In Search of a Cheap Altitude Record - Going high and fast with 3D-printed components

In some discussion on this thread, I mentioned a project I'm working on where I'm flying some high-performance (but small) HPR rockets using largely 3D-printed components, and there was some interest in seeing a full thread on my project, so here it is.

The project started out as a bit of a resolution after I got my L3 this summer - continue to push my own boundaries in rocketry, but maybe try to avoid completely breaking the bank for a little while (gotta love student budgets). Out of that wish, came the idea of trying to push for as high an altitude as I could with as cheap a rocket as possible. I settled on 29mm (at least to start), and seeing what I could do with a cheap, largely 3D-printed airframe, and an Aerotech I205 motor (the largest DMS motor available - didn't want to risk any of my or the school club's hardware on the first proof of concept flight).

In terms of printers, I have a bunch of stuff available to me (see rpiforge.dev if you're curious), including a Markforged Onyx Pro, a (brand new!) Formlabs Form 3, and a full wall of Prusa mk 3's (plus a bunch of other cool things like a full-color 3D printer that don't really benefit this project). For the moment, I've written off the Form 3 as not really being suitable for this project (resin prints tend to be very brittle), and focused on the FDM printers.

From there, let's get on to some actual rocket design! It's a pretty simple 3FNC 29mm minimum diameter rocket, with a couple interesting features. I didn't want to have a fully 3D printed body tube, since that would present a really obvious point of failure along the layer lines, so I'm using phenolic tubing - at this size, I'm not really worried about it being brittle, and it's plenty strong, plus it's cheap! I went with a 12" section for two reasons - first, it's just long enough to hold the I205 (after including the fin can), with enough room for recovery; and second, 29mm phenolic comes in 36" lengths, I was always planning to make at least two of these rockets, and another project of mine needed a motor mount. Perfect!

Now for the part people actually care about - a 3D-printed fin can and nosecone. Design for the nosecone was straightforward enough - a 6" long (tallest the Markforged can print) Ogive nosecone, without any shoulder on it - I'm doing this like FWFG nosecones with a separate coupler that gets glued in, so it's a bit easier to print. I also gave it a 0.75" long straight section at the bottom for this coupler to nest into, so the actual ogive shape is only 5.25" long. The fin can was a bit trickier - it's a 4" long piece of 29mm tubing (standard PML sizes) with three trapezoidal fins attached - I went with something approximating my usual design - long root chord, small tip chord, and a trailing edge slightly swept forward to avoid breaking on landing (which is probably not as much of a concern with this rocket - I may revisit the exact fin design in future versions). I've attached STL files for both the fin can and nosecone to this post in case anyone wants to take a look or try printing them out yourself (they're in mm).

Now we get to printing - I tried 3 different filaments (in two different printers) in an attempt to see which might work best for this application. The first version I printed was a test print on a Prusa in PLA, with the standard print settings we've used in the shop (0.2mm layer height is I think pretty much the only standout feature), except for printing entirely as walls (for the added strength - it really only mattered for a tiny portion near the tip of the nosecone where it wanted to lay down some tiny amount of infill). I printed the fin can upside down so I wouldn't need supports, and both it and the nosecone came out looking great on the first try. I then decided to print them both on the Markforged, since I was slightly concerned about the structural integrity of the PLA parts, especially the fins - I was worried they were going to snap along the layer lines if I dropped the rocket, or potentially in flight. The tip of the nosecone on the Markforged print didn't come out great due to some issues with oscillation and trying to print a small tip like that, which were eventually resolved, but I didn't bother to re-print, since the bigger issue with that print is that, because the parts are so thin, I wasn't able to use ANY of the continuous fiberglass reinforcement material. This resulted in the parts being made entirely from the proprietary Onyx filament, which is something like 80% nylon. This resulted in substantially more flexible fins than the PLA, and I started to worry about fin flutter, since this project travels somewhere in the Mach 1.5-2.2 range, depending on the configuration. I had similar rigidity concerns with the PETG print I tried on the Prusa for the heck of it - I wasn't really expecting it to be better, but since it was only a couple cents of filament (and a couple hours of print time), I went ahead with it to see if it might at least have better layer adhesion than the PLA (answer: nope! I actually did break a couple layers apart on one of the fins of the PETG test fin can, and had to melt them back together with a torch).

In my next post, I'll have some pictures of each of the test print parts, as well as the fin flutter analysis I did for each of the 3 materials (and maybe I'll even have time to dig into some of the thermal analysis I did, which might help explain why I was so nervous for the first flight). Hopefully I can get all that stuff together while I'm sitting through my little brother's dance competition tomorrow!

I'm very interested in seeing how you were able to break apart the builds between layers. Specifically...
1) How similar was the test loading application to loading seen in flight.
2) How much optimization was done with the filament and print parameters to ensure optimum layer adhesion prior to building the test articles.

wsume99 said:

I'm very interested in seeing how you were able to break apart the builds between layers. Specifically...
1) How similar was the test loading application to loading seen in flight.
2) How much optimization was done with the filament and print parameters to ensure optimum layer adhesion prior to building the test articles.

Click to expand...

It wasn't a controlled test - I accidentally pressed too hard while sanding it prior to painting. As for the settings optimization, I used our standard makerspace PETG settings for the Prusas, which were tuned by the leadership there, but haven't had quite as much time spent optimizing them as the PLA settings (since PLA gets printed way more often). I have done some more controlled destructive testing on the PLA parts (which I'll talk about later when I can think straight), but currently don't plan to extend that to PETG because of the fin flutter issues (even given its better thermal performance).

jjwb22101 said:

For the moment, I've written off the Form 3 as not really being suitable for this project (resin prints tend to be very brittle)

Click to expand...

Don't discount the resin prints just yet. The strength of various resins varies widely. Refer to some measurements here for some of the FormLab resins I have worked with:
https://forum.ausrocketry.com/viewtopic.php?f=56&t=5324&start=30

The resin prints were generally the same or better than the expensive FDM print (in ABS).

jjwb22101 said:

The tip of the nosecone on the Markforged print didn't come out great due to some issues with oscillation and trying to print a small tip like that, which were eventually resolved, but I didn't bother to re-print, since the bigger issue with that print is that, because the parts are so thin, I wasn't able to use ANY of the continuous fiberglass reinforcement material.

Click to expand...

Very interesting. What's the minimum thickness/width for which it will use the reinforcement?

Do they have a story for increasing inter-layer (Z) strength (since the reinforcement presumably only helps strength in X and Y)?

This is the kind of information we hoped to get.

A little bit about fin flutter. My observation is that it’s worst just before M1 and then drops off a lot. It doesn’t necessarily get worse the faster you go.

Try are larger nozzle with wider extrusion width. Gives more surface area to bond.

And, maybe ABS or ASA (modded abs for UV resistance). Might be a decent compromise between stiffness/strength and heat tolerance.

ASA is strong. You can also get ASA with CF in it.

OverTheTop said:

Don't discount the resin prints just yet. The strength of various resins varies widely. Refer to some measurements here for some of the FormLab resins I have worked with:
https://forum.ausrocketry.com/viewtopic.php?f=56&t=5324&start=30

The resin prints were generally the same or better than the expensive FDM print (in ABS).

Click to expand...

I may give them another try with the Form 3 - we just got it after our Form 1+ broke its stepper motor for the third and we gave up on fixing it. The other downside is going to be figuring out a print orientation in which it'll have enough area to adhere to the supports, but also won't cause pooling of the resin inside the part (especially with the nosecone I could see that being a potential problem). It'll also have the advantage of giving finer layers, so the bevels on the fins should come out cleaner.

JohnCoker said:

Very interesting. What's the minimum thickness/width for which it will use the reinforcement?

Do they have a story for increasing inter-layer (Z) strength (since the reinforcement presumably only helps strength in X and Y)?

Click to expand...

The thickness depends on whether the part being reinforced is a continuous ring or a protrusion - since the fiber has to be laid continuously, for a protruding part it must be able to go out and back. For a continuous ring (like the airframe tube) the minimum thickness is 2.9 mm (or 0.114 inches), and for a protrusion it is 3.8mm (or 0.15 inches). As for increasing Z strength, I'm not sure there's much argument that it'll actually do much, though a fiber run is significantly thicker than the individual layers, so it might do a little bit. The bigger advantage there is that the markforged machines are professionally tuned for industrial use, so their layer adhesion is excellent (I've actually never seen a print from that machine break cleanly along the layer lines, they tend to be nearly isotropic in their failure modes).

Wallace said:

Try are larger nozzle with wider extrusion width. Gives more surface area to bond.

Click to expand...

That's something else I've considered, but I'm already dealing with at most 4 walls for the fins, and the beveling turns out not great as a result (since it can really only take out layers in multiples of 2, there's a bit of a sharp transition where it goes from 4 walls down to 2). Also, in my experience (we've got a volcano nozzle that's almost always attached to one of our TAZes), the layer adhesion isn't significantly improved by increasing the nozzle size, since you also have larger gaps between each wall. I'd rather focus on getting good layer adhesion with our standard 0.4mm nozzles than having to re-start the process with a different nozzle size.

Wallace said:

And, maybe ABS or ASA (modded abs for UV resistance). Might be a decent compromise between stiffness/strength and heat tolerance.

Click to expand...

cwbullet said:

ASA is strong. You can also get ASA with CF in it.

Click to expand...

ABS or ASA filament is another thing I'm excited to finally be able to do some experimenting with this semester - the makerspace room is an old converted server room, so the temperature is fairly inconsistent, and always cold, which made printing with ABS, nylon (aside from on the markforged, which has a heated chamber), or any other temperature-sensitive material basically impossible. However, at the end of last semester we got a new shelving unit with individual enclosures for each of the Prusas, so hopefully that will be less of a problem moving forward!

Steve Shannon said:

This is the kind of information we hoped to get.

Click to expand...

jjwb22101 said:

That's something else I've considered, but I'm already dealing with at most 4 walls for the fins, and the beveling turns out not great as a result (since it can really only take out layers in multiples of 2, there's a bit of a sharp transition where it goes from 4 walls down to 2). Also, in my experience (we've got a volcano nozzle that's almost always attached to one of our TAZes), the layer adhesion isn't significantly improved by increasing the nozzle size, since you also have larger gaps between each wall. I'd rather focus on getting good layer adhesion with our standard 0.4mm nozzles than having to re-start the process with a different nozzle size.

Click to expand...

@Steve Shannon - IMO this is the most important snippet of the whole discussion. While this information is certainly useful and will help people understand a few of the many considerations that go into 3D printing the problems that are being encountered are likely a direct result of the fact that thin fins are being printed. 4 walls at 0.4mm is only 1.6 mm fin thickness or ~1/16". (Note - I'm assuming that if a 0.4 mm nozzle is being used that the line width is also 0.4 mm which using a line with that matches the nozzle diameter is common practice but OP can confirm). I doubt you are getting many people flying L3 rockets with 1/16" fins (perhaps CF).

What is being laid out for all to see here is that 3D printing parts is not easy and there is a very important step in the process that most people are not aware of - setting the build parameters. Even if you download a design it's just a 3D model (in almost all cases), someone still has to figure out how to manufacture that design on the 3D printer. Just like if you wanted to machine the part on a CNC mill someone would still have to program the machine to produce the geometry. When it comes to 3D printing a fin can I would suggest this be viewed as the assembly process. This is equivalent to purchasing a rocket kit and then the builder figuring out how to assemble the parts to ensure they can withstand the forces of flight. You are figuring out how the filament will be assembled into the shape of the model to produce the part strength you desire.

@jjwb22101- What temp are your build temps for PLA? What is the print speed? I would suspect that these parameters are going to influence layer adhesion.

jjwb22101 said:

I may give them another try with the Form 3 - we just got it after our Form 1+ broke its stepper motor for the third and we gave up on fixing it. The other downside is going to be figuring out a print orientation in which it'll have enough area to adhere to the supports, but also won't cause pooling of the resin inside the part (especially with the nosecone I could see that being a potential problem). It'll also have the advantage of giving finer layers, so the bevels on the fins should come out cleaner.



The thickness depends on whether the part being reinforced is a continuous ring or a protrusion - since the fiber has to be laid continuously, for a protruding part it must be able to go out and back. For a continuous ring (like the airframe tube) the minimum thickness is 2.9 mm (or 0.114 inches), and for a protrusion it is 3.8mm (or 0.15 inches). As for increasing Z strength, I'm not sure there's much argument that it'll actually do much, though a fiber run is significantly thicker than the individual layers, so it might do a little bit. The bigger advantage there is that the markforged machines are professionally tuned for industrial use, so their layer adhesion is excellent (I've actually never seen a print from that machine break cleanly along the layer lines, they tend to be nearly isotropic in their failure modes).



That's something else I've considered, but I'm already dealing with at most 4 walls for the fins, and the beveling turns out not great as a result (since it can really only take out layers in multiples of 2, there's a bit of a sharp transition where it goes from 4 walls down to 2). Also, in my experience (we've got a volcano nozzle that's almost always attached to one of our TAZes), the layer adhesion isn't significantly improved by increasing the nozzle size, since you also have larger gaps between each wall. I'd rather focus on getting good layer adhesion with our standard 0.4mm nozzles than having to re-start the process with a different nozzle size.




ABS or ASA filament is another thing I'm excited to finally be able to do some experimenting with this semester - the makerspace room is an old converted server room, so the temperature is fairly inconsistent, and always cold, which made printing with ABS, nylon (aside from on the markforged, which has a heated chamber), or any other temperature-sensitive material basically impossible. However, at the end of last semester we got a new shelving unit with individual enclosures for each of the Prusas, so hopefully that will be less of a problem moving forward!

Click to expand...

I've had many successes (even some fairly large parts) printing ABS w/out any type of enclosure when it was warmer out. Just using blue tape and glue stick. Now that winter is here I won't even try, at least not 'till my enclosure is finished..

wsume99 said:

@Steve Shannon - IMO this is the most important snippet of the whole discussion. While this information is certainly useful and will help people understand a few of the many considerations that go into 3D printing the problems that are being encountered are likely a direct result of the fact that thin fins are being printed. 4 walls at 0.4mm is only 1.6 mm fin thickness or ~1/16". (Note - I'm assuming that if a 0.4 mm nozzle is being used that the line width is also 0.4 mm which using a line with that matches the nozzle diameter is common practice but OP can confirm). I doubt you are getting many people flying L3 rockets with 1/16" fins (perhaps CF).

What is being laid out for all to see here is that 3D printing parts is not easy and there is a very important step in the process that most people are not aware of - setting the build parameters. Even if you download a design it's just a 3D model (in almost all cases), someone still has to figure out how to manufacture that design on the 3D printer. Just like if you wanted to machine the part on a CNC mill someone would still have to program the machine to produce the geometry. When it comes to 3D printing a fin can I would suggest this be viewed as the assembly process. This is equivalent to purchasing a rocket kit and then the builder figuring out how to assemble the parts to ensure they can withstand the forces of flight. You are figuring out how the filament will be assembled into the shape of the model to produce the part strength you desire.

@jjwb22101- What temp are your build temps for PLA? What is the print speed? I would suspect that these parameters are going to influence layer adhesion.

Click to expand...

Thanks!

wsume99 said:

@Steve Shannon - IMO this is the most important snippet of the whole discussion. While this information is certainly useful and will help people understand a few of the many considerations that go into 3D printing the problems that are being encountered are likely a direct result of the fact that thin fins are being printed. 4 walls at 0.4mm is only 1.6 mm fin thickness or ~1/16". (Note - I'm assuming that if a 0.4 mm nozzle is being used that the line width is also 0.4 mm which using a line with that matches the nozzle diameter is common practice but OP can confirm). I doubt you are getting many people flying L3 rockets with 1/16" fins (perhaps CF).

What is being laid out for all to see here is that 3D printing parts is not easy and there is a very important step in the process that most people are not aware of - setting the build parameters. Even if you download a design it's just a 3D model (in almost all cases), someone still has to figure out how to manufacture that design on the 3D printer. Just like if you wanted to machine the part on a CNC mill someone would still have to program the machine to produce the geometry. When it comes to 3D printing a fin can I would suggest this be viewed as the assembly process. This is equivalent to purchasing a rocket kit and then the builder figuring out how to assemble the parts to ensure they can withstand the forces of flight. You are figuring out how the filament will be assembled into the shape of the model to produce the part strength you desire.

@jjwb22101- What temp are your build temps for PLA? What is the print speed? I would suspect that these parameters are going to influence layer adhesion.

Click to expand...

One of the advantages of 3D printers is that you can print the same parts multiple times and stress test them to the point they break and then improve as needed. This is something that is much more difficult to do with conventional methods. Whenever I change the design, materials, etc. I stress test the parts in my garage by hanging weights or putting pressure in critical areas. It would be great to get some help from experienced builders on other stress testing techniques including how to calculate limits, tolerances for testing different rocket parts. Something like an Excel spreadsheet where I could input the size/mass of the rocket, motor size and will tell me what forces and where I should apply them to test the strength of my parts would be awesome. It is not just about building techniques but more importantly I think how we test the components and I think this is the area we should emphasize.

Formlab now has a glass-filled resin. We have printed parts at work with it and they are dimensionally accurate and very stiff. This would mitigate fin flutter that might occur on more flexible materials. I have not done any strength testing on that resin. Maybe I should .

For ABS prints from the Stratasys printer I have had no problems with heat, either with static loading on the pad or dynamic loads during flight. I have used it for nosecones and transitions on M-class flights. Fastest so far is Mach 2.14.

Steve Shannon said:

A little bit about fin flutter. My observation is that it’s worst just before M1 and then drops off a lot.

Click to expand...

aka the "Transonic Region" ( MACH 0.8 - MACH 1.2 )

https://media.neliti.com/media/publ...mic-characterisation-of-rocket-f-a5fda0b0.pdf

https://apps.dtic.mil/dtic/tr/fulltext/u2/a502110.pdf

Dave F.

Remember that figures for the "transonic" region vary for each design. It is, by definition, the points at which some part of the airflow around the aircraft are simultaneously subsonic, sonic, and supersonic. This can happen in some designs at speeds below what you have indicated, typically in areas of high turbulence. The numbers are always rubbery, and general in description, unless you have some CFD or experimental results to pin them down better.

Angle of Attack also comes into play.

AoA creates turbulence. Many airframe and flight envelope factors affect the numbers.