Contraption: Filament winder build

[eggplant]

I've wanted to build a filament winder for years now but always considered it to be too daunting a project, given that building one includes mechanical, electrical, and software work, not to mention lots of messy composites. Examples of homebrew winders are few and far between with no real community compared to 3D printers so it is harder to stand on the shoulders of giants. If you dig for a little while though, you can find plenty of examples, from videos of industrial winders in action, research papers, to discussion about the X-Winder, the main hobby-grade winder available. I started doing research in earnest when I came across this excellent thread where @Scrapmaster87 did a great job documenting his filament winder build. From there I was committed and started to CAD:

I ended up with a largely X-Winder inspired 3-axis machine that's designed to be easily upgradable to 4 axes. It should handle up to 5" mandrels and though I've built mine with 4' travel, should be able to handle more length. It is built entirely out of off-the-shelf hardware and 3D printed parts, so it is quick to iterate on and should be straightforward for others to replicate. If you want to follow along with the CAD, you can do so here. I still need to finish the design of the resin bath, but most of it is there.

I'm using NEMA-23 steppers for the carriage and mandrel and a NEMA-14 for the delivery head. All driven by TMC2209 drivers. The machine is currently running off of a standard BigTreeTech Octopus board using Marlin.

I'm also working on software for generating toolpaths, plotting them for evaluation/debugging, and running the machine. Definitely a work in progress, but it can generate gcode for helical and hoop winds and stream them to the machine. The plotting functionality has been super useful for debugging so far:

Bad gcode due to a typo in the code	Good gcode!

I plan to make the plotter show tow at the real width, draw overlaps in an intuitive way, animate the output, and potentially wrap it onto a 3D cylinder. Already very helpful though. There isn't much documentation yet, but the software is located here.

Initial results in the next post!

 

[eggplant]

There was plenty of Marlin tuning to do at first given that all of the defaults are aimed at 3D printers. I eventually got all of the axis currents, accelerations, and max feed rates happy enough that the machine could consistently run through a full wind without skipping steps. Once I got there, I realized that my gcode generation code had some bugs (shown in my last post) that were preventing me from getting even coverage of the tube. Once fixed, the results are pretty nice:

As of today I've been winding thread to avoid wasting CF/FG while I get the machine up and running. I've been generating the gcode as if I'm using actual tow with a width of 11mm, and the diamonds you see in the picture are all really consistently 11mm! Other than at the ends, which I still have some tuning work to do on. Still, I'm happy enough with the results so far that I have ordered some CF tow to experiment with. I plan to do some dry winds while I get the resin bath design finished and the last path planning kinks worked out, but I hope to be winding with CF within a month or two.

Video of a practice wind with thread, near the end:

 

[Scrapmaster87]

This is looking really awesome and I feel your pain and joy!

One thing I spent a lot of time on was smoothing my end transitions with arc moves. This helped with the speed of the wind greatly and also stopped filament slip. I set my arc radius to a multiple of the mandrel diameter.

Hint: look at M205 for junction deviation, I use M205 J0.1. Without this Marlin tries to exactly follow the trajectory at junctions, this allows "corner cutting" to keep your speed up.

I'm currently working on version 3 of my winder, which I'll eventually show on my thread. I've been held back by some job transitions and now a move.

 

[Sandy H.]

So now there are !2! of you crazy guys on the forum. . .did somebody leave a door open or something!?!?!

Hoping to see progress on this build and SM87's V3. I'm involved in computer controlled machinery and would generally be intimidated at what you guys are doing, even though multi-axis moves are somewhat trivial. I'm not sure what abut the whole process seems intimidating to me, but my knowledge of winding and take-up from the paper industry might be part of that. That was my first job out of college and the paper tension, winding and take-up were treated as black-magic, so maybe that's it.

Regardless of my hangups, I look forward to many more cool posts from you both. "Now, somebody, go close that door! Too many of them crazy's doing them home made winders 'gettin' in here!" J/K, of course!!!!

Sandy.

 

[Kelly]

Just curious, why aren't you guys using some kind of cog at the ends of the mandrel to hold the filament, rather than piling it up at one point?

These are awesome builds, BTW!

 

[FredA]

Just curious, why aren't you guys using some kind of cog at the ends of the mandrel to hold the filament, rather than piling it up at one point?

Because the cog then gets embedded and you may not want that in your airstream....unless you're cutting the ends off.
Plus hitting between the teeth of the cog is an added complication when using a tow (with width) instead of string.
Lastly, for our process, we wind over the closures so they make a step to wind onto and a little pile-up beyond that step isn't an issue.

 

[Kelly]

Ah, I guess I'm more lost than I thought I was. I assumed you were cutting those ends off anyway, And you say 'closures' - these are motors that are being wound? I assumed they were airframes.

Thanks for the info!

 

[Scrapmaster87]

Ah, I guess I'm more lost than I thought I was. I assumed you were cutting those ends off anyway, And you say 'closures' - these are motors that are being wound? I assumed they were airframes.

Thanks for the info!

I might try it some day. The more longitudinal the passes are, the more space I need to change direction. Therefore a bigger "dog bone" I have to cut off. I go up to about 75 degrees from hoop winding, it's really dependant on mandrel diameter and filament width. Sometimes just adding another pass isn't possible, though my g-code generator only works with an even number of passes.

 

[eggplant]

Yeah, overall what Fred said. I've seen pictures/videos of winders with "combs" at the ends of the mandrel and it definitely does get you a bit more useable length out of the tube, but seems annoying to manufacture and free the tube from. I intend to make both airframe tubes and motor cases with my winder, so I'm focusing on strategies that work for both.

I do enjoy making my own motors, especially liner-bonded monolithic ones. Tons of work left on the winder before I get to this point, but I plan to cast a motor into a liner, install a single-use phenolic nozzle on one end and an aluminum extension/mandrel on the other, wind over it, and remove the extension. From there I would slide a forward closure down the extra wound tube forward of the liner until it seats against the liner and glue it in place. The avionics, recovery, and nose cone coupler could all live in the extra length beyond the closure. Should be a crazy mass fraction!

 

[FredA]

I assumed you were cutting those ends off anyway, And you say 'closures' - these are motors that are being wound? I assumed they were airframes.

Yes, we use our winder exclusively to wind motors at this point.
We wind with the propellant already inside so we can wind the complete pressure vessel.
No trying to open and re-seal as we use it.

 

[Kelly]

We wind with the propellant already inside so we can wind the complete pressure vessel.

Ok, mind officially blown. I thought you folks were over the top just going through this effort to make your own body tubes! Anyway, fun stuff to see; love seeing these threads.

 

[eggplant]

More progress! I've touched up the path planner quite a bit and made some hardware changes including a set of centering rollers that prevent tow from wandering on the far roller. I also received some tow, a small spool of Hexcel 12k IM2 and two medium spools of Mitsubishi 18k. I'm going to use the Mitsubishi for my first wet winds, but I've been messing around with the Hexcel to see where the machine is at.

It is pretty close! I bet this would be able to make a tube if I put some mylar on the mandrel and finished the resin bath. Before I do that though, I also want to increase the fiber tension to closer to what I was getting with the thread spool, increase the speed so I can make bigger tubes within the resin's pot life, and try arc moves at the ends.

 

[FredA]

Resin pot life was our nemesis
Also controlling how much resin was applied to the tow.
Our first winds were messy and had way too much.

Good luck as you move to the wet winds.
Looking good so far.

 

[OzHybrid]

Because the cog then gets embedded and you may not want that in your airstream....unless you're cutting the ends off.
Plus hitting between the teeth of the cog is an added complication when using a tow (with width) instead of string.
Lastly, for our process, we wind over the closures so they make a step to wind onto and a little pile-up beyond that step isn't an issue.

If you need a cog, 3d print it and cut off.

 

[alexzogh]

Somewhere here on the forums, I posted an excel spreadsheet and script that can help you build your winding gcode. I originally wrote it for the x-winder, but since it outputs gcode, it should be easily adaptable to yours. There is a lot math behind the correct winding angles for a given application. The hardest one I worked on was making a VK nosecone. The haak formula requires logarithmic changes in rotational and tow feed speeds to get a proper winding.

 

[OzHybrid]

Resin pot life was our nemesis
Also controlling how much resin was applied to the tow.
Our first winds were messy and had way too much.

Good luck as you move to the wet winds.
Looking good so far.

There are uv cure resins. Have you considered trying them and cure directly with UV. Clearly with CF that's going to be tricky as it might prevent uv penetration. But GF should be good.

 

[Rocket501]

Yes, we use our winder exclusively to wind motors at this point.
We wind with the propellant already inside so we can wind the complete pressure vessel.
No trying to open and re-seal as we use it.

Would you happen to have any photos or video of this process? Someday I'd love to try the same thing and any inspiration would be helpful.

 

[FredA]

Here's a couple of photos - more should be in a thread elsewhere....

To start - two that are ready to wind - complete assembly on the right.
These are 54mm liners with propellant cast inside.
The forward closure is a circle of 1/2" plywood with a groove on the forward side to fit a 38mm FWFG tube used as the drogue bay.
There is a small loop of harness glued in the forward closure for the recovery attachment point.
The nozzle is a used phenolic that I cleanup and re-drill the throat and adjust the lip to fit the liner.
Motors are L-1000's when done.


Here's the wind in just getting started:



Here's the finished product - Dual-deploy with the chute in the NC.
Include a BRB tracker and Stratologger CF in the Ebay.

Winding onto the 38mm FG and the nozzle complete the pressure vessel.

These are our first attempts - some refinements have been made.
Fin-attach is our problem - we've got another solution that should do better but is unproven.
I'll keep that private for now.

 

[Chad]

i'm fascinated with filament winders, keep it coming

those lightweight nitrogen bottles used for paintball are filament wound. I'm guessing it's filament around a thin metallic container. i've always wondered if the people working on liquid bi-prop motors could benefit from filament wound tanks and combustion chambers. ( on example of "people working on bi-prop motors" = halfcatrocketry.com )

 

[dhbarr]

i'm fascinated with filament winders, keep it coming

those lightweight nitrogen bottles used for paintball are filament wound. I'm guessing it's filament around a thin metallic container. i've always wondered if the people working on liquid bi-prop motors could benefit from filament wound tanks and combustion chambers. ( on example of "people working on bi-prop motors" = halfcatrocketry.com )

Type 1 : All metal
Type 2 : Metal cylinder with composite hoop wrap
Type 3 : Metal liner, fully overwrapped
Type 4: Thermoplastic liner, fully overwrapped
Type 5: Fully composite, no liner

 

[boatgeek]

Resin pot life was our nemesis
Also controlling how much resin was applied to the tow.
Our first winds were messy and had way too much.

Good luck as you move to the wet winds.
Looking good so far.

Did you have some kind of wiper to reduce the amount of resin on the tow after it came out of the bath?

 

[FredA]

Did you have some kind of wiper to reduce the amount of resin on the tow after it came out of the bath?

First wind was with the X-winder's system that did little.
We made a new resin tray with a triple wiper system that helped a lot.

 

[Cl(VII)]

Here's a couple of photos - more should be in a thread elsewhere....

To start - two that are ready to wind - complete assembly on the right.
These are 54mm liners with propellant cast inside.
The forward closure is a circle of 1/2" plywood with a groove on the forward side to fit a 38mm FWFG tube used as the drogue bay.
There is a small loop of harness glued in the forward closure for the recovery attachment point.
The nozzle is a used phenolic that I cleanup and re-drill the throat and adjust the lip to fit the liner.
Motors are L-1000's when done.
View attachment 515974

Here's the wind in just getting started:
View attachment 515975
View attachment 515976

Here's the finished product - Dual-deploy with the chute in the NC.
Include a BRB tracker and Stratologger CF in the Ebay.
View attachment 515977
Winding onto the 38mm FG and the nozzle complete the pressure vessel.

These are our first attempts - some refinements have been made.
Fin-attach is our problem - we've got another solution that should do better but is unproven.
I'll keep that private for now.

This is officially badass.

 

[kramer714]

Couple of suggestions (from a trained professional on a closed course)
1) Use a series of rollers including 1 inside the resin bath (submerged) helps wet the fiber and also creates a flat ribbon not rope
2) a plate with nails on it, works well for the turn around, you don't need the third axis if you to that for tubes, just cut it away the end and through them away. Dont try and get the fibers to fall between the nails just get them to catch it,
3) a weight on a swing arm with an 'eye' in it that the fiber feeds through helps keep the tension at turn around.
4) use a wetting agent and a VERY thin resin, not the best time to use west, you want something with along pot life. Wetting agents get rid of the bubbles and help wet out the fibers.
5) run multiple tows together, again good for forming a 'ribbon'
6) box the fiber and use an air filter, small graphite fibers will come off the rolls and short out EVERYTHING electrical. (Plus better for the operator too)
7) slow roll for cure or hand vertical
8) For pressure, you need a lot of 90 degree wraps (hoop), for axial loading you need lots of +/- 20 ish and a little 90 for buckling stability

Mike (been filament winding for 30 years) K

 

[eggplant]

I have been busy with work and other projects, but I have found some time to continue tinkering on my winder. Things are looking pretty good, and I'm hoping to do a wet wind soon! I think this dry wind would have been just about perfect if I had told my planning script that the tow was about 1mm narrower to eliminate the small gaps.

 

[eggplant]

I did another test run today. I'm going to speed up the feedrate a bit (real time for this layer was 12 minutes) and try it with resin soon.

 

[Scrapmaster87]

I did another test run today. I'm going to speed up the feedrate a bit (real time for this layer was 12 minutes) and try it with resin soon.

That's a beautiful wind sequence! Prepare for some more self-taught lessons when you get resin in that, though I don't think you're gonna too much trouble.

I'm all moved into my new place and the old place is closing on Friday. I'm starting get back into my winder with some control theory, I have some issues to resolve/ go around with using an induction motor for the mandrel drive... Should be an interesting update, but not as pretty as yours.

 

[G_T]

The small gaps are not as bad a thing as you might be thinking. The weakest part of any good layup is the locations where a fiber layer goes from below to above, or vice-versa. That is due to the abrupt changes in direction. That's particularly something carbon fiber doesn't like. Having a small gap allows the angle changes to be much less, thereby causing less strength loss at these intrinsically weak locations. So, IMHO perhaps narrow the gap a little, but don't eliminate it entirely.

2nd suggestion - you are not making a driveshaft, where the optimal winding angle is 45 degrees. You are making a rocket body tube, where the least optimal angle is 45 degrees! Inner and outer layers - only - should have a winding angle which looks more like a close packed spiral pass over the length of the tube, then back. These innermost and outermost layers are to improve the pinch resistance of the tubes to bending loads. The layers in between should be run with the tows put down as close to parallel to the tube axis as you can manage. This combination would greatly improve your bending stiffness and strength, for the same overall weight.

A winding which is just 20 degrees misaligned with the applied stress only contributes half its strength. Anyway, this is the reason for those discardable domed end caps in commercial tube making, that get cut off. It allows one to wind tows quite a bit closer to parallel to the tube axis. With end caps, you can get very close to parallel.

There was one video posted showing a commercial tube winder in operation. The person who set up that wind understood the loads and how to deal with them with composites. The inner and outer layers were put down like threads on a spool of thread. Right next to each other. The inner layers were nearly parallel to the tube long axis. Beautiful! Unfortunately the filament wound tubes we can buy are not wound correctly for rocket body tubes. They have a winding more appropriate for transmitting torque. That's what 45 degree windings are best at... But they are one angle to completely avoid for rocket body tubes.

Gerald

PS - Not meaning to be critical, though I know it sounds that way. Keep up the good work!

 

[eggplant]

The small gaps are not as bad a thing as you might be thinking. The weakest part of any good layup is the locations where a fiber layer goes from below to above, or vice-versa. That is due to the abrupt changes in direction. That's particularly something carbon fiber doesn't like. Having a small gap allows the angle changes to be much less, thereby causing less strength loss at these intrinsically weak locations. So, IMHO perhaps narrow the gap a little, but don't eliminate it entirely.

2nd suggestion - you are not making a driveshaft, where the optimal winding angle is 45 degrees. You are making a rocket body tube, where the least optimal angle is 45 degrees! Inner and outer layers - only - should have a winding angle which looks more like a close packed spiral pass over the length of the tube, then back. These innermost and outermost layers are to improve the pinch resistance of the tubes to bending loads. The layers in between should be run with the tows put down as close to parallel to the tube axis as you can manage. This combination would greatly improve your bending stiffness and strength, for the same overall weight.

A winding which is just 20 degrees misaligned with the applied stress only contributes half its strength. Anyway, this is the reason for those discardable domed end caps in commercial tube making, that get cut off. It allows one to wind tows quite a bit closer to parallel to the tube axis. With end caps, you can get very close to parallel.

There was one video posted showing a commercial tube winder in operation. The person who set up that wind understood the loads and how to deal with them with composites. The inner and outer layers were put down like threads on a spool of thread. Right next to each other. The inner layers were nearly parallel to the tube long axis. Beautiful! Unfortunately the filament wound tubes we can buy are not wound correctly for rocket body tubes. They have a winding more appropriate for transmitting torque. That's what 45 degree windings are best at... But they are one angle to completely avoid for rocket body tubes.

Gerald

PS - Not meaning to be critical, though I know it sounds that way. Keep up the good work!

Actually, a wind angle of ~55° (as shown in the video) IS optimal for my primary intended application of motor cases. You primarily need hoop strength so you want a wind angle not too far from a typical hoop wind, but you also have to have some axial strength to keep the closures in. I can dig up and link the paper I found on the subject later.

 

[eggplant]

That's a beautiful wind sequence! Prepare for some more self-taught lessons when you get resin in that, though I don't think you're gonna too much trouble.

I'm all moved into my new place and the old place is closing on Friday. I'm starting get back into my winder with some control theory, I have some issues to resolve/ go around with using an induction motor for the mandrel drive... Should be an interesting update, but not as pretty as yours.

Excited to see your progress! And yup, I expect the first wind or two will be messy and involve retuning a lot of what I just tuned.