Form 3 accuracy

Support & Troubleshooting
1

I’ve been trying to make a hook that will snap onto a 5mm wire frame. I found that when I specify an inner diameter of the hook in Fusion 360 and then print it on my Form 3 the inner diameter winds up being a few tenths of a millimeter less than I specified. I’ve mainly used black resin. I find myself having to tweak the size repeatedly and locking it in rather than basing it on real-world dimensions plus a tolerance.

Is that normal? Should I expect more accuracy from the Form 3? Should I expect different results from different resins?

Thanks

2

the only process in manufacturing that results in dimensions on spec is Machining.

Casting, printing, forging, stamping, bending and breaking, and rolling all will have some inherent variation from specifications that have to be accounted for in design.

for example- I make a lot of bronze figurative sculpture. from the original Clay ready for mold, to the final bronze figure- i will lose approximately 4% in size. So we have to plan to make the original that much larger.
if the figure has to, for example, sit on a concrete or steel bench- we have to create a model of the bench that is 4% larger to check for fit, prior to molding.

in printing- its even worse, because variation from spec can change depending on the resin you are using, and even with a specific resin, it can change depending on the layer height you are printing at, and the ambient temperature/humidity in the room the printer is in- AND from variations in how longs the part is exposed to IPA- and the power of the UV lights used to cure it and what temperature at which the part is post cured.

your best bet is to measure your outputs under various conditions and keep a graph of percentile variations from spec for each different resin. Over time you can determine an approximate percentage you might need to vary the model in order to get much closer to your target specification.

Or- for something precision, like a very specific internal diameter of a hook, you may prefer to simply design a part you can efficiently and readily drill out to the exact dimension you need.

When I design parts that have to have pivots or other assemblies with screws or other close tolerance parts- I will design a hole knowing the part will come with a smaller hole than I modeled- this will act as a pilot hole for drilling the parts with much better alignment and dimension.

1 Like
3

I thought that FL claimed in a video somewhere that PreForm adjust the sliced model dimensions to allow for shrinkage. Did anybody else see that or am I imagining things?

4

I believe Sculptingman is correct, and he lists the right reasons.

Your expectations are beyond what the technology can deliver.

Different resins will give different results, especially over time. I would expect rigid, given its inclusion of glass fibers, would give you the best results and stability.

But I’d be surprised if it’d deliver what you require. The solution is to design to the machines abilities, as he suggests.

5

It tries- but again- because of all the variables it is impossible for software to accurately compensate for every possible variable.
Its not a linear shrinkage percent.

For example- lets say you have part that starts at 5mm thick ,and gets steadily thicker to 20mm- like a wedge-
with straight sides.
The problem is that as it gets thicker, the shrinkage compounds. So a .01% shrinkage is going to change the dimension of the thin end by .05mm but its going to change the dimension of the thick end by .2 mm.
Scaling the part up is not going to solve this issue because its only going to make the discrepancy larger.

This effect on plastic parts is called Attenuation. Thick areas shrink out of proportion to thin areas.
This is part of why SLA prints and injection molded parts generally warp.

Moreover- some of the effects on SLA printers are due to scattering of the laser and the ‘bleed’ of long chain polymers. So when printing a very small hole, the laser in drawing a tiny circle is more likely to cause the polymerization to spread Inside the arc than outside the arc because scattering light is closer to the drawn line toward the inside of the arc.

It is impossible to perfectly predict all the various ways in which a print is going to deform as it prints… and moreover- just changing the orientation of the model slightly is going to entirely change the effect that shrinkage will have on the print… Preform simply does not have the AI required to make those kind of iterative simulation runs that can deform and modify the model to compensate for every possible variable.

So they do a simple gross scale adjustment, only.

And as a designer, you have to realize the limitations of whatever technology you are using to produce a part and simply build in the tolerances that allow the design to work within the variations in model shape and dimension that any given method of production has.

1 Like
6

Thanks for the helpful replies. Sounds like the machine is functioning as intended. I was able to get a hook design I’m pretty happy with through a few iterations, just wondering if there was some adjustment I could make to get there in one iteration. I’m sure I’ll get the hang of when to add or remove a bit to make it work over time.

7

@Sculptingman - thanks for that detailed description of the issues involved in 3D design scaling and modeling. At least I’m better prepared now for how to compensate my designs. I tend to print either my own mechanical parts or interesting organic and mechanical parts/assemblies from Thingiverse. The latter are usually only available as STL files, which makes revision for a lot more difficult (if it is even necessary), but the former are entirely in my control.

How do most designers deal with the need to model the model differently than how the 3D-printed part will turn out. Often parts need to be modeled so that the final forms fit together properly or meet some tolerance criteria. Do most designers maintain two different sets of modeled parts and assemblies - one as input for the 3D printer and the other as the actual physical model?

I use either Alibre Design or Onshape for CAD work, if that makes a difference.

8

well, if you were doing design for, say, injection molded parts, there are sources with detailed guidance on most of the materials you could be making your part out of-
but in truth- you learn mostly thru practical experience.

This is especially so for a novel technology like SLA- where the variables are so great.

Car parts that are stamped or vacuum formed, - for example- are mostly designed to accommodate a fairly wide latitude because there is a lot of potential variation from one part to another so you often parts with wide overlaps, or with slots instead of round screw holes- or with mounts, like car door hinges, that can be adjusted to optimize fit with surrounding body panels- injection molded parts that at small can have very tight tolerance- but as they get larger, so do the potential variations due to warpage.

Being a designer is essentially an exercise in understanding the limitations of production technologies, and designing something that can be made using that technology.

Another example is Wrenches- these are match plate, rammed sand foundry cast in steel… but there is no way to get a precise tolerance on the business ends of the wrench in a foundry process- which is why you can see the sandy texture where the sand casting was adequate for the handle- but the actual working ends are clearly machined.

if you look up the white papers that formlabs had put out about fitment and tolerances for such things as snap fits and such- they can give you a good starting point of understanding how much slop you need to design in to get parts that a functionally good enough. Rough ideas of how much larger a hole has to be than a pin in order to fit freely- closely, or tightly … that kind of thing.

as to whether designers keep different versions of one design- I dunno- I never do-
As A designer, I consider what I am making, for what purpose and out of what materials- and from there I select the method or methods of manufacture that best suits the objective of the design…
It might be a mix of parts- some of which are off the shelf- others parts may need to be injection molded, and still others stamped in sheet metal…
And once I have a plan for production, I do the modeling work for each part based upon those design decisions.

while I might 3D print a part as a prototype of a part I intend to injection mold… I am not interested in a version designed for 3D printing- I want the prototype to represent the injection molded part, for evaluating how well it is designed for injection molding.

I know a 3D print will not be identical to the final injection molded part- but that does not matter, its still close enough to be useful as a prototype.

If I am designing something for 3D printing as final means of production, then I will try my best to design around the issues of 3D printers- but I am also designing to play to the strengths that 3D printing offers- like I can print shapes I cold Never injection mold.

1 Like
9

Thanks for the insight. I mostly design for the parts to be used as printed, possibly with a bit of post processing, like tapping holes that need to be threaded, and the jobs are almost always short run, say 1 to 10 parts. Final dimensions are sometimes important, especially where the 3D printed parts will be fitted to off the shelf or machined parts, in case tolerances between the various types of fabrication can be expected to differ to some extent. Often, though, the tolerances are not critical, so no worries in those cases. Where it does matter, though, I’ll look into ways to automate adjustment between as-designed and as-printed versions.

10

I thought I’d link to a discussion on another thread comparing prints w/ the Form 2 and 3 (including some I’ve printed). It seems like the Form 2 is getting much cleaner results on that part. I suspect this may be what I’m experiencing w/ my hooks, etc.

11

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.