Shimano Derailleur Print - mechanical test

Here’s a test of a stock CAD-designed 20-part assembly of a Shimano bicycle chain derailleur ( created by Derek Recker, on GrabCad).

I did this print to see how well the printer would do with the fully assembled part. I loaded a single STL file with a total of 129K triangles.

I set up the supports manually. This was my first vertical print and I wanted to get a sense of what was required.

Here’s the preform set up I used:

This took me about 30-45 to do set up. Print time was about 7.5 hours at 0.05.

I DID have to scale the part down slightly to fit the build volume as a whole and that probably introduced a few issues.

Worked ok first time. Only two blow outs in some really thin stuff (more on that below). I did have some issues with ‘textured’ surfaces where the model actually included super thin grooves… there was really no way to support those so I just punted and hoped for the best. Same with some screw threads that were exactly horizontal (the screw was horizontal that is… no way to support those threads). Good news is that stuff seemed to come out ok).

The screw

Blow out on really thin wall

You can see here above where I had some issues around the hex nut. Lost some geometry here. Going back to look and see what was going on there.

Other than that I was pretty pleased with the results.

My NEXT test is to print all of these parts separately in a single print session, at their proper modeled size, and attempt to assemble the working mechanism from them. I’m dubious the screws will work, but the rest should assemble. I may run it at 0.05, and then 0.025 just to see if I get fewer failures.


1 Like

very nice !
Where did you get the model from?

Thanks Cesar.

Lots of great stuff to test out at It’s sorta a thingiverse on steroids for CAD engineers.

I’ve heard of it. Need to start borrowing some models from there.

I find the workflow of placing support manually, “Apply Changes” then go back a little clunky. Needs some improvement software wise. I can see how you can spend 45 minutes going back and forth.

I am still trying to warp my head around identifying where I need support and where I do not.

ROUND TWO! - Here are some notes on my attempt (!) to print out the individual parts from this mechanism. I learned a few ‘what not to do’ things here, with some successes, and more tests coming. Read on if you are interested…

For this test, I took the original CAD parts and brought them in to a single Preform 1.6 session. It took me a minute to realize I could multi-select them all - that was a huge help :slight_smile:

Once I brought them in, I manually spaced them out and decided (for this round) to manually set up supports. (not entirely successful… more on that below).

I also decided to attempt to print some flat-based objects directly on the build platform with NO supports. (also not entirely successful, but educational - and somethings did work).

I should also point out that I used the native scale of these parts (where in the assembled print run I did first, I had to scale things down to 75% of native to get it to fit the build volume. It was interesting to see that the assembled part did better in some ways - though smaller - than the un-assembled parts printed individually at a larger size. I think this was because the assembled part fused super small parts together and they maintained some integrity. Where printing those small parts separately resulted in failures… more on that below

In all, set up time was 45-60 minutes. NOTE: as indicated by Preform support folks, this is NOT strictly necessary to do - I’m on a quest though to understand how/what/why and so I’m going this route for a bit longer).

I printed still at 0.05 with Grey Resin. VERY tempted to try .025, but not ready for that yet.

Here’s the Preform 1.6 layout:



Print Time was just over 7 hours.

As you can see - it is perhaps a crazy layout. Pretty tight. I had no clue if I would have ‘resin flow’ issues or what. Also as this was my third or fourth print, I was hoping I would not start to see any issues with peel as the tray is starting to get some built up use and I can SEE where things have been printed on the tray floor when scraping the resin tank…

Furthermore, in one case (below), I kept getting RED in the support editing mode. And though you can see I did not add a lot of supports, even when I added more, it stayed red. I was unsure where the magic missing support was, so I just went with less. Which you will see below may have been the source of the failure.

Flirting with Disaster - manual edits and too much red.

So here’s how it turned out on the build platform:

Definitely some failures here… but some successes too (some better than in the assembled part).

Here are the cleaned parts… some good. many bad:

and some close-ups of failures (note - these are probably bad because they were simply too small, on the build platform directly, or not angled in their supports… again, I was testing the bottom here and not suggesting this is the right way to go… )

These two were manually mounted the same way. Still, one printed fine and the other failed. Not sure why. Will try again in a future print.

Not enough supports? Likely. :smile:

This simple post or pin was directly on the build platform. Seemed pretty straight forward (a simple cylinder), but the sides came out quite badly. Odd contrast to the screw (configured the same on the platform, no supports) not far away that printed perfectly - threads too. So maybe this failure had more to do with proximity to other parts? Not sure.

This one was nearly flawless except where this base touched the build platform. I am just curious how/why the auto-support bases come out so clean, but this one failed. Alas it will need a reprint. :smile:

And lastly this barrel, which printed very well - the grooves are geometry. wow. And yet the super thin transition this next region cracked. The mounting for this piece was manual, but off the build platform (it had supports) and its was oriented so that those grooves were vertical and needed no supports. I thought that worked well, but the flat neck area failed. Maybe just too thin. But from inspecting the thickness in Grabcad (they have a nice measuring tool btw), this failed wall was .5 mm thick. Is that too thin for this printer? Seems it should be able to do that. Note also that this part is NOT closed. So this failure would not seem like a blow out. I will probably reprint this either at an angle, or at a finer resolution (0.025) to see if I can get it to work.

In the end, I’m not convinced I can get all these parts to print successfully at this scale - even though the thinnest walls appear to be about 1/2 mm thick. That may be the practical limit at 0.05 with grey resin, or my crazy approach to mounting may be to blame.

Next up: auto-generated supports for failed parts. I will go the auto-route on a few of the key parts next and see how they turn out. Still at 0.05 resolution setting with grey.

keep your testing coming, nice read. I am too going through the learning curve.
wish resin was sold is smaller bottles to test they all.


I went back to reprint a few of the failed parts. Many had issues and as it turned out (from observation), auto-support generation definitely did out-perform my former minimalist approach - especially compared to doing things right on the build platform. That said, there were still a few failures and I continued with an extra pass to address those. And of course I used a good bit more resin. But considering I had failures, you could argue that the auto-generate approach is pretty competitive. More details below.

In the second pass, I just did auto-layout on the failed parts and tried to print in two groups:

group 1

group 2

I actually got pretty good results this way, with a few exceptions.

The large flat plate called the Tensioner Body printed quite badly with bubbles. Even though I used auto-generation of supports. Granted I had oriented it vertically so that the body itself supported the sections above it. Nothing horizontal. Just vertical. Still I got a large number of bubbles on one side. The print position was close to the hinge (left side, when you are facing the front of the printer). I reprinted again in a different position with the same supports and things worked out a bit better - no surface bubbling, but definitely some anomalies inside the material. Perhaps this is build up front other failed prints? I have had more than a few now, and perhaps there is cruft in the resin that I have not effectively strained out. Anyway, I was happy enough with what I got on this part for now.

Bubbly Tension Body

The other odd failure that still need to fix is one of the Beams. There are two identical beams and one has consistently failed. Not sure why as the geometry looks good. I used the same orientation on both and used auto-generated supports as well. The failure occurred in the same place each time as you will see below.

Failed Beam

The rest came out pretty good.

And the uncleaned GOOD parts:

Barrel - no cracks this time. I have to believe the angled orientation helped.

Way better results on this one… the last one did not really print at all

That’s it for now.

I still have some cleaning to do - I’ll clean everything up and see how the assembly goes. I honestly do not know how well things will fit together or if they were modeled with proper tolerances to support assembly.

cool. I am starting to wonder if the bad surface finish or bubbling is the resin. Seems like you are using grey ? - I am and in my prints I get several bubbles on planar surfaces even at an angle. Do you have any other resin to try the tensioner?

Got some black resin on the way and a spare tray. Will definitely try some comparison prints to see about the bubbles. I may also attempt to strain my left over resin in the tray in case I have bits of failed print floating about. Honestly not sure how/if that stuff build up over time.

I am making it part of the setup to stir and run the “scraper / putty knife” across the tray before every print. I feel it helps the resin loosen up and cleans the bottom of the tray.

I am thinking about clear or black. I wish their resin was bit more affordable.


Closing out this test: I just wanted to add that it turned out the individual parts of this Shimano Derailluer test were NOT modeled to a physical tolerance that would allow proper assembly. Essentially the inside of some objects were the same diameter as the outside of others. So they print assembled nicely (and that is probably why they fuse so well together), but they will not assemble if printed separately.

I’ll have to find or make a separate test in the future to see how well toleranced parts hold up after curing. TBD!

The the cog into inventor or soliwords (or whatever your tool of choice is ) and make a bearing for it.
I have measured some parts they are usually +/- 0.005.