I've gone off the deep end: working on a filament winder

The Rocketry Forum

Help Support The Rocketry Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Did my first real wind yesterday! Surface is just a little lumpy, the tow is not perfectly even coming out of the spool.

The tube here is for a 38mm motor mount and is 3 layers thick. My mandrel was a 1.5" aluminum tube and the OD is measuring 1.558"-1.569". The wind was finished with a 50% overlap of Dunstone shrink tape. I have to sand off the excess resin on the surface yet, but I'm very happy with the result and ease of the wind.

Tensioning out of the center pull spool still needs some work and I could use a length of filament calculation to estimate resin usage.

Next up is a 29mm motor mount and some 5/8" guide tubes for a 2-stage!PXL_20210502_125731767.jpgPXL_20210502_125719387.jpg
 
It would be impossible for me to do what you do. But I love reading about others who can, and who do it! Keep posting results, whether good or not-so-much. Encourages others to get into the activity.

Best -- Terry
"More education is almost always better than less" -- T. W. McCreary ;)
 
I've done the 2 more winds since the 38mm motor mount. On the 38mm I found that I had the angle reference wrong, so the winds were more circumferential than longitudinal. The 29mm wind I had an error in my warnings and I had winds laying over top of each other within a layer (you can see the thickness difference in the picture below). Last one I did was a 1/2" ID tube for a set of 4 guide tubes for my SA-2/ S-75 2-stage. The 1/2" came out fine with the exception of the belt loosening on my carriage about 1/3 through. I saved the wind by manually holding the belt to the carriage. I'm sure there are timing artifacts from all of the slips, but most people won't be seeing the ODs of these guys.

PXL_20210521_182449454.jpg

Having a lumpy OD is a common characteristic of all my winds so far. For the sizes I'm running right now, I'm thinking a narrower shrink tape will help a little (I'm using 1.5"). I'm thinking a half- layer of a circumferential wind would do nicely to smash everyone down and serve as a sacrificial layer for finishing.

Speaking of finishing, hand-sanding is brutal and has its own set of issues with getting a nice looking tube. My lathe is available, I just need to print the end flanges to mount a wound tube for finishing. ...I do have a strong working knowledge of centerless grinders... I'm toying with the idea of making one out of either a belt sander, belt grinder, or a bench grinder...
 
Last edited:
Figured I owed you all some better content. I wound another 1/2" ID tube today. I wanted more ID clearance and a smaller OD for my guide tubes. Went from 1.5 layers of parchment paper to 3.5 and wound 2.5 layers vs 3. Figured it was a great opportunity to trial a spiral wound half layer to finish and to document the process.

3 video parts, enjoy some narration for once:


 
You are doing amazing work! I really can't say more than that, as it is just great. Keep posting and I'll sure keep watching.

Sandy.
 
Well, I've updated my carriage with a tension wheel that uses a rubber band tire. Early dry tests look great! I also tightened up the spacing.

These dry test winds were too pretty not to share. I'm making a new 29mm motor mount to replace the one with the messed up sequence.

I have 3/4" shrink tape to try with this next one. The half layer wind definitely helps, but the surface artifacts I'm seeing are coming from the shrink tape. Being a motor mount, I won't be doing a finish half-layer.
 

Attachments

  • PXL_20210627_200116920.jpg
    PXL_20210627_200116920.jpg
    225.1 KB · Views: 61
  • PXL_20210627_212007289.jpg
    PXL_20210627_212007289.jpg
    116.9 KB · Views: 56
Another update: worked on dialing in the speed. The 1.125 x 18" x 3 layer winds were taking 40+ minutes which is unacceptable. Looks like I'm now around that time for a 2.125 x 24" x 4 layer wind!

Here's a video of a dry wind test! Much noiser from a "downgrade", but boy does it sling glass now compared to before.

 
Are you using NEMA17 or 23 stepper motors?

I'm using nema 17 running on 12V right now. The torque is really quite anemic on the spindle with my 54mm mandrel. I'm working on designing a polisher to easily finish the tubes right now. When I take a break from that I'll either work out a belt drive reduction or nema 23 upgrade.
 
Another update: worked on dialing in the speed. The 1.125 x 18" x 3 layer winds were taking 40+ minutes which is unacceptable. Looks like I'm now around that time for a 2.125 x 24" x 4 layer wind!

Here's a video of a dry wind test! Much noiser from a "downgrade", but boy does it sling glass now compared to before.

With a dry run, is that filament just going to waste? I see you stopped it early but no reason to waste more material. It looks like you were satisfied for the performance.

I've bought some tubes from hobbyist "winders" and the quality was great. If you get the skills and some stock builds up to more than you can use, you can always post and sell it on TRF if so inclined.

Kurt Savegnago
 
I'm using nema 17 running on 12V right now. The torque is really quite anemic on the spindle with my 54mm mandrel. I'm working on designing a polisher to easily finish the tubes right now. When I take a break from that I'll either work out a belt drive reduction or nema 23 upgrade.

I would recommend upgrading to a NEMA 23. I do not recommend a belt reduction. Steppers lose their torque the faster they run. My winder has gearbox’s with a 4.25:1 ratio and 1.6:1 gear and pinion set up making the motor spin 6.8 times more than the original speed. This causes my motors to lose a lot of torque and I can not run my winder more than 40 RPMs
 
Time for another update, I wound 2 tubes this week. The first one is a horrifying failure. The second is looking like it's going to come out great! This one is a 2.167 ID x 25" long target length (need a 23" section). This is to replace a kevlar roll wrapped tube that my dad and I rolled 20 years ago ( and then my dad stole 10 years ago for a high school robotics project). The goofy ID is thanks to the mandrel prep we worked out back then:. Craft paper rolled directly on the mandrel (2" copper pipe), followed by several wraps of a carefully cut trash bag. I'm using my 2.125" aluminum mandrel (same OD as 2" copper), 2 wraps of craft paper, and the full 15" width of parchment paper.

This is a 4-layer wind, alternating between 45 degree and 65 degree triple helix patterns. My failed wind I finished with a straight helical half-layer and pulled the shrink tape stupid tight. Turns out I was pushing the filament layers all over the place with the shrink tape, no to mention I lost steps/ got pattern shifts from running the tension too high.

Anyways, here's the content:

Winder in action:


Filament lays mid-wind
PXL_20210813_024856571.MP.jpg

Applying shrink tape by hand:
PXL_20210813_030301726.jpg

Shrink tape in place:
PXL_20210813_030617217.jpg

Resin weaping as I'm shrinking the tape:
PXL_20210813_031507466.jpg

More weaping:
PXL_20210813_031847129.jpg

Shrink tape removed:
PXL_20210813_121249947.jpg



My previous issues with "lumpiness" were due to wrinkles from not enough tension on the shrink tape, this time I got it! This tube will be going on my lathe (what the winder is sitting in) to get those resin ridges sanded down. On the lathe I have the perfect opportunity put on a thin layer of epoxy if I start sanding into the fibers. I'm wondering now if I need to go to a 75% overlap on the shrink tape or go down to 1/2" wide if it's available. A perforated shrink-tube would be an interesting solution...but I'm not doing anything new here to warrant an exotic solution...
 
nice work on the glass tube, I will recommend more overlap on the shrink tape, however, ask for the 120R series from Dunestone, it is more flexible and doesn’t require so much tension to wrap around.

I recommend getting a squeegee to scrape off as much of the excess resin as possible, this also helps when applying the shrink tape.
 
Last edited:
Well I finished the tube in my lathe. 2 skim coats of resin made this the highest quality tube of my making to date. The 2nd to last picture shows the rest of the airframe for this build (sans tailcone). That's a kevlar roll-wrapped tube with a fiberglass veneer layer circa 2002! Way back then carbon cost a fortune so my dad and I used it sparingly.

This build is actually a clone of our favorite rocket from back in the day. We got it here to 80% while the original was camping out in a tree at the MDRA sod farm for 2-3 months. We pulled it down and only cleaned it up, and replaced the electronics. It got 2-3 more flights before my ~19 year hiatus. It's in the ceiling of my basement right now waiting on some minor work to rejoin the fleet.
 

Attachments

  • PXL_20210814_154221778.jpg
    PXL_20210814_154221778.jpg
    307.6 KB · Views: 39
  • PXL_20210814_154229625.jpg
    PXL_20210814_154229625.jpg
    114.2 KB · Views: 42
  • PXL_20210817_014842064.jpg
    PXL_20210817_014842064.jpg
    163 KB · Views: 41
  • PXL_20210817_014907323.jpg
    PXL_20210817_014907323.jpg
    168.3 KB · Views: 40
Do you think these tubes would make good cases for disposable ex motors?

Been doing it for a few years since we got our X-winder -- best use we've found for it.
Here is an early 54mm being wound - Areotech nozzle re-used - propellant and closures already glued in place prior to winding.
Clearly single-use.
WindShow.jpg


Bitchin Flights -- great thrust-to-weight ratio's:
Here is an L-1000 on takeoff:
WoundFlight.jpg

The hard part is keeping the fins on......
 
Been doing it for a few years since we got our X-winder -- best use we've found for it.
Here is an early 54mm being wound - Areotech nozzle re-used - propellant and closures already glued in place prior to winding.
Clearly single-use.
View attachment 477883


Bitchin Flights -- great thrust-to-weight ratio's:
Here is an L-1000 on takeoff:
View attachment 477884

The hard part is keeping the fins on......

That is awesome, are you guys using high temperature resin or are you using something else? Also do you just do hoop winds on the liner or do you also add helical winds?
 
https://tcrcomposites.com/product/product-tow.php You might consider looking for something like this. Right now you are using way too much epoxy. That much excess guarantees the fibers will swim and not be properly compacted. IMHO, of course. It looks to me like you are using easily way more than double what you should be using.

You might be able to come up with an automatic epoxy applicator to go between the top wheel on your carriage and the hook. That might be one way to improve the control on the process.

Another approach you could consider is to do the complete winding dry with no epoxy added. Then use a resin diffusion epoxy (likely any low viscosity laminating resin with the properties you want) on the the outside of the dry tube, and let it soak towards the inside. Weigh how much epoxy you use - you want somewhere in the ballpark of 2/3 the weight of your glass. Once the epoxy is taken up by the fibers, then wrap with shrink tape and shrink, or use your filament winder to wrap mylar strip around the outside with just a hair under 50% overlap. That's if your winder can achieve a little tension. Then put the mandrel and tube in a hot box for the first cure.

A hybrid of that is to do a full layer, stop the winder, epoxy coat with thin foam paint roller just enough that the glass becomes transparent but NOT wet, then resume winding. Do in layers, dry winding over damp fabric. The new winding will take up excess epoxy from the previous layer. Leave the outermost layer dry. Put on your shrink, and squish it solidly! The dry outer layer will take up the excess and you should have very little ooze. Done this way the viscosity of your epoxy won't matter as much. And for that matter, mix a fresh small batch of epoxy for each layer. With layups it is generally better to work with multiple smaller weighed and well mixed batches, rather than one single batch which is thickening on you by the time you are done. Thickening epoxy doesn't wet out materials nearly as well, and leads to excess epoxy being used. That makes for weaker layups - possibly much weaker.

Just some ideas for your consideration! Have fun!

Gerald
 
You might be able to come up with an automatic epoxy applicator to go between the top wheel on your carriage and the hook. That might be one way to improve the control on the process.

Funny enough I bought pastry piping bags to use exactly for this purpose. My inspiration was this winder from the start:

I have a ways to go before I want to tackle tow-preg. For one, I don't have a means the get the layup hot enough to cure. Tow-preg machines are insane and I would love to eventually be on their level: Just listen to those servo motors sing!

I'll absolutely try knocking down the amount of resin I'm using, I just have a huge fear of getting dry spots.

My mandrel motor only has 85oz-in of holding torque (65oz-in dynamic?) and the 2.125" mandrel is the largest that runs comfortable so far (haven't tried my 2.5" mandrel yet though). I can spin the 4.5" mandrel but the reduced accelerations have a huge affect on the wind speed, though I've made a couple of optimizations since I last tried. Regardless, the reserve torque on the 4.5" is abysmal, such they I'm afraid I'll stall it applying resin.

I'm doing some background work on v3 of my winder. I want to double the torque on the the traverse so I don't ever have to worry about stalling it. For the mandrel motor, I have a NEMA 34 motor picked out that has ~1,200 oz-in of holding torque... However, my lathe has a 1hp 3-phase motor on it... The VFD I have running it can't accept an encoder for position control, but the Arduino nano I have in front of me can! Combining the winder with the lathe would be a massive improvement in handling layups (and floor space!). The 1hp 1800rpm motor is only good for 576 oz-in, but there's no reason why I cant apply a reduction and get more torque eventually. I don't run over 900 rpm anyway when I'm sanding! I can run the motor up to 2,250 rpm safely so a whole world opens up with reductions!

I still have a each iteration of tube winding fixtures, at some point I'll share how I've advanced since 2019. I may have a picture of the layup lathe my dad and I used (it still exists!) back in 2000-2002.
 
You probably don't have the torque to wet the tow as part of the application process. But if you can pause the winder at the end of each layer you may be able to do the hybrid approach I mentioned.

A steel pipe, wrapped in a heating element (can get as cords) connected to some sort of temperature control circuit (could be a light dimmer switch) wrapped in some high temp insulation and another tube, could suffice for your curing oven for higher temp stuff. You can achieve much higher temperatures than you would need. Think of it as a tubular oven. With and end cap as a door. You'd have to calibrate or have temperature measurements from the inside, or make some test parts. That would be for post-cure of a high-temp system. The initial cure temperatures are not that high.

Gerald
 
I should mention that I've made tubing before, two different ways. But I've never worked with a filament winder. Any info I have there is second or third hand. A filament winder is a great way to make shafts that need to transfer torque, but it is not a great way to make rocket tubes since such tubes take completely different loads. The fiber orientation from a filament winder is nearly optimally bad for the job. The net result is such a tube needs to be much heavier to take the same stresses. However filament winding does give lots of flexability on what you do with it. It is an excellent tool.

One tube I made weighed 13.5g, was about 20" long, and would hold 7 1/2# weight cantilevered on its end without breaking. Diameter of the tube was about half an inch tapering to a quarter inch. All of the fibers went lengthwise or radially. Actually both tube methods I've used achieved that result. The methods were quite different though. If interested I can give some details, but the methods are not relevant to using a filament winder.

For a rocket body tube, think of it as a long thin cylinder of thin wall. This cylinder should be designed to take axial compression. When a long thin cylinder is compressed axially, it wants to bend out to a side - flex a little. As it flexes, the tube will ovalize such that the difference in the stresses on the inside of the bend and the outside of the bend decrease. It gets thinner in the direction of the bend and wider perpendicular to that direction.

As the stresses approach the failure point, the ovalization increases. The resultant decreased separation of the opposite sides of the tube decreases the stiffness of the tube resisting the bending load. This leads to stress concentration at the bend - the ovalization becomes more localized and more pronounced. As the load is increased, this eventually leads to catastrophic failure.

Fibers should be optimally oriented to resist the stresses.

End-to-end compression resistance - fibers parallel to the center axis of the tube. Where they are situated within the thickness of the wall of the tube has little practical effect. It is after all a thin walled structure.

Resistance to ovalization - that is more interesting. Take a small patch of the tube, cut it out, and magnify it. What we are looking at is a nearly flat sheet of some thickness. Resisting ovalization around the tube is equivalent to resisting bending in one direction on that thick sheet. The optimal fiber layout to resist such bending is to have fibers parallel with the bending direction of that sheet, and on both surfaces of the sheet. Fibers towards the center of the thickness of that sheet do essentially nothing to resist the bending.

So the optimal layout for a rocket body tube, in a generic sense, is an outer and inner layer of unidirectional fiber perpendicular to the axis of the tube, and a layer of unidirectional fiber parallel to the axis of the tube sandwiched in between.

(How much of each fiber orientation is needed depends on the ratio of length to diameter for the tube, and on the ratio of wall thickness to diameter)

A filament winder can't achieve this ideal fiber orientation - but it can try to approximate it. Inner layer with windings using a very slow traverse. In fact, make the filament wraps lay right next to each other, no gaps. Middle layer(s) have a rapid traverse to get the filaments as close to parallel to the tube axis as you can manage. Then the outer layer is back to having very slow traverse with the filament wraps touching each other.

For the same stiffness and strength, you'll need somewhat less material that way. AKA a lighter tube.

You might even find it advantageous at some point to consider putting in a structural foam such as Rohacell 1mm sheet between two of the layers. That increases the separation of the inner and outer walls - increasing the resistance to ovalization - without having to add the mass of fiberglass in there to achieve the separation. That would improve your strength to weight ratio. But it does have limitations for post-cure temperature.

Have fun!

Gerald
 
Last edited:
Just using TAP's highest temp resin. Various winds used - usually a hoop for the first and last.
It's an evolving art.

Is the resin room temperature cure? Most high temperature resins I know of require a post cure. I am curious as to how your curing it with propellant loaded o_O.
 
Just room temp cure - but you could do the elevated cure as most cycles are way under ignition temp.
Remember - the burn is only 3 seconds long.....
 
Last edited:
BTW, the towpreg winding machine video above is showing the winding pattern I'm talking about.

Gerald
 
Development action still going on, I have a terrific lesson on PID loops going on here.

I could have just bought a big NEMA 34 motor and drive, but there's no fun in that. Besides, if I get this induction motor to servo conversion working, I'll be condensing 2 machines and a large amount of electrical junk I've collected over the years into one machine!

PXL_20210919_014530790.jpg
 
So the optimal layout for a rocket body tube, in a generic sense, is an outer and inner layer of unidirectional fiber perpendicular to the axis of the tube, and a layer of unidirectional fiber parallel to the axis of the tube sandwiched in between.
Very interesting information; thanks for all the detail. It's amazing how there's always more depth to any subject.

If course there is a question of what the major stresses on the tube are. I suspect that surface-mounted fins and landing impacts are more significant.
 
So it has been quite a long time since my last update, I've had a job change, some pet problems, and a move all happen in the last year. The dust is finally settling and my work on filament winder V3 is being wrapped up.

Unfortunately V3 is a dead end: I got the induction motor work as a servo, but the 200Hz update rate in the VFD is far too slow to hold decent position. I was going to rig up winder V2 alongside V3 to demonstrate control of the induction-servo setup, but new problems came up. I've been having encoder and Arduino issues where the system is effectively broken.

The good news is that I learned a lot and failure was an acceptable outcome from the start. V3 was only built from junk I had laying around and some minor things off amazon to tie everything together. In the pictures I'll be sharing, you'll see I only built head stock. There was no sense in building the tail stock due to the amount of space and development time it would take up. The x-axis was also bulky and low risk, so it also never left early its early CAD stage.

Here are the specs/design intent of V3
* 1hp (750W) induction motor spindle drive
* 145W Integrated servo for x-axis
* 2-speed timing belt setup for winding/polishing modes (215/1200 rpm modes)
* Manual operation with position hold and pedal jog for roll-wrapping
* Continuous run mode for curing
* Target 12" dia mandrel with 20lb filament tension
* Marlin 2.0 control adapted from V2 (LCD screen and run programs from SD card)

I'm beginning a clean sheet design for V4, so far I've determined:
* 1.0+kW servo for spindle drive (A-axis)
* 145W integrated servo for X-axis (same from V3)
* Carriage mounted x-axis drive (more rigid, simpler setup)
* B-axis payoff guide (like eggplant's winder)
* Wet wind capability via draw-through resin bath
* Marlin 2.0 control adapted from V2 (LCD screen and run programs from SD card)

Pictures for follow
 
Back
Top