Jump to content
Ultimaker Community of 3D Printing Experts

Zero gravity direct drive extruder

Recommended Posts

Maybe a stupid question, but how does the squared long rod glide through the extruder part when the head moves in XY ? will it wear down over time ?


Is rotates over bearings, they should probably outlast your printer. Or at least I think that is more probable that the printer x/y bearings where the shaft spins get wear down long before than this bearings.

Share this post

Link to post
Share on other sites


Thanks foehnsturn !

I guessed that was for that , but I was not sure, 3mm is quite a large security lifting distance :)


Well the 2nd nozzle sits approx 2.5mm lower so it's just 0.7-0.8 mm above the bed when the 1st head is printing the first layer.


Yeah! I understand better thanks ! ;)

Share this post

Link to post
Share on other sites

No additional work for you guys :)

There are some really smart people in the tool changer beta group who could adapt a ZGE UM2 head mount for the extra head if there's interest.


I still have one unmodified (well it has a bondtech) UM2, so I'm up to testing the ZGE on there first, get to know it, and than later see if i can modify it to the magnetic changer.

@gudo just let me know when you want me on board to test a version on UM2. If I can just get the files first I can already start by printing the parts....

Share this post

Link to post
Share on other sites

Again. First things first. We have a lot to do, and we will release versions after finishing the whole process, making manual, etc. We don't want to open a ton of paths and get many unfinished low quality products.

So, after we get the feedback from the betas and we finish all the other stuff, then we can take a look into other versions or adapter parts.

Edited by Guest
  • Like 1

Share this post

Link to post
Share on other sites

Recreus Filaflex is a very hard filament to control. There's a point where you can actually push more filament per mm/s but also it has the drawback that drips very fast. Today made a long distance retraction test, 'testing' not just how long I could retract the filament without issues, but also if it was worth.

Clearly there's a curvature on what's enough, what's too much, and what's too little.

2.0mm yields the best effects for long distance retraction, better than 3.0mm, but 4.0mm does clear a really good job when the distance from A to B is very small (5-7mm) leaving a close to zero string (as close as is possible with this flexible filament).

Retraction speed was at 100mm/s.

Basically there's a curvature on how effective is retracting more or less, the amount of time used to do 2.0mm is less than 3.0mm, but when you retract 4.0mm clearly there's an improvement on 'less' string when moving from A to B. So 3.0mm is worse than 2.0mm but 4.0mm is better than 2.0mm. Is a battle between speed/pressure and distance.


I leave the FFF profile on S3D if anyone wants to play with it. But mind that without a DD you won't be able to do 1k retractions and without a Bondtech just even don't try. Profile is for UMO+ with um2+ hotend, so change start gcode if you try on um2 and change filament size if you use 2.85 since this profile was done using 1.75mm.



  • Like 1

Share this post

Link to post
Share on other sites

A video showing some of the Tortures I been doing with this ZGE


Clearly I did learn some things about flexibles.

a) You can't stop the dripping unless you print VERY VERY slow, even then, is hard.

b) You can cut the dripping with long retractions of 4mm, but it highly depends on the viscosity of the filament. Semiflex seems the most promising.

c) Some flexible materials (smartmaterials specially) clog if you don't keep pushing and pushing, so they make partial clogs of filament.

d) Too much fan for some brands can be problematic, since the tip seems to cool down a bit too fast making a tinny bit more of pressure. This makes some patterns on flexibles that come from the extrusion flow. My fans are way too powerfull (my 50% is like 100% of um2 head fans). Installing a Nozzle sock is mandatory to keep the tip of the nozzle as hot as possible. This instantly helps a lot.

e) Way too flexible materials like Recreus filaflex hardly allow to print tiny details just because they are TOO flexible, even the nozzle tip can bend a part of the print, making errors. Is very important to use Z hop if the print needs small details, but the dripping suffers due more time idle from A to B.

  • Like 4

Share this post

Link to post
Share on other sites

Yesterday made a test to check the accuracy and quality. Zscar is practically gone with @Gudo ZGE

But, check it by yourself



This test is made to check while under at least 50mm/s at 0.15 without any Acceleration/Jerk trick. Machine settings where 3000 acceleration 20 Jerk

Profile with multiple process (this also show the tricks I use to get perfect top layers)


The profile is for UMO+ with UM2 hotend 1.75mm. Adjust accordingly before print (Specially the start/end gcode if going to print on a um2)


  • Like 4

Share this post

Link to post
Share on other sites


A small update, we are doing a lot (A LOT) of testing to optimize every single part, from print time to assembly to tutorial videos. Is quite a lot of work, but so far we are on a really good point print quality. Ofc like any beta testing we found stuff to improve. Right now we have 3 versions and we are testing every single part as much as we can. Basically from the first design to the new we had 5 versions, and atm the last 3 versions are under stress testing.

Also please, if someone has a REALLY difficult print, something that requieres finesse and precision, I would like to print it and test it to see if there's any weak point. Ofc if someone want to share it with me and keep the files form being public that's ok too. I would really like have an architecture file or something really challenging apart of voronoi and very thin walls.

Also after talking to @bondtech and we just got 2 ideas that could be really interesting for the guys like us that want as much precision as possible from the extrusion, up to the decimal points of estep calibration for each kind of filament. Very interesting stuff, we will publish more when we have more tests of this ideas.


  • Like 12

Share this post

Link to post
Share on other sites

Gudo and I wanted to give an update on the status of the ZGE

So, long read ahead! (but interesting)

Why we made it

Gudo and Neotko are moving forward this extruder because we are always pursuing quality, speed and precision. Bowden is a lightweight solution that creates quite a number of handicaps, like for example hysteresis (filament compression due pressure) making it impossible to actually control the extrusion. With ZGE Direct Drive this is now posible, for 1.75mm filaments and 2.85mm, without any compromise on Extrusion horse power.

Why don’t we sell it atm

Gudo and I (Neotko) have other interest than just selling upgrades. Is a nice idea, but we been crunching numbers and isn’t something we would love to be doing on the short time. I have a business and is starting to work and it demands my full time (except weekends where I try to keep improving every bit of the printer). Also isn’t that easy, to make a comercial version of the ZGE would requiere most probably a mold, and that’s freaking expensive, delivery, stock, gudo making sliders, me printing all the other parts, and that could be a nice dream but we don’t have a 100 printers farm to do it right, and sincerely is great to not need to defend something that you want to 'sell' it keeps the mind objective and fresh to new ideas.

So we have choose the YDIY (You Do It Yourself) path. I will make videos of how to print and assemble the key parts that requiere extreme love and care to do them right. Gudo I think can provide for some individuals the key parts, but don’t overwhelm him, if you don’t get one, you need to DIY.

The good the bad and the ugly

The good

It freaking works, it extrudes beautifully, it keeps working and working and delivering repetition and amazing precision even with flexibles. Bondtech dualdrive has a lot on that, it could be done with a lesser quality extruder gear, but it would give troubles, bondtech does work and work and work. It grinds a bit, NO, it pushes the filament over and over (and we get Zero commission from bondtech on this!). The system moves a 390mm square shaft of 4mm square. This was the first challenge. I have been testing a 3x3 thinner rod and it works beautifully, but it will only be of use for 1.75mm guys.

1.75mm requieres less heat to extrude, less pressure, but also it has a drawback, it need’s more rotation on the drive gear, this means more stuff moving. That’s why a 3x3mm is better, because it delivers less noise to the weak um head (more on that latter)

2.85mm requieres more heat to extrude, more pressure but also has the advantage of less rotation on the drivegear, this means less noise translated to the printhead, but more force. So for this the best is a 4x4 square shaft.

The Bad

The punny UM2 hotend. We are using a hotend designed specifically for bowden, and bowden has one advantage, remove pressure and movement on the printhead, that’s why any single force done on the printhead (even a wooble bowden shacking) can add noise to the prints. And for the ZGE this means that every rotation on the drivegear can, since the head has very easy to bend and flexible shafts, translate tinny bits on Z noise. Does this Z noise is perceptible? Well, is hard to see, but is there, most users that don’t have a perfect Z might not notice this, but since I’m a crazy guy I did noticed it. The first version was much more dependent of a perfect square shaft. Gudo made an amazing improvement to this problem developing a printable UJoint system that allows to remove the imperfections of a inexpensive rod (more on this later). And just yesterday, I got an idea (star medal for me on this since I’m really proud) we changed a part to be printed on flexible material removing almost 50% of the noise and making the prints look even better.

So… Does it print better than a Bowden? It freaking does. Does it has a bit more of noise on Z? Yes it does, sorry, that’s how things are. UM2 head, even the 8mm shafts can ‘bend’ and they keep wobbling making them very susceptible to Pressure from the drive-gears to the head. Every retract/extrude is a sudden move, that translates this sudden moves to the printhead. But with this last update the problem is hard to see, and most of normal users won’t ever see it.

The Ugly

To print a ZGE you need aprox 5h for the most delicate parts (slider+driveshaft) and they will need some post-processing. Add one hour of assembly for this part, that must be done with love and not by brute force. Also you do need a hightemperature material, like greentec, platec. Just because this part get’s heat from the bed, so if you print at 60, the air will get there to 35C, and that + extrusion forces = bends. The Extruder head also should be greentec, platec. Is a very amazing material (we also get zero commission on this), but need’s hairspray on the bed and brims, and cleaning. I made a good article about how to add custom brims to the support, read it, read it again, and use it. You need all the parts to be printed perfect. More than 50mm/s can ruin the print and will add time to manual cleaning, that is less reliable than a 3dprinter.

You will need to calibrate your printer to the 0.02-0.03 closeness to the measures we will show as base. Just like Gudo and I did, we used one of the parts, print it and with a digital caliper of at least 2 decimal points, I did print the object 3 times, with different Horizontal Expansion values of -0.02 -0.03 -0.05. For me the winner was the 0.02. So as you can guess, the difference from printing it right or wrong is very very small. You need to keep that in mind for the main part of the ZGE, since is the most delicate part. First time I print one I did it horrible, and after printing 20-23 I thing I have enough practice to make a perfect ZGE slider+shaft.

The rest of the parts, need screws, nothing hard to find. There are two types of heads, one is more compact but need’s some special small 4mm mini shafts. The other, is more complex to assemble and is based on my Neotko-FatIRobertI extruder. Gudo also added a tension selector (very cool) to have 3 basic positions easy interchangeable. But is much complex to print. Also is taller… That… Well, shorter means less vibrations, taller more vibrations… Yea there are many many factors that we have take into account before posting this.

We still need to talk Gudo and I of what basic version we release. Most probably the slider, they key part, won’t have a step file release, but the others might, or not. The important is that all the parts will be easy to print, except the slider (the slider, as I said over and over here must be printed with love and finished more or less by hand).

The Main key parts you will need to get to DIY.

First of all, the square rod of 390mm, there are shops on ebay that cut on demand, but try to be ware that the tip you need to soft it (file it) for safety! You don’t want to loose an eye because you didn’t hear us out right?

For precision square shafts (they call them Key Steel or Parallel Key) the max length we where able to find is 350mm (on misumi) so that means that it would not work. It need’s to be a minimum of 390mm. Gudo might be able to provide this to some users, but again, ask once, if he doesn’t answer he is busy doing something more interesting, don’t hammer it. And you will need to pay for it, you know, square shafts don’t grow on trees.

A shorter smaller screw. You know the four screws that hold your hotend? Well, one must be shorter, like 4mm taller, you can ‘sew’ it yourself indeed. Again, not easy to find. Gudo is amazing and he made me a few with the first kits he made for me.

Bondtech Drive gears for 1.75 or 2.85 you need the 8mm shaft dia, with 4mm - 24-25mm long shaft for the mirror gear.

Motor. I have use, and I keep using, the UMO motor without any geared solution. For 1.75 works perfectly. For 2.85 the best (until one of the beta testers install it and test the motor I sent) is a Bondtech 5.18:1 geared nema. It works and works.

I have a shop, I want to sell it!

Contact Gudo for this, you two might get along and do some business. I personally would love to see this extruder on every printer out there. But mind one thing, no shortcuts, no china quality. If you thing you are decent, come talk with Gudo.

Where the BOM ?

Soon, soon. Don’t worry. We have it on a txt file, we did it long ago, and it might need to get a few final touches.

I want it for my printer!

Our dream is to have this on as many printers as possible, a Bowden-free world. If you are a 3d printer maker contact us and we will study making a version of our ZGE for your machine. But remember this is Open Source, so we will focus on Open Source machines like UM, BCN, and probably others.

If we don’t support your printer, give us time, we are just two guys that love 3d printing and making our printers as precise as possible. Contact your printer company, if they give us access to your files we might do it! And ofc if you have a Solid Robust No Woobly printer, we are interested 8)But ofc, Maybe Gudo just make's a better printer out of this UM.... But that's a history for other day ;)

So what now?

Well, we will post the files soon. We need to short all the files, we had a lot of beta test, many iterations and now is time to arrange that.

The following weeks I'll try to make some videos of how to assemble, deblur it, etc. What can work, what will give problems, where it need's to be smooth and where the play must be zero. A bad print of the Slider part (the thing that moves) just translates into more noise or backlash.

So, more soon! (but give us a few days)

Edited by Guest
  • Like 8

Share this post

Link to post
Share on other sites

Exciting! And a very interesting read. You refer to an article you wrote about custom brims, can you post a link here too?

Have you used both Cura and s3D with ZGE? What is your verdict on this?

Curious to see more of it!

Sure here is the link:


Oh Indeed I use S3D, but the changes from a profile X to zge are very minimal. The thing is that for zge the best (easiest) is to ise the @tinkergnome firmware to change the amount that's retracted at the end of a print. Just because when using a directdrive you need a maximum safe of 2mm, and the end retraction for standard UM could make a flexible material clog.

I will write down more about that, is very interesting indeed.

Also ofc Cura works too, Gudo uses the old cura afaik.

The hardest part for me was to be able to get near zero z marks when using flexibles. One of the most instantaneous improvement from bowden to directdrive is that there's no pressure from the filament being flex inside the bowden (hysteresis) so for example while on a 2.85 the user needs a 4.6-5mm retractions, for directdrive you can be more than ok with 2mm (or less depending on the material). This means that a) You retract faster and b) you have less pressure when doing it. Also you need much less the common tricks like coasting or wipe, but for some materials like flex ones, it allows you to retract+wipe and do tons of retractions without any fear of the filament bending inside or the classic bowden problems. Also since the distance from extruder to hotend is very short, you can actually create pressure to keep pushing the flexibles and still be able to retract.

First test (some pages ago) do show that there's a middle point of happiness between distance retracted and time that takes to retract it. I made a lot of profiles for my 1.75 setup, but for 2.85 is even easier to print flexibles just because the filament is bigger and it allows more stiffness when creating pressure to extrude.

There's much I need to write down but basically is very easy to get nice prints, with less tricks. And it gives you a 'realtime' feedback from motor movement to actual extrusion.

I was thinking on releasing a set of 'readytogo' firmwares, with the basic changes like prime amount + retraction at the end, but since tinkergnome already covers all that it would just be more confusing to make firmwares.

For umo ofc, all this is 'easier' since the users ave direct access to the prime/end gcode. But also I will share some starting sequences for the prousers.

One of the funny side effects of having a ZGE is that if you have a dinanometric tool, you can actually make graphs of what's the 'ok' extrusion vs Xnm. Allowing a word of precision that is very hard to get. But that's for other day ;)

Edited by Guest
  • Like 1

Share this post

Link to post
Share on other sites

Good read, I will build this as soon as its released and as fast as I have time.

I looked quick on ebay, its not called key steel, but just stainless square bar


Its supplied in 400 and 500mm lengths in 4x4mm

as long as its pretty straight, would it work ?

Share this post

Link to post
Share on other sites

Good read, I will build this as soon as its released and as fast as I have time.

I looked quick on ebay, its not called key steel, but just stainless square bar


Its supplied in 400 and 500mm lengths in 4x4mm

as long as its pretty straight, would it work ?

Indeed that should work. @Gudo might now much better than me if that's ok or not.

What I do know it's that you can 'rectify' them by hand, and love (always love ;) )

Once I did bend one, of the 3x3 shafts with one of the versions of the topdrivegearhead, and I was able to rectify it using a misumi shaft as guide. Most of the invisible imperfections of the shaft are corrected by the fantastic UJoint system. But ofc there's a limit to what can be fixed.

I will print a new set of printhead to replace two of the beta ones, and probably past tomorrow I'll start with some print-assemble-tutorial to everyone can see where are the challenges when printing this, a basic set of settings and an assembly video that I made with the beta2 that on it's core is the same to the ones that will be released (except that internally they are really different).

So yea that should work. If you prefer ofc we will release the BOM probably also this week. There's some set of bearings that must be exactly one size etc etc.

I will also take this moment to start explaining some questions that users might do (and I did also at first). Questions like:

- Can I use other bearings? NO. There's much math and simulations and beta testing on all the parts, if you fell like changing something just to save a buck, then this isn't the extruder for you, sorry!

- Can I use a shorter square rod? NO. I did bang Gudo with this many times, he is just so patient and explained me over and over how and why it can't. Basically that's the correct size for a umo/um2. Can you use 400 instead of 390mm, yea you can. But 390mm is the perfect length so it doesn't pick an eye while moving or hits you, and also is the length that allows the slider part to have always full contact with the square rod. So, like I said before, file the tip, because metal can hurt. (It never ever happened to me and I do look at the printer like a mad men when printing to debug stuff).

- Could I move the holder to the other corner? Yes, print it mirroed (that's something I did ask at the beginning, but after using it on the Left side I would not change it, it just make's sense... Hard to explain..

- Can I use a cheaper drivegear? NO (well you actually could if you redesign most of the stuff, but if you just want to extrude without quality you can do it by hand :p )

- Can I just print it on ABS or PLA, or China cheap PLA?

NO. ABS for example size changes a lot, it can't get as much flex torsion without deforming as platec/greentec. Ofc if you thing you can, just do it, but don't come here asking why it fails ;) Also the material used must be as precise as possible, that something that only good brands give. You wan't to have as many things under-control as possible when printing printing this with a bowden since the tolerances are quite hard.

- Could I avoid to print the Flexible material parts?

Oh yes, I forgot, there are parts that actually need to be printed on flexible materials like ninjaflex, recreus and a semiflex material. WHY? Just because ;) (will explain all this on the video/tutorial). Most of them you can actually print them after installing the system, but also, well you can print them one by one very slow until you get the zge working ;)

- Why the motor holder moves a bit when printing?

Because the system compensates the extrusion forces to avoid as much motion transmission to the printhead. Also, better the motor holder than the printhead, right?

- I have a dual printer, I wan't to use this!?

That's actually doable, @Gudo even made a design that could work on a UM3. But... More or that someday. Basically we don't want to just make stuff that could work, but stuff that actually put something interesting on the table. There are many factors beyond making the extruder work on a um3 than just making it. Gudo actually I think did cover all of them, but then there's the part of the firmware. The UM3 firmware need's quite a lot knowhow to actually change parameters like Extrusion retraction while changing tools, prime amount, retraction amount, esteps, etc etc etc. So change this parameters on a um3 the user would need to know much knowhow to do stuff on linux without bricking the machine (editing stuff is really easy in the end but is a giant leap for most users). So for now we choose not to, but is there, if UM or maybe a modder makes a easy way to access the Machine Json files and edit them, then we might release it. Just because, it doesn't make sense to release stuff just for releasing stuff. Also, there's a big problemo for flexibles on um3, and that's the gap of air on top of the cores while they rest on the printhead.. So, could it work? Sure, but would it work everytime? That's the question... So, short answer, not yet.

And that covers for today. More soon :D

  • Like 2

Share this post

Link to post
Share on other sites

A fast overview of Gudo's ZGE, sorry for the bad sound quality but the holidays are almost over!

Also, Gudo passed me the BOM of the ZGE, so you guys can start collecting stuff for it. I personally recommend the Shortzge over the Neotko-FatIRobertI version, just because is smaller and translates a bit less vibrations, but ofc it has the advantage of a tension selector for the filament.

Let's go with the BOM · Written by Gudo


- BONDTECH Dual Gears kit 1.75mm or 2.85mm/shaft dia 8mm

* important if use

FatIRobertI order the 4mm x 22mm driven gear pin

Short ZGE order the 4mm x 25mm driven gear pin


- 4x4mm precision Square Rod

Steel or Stainless steel - 390/400 mm lenght

- ZGE Bearings

8x - 3x8x3 ZZ (slider)

2x - 8x16x5 ZZ (extruder)

2x - 10x19x5 ZZ (frame motor holder)

2x - 12x21x5 ZZ (print head extruder holder)

- Fastener metric M3 SOCKET HEAD BOLTS & NUTS



2x 20mm (UM2 frame)

2x 18mm (UMO frame)

2x 8mm + nut

1x 35mm + nut


4x 12mm + nut


4x 10mm

3x 16mm + nut


10x 18mm + nut


2x 18mm + nut


- FatIRobertI version

* 3x 8mm Hexagon socket button head (or Philips equivalent)

3x 30mm + nut

- SHORT version

* 4x 8mm Hexagon socket button head (or Philips equivalent)

1x 30mm + nut


- SPRING SCREW for the 2 Extruder versions

1x 35/40 mm HEXAGONAL head + nut

- Short ZGE Spring pressure guide

2x 4mm x 18mm pin

  • Like 7

Share this post

Link to post
Share on other sites


I looked quick on ebay, its not called key steel, but just stainless square bar


Its supplied in 400 and 500mm lengths in 4x4mm

as long as its pretty straight, would it work ?


Indeed that should work. @Gudo might now much better than me if that's ok or not.

What I do know it's that you can 'rectify' them by hand, and love (always love ;))


Don't expect them to be 100% straight...they are shipped in some bubble wrap and a plain taped envelope so even if they are straight when posted...you are at the mercy of the delivery service...I ordered 2 just to have one spare and hoping that 2 packed together would not bend as much as one...thinking that they would not be packed that well :p

That said...mine were not that badly out of shape...tested them on the glass bed of our stove...maybe max 1 mm off on the whole length...and at least one of them I could straighten out pretty good with little effort...didn't try the second one yet

But time will tell how good it actually works :)

Share this post

Link to post
Share on other sites

Atm Gudo has prepared and short all the step files and imgs to show how to assemble all. I made a few photos of how to ready the shortzge and the motor holder. Also Gudo is preparing some literature about how much/little must the slider be filled/deblur since is the most delicate part to make.

About the shaft, I think the system can compensate for a good margin of error on the shaft but a visible curve will add friction, that on the prints shows as vibrations while doing straight lines.

I won't be able to post much until Friday/Saturday since I'm a bit overwhelmed by my job, but this weekend I'm sure we can post stuff.

Also remember that apart of greentec/platec some parts are printed on filaflex/ninjatek because they work as dampener, but I think that all that parts could be printed after installing a basic unfinished zge without them.

Share this post

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

  • Our picks

    • Ultimaker Cura | a new interface
      We're not only trying to always make Ultimaker Cura better with the usual new features and improvements we build, but we're also trying to make it more pleasant to operate. The interface was the focus for the upcoming release, from which we would already like to present you the first glance. 
        • Like
      • 130 replies
    • "Back To The Future" using Generative Design & Investment Casting
      Designing for light-weight parts is becoming more important, and I’m a firm believer in the need to produce lighter weight, less over-engineered parts for the future. This is for sustainability reasons because we need to be using less raw materials and, in things like transportation, it impacts the energy usage of the product during it’s service life.
        • Like
      • 12 replies

Important Information

Welcome to the Ultimaker Community of 3D printing experts. Visit the following links to read more about our Terms of Use or our Privacy Policy. Thank you!