe your design is going to be better than my abilities. As I was thinking about designs for this type of system, I had originally put the gearing back at the stepper motor to reduce weight even further. Would this result in too much play at the extruder gears?

Thank you, I think some gearing help to reduce the twist on the flexible shaft. Some even working with higher gear ratio like 30:1 plus a gear up at the extruder motor like 1:3 this reduces the effect of the shaft twisting.

Great design... i'll watching for nema 8 options..

  • Posted with 3D Geeks Thingiverse Browser App

Thanks, The nema 8 version is not tested yet, so any feedback would be great :)

Im going to try to build one as well, but what's the inner diameter of the ptfe tube for the flexible shaft? 3mm or 4 mm?

Cool. I have been designing something similar but with a "flying" stepper with a shorter shaft and a lower gear ratio and bevelled gears rather than a worm. It's cool that you got the printed worm to work.

I'm curious whether you think a flying stepper with a very short 30-100mm shaft and a lower ratio <10:1 is a good idea. One problem is allowing the motor to sway while keeping the rotation locked, not allowing it to twist. Ideally a flexible hollow shaft. But I thought to reduce artifacts due to effector mass you only have to decouple the mass from the fine movements, not remove the mass entirely.

Yes, you are right. a flexible coupling would decouple the high frequency vibrations which causes the ringing. I have seen designs with double XY axis o top of each other the top one was carying the bowden extruder with a very short bowden tube. For a delta printer maybe a good idea but for a core xy where the X axis assembly is towards 1kg...100 grams would not make considerable difference...better try reduce to half the 1 kilo weight.

I have seen those double xy ones too. Cumbersome, but I guess good for heavy multiple extruders. Could potentially bear all the weight of the effector axes.

Yes, I don't know why people use MGN12 on corexy x axes. I plan to use MGN7 with two carriages, 1/3 and 1/2 the weight on the y and x axes, respectively. Trying to come up with a lightweight support, maybe a square carbon tube, or 4mm aluminum plate. Hopefully with such efforts a remote direct drive will be helpful. I wonder whether a heavier flexible shaft could be sourced that would allow a lower gear ratio?

BTW I was testing this globoid worm gear for such an extruder, I got it running very well with petg at dremel speeds. Very large mating surface! It's even hard to separate the gears. I think it was 1:10 ratio. https://www.thingiverse.com/make:708592

globoid worm gear drive - parametric OpenSCAD code with large ratio range
by haleyp

Hi, I also tried the globoid worm-gear generator you have, my issue was that given the very small size and the gear ratio of 10:1 the globoid actually was worse than the normal worm gear I have.
About the MGN7 interesting idea, but maybe it will not lasts so much...too small balls inside...What I plan to do now is to make my X axyx parts out of 5mm carbon fiber plate instead of aluminium and plastic, and go for the short carriage version the MGN12C instead of MGN12H, this is about 35g so it's like two MGN7 carriages., Another Idea I have since a long time, is to cut out a part of the MGN12 steel rail core and replace-it with carbon fiber rod. But I would need a very fine cnc machine to do that...but maybe with the hiwin rails I have is not even possible, I have not measured how strong the material is.
Why you would like to go bebelow 10:1 ratio?

Interesting thoughts. I wondered about MGN7, but I don't know why it would wear out, with all steel and only a light preload. With two carriages only one moment of the 3 is significant, and the belt pulls more parallel. I don't like the MGN12C idea because the rail is by far heavier than the carriage, it loses rigidity but not much weight on the y axis. MGN9?

I don't mind 1:10 but at that ratio I worry about the play in the flexible shaft. At 1:30 I worry about retraction and extrusion acceleration and speed. Maybe I'm wrong and it's fine. How about using a 1:30 at the extruder but a 3:1 at the stepper, so the shaft rotates faster? More friction losses but with a big enough stepper, could help diminish the effect of the shaft.

I'd be interested in where you source your long carbon plate. I thought of building a composite sandwich from aluminum sheathing (0.2mm) and extruded polystyrene foam. I want a C-shaped plate because I have two MGN9 carriages at either end of the y axis, because I'm building a H-bot instead of corexy. Simpler, shorter 9mm belt. Just need the spaced carriages to counteract the one disadvantage of shear forces, and the plate to span them.

I thought of boring out the rail too. True, I think very difficult. A routing table? Maybe one could convince the Chinese manufacturers to do it.

From my experience the most important in the X rail is how much it allows the printhead nozzle to "play". Higher preload and wider rail gives lower movement or play of the printhead nozzle. Maybe an MGW7 would be a better option...its 2mm wider than the MGN12 and less weight, the carriage is also about half the weight.

Increasing the shaft speed would also a good idea. would reduce the flexing force on it, I have not tried it..

From the carbon fiber plates I want to make the X rail attaching parts to the Y axix. The X rail will remain naked. But you can find many carbon fiber square or rectangular rods on aliexpress to use for stiffening the X rail.

globoid worm gear does look helpful here. Just asking, is it easier to simply buy metal gears from aliexpress/ebay?

I bought this (at half the price) and made a working prototype extruder without the shaft. I wasn't totally satisfied with it. You might be right that its better than printed though. no grub screw in the worm though.


The problem I faced was that I have not found any which can be integrated for this purpose, specially the worm, some bearings must be attached to the two ens plus a way to insert the flexible shaft... I have fount no good solution up to now, beside making one on a machine.

Yes, the brass worm above I had sandwiched between two 10mm bearings, tightened with an m3 bolt. Unfortunately the flexible shaft is 4mm and the worm hole 3mm. I had planned to use a coupling.

I am very surprised that PETG works for the worm. I will give this a try. Any reason why you don't use a speedometer cable?

Well I tried many materials and PET-G lasted best. I even the very expensive IGUS filament, was working the smoothest without any lubrication but did not lasted much, the gears gave up pretty fast. The cable I used is actually a kind of speedometer cable. I used the one from the BOM because I have no speedometer cable I have the one from the BOM :).

I think I would have tried Nylon instead.

Yeap, from material properties perspective that's maybe is a better option I have not tried yet, I have no Nylon filament.

Ty Robert for great job! I integrate your remote extruder to my printer.

Thank you! Very nice :).

Hi Robert,

following your recommendation I ended up printing the gears in PETG :-)
I´m in progress building the extruder right now and it´s looking good so far. There is one question remaining:

How do mount the extruder to the parts holding the hotend?
Are you using the heatbreak, only? OR do you use the M3 screws of the extruders housing as well?

Some pictures here might be helpful, too.

Thanks for your response and help in advance.
Best regards, René

Hi, have you had the chance of testing the extruder?
By the way the gears printed out of the IGUS material is a big dissapointment. It does not last much. I changed back to petg plus lubrication...with this opion I never had issues.

Hi Robert,
I´ve finished the build of your extruder with PETG gears and it is looking promising :-)
BUT.... my BLV cube build is far away to be ready for any tests.... :-(
I´m running out of time at the moment, so it will last some more weeks, I guess. Will keep you updated.
Best regards, René

Hi. The heatbreak is not enough. I have attached it via the two M3 screws. but my printer is custom made. What printer you have?

It´s a BLV cube in general, but the printhead and mount will be custom made.
Thanks for your hint. I know where I do have to look for.
Guess I will use longer M3 screws using these to mount the extruder to the other parts.

There is also an additional hole on the bottom housing part. I use that and one of the long M3 screws. The other M3 screw hits an aluminium part in my hotend design. I have attached it using insert nuts.

Hi Robert,
how does the 5mm carbon rod fit in the PTFE tube? all the PTFE tubes I found was 6mm OD and 4mm ID...
If I have a tube like that, it will be ok to get a 4mm carbon rod?
Amazing work by the way!

Hi, I do not really understand your point....The 5mm carbon rod is needed to manufacture the shaft on which the gear and one the mini bondtech gears are fitted. In the PFTE tube hods the flexible shaft between the Extruder motor and the extruder itself.

Hi Robert,

first of all: "Thumbs up" for all your work to this extruder :-)

Would it be possible to get the gears as CAD files, please?
Reason: IGUS is offering a print service, accepting STP or STEP files only.
They are using the filament you´ve suggested. Prints are possible in FDM and SLS.

So this should be the best solution for long lasting gears and a success for the rebuild.

Best regards, René

Hi Rene,

I have uploaded the gears step files, I will share the whole poroject source file from Fusion360 just I have not the time now to prepare it.

I would recommend first to try it out using printed out of PETG. Check if you are satisfied with the Printing quality/ I have to admit my printer produces better quality at low speeds (50mm/s) using the direct extruder . I still prefer this one since I can have 95-98% same quality but at 150-200 mm/s which is not even possible with the direct extruder.

I just finished printing the gears out of IGUS Iglidur I180 filament. After fixing warping issues - warps worse than ABS...I got better surface quality compared to the PETG. But the strength of the gears is less.
Especially for the worm...I printed the two halves and glued together.
With one of the warped prints I tried some brake test ...I was able to brake the worm gear fins with my finger...The PET-G printed ones I was not able to brake.
Up to now it seems that it make sense to print the gear out of IGUS and the worm out of PETG.
Now I have both out of IGUS it works much smoother compared to PETG Let see how much it will last. With the PETG I have more than 48h of printing and not even sign of wear.

Hi Robert, thanks for your reply and help.
Igus is preferinging the SLS filament iglidur I6-Pl for worm gears especially. I will send the gear files asking for their expertise and quote.
I will keep you informed...

Thanks. Printing it with sls would definitely give a better result.

Hi Robert, here I´m back again.
Just checked the printability in SLS on the IGUS webside..... The worm gear is showing areas were it will not be printable successfully (see picture attached), because the thickness is below 0.7mm.
Do you see any chance to change this in the design? The areas in question seems not to be used in the gears operation, so it might be possible, just extending the main tooth in its original size !?

Beside this, a single gear in this size is approx. 10 to 12€ + shippment. It´s going to be cheaper when ordering more. Who wonders... :-)

Hi Renner,
Unfortunately I cannot modify the tooth. I got them already made, I just adapted them to this design. I'm not so good in gear design especially worm gear which is a bit more complex if you want to do it right.


I love how clean it looks with direct drive mounted vertical! Fingers crossed they have enough torque.
Is there any changes to the main housing from V1? Thanks for your work.

I'm waiting for the motors delivery to test it. With the 20:1 the overall torque is at the limit maybe will work a bit slower but perhaps more precise. The housing is changed because the wormgear is 2 mm wider and the diameter is smaller so I have adapted the dimensions of the housing accordingly.

Machbox v2.0 , can you send a link please.

Great to see others implement an idea i had.
Good work, keep going!
Wondering if this wire gives you any backlash? And how does one measure it?

Hi Thanks!
I have not measured the backlash, however from my understanding the phenomena is the same with the bowden tube. In the bowden tube due to the elasticity of the filament and the PFTE tube pressure builds up and delays the extrusion. The same we have here there is a delay because of the torsion in the drive wire but this can easily compensated by pressure advance. I have not done yet a comprehensive check of it just setting different values and checking the result. Looks like 0.1s is a good value.
Nevertheless the backlash amount is linearly increasing with the printing speed which is not the case for the stech/compression of the bowden tube.