The initial delay was that I was thinking that the Formlabs IG account was going to want to publish the CAD but then that didn’t pan out. But then I realized that the CAD I had was built around a non-standard 45T bevel gear that TJ2 just happened to have a few dozen of (we were going to start selling our Diff Swerve Modules and by coincidence the SDS MK4 gear used an almost identical gear – not the MK4 wasn’t even a thing yet when we designed our Diff Swerve Module – great minds…) So I needed to spend a bit of time changing the design to be compatible with the SDS MK4 bevel gear. Plus TJ2 was busy getting ready for the WPI Battlecry (awesome competition by the way – really a well run event). Long story short, it too much longer than I hand planned.
If you haven’t seen the original IG post please go check it out. Formlabs IG featuring the TPU90A tire . The Formlabs social team is closely tracking engagement and if the posts that they do that relate to FIRST get enough interest I get to keep making cool stuff to share with the FIRST community.
FWIW, I think I can print 13 TPU90A tires on a Fuse1+ in one build volume that takes less than the 23hours you are showing. I know a Fuse1+ is not super available for most FIRST teams but it isn’t outside the realm of possibility for a lot of teams. Any team that can afford a reasonably capable CNC router could probably stretch themselves to get access to a Fuse1+
(I know I sound like a Formlabs homer but even after 5 years there, I still get blow away by the prints that a Fuse1 and Form3 can produce)
Thanks for updating & sharing the design! Thinking about 13 tires being produced within the Fuse1+ build volume (6.5x6.5x11.8")… The key seems to be that “The unused material supports the parts during printing…” as stated on this page, where the part packing feature is also shown.
A dual extruder (IDEX) FDM printer set up with dissolving supports could accomplish the same thing, but end-to-end production time would be a LOT longer, and there would be a lot more wasted material. (Plus, I don’t know of a “great” IDEX printer. And the Bambu Labs-style filament switch tech seems super wasteful.)
I agree with you that the Fuse1+ is within reach for some teams (cost is comparable to Markforged printers). Thinking about enabling students to use the printer though, how’s the learning curve, reliability, maintenance, etc. for the Fuse1+ vs. the much more common FDM printers?
It would be interesting to see a side-by-side comparison. I assume most parts would still be more economical (& fine) to make using FDM, but the opportunities of the SLS tech are intriguing. Thanks again for sharing.
The only issue I’m seeing with FDM (Bambu Lab) is when I print the “leaf springs” where they cross the filament sags enough to cause the centers to bond lightly. I think this is leading to the wheels being stiffer than they should be.
Trying to get the V40 model to post in the slicer to test it. (Old computers and complex models don’t mix)
Overall print time is down to around 13 hours for me for 1 Tread. Still a rookie at this 3d print stuff though.
Would it work the same if printed on a Bambu or Prusa or does it need to be printed on a Fuse 1? If it can be FDM printed, what materials need to be used?
I too am curious about @gxxk 's questions. Also, as a follow-up, for the TPU, is it safe to print it without a vented hood? Our printer does not have an enclosure.
Our Velox VR5050 (4’ square) was $17k fully loaded plug&play (in 2016), our 150w full 4’x8’ Lightobject laser was $18k fully loaded plug&play (in 2019), the Fuse1 starts at $19k for only the core machine - a minimum “workflow” setup with powder recycling is just clear of $31k. (To be fair… A Shopsabre PRO408 runs $32-40k, but I would call that more than moderately capable as a router with a vacuum table, atc, etc for 4x8 sheets)
From that I get minimum of $19/tire in powder consumption.
From first order analysis sanity checking, I can’t actually fill the build volume of 6.5x6.5x11.8" with a single 6kg powder purchase? I’d actually need about 9.5kg to get a complete build volume? This is taking a rectangular build volume and multiplying by the as-fused density, which I think is flawed.
Buying the ~$10k Sift “workflow” enables the “30-50%” refresh rate on powder, which still doesn’t seem great, but it definitely seems like a required purchase to keep material cost anywhere near reasonable.
To run one full-volume batch of 13 tires, I’d need two 6kg sets of powder. An argument could be I then recycle the vast majority of that powder into a second batch of tires… that’s still $60/tire for 26 tires, before we get into amortizing the machine cost.
I need to start replacing a lot of other purchased components before this thing pencils.
I don’t have experience with FDM printing. It might be possible to print in a TPU on an FDM printer. I just can’t say.
If you are asking about FDM printing with TPU, I don’t know. On a Fuse1 or Fuse1+ it is fine to print in a normal shop area. I don’t think it is unsafe for an office but there are a number for fans and motors that make noise plus some powder does get on the floors and such. Neither are great for an office environment.
Our kids print on Fuse all the time. We have one in our offsite facility (an empty strip mall shop that we’ve acquired during covid when meeting that the school was essentially impossible) and the students pretty regularly have prints that get printed there. If a student wants me to print a job in TPU or Nylon11CF (Carbon Fiber) at Formlabs, I have them prepare the “Form File” and all I do is download the job to the printer, start the job, and post process the parts (which is not hard but if there are lots of parts can be tedious). The process for preparing a job for a Fuse is pretty straight forward.
@Grant is our team captain. He has an FDM printer at home and uses Fuse about as much as anyone. I hope he will jump in an talk about his experience
The Onyx One is $6500 on Matterhackers, which is the highest I’ve ever seen - we got ours in the $5k range a few years ago. That compares to $30k for the Fuse workflow. Onyx material is $0.25/g, which does run about double the Formlabs $0.13/g TPU powder, but I don’t have a good sense of the Fuse waste ratio - the Markforged is pretty good about waste if you can design for a decently reduced number of supports and avoid spoiling whole spools.
The “springs” would definitely need to be redesigned for FDM due to the process limitations. As I noted in the previous thread, one possible approach for FDM is to use very sparse spokes in TPU and then fill the whole cavity with a soft 2 part casting polyurethane. I messed with the v33 wheels doing just that and got print times under 10 hours. I only had some old 20A durometer silicone to play with that was too soft, but the idea showed promise with a lot more tuning.
For context, I’m printing the tire out of Overture 95A TPU on an Anycubic Kobra Neo. It’s still mid-print and very much a long time investment for just the one, but it seems to be coming out okay so far. I’ll be happy to share more detail once it finishes tomorrow.
I’ve never heard anything that suggests you would need a vented hood for TPU. You will absolutely want a direct-drive extruder though. TPU is just too soft and stretchy for bowden style extrusion.
As I’m sure many of you are finding out, there are a couple issues with printing V40 using an FDM machine. First off, the Bambu and Prusa printers are high quality printers that yield amazing results. That being said, no FDM machine can emulate SLS, no matter how hard we may try.
The main issue is supporting the thin studs on the tire as well as printing the leaf springs on the inside. Here are a few ways you could do it, and why they won’t give you the same result as the Fuse 1.
Use a dual extruded machine with dissolvable supports. While everything will be supported, there is a massive difference between the way FDM and SLS supports work. FDM places material onto support structure that is either broken away, or for the best result, dissolved. The issue with this is that you compromise the next few layers after the support structure. The printer will often dispense material higher than its normal layer height to make the supports break off easier. Because of how small each spike is, this will likely significantly degrade the strength and therefore durability, and make the tire slightly directional. The support is also melded to the material placed above it, which changes the way that part of the stud will behave/wear after the supports are removed. You could use the same technique to support the leaf springs, but it will again loose strength from the imperfect layer, and will be inconsistent from tire to tire, or even from one side of the tire to the other.
Print unsupported. While some machines may be able to do it, when FDM printers print bridges (dispense material over air between two printed surfaces), they either sag or at the very least will have incomplete layers with thinner material lines until there is enough structure to support a normal layer again. Either of these outcomes will change how the tire behaves and possibly how long it lasts. The studs will have the same issue and will likely become very directional, prone to tearing along the layer lines, and/or will not poke into the carpet as they are too, for lack of a better word, droopy.
Attached is a gif of how the Fuse 1 prints from a top down view. The black is sintered powder, and the light gray is unsintered material. All edges surrounding the part are touching an equal amount of unsintered material, so even the supported areas behave the same as any other area on the edge of the print. Pick up any SLS part and it is nearly impossible to distinguish the orientation in which it was printed in, and no SLS prints are printed off of the build platform. They are all suspended in powder.
An FDM printed tire may behave differently depending on which direction the print head dispensed lines to create the part. Fused parts act more like injection molded parts than 3d printed ones in the sense that layers are bonded significantly better. Powder in the fuse is completely melted and acts as one continuous piece, which we believe is why the thin studs are so durable on our tires. FDM prints often fail along layer lines as this is typically the weakest part. If you were somehow able to perfectly and uniformly print the tires with FDM, they still wouldn’t be as durable as SLS tires because of the far inferior layer adhesion.
TLDR: No. You would likely lose the consistency in the way they drive, as well as significantly lower the durability. It’s very possible they would not be nearly as grippy either. The second image I attached shows a tire we had on the robot in worlds after a failed auto made the wheel hit the edge of the charging station like it hit a blade. We drove on this for another few matches and the damage did not spread. If this were FDM, I can’t say it would look the same.
I’m not really sure what I was expecting from this, but a print that finished on the first try was not one of those things. As Grant suggested, the SLS version is almost certainly going to be superior, but I can’t help but wonder if you could get a workable alternative on an FDM machine. If nothing else, it does look and feel somewhat like a tire.
V38’s suspension is significantly stiffer than V40. It’s essentially a thinner more compliant version of V33 with small stud angle changes. Suspension wise it didn’t add too many benefits so we went with the pattern found on V40 on all future tires.
V38 should be easier to print in FDM for a multitude of reasons, and may initially provide similar CoF numbers to V40 (tested with SLS, mileage may vary), but the durability even on a SLS printed V38 just wasn’t there. After one pushing event, a failed auto, a few times over the charging station, or really anytime the wheels spun over the carpet without the robot moving, the studs were eaten away. V33 is similar, and I believe is currently public, so feel free to try and print that for now if you want to feel it yourself. If you were to use them though, they get destroyed very quickly, and that was with a more durable SLS version.
I’m personally making a copy of this that will hopefully perform about as well as these, but are fdm printable, and also are for MAXSwerve (3" tire)
I don’t think they will be anywhere close and may underpreform tread by a significant margin, but it’s a fun design challenge.
Anyway, I’m making it out of four slices that have half of a leaf spring, so I don’t need any supports to print it.
The tread is not really close at all, it’s more of a straight line in a v shape. However, I figured that printing with FDM would not give me the ability to print the hair-like tread, so that’s the best that I can do as far as I know.
It’s printed on a Ultimaker S5 with TPU 95A filament. The hub is printed in PLA, but I would make it in markforged onyx for the final design. The TPU is really slick on my hand, so I don’t think it will perform that well on the field.
After printing one in 90A TPU, I don’t really think the leaf springs are much of a problem. The response when dropped isn’t nearly the bounce shown in the video but it’s definately compliant enough to increase the contact patch.
The texture on the tread is workable but definately not the design intent. This would be the real trouble. Might try printing them in just straight lines when i get more filament.
Definitely gonna test them for the sake of it, but it’s probably at best only 70% of sls.
