Final product orientation & workflow

Hi folks,

I posted about this a month ago, but after an interesting discussion on the Feature Requests & Ideas forum regarding internal supports, I want to raise the question again with the benefit of further data & background detail. This is an absolute brain dump of where I’m at with this project, so be warned it’s a long read!

After a long journey prototyping and testing my Irish pennywhistle design, I’ve come to the point where I’m ready to start selling them, with the eventual goal of making instruments and doing music performance full-time. The final parts are 3D printed in Formlabs clear resin, given a base coat of acrylic paint, and finally sealed with food-safe 2-part resin. I’m trying to optimize my final production workflow, and while I’ve become a very good woodwind maker over the last half decade, I’m still relatively new to 3D printing. I got into 3D printing specifically to make woodwinds, not as a hobby based around the manufacturing method, so there’s been a significant learning curve, and I’m curious for suggestions based on your experiences.

Key Goals:

  • Preserve areas of critical accuracy at the windway face and airblade
  • Orient print for straightest airstream between windway entrance and airblade
  • Optimize support removal to maximize speed of descaffolding, minimize pitting on aesthetic surfaces, and minimize care needed when removing supports

The Whistlehead:

Meet the Irish pennywhistle! Although based on the same mechanism of action as the renaissance & baroque instruments known as recorders, it sounds quite different, and the geometries of the key areas aim to achieve very different outcomes.

Air travels down the windway (1), which is slightly convergent (larger footprint on the entrance than exit), and enters the voicing window (2) via the windway exit (3). The airstream is then split by the airblade (4), which induces alternating vortex shedding. This excites the air column in the body tube, producing a tone. The plug backset (5) plays some cryptic role in improving the “punch” of the sound, but I don’t have a clear understanding why.

There are many important factors, too many to list here, but the critical areas of accuracy with regard to printing them are the windway exit and exit face, the airstream angle relative to the airblade, and the position of the airblade itself. The airblade surface must be smooth and without bumps or pits on both top and bottom surfaces, and the windway exit face should form a crisp 90 degree angle with no beveling of the windway ceiling or floor relative to the vertical axis.

General workflow:

  1. Print at 100um, wash for ten minutes, using a pipette to vigorously blow IPA in and out of the windway, and cure for 15 minutes at 60C
  2. Coarse support removal - remove structural supports and non-critical internal supports with flush cutters and x-acto knife.
  3. Fine support removal - using sections of jeweler’s sawblade and needle files to avoid putting stress on the supports applied to critical areas. This part really sucks if there are default supports on the blade, though I’ve had more success with custom ones. Discussed further down.
  4. Structural sanding - shape airblade, windway face, and windway exit where necessary. Typically there is some flashing on the windway floor and ceiling. “Stepped” layer flashing is preferred because it is easier to see and is a thin layer right at the windway exit. Flashing due to deformation penetrates farther down the windway, is more difficult to inspect visually, and harder to remove.
  5. Fine voicing - play the instrument and apply adjustments to shape the tone. Usually involves finalizing airblade length, sharpness, etc. Small changes to the windway are applied with 1000 grit sandpaper if necessary.

FAQs

Q: What precision is required for the critical surfaces?
A: Five thousandths of an inch for the airblade position vertically, five thousandths of an inch for total distance from airblade edge, and 2or 3 thousandths of an inch for the windway floor & ceiling vertical positions at & slightly inside the windway exit.

Q: Why are you going to so much trouble for a toy?
A: Go look up Burke or Carbony and tell me it’s a toy :rofl:. These are high precision instruments for serious players, and mine are endorsed by some of the best musicians in traditional music. Plus, with the ability to make around 70 or 80 whistleheads with 1 liter of resin, change my design whenever I want for free, and use simple brass body tubes like the vintage instruments mine are inspired by, my material costs are exceptionally low.

Q: What do they sound like?
A: Like this. It’s quite an old model, but the new ones optimize playing characteristics and tuning across the octaves more than changing the tone.

Q: Why don’t you injection mold them?
A: I’m not rich.

Q: Why don’t you machine them in a high-precision shop?
A: See previous answer.

Q: Why don’t you cast them in some other kind of resin or metal?
A: Because they shrink. See answer to question 1.

What I’ve Tried Most Recently:

One piece, steep upward angle (relative to PreForm display), edited supports but no custom ones.

Benefits: good airstream angle accuracy, easy support removal once windway exit face supports are cut, minimal supports applied to aesthetic surfaces

Drawbacks: windway exit face deformation caused by supports, sometimes arching supports get fused to blade, causing a chip upon removal.

Windway exit and exit face are one of the hardest things to assess visually, and difficult to measure precisely as well during final finishing. I tried outsetting the windway face and then filing it back to final position, but it was exceptionally difficult to keep a crisp 90deg angle on the face. I made a jig, but it didn’t produce great results, so I suspect there was further deformation inside the windway. These didn’t play great, frankly, and had a rattle on the lowest note which is indicative of something being off in the windway, even if my corrective measures have made the other notes play well.

One piece, 50ish degree upward angle (the steepest I could get without PreForm basing internal supports for the windway exit on the blade.

Benefits: Dead-on airstream angle, no stairstepping on any face, no supports to remove from the airblade, no supports applied to forward aesthetic surfaces.

Drawbacks: tons of internal supports with thick bases that took ages to carve/file/sand smooth. Supports on windway face still introduced some deformation, and thick flashing took ages to remove from the windway exit face. Sanding inside the windway on a one-piece model is suboptimal as it’s easy to accidentally bevel what must be a crisp 90deg corner at the aperture. I produced two very good Ebs on this print, but they took hours to finish. the Ds were disappointing, and had some layer abnormalities in the middle of the mouthpiece section I wasn’t sure the origin of. They originated at the same layer that wierdness showed up in the next print.

One piece, steep downward angle, custom support applied to blade, then PreForm defaults, no internal supports

Benefits: Easy support removal with no laborious internal support filing. Reinforced airblade support successfully protected it from chipping during support removal, and filing off the buildup wasn’t too difficult. In general the airblade is easier to correct because I can see & measure it much more clearly than the windway.

Drawbacks: Some possible banana-ing of the head, maybe affecting airstream angle. Pronounced stairstepping on windway face, but oddly not in the windway itself. Required filing and laborious internal sanding to get windway face nice and flat. This lengthened the window, and as a result the final blade position was on the edge of being too far away. Still played well, but I could tell the high notes were needing more air, and there was still a bit of a rattle on the lowest note. There was layer weirdness again at the highest section. Under supported?

Two pieces, barely 3ish degrees off straight-up and down with internal supports. The corners of the indexing rails are beveled slightly, which I read helps with fit.

Benefits: really nice aesthetic surfaces; barely any outer sanding needed at all. Easy access to windway surfaces and airblade. Of all the whistles I’ve made, these are probably some of the best sounding.

Drawbacks: laborious internal support removal, but made easier by the now-open section in the middle. The fitup between the parts just sucks, with lots of sanding, checking the fit, sanding, and checking the fit, with the risk of cracking the every time. I did a lot of research on indexing and engineering fits, but never really succeeded in getting a consistent transition fit, likely due to my own lack of experience. The one I did in purple resin is an improvement, but still required a lot of scraping to fit properly. (I wouldn’t recommend the color kit for final production work, by the way, just product protoypes - it’s like making a whistle out of a Necco wafer. Chalky texture and quite weak.)

Anyway, the biggest questions I have are these:

A. Does printing at increased resolution (say 50um or 25um) improve overall precision, or just at the layer level (i.e. for very small surface features)? Would I benefit from tweaking this? I can’t say I noticed any improvement printing on “adaptive” mode, but since PreForm doesn’t actually INDICATE what areas are “adapted,” who knows…

B. Is there a better way to index the part containing the windway floor and plug backset to the main body tube, such that it preserves the angle of the airstream with high precision? I have zero engineering background and have had to learn on my own, so I might be missing something obvious. Long rails seems necessary for this, but are difficult to fit. Most recorders and handmade whistles are made in two parts to get at the key surfaces.

C. Related to B, how much do you backset your surfaces by for a light transition fit? I’d prefer to just give these a wee knock with a dowel to get them back out, but currently they require a bit more persuasive force, which I’m sure my downstairs neighbors appreciate…

I really appreciate all your comments and suggestions, they’ve been really helpful for my own learning process and getting to where I am with this project today. If you’re interested in the broader world of 3D printed Celtic woodwinds, check out the Lindsay System Chanter, and the 3D Whistle and Bagpipe builder’s group on Facebook!

1 Like

You could try reducing the touchpoint size, especially on interior faces.

This is what you should be aiming for:

Sorry, I should have included a scale - that is about the size that they are. The issues with touchpoint size tend to stem from the “growing” surface of the internal supports rather than the surface they are actually supporting.

I love the fusion of engineering precision and aural discrimination you’ve brought to bear on this project.

To be honest, reading your requirements I was hesitant whether Formlabs SLA printers could achieve the level of accuracy you’re after, but you seem to have proven out the concept with some well-performing test prints. You’re taking the right approach in terms of iteration and refinement.

A. Larger resolutions give better overhangs (due to the thicker, stronger layers) and are a little more forgiving of sparser supports. I expect 100um or 50um would be appropriate for your project. You’re correct, the main advantage of thinner slices is to reduce layer line artifacts. You might not have noticed, you can “preview” the adaptive layers by walking through the slicer in Preform (PGUP/PGDN) and paying attention to the thickness shown next to the slider.

B. Snap fit parts are always fun. You need a slight gap to account for varying tolerances between the parts. I usually start around 0.1mm but increase that if the mating surfaces are large. Large flat surfaces fresh from the printer almost never end up perfectly flush, unless you sand them both flat. Wall thickness can also play a factor in warping. While it may not be possible in your application, one trick is to extrude a small lip around the outside of the joint to reduce the amount of contact area. Picture the “fit” between the bottom rim of a paper coffee cup and the table it rests on.

I know you want the transparency of Clear resin. It’s a pretty good material. But if you hit limits with it, you might also want to consider Rigid for this application. You’ll get stiffer prints, and more crisp edges.

When you’re requirements are really exacting, I’ve found custom supports are the way to go (or a hybrid like you’ve done of custom supports on critical surfaces mixed with Preform generated ones for the rest). I’ve made them directly in my modeling software, and also found Formware helpful (even though some of the UI can be a little unintuitive).

I’m really looking forward to seeing how your project turns out.

1 Like

Wow, a TON of great stuff in here - thank you so much for your thoughts!

I was hesitant whether Formlabs SLA printers could achieve the level of accuracy you’re after

It’s definitely pushing the razor edge of what they’re capable of! Funnily enough, though, they’re also the sole reason I can develop these at all! There is a growing community of people making instruments with 3D printing, because the Venn diagram of “people who are interested in / can play historical and traditional instruments” and “people who own / can access a shop with tens of thousands of dollars of machinery” is a pretty narrow slice at best. I can make whistles, or even an entire set of bagpipes, on my 3x5 ft kitchen table.

Before the pandemic I had worked out of a makerspace to try out my design, and that’s how I learned about SLA technology, and Formlabs in particular. A few months into the pandemic I took most of my instruments fund and got the printer. I thankfully seem to have avoided some of the early lemons that went out; aside from occasional “screen jank” that doesn’t affect the operation of the machine it’s been great.

Using 3D printing to make “blanks” and save only the final finishing work, the part that requires skill playing the instrument as opposed to just shop work, is enabling TONS of incredible discoveries to be made about these instruments, which haven’t been played or heard in hundreds of years. I don’t want to get sidetracked talking about historical bagpipes, but there’s university-level research findings being made in people’s laundry rooms.

Anyway.

Larger resolutions give better overhangs (due to the thicker, stronger layers) and are a little more forgiving of sparser supports. I expect 100um or 50um would be appropriate for your project.

I wondered if this wasn’t the case! I think the last of the D standard bore heads I printed was actually on 50um resolution. That might be a happy medium because I believe it improved some of the flat faces; but honestly it’s hard to get my head around how the different dimensions build - how can the vertical surfaces be so smooth even at sharp corners like the turnings for the O-rings if the layer height is 1mm? Is it “tweening” the layer profile or something?

You might not have noticed, you can “preview” the adaptive layers by walking through the slicer in Preform (PGUP/PGDN) and paying attention to the thickness shown next to the slider.

I didn’t notice that! I’ve been using the slider to preview how the piece builds, but a change in the numbers didn’t register! I have dyscalculia; basically dyslexia for numbers, so I guess I shouldn’t be surprised I missed that. It’s not that much of an impediment other than having to re-do acoustic algorithms due to dumb errors in arithmetic, and occasionally writing my phone prefix instead of my apartment number on online orders and then wondering where all my stuff is (sorry, McMaster-Carr). I wish PreForm would color-code it… Might be nice to let the users decide which portions are printed at what layer height, too.

Snap fit parts are always fun. You need a slight gap to account for varying tolerances between the parts. I usually start around 0.1mm

Whew - my wild guess of taking 0.002" each off the plug and the body tube wasn’t far off!

One trick is to extrude a small lip around the outside of the joint to reduce the amount of contact area. Picture the “fit” between the bottom rim of a paper coffee cup and the table it rests on.

That’s a really good idea - thanks! I don’t think it’s strong enough for the rails, but for controlling the final depth of the plug portion it could be useful. Really neat concept just in general - definitely going in my design notebook!

I know you want the transparency of Clear resin. It’s a pretty good material. But if you hit limits with it, you might also want to consider Rigid for this application. You’ll get stiffer prints, and more crisp edges.

Oh, cool! I was looking into some of the more engineering-focused resins for selectively-deformable reed staples (tube that double reed blades go on), but I honestly defaulted to clear because that’s just what the makerspace had when I was learning about it. The point about crisp edges is really helpful if I keep having windway exit face issues.

Again, thanks so much for all your feedback here - it’s really helped with the design considerations for future projects. I’m trying to catch up on all this stuff so I can contribute more to the real titans of 3D instruments, like Donald Lindsay and Zexuan Qiao (makers of my 3D printed Scottish smallpipes, with loads of extra notes normal bagpipes don’t have). I occasionally test high-precision stuff for them but they’re on a different level entirely.

I’ll post some sound samples and video soon - was planning to do it today, but my old Surface finally gave up the ghost and inflated like a balloon. It’s currently sitting in my bathtub between two baking sheets because the city household hazardous waste place isn’t open.

1 Like

The easiest way of making these would be to split the print into two equal parts so that the top face of your print would the same as the section you posted above.

I suspect that the original, injection moulded, pieces were made that way and, thinking about it some more, I think it would be the only way of getting that shape out of the mould.

If you have one of the moulded pieces a close examination of it, and its mould lines, may give you some clues about how it was moulded.

All right, update time - my new indexing system for the plug is working great, and the “coffee cup” suggestion by @rkagerer worked great to bring the plug to final depth. I offset all contact surfaces except the sides of the windway to -0.002, so all told a 0.004 gap between the pieces.

As @JasonC4 suggested in the other thread, I used BC9 Creator to manually add a veritable Cthulhu of supports to the airblade and windway exit face, as that program permits editing of where internal supports originate. Preform was then allowed to perform its normal support analysis. Still printing on 100um for the time being, since so far resolution hasn’t seemed to be a problem.

Printing now!

Progress update! The custom supports worked great, and my D and Eb models are pretty much done. I did end up going for a slightly different, and marginally less buttpluggy, look after finishing my O-ring adjustment tool. Whistles have a tendency to crack at the collar where the tube goes in, so I reinforced that area for when the O-ring grooves came out tight, but now I can adjust it so I don’t think it needs quite so much bulk down there.

Also, I just wanted to “spread the Good News” about the existence of dovetail o-ring grooves :grin:

Interesting, I wasn’t familiar with dovetail o-ring grooves.

For anyone else wondering:

Here’s a good guide that talks about it along with other configurations of O-Ring seals: https://practicalmaintenance.net/wp-content/uploads/Working-Design-Considerations-and-Maintenance-of-O-rings.pdf

I’d love to learn more about how it’s helped in your specific circumstance and why you chose it over other alternatives.

1 Like

It just really capitalizes on the unique kinds of advantages that additive manufacturing offers; those would be a pain to machine, and they’re cool as hell!

Truth be told though, I don’t bother on whistles because sizing those grooves was way more difficult than I want to admit (dyscalculia really, really, really sucks for stuff like that). My friend Donald uses them on bagpipes, including mine, for adjustment areas like tuning slides, and the chanter mainstock, which is disassembled after playing.