12/12/2025 2:12 AM - The Beginning

To start off, I have just been doing a lot of research into filament winders. I have spoken with the designers of a few winders, including Andrew Reilley of Contraption (https://www.reilley.net/winder/) about what designing a winder entails. I have also been looking at designs and attempting to find any available educational/research papers on the design aspects, though the latter has been somewhat unsuccessful.
This is a somewhat ambitious project design wise to complete in the roughly 2.5 weeks left of blueprint, but I think I can pull it off if I sacrifice my sleep.
Here is a photo of Contraption, which will be my main inspiration for this project, but I have some modifications in mind.

So much research.

12/12/2025 2:16 AM - Purpose

I figured I should touch a bit on design purpose. Filament wound CF tubing can be, and often is, a lot stronger than cf tubing made through other means (such as wrapping). This is specifically prominent in composite motor cases, where properly made wound tubes can withstand a LOT more pressure inside, making for a great pressure vessel. AKA, it won't explode. They also typically have lower mass for a given size due to how resin efficient they are.
This machine will need to be able to feed both CF filament, heat shrink tape, and attach a heat gun for shrinking the tape. This allows for winding and then proper compression to ensure good bonding and minimized resin content.

In theory I could even wind nose cones, but I have more interest in pursuing compression molding using gores of fiberglass, as wound nose cones take a lot of effort and post processing.

Image from Andrew Reilley.

12/12/2025 11 PM - Frame Layout

I began with the simple frame layout. For simplicity's sake it will be full 2040 extrusion. It took a while to size properly as I had to account for all the future rocket builds I have planned - aka, what tube sizes I will need. My main reference for this was Contraption and another winder I found: https://www.hackster.io/news/carbon-fiber-filament-winder-974ad994e5a5

It's 1280mm wide!! Sheesh.

12/13/2025 - Linear Motion of the Delivery Head

After researching common practices, I decided to use a modified servobelt system with 20mm linear rods. This allows for rigidity and simplicity, as I don't have belt loops, long spans, etc.
Servo belt means esentially that its a mock rack and pinion system, but using a belt for the rack and a pulley for the pinion.

Above is my version, below is the traditional version.

Main difference is the distance between my pulley and idlers (for mounting purposes) and the lack of a lower belt span, as it seems to be superfluous.

Quad carriages because I want hella rigidity.

12/14/2025 9:07 PM - The Spinny Bit: Part 1

This is the powered side of the rotating mandrel assembly.

That's a lot of big words. Let's break it down:
In a filament winding system, there is a part called a mandrel. Actually, this part exists in pretty much all tubular composite work. This is the base form that the carbon fiber is wound onto - it's like a mold. On a winder, this part has to spin!.
Because they can get rather heavy, the spinny bit is powered by a high torque nema 23 stepper motor. To keep things compact I have the motor offset, and the power to the mandrel will be transmitted via two pulleys and a loop of belt. You can see one pulley at the top and one on the motor shaft. I have these pulleys in double shear (two bearings, one on either end) to keep things rigid.

Here it is mounted. To cut down on screws, I added t slot conformal sliders onto the part. That way it slots into the extrusion.

This is an elaborate part and took a while.

12/14/2025 9:11 PM - The Spinny Bit: Part 2

The spinny bit also needs an inactive side on the opposite end, called an idler (because it is idle).

Both sides are double bearing for the mandrel shaft, as this is a big load on a very long shaft and I just want to ensure stability.
The shaft is an 8mm threaded rod so that I can use nuts for retaining.

I ensured to make this part all mounted in one axis, so that I can just slide the whole thing off when I need to swap mandrels.

Now we have an assembly that looks like this. Both sides of the mandrel spinning system are complete. All that is left is end caps to hold the mandrel onto the center shaft.

12/14/2025 9:17 PM - Design for Manufacturing

I wanted to touch a bit on design for manufacturing. These parts are being 3D printed. So, as a responsible designer, it is my job to make them printable. I also happen to be the manufacturer but that isn't always the case.

You can immediately notice that there is a big flat side. This is the side I intend to be against the bed, and the entire rest of the part was designed in relation to this side.

For example,
All overhangs are curved so they don't need support!

Overhangs in general are actually just minimized, and the ones present don't need support. This is the main factor when it comes to designing for 3D printing: overhangs.

This can be seen on the other side too!

The right side is the flat side intended for printing - this part actually manages to have no overhangs.

So it prints like this. Nice and easy!

This is something you can do with any design (and should do). The considerations change for if you are printing, milling, bending, carving, etc. But each option has constraints!

The time added here is for the time it took me to optimize these parts for 3D printing.

12/15/2025 - Mandrel Endcaps + Material Refresh

I designed the endcaps for the mandrel, which will consist of a printed part bolted to an aluminum shaft coupling thingy.

There is one on each side. The printed bit can scale up to however big my mandrel is (theoretical max of like, 8" diameter!).

I also took a little to assign everything the right materials and colors, as I was getting sick of the monotone. So now it's monochrome instead of monotone!

I will be clamping these end caps with a hose clamp so it locks onto the mandrel but is still removable fairly easily. I did not model those as it would be really annoying, but I may model them at a later date if I am feeling particularly motivated.

12/16/2025 2 AM - Belt Tensioner

So I may have forgotten that belts need tensioned. Whoooops.

Simple tensioner. I played around with a few ideas before this, such as having two offset idlers in the center on a cam system so that I could rotate the assembly and it would push the two sides of the belt apart equally. Ultimately I decided to just keep things simple.

12/16/2025 4 AM - Carriage Things

I did two things:
Tow spool mounting
and Servobelt drive mounting.

The first:

A pretty simple and elegant spool holder. You just slot in a spool with a 1/4" center axle and you're good to go!
Most designs I see do this fully printed, but I like having the base as an extrusion. More rigid.

The second:

I addressed the mounting for the servobelt module. By the way, I just learned that this is actually called omega drive because of the modification I made! Normal servobelt is actually patented.

Basically, it just slots into the two side t slot extrusions and then gets bolted in with the 4 bolts at the top.

The t slots are weirdly sloped like that because of... you guessed it! Design for manufacture! This part will print standing in the orientation of those photos. So, the angle allows for no supports, as it is an angled overhang (easy) versus a floating flat surface (impossible without supports). Fun fact, optimizing that one feature, the t slot, for printability took 45 minutes!

12/17/2025 - The Tow Delivery System - Part 1

I began actually working on the delivery part of the delivery head today. To start off, I made the rollers, which tension the tow.

I went through about 4 revisions of these as I kept running into clearance issues.
This was the first revision:

Center mounting, wider roller. However, the wide base + 2040 extrusion meant that the tow path was obstructed. So not only did I have to change the base, I had to alter the frame and the roller mounting.

Here is a close up of the final roller mount. The roller has one 688 bearing on either side, with an 8mm threaded rod as the shaft. This allows me to retain the roller with a nut on either end. Ideally I would have a smooth shaft for the bearing, but that would raise cost significantly.

Next steps are:
Tow alignment (gotta get it all neat and centered)
Resin Vat/bath (otherwise I am making expensive balls of yarn)
Motorized delivery head (the last roller before the mandrel, aka the delivery head, needs to be able to rotate in a controlled manner to set wind angles. very important!)

12/21/2025 5:33 AM - Resin Vat

Stopped procrastinating and made the resin vat. The tow strand feeds through this to get soaked with epoxy. It still needs a roller, which will sit in the bottom and be what pulls the tow in, but it 5:30am so I am going to do that tomorrow.

That is it in place on the carriage. It is held in by 2x m3 screws as this needs to be easily removable, as they are single-use (the epoxy can't ever be cleaned out fully).

Now you may be thinking: "Parker! That looks hard to print! What happened to designing for manufacture?!?!" (I doubt you were thinking that but just play along).

Surprise! Bet you didn't expect it to print upright like this.

Just like that, it's printable!

This took a while because I am tired and its a bit of a weird shape. I had to angle match to the previous roller so that the tow could feed in without scraping! That was so tedious.

12/21/2025 5:59 AM - Resin Vat Roller

I couldn't sleep so I ended up just making the roller. These shouldn't have to be single use as I can do a lot more aggressive methods of cleaning when using metal.

Basically, it's just an 8mm metal dowel that snaps into place. You can see the retaining bump thingies.

I also angled the other side of the vat so that the resin would all be funneled to where the tow is. This makes it a bit harder to print, but still completely doable - it just goes off an angled face now. Still should not need supports.

12/22/2025 1 AM - Alignment Pins

This is a simple part that hold 2x 3mm alignment pins that the tow passes through. This helps center it on the rollers and keep everything running smoothly.

Didn't take super long.

12/22/2025 2 AM - Resin Wiper

I forgot that I need a wipe over the vat to scrape off excess resin. This will be either cut out of neoprene or printed out of PP filament.

Also meant I moved the mounting of the vat so I could clamp both the wiper and vat down with only two screws.

12/23/2025 11:13 AM - Delivery Head

It was time for the last big part of the puzzle: the delivery head.
This is the last thing before the cf is on the mandrel and controls the wind angle, so it must be powered.

It ended up looking like this.
There is a nema 17 with gearing to spin the final roller, which I decided to make a fancy mount for. Fairly large gear reduction for accuracy.

Rather than a bearing, its essentially just a printed bushing.

So theres the main cylinder on the big gear and then that further back ring, a retaining ring, that bolts on.

That essentially sandwiches the gear-shaft assembly into this bore, which will be greased.

Pretty elegant solution.

The roller mount was made with a simple double loft, once for the main structure and once for the cutout, and then I did a lot of fillets.

The roller is at a slight angle from flat but overall is aligned with the mid axis of the mandrel.

See? Very close.

On the small gear I put the term "Straight Windin It". And by it, well, let's just say, my mandrel.
(please someone get this reference).

All things considered this section was actually pretty fun to do, despite the fact I pulled an all nighter.

12/23/2025 11:15 AM - Linear Belt

I decided to model out the linear belt so that it was a bit easier to see what was going on. Fusion fought me really hard on this.

I also modeled the belt clamps, which are just little end caps that clamp the belt with the edge of the extrusion.

12/23/2025 11:19 AM - Screwed.

I don't know if you noticed in the last two entries, but I added all the screws! I was considering also adding all the T-Nuts, but there probably over a hundred... so if I feel like it, I will add them, otherwise any time a screw extends into the T-slot, you can assume it goes into a T-nut. Still just doing these screws and sorting them all out took 2 hours. So, you can imagine how much time the t-nuts would add...

It's so many screws.

12/23/2025 11:26 AM - Motor Sheath

Forgot I wanted to add this. Just so my motor doesn't get trashed by epoxy, I am going to print it a sheath out of polypropylene. Because it is a flexible material, I can kinda just make a form fitting "glove".

12/23/2025 11:41 AM - Feet (and now the machine itself is done!)

Decided to make some quick simple feet. With that, it's basically done! I just need to decide on what to do for the electronics enclosure, but I think I will keep it simple. Need to do the BOM first so I know for sure what electronics I am using!

Hard to get photos of the full thing because of how big it is.

12/27/2025 7 AM - Writing a BOM

I spent about 4 hours (3am to 7am) writing up a bill of materials for this thing. About 150 screws!
Took a long time to cost optimize and find everything, but I got it done. One step closer to submission!

Such a big list! Everything is either aliexpress or misumi (misumi is just the frame, and is cheaper than buying stock and cutting it myself at this scale).

12/27/2025 9 AM - Electronics + Electronics Enclosure

This is the last part needed before I can ship the design!!

Going to be using an skr mini and a 150w 24v psu. Super affordable and simple.

It's just a simple box held on by 3x m5 screws with T-nuts (t-nuts not pictured). I also included a fan just in case things get a bit warm powering nema 23 motors. However, since this is not a 3D printer, and I am using sensorless homing, the only wires coming out of the enclosure will be motor cables and the power cable! So simple and elegant.

With that, I just have to make the repository look all nice, and submit!

12/27/2025 2 PM - Swap to V wheels for the Embiggening

So, after talking to some rocketry friends, I decided it would be beneficial to have a bit larger usable size. To allow for this size increase with the $400 budget, I had to swap from the linear rods to v-wheels.

So now I can wind 60" long tubes! Which is as long as I will literally ever need. Super cool!

I also had to redo parts of the BOM to match the changes. Mainly just some screws, the frame, and the removal of the linear rods in exchange for cheaper vwheels.

12/27/2025 3 PM - Repository Formatting and Renders

I spent a few hours doing formatting and renders, writing my readme, etc.
I think it turned out pretty good! The renders took a bit as my computer was misbehaving. Ended up with very similar formatting to my repository for BECCA. It's clean and simple!

Some closeups of the final renders

And the repository!
I guess it's time to submit... scary

12/28/2025 1:42 PM - Fixing things for submission

I was asked to make some changes by a reviewer, so I did.

• I corrected the BOM to be a .csv I have no idea how it became a .txt.
• I created and uploaded a wiring diagram
• I added links to Marlin and Cyclone in my main README as well as added a basic marlin configuration file in the Firmware folder.

Overall this didn't take super long - here's what the wiring diagram looks like!

Made with Excalidraw.

12/28/2025 1:55 PM - I forgot the fan on the wiring diagram

Completely forgot I had a fan for the electronics.

Now that I added that, I should be good to go! With the new project extension I am considering trying to build a proper composites oven? But I am not really sure what would be the best way to get a box big enough to hold 60" tubes to hold a temp of like 160 celsius and not cost a bajillion dollars.

1/6/2026 - Update BOM to meet reviewer requests and out of stock parts

I got told I couldn't use the grant to pay for the carbon tow, which was something I initially was concerned about so I am not surprised. The skr mini and nema 23 stepper motors I was planning on buying have gone out of stock, so I had to switch to the next cheapest listings that would still fit my needs (and was comfortable buying from), which unfortunately raised costs a little. So, while the carbon tow has been removed, that actually only ended up removing a few dollars from the grant.

First image is the render for the sake of my project thumbnail - heres a huge zoomed out screenshot of the BOM.

Also this took a while because I quadruple checked the BOM, which involves going through the aliexpress cart and the BOM list and manually marking off every component (in the alix cart I would just select the item with the little check box, in the google sheets for the BOM I would change the cell fill color to green).

Hopefully this is the last resubmit.

1/14/2026 - Received the grant

Grant acquired ! Only took a little.
I placed the aliexpress order after spending 15 minutes double checking everything. Unfortunately, Misumi’s website is currently not able to process debit cards… So I will order the frame later !!

1/15/2026 - Misumi Website Fixed

Misumi fixed their website, meaning I can now order the frame (which I did). Their ordering process is a little tedious as it is very catered towards industrial customers but I got it done. They asked for a company and department so I said I am the Doohickey Department of Bagel Industries.

I also discovered that you can share PDF receipts directly to the HCB app and it can store them in a “receipt bin”. So that’s neat.

All parts ordered! Now we wait and print stuff.

1/25/2026 - Some Stuff Arrived

The Nema 23 motors and aluminum extrusion frame arrived. I will need to tap and drill blind joint holes into the extrusions but those are quite easy to do at my school shop, I’ll get it done some time very soon.

Kinda crazy how big these extrusions are. The super long ones are at least half a foot taller than me and the carriage extrusions, as in the part that is actually moving, are taller than those on my ender 3 or lulzbot. This is a really big machine.

Here’s a sense of scale on the carriage extrusions, aka me holding them, as I feel like the door comparison doesn’t portray it super well.

0.3 hours, or around 20 minutes, is how long it took me to unbox the extrusions!

Oh, an 8mm pin also arrived for the roller inside of the epoxy vat. I did not go for the M8 threaded rod that I am using for the other rollers here as I didn’t want to deal with epoxy in the threads. This roller will likely get a polypropylene sheath.