Almost every prototype conversation we have starts the same way: should this part be printed or machined? It is a fair question, it has a real answer, and the answer is not always the one that makes us money.
We run FDM and resin printers in the shop alongside the mills and the lathe. We do not sell printing as a service - it is a tool we reach for when it beats cutting metal. That means we have no particular reason to push you one way or the other, so here is how we actually decide.
The short answer
| Print it when | Machine it when |
|---|---|
| You are checking fit, clearance or ergonomics | The part has to hold a tolerance tighter than roughly ±0.010" |
| The part carries little or no load | The part carries load, pressure or heat |
| You expect to change the design next week | The design is settled and you need the real thing |
| The geometry is complex, internal, or awkward to fixture | Surface finish matters, cosmetically or functionally |
| You need it tomorrow and close enough is close enough | You need the actual properties of the actual material |
What printing is genuinely good at
A printed part answers questions about shape faster and cheaper than anything else. If what you need to know is whether the bracket clears the harness, whether the handle feels right, or whether two parts go together in the order you assumed, print it. You will have the answer this afternoon for a few dollars of filament.
- Fit and clearance checks before you commit to metal
- Ergonomic and form models you can hand to someone
- Jigs, fixtures, soft jaws and assembly aids for your own bench
- Geometry that would need five setups and a custom fixture to machine
- Any part you expect to revise three times before it is right
Where printed parts fall down
Tolerance
A well-tuned FDM printer holds somewhere around ±0.2 mm, call it ±0.008", and it will not hold that identically across a whole build plate or from one material to the next. Resin does better and still moves as it cures. For comparison, routine machining here is ±0.001", and the Hardinge RS-51MY is a super-precision turning center built to go to ±0.0001" when a drawing calls for it. If your print has a bearing fit, a thread that has to gauge, or a mating surface with a real callout, printing is the wrong process - not a cheaper version of the right one.
Strength has a direction
A printed part is built in layers, and the bond between those layers is the weakest axis. Load it the wrong way and it fails well below what the material's datasheet suggests. A machined part starts as solid stock and has the same properties in every direction, which is why a load-bearing prototype that gets printed often tests worse than the design deserves - and sends people back to redesign a part that was actually fine.
Heat, chemicals and time
Most shop-floor printed plastics soften at temperatures a machined aluminum or steel part does not care about, and many do not like solvents, fuels or sustained UV. If the prototype has to live in an engine bay, a wash-down area or outdoors, print it to check fit and machine the one that has to survive.
Surface finish
Layer lines are visible and they are a sealing surface's enemy. Machined finishes here run to 16 Ra. If the part seals, slides, or gets photographed for a pitch deck, that difference matters.
The cost curve crosses sooner than people expect
At a quantity of one, with simple geometry, printing usually wins on price. But machining cost is mostly setup and programming, and that is a one-time charge spread across the run. By ten or twenty pieces the machined part is frequently cheaper per unit than the printed one - and it is the correct part rather than an approximation of it. If you are quoting a prototype and quietly expecting to need fifty later, say so when you ask. It changes the answer.
The hybrid that actually works
The best outcome is usually both, in order: print the thing that answers your question, then machine the thing that has to work. Print the fit check on Monday, find the two dimensions that are wrong, and machine the real part on Thursday with the correction already in it. That sequence costs less and takes less time than machining twice.
It also applies to tooling. A customer came to us wanting fixtures built. The honest answer was that he did not need a machine shop for them:
Inquired on building some fixtures for our business. Contacted them and they insisted on not building the fixtures, instead to get some 3D printers and make them ourselves. After that they aided in design of the fixtures. Awesome help from this team!
We lost that job on purpose. Fixtures that hold a part for inspection do not need to be milled from billet, and a shop that tells you otherwise is selling you machining rather than solving your problem.
How to decide in thirty seconds
Ask yourself what the part has to do. If the honest answer is "show me whether this shape works", print it. If the honest answer is any of the following, it needs to be machined:
- There is a tolerance on the drawing tighter than ±0.010"
- It carries load, pressure, or clamping force
- It gets hot, or lives in fuel, solvent or weather
- It has to seal, slide, or thread into something real
- It is the part you intend to actually ship
What happens if you ask us
Send the drawing or the CAD file and tell us what the part has to do. We will tell you which process fits and why - including the times the answer is that you do not need us, or that the part you are about to pay for should be three simpler parts instead. We would rather send you a smaller invoice you were happy with than a bigger one you were not.