Thursday, November 12, 2026

Your FDM Prototype Isn't ABS. It's a Different Part.

Here's a question I want an honest answer to: do you prototype in FDM and then mold the part in ABS?

If you do, you're in good company. It's how a lot of products get made. It's also how a lot of products end up with a surprise at first article inspection.

My unpopular opinion: an FDM ABS prototype tells you almost nothing about how your molded ABS part will behave. They share a name. They are not the same part.

[Insert image: fdm-vs-mjf-same-material.png]

Same name, different part

The word "ABS" on the spool and the word "ABS" on the molding quote are doing a lot of heavy lifting. Here's what's actually different:

  • Layers. An FDM part is built from strands of plastic laid side by side and stacked. Where one layer sits on another, the bond is weaker than the plastic itself. Pull across the layers and the part can fail well below what the datasheet says. A molded part is one solid piece with no layer lines to fail along.
  • Gaps and infill. Even at 100% infill, FDM parts have small voids between strands. Most prototypes aren't at 100%. A molded part is fully dense, so stiffness, weight and the way it breaks are all different.
  • The grade. Filament ABS is formulated to print well. Molding ABS is formulated to flow into a mold. Different additives, different flow, sometimes different impact strength and heat resistance.
  • The geometry. To mold the part, it gets draft angles, even wall thickness, ribs instead of thick sections, and gates and ejector pins. The part you tested in FDM often isn't the shape you end up molding.
  • Stress in the part. Molded parts carry stress from cooling in the mold, and they can warp or crack differently near knit lines and gates. Your FDM part had its own stresses, in different places.

So the snap fit that worked in FDM breaks in the molded part. The boss that held up cracks at the knit line. Or the opposite happens: the FDM part failed a drop test, someone added material, and the molded part ends up heavier and more expensive than it needed to be.

Either way, you validated a part you're never going to sell.

Start and finish with the same material

Here's the alternative I keep coming back to: prototype in the same material and process you'll produce in.

With HP Multi Jet Fusion, that's realistic. The PA12 part you test on day one is made the same way as part 500 or part 5,000. Same powder, same process, no mold to design around.

  • Much closer to equal strength in every direction. MJF fuses a whole layer of powder at once, so the layers bond far better than FDM strands. The Z direction isn't the weak spot it is in FDM.
  • Fully fused, not stranded. No infill pattern and no gaps between strands.
  • No redesign for tooling. No draft, no worrying about undercuts, and wall thickness rules that are much easier to live with than molding's.
  • Production without a mold. For many parts, runs of dozens to thousands make sense in MJF, so the prototype design is the production design.

When the prototype and the production part are the same material made the same way, your test results carry straight over. That's the whole point of a prototype.

The other side

To be fair, FDM is a great tool. For a quick fit check, a fixture, a jig, or a part you need on your desk today, it's hard to beat. Some FDM materials, like ULTEM 9085, are used in certified production parts, printed and qualified on that process.

And MJF isn't magic either. If your part is going to be molded in ABS at 100,000 a year, an MJF PA12 prototype is also a different material from your production part. My point is the same in both cases: test the part in the material and process you'll actually sell. If that's molding, get molded samples or a bridge tool before you sign off. If the volume fits printing, prototype in the process you'll produce in and skip the redesign.

The mistake is testing in one material, producing in another, and treating the two as the same because the label matches.

Questions to ask before you sign off on a prototype

  • Is this prototype the same material and grade as production?
  • Is it the same process, or will the part be redesigned for molding?
  • Which tests did it pass, and do those results transfer to how production parts will be made?
  • What volume do we actually need in year one? Could the production process be the same one we're prototyping in?

The short version

  • FDM ABS and molded ABS share a name, not properties.
  • FDM parts are weaker across layers and have voids; molded parts don't.
  • Molding usually means redesigning, which means the tested part isn't the sold part.
  • MJF lets you prototype and produce in the same PA12, no mold, no redesign.
  • Whatever you choose, test in the material and process you'll actually ship.

So, tell me in the comments: do you prototype in FDM and mold in ABS? Has it ever bitten you, or am I overthinking this?

More from me

Picking the right process before you spend money on it is a big part of my book Print It. Price It. Sell It., out now on Kindle and in paperback.

Tim Ruffner has spent almost 20 years in additive manufacturing and is an Account Manager at 3D Parts Unlimited, a service of MSC, which runs HP Multi Jet Fusion production in house. Connect with him on LinkedIn.

Thursday, November 5, 2026

Your $2,000 Bambu Can Print Carbon Fiber Nylon. That Doesn't Make It an Industrial Printer

Desktop FDM printer vs. industrial FDM printer comparing materials, cost, quality control and security

I'll say the quiet part out loud: a Bambu Lab is the best desktop printer I've ever seen, and I still wouldn't put my name on a production part that came off one.

Both of those things are true. That's what makes the "Bambu vs. industrial" fight so loud right now. One side says a $1,999 H2D does 90% of what a six-figure machine does. The other side says desktop printers are toys. They're both wrong, and the engineers caught in the middle are the ones paying for it.

I've been in this industry for almost 20 years. Here's where I think the line actually sits.

Let's give Bambu its due

Bambu changed the game. Fast, reliable, almost no tuning, and the newer machines have a heated chamber and a hot enough nozzle to run PPA-CF and PPS-CF. For under $2,000 you get a printer that would have been science fiction ten years ago.

If you're an engineer and you don't have one on your desk for fit checks and quick iterations, you're slower than you need to be. I mean that. Print the ugly version tonight, find the mistake, and fix it before lunch.

But "it can print the material" and "it can make the part" are two very different sentences.

1. Materials: same name on the spool, not the same thing

A spool that says "carbon fiber nylon" on a desktop printer and Onyx on a Markforged machine are not interchangeable just because the label sounds similar. On industrial FDM, the material and the machine are developed and tested together. The print settings are locked to a material that was characterized on that exact platform.

Then there's the top shelf. Stratasys Ultem 9085 is certified to an Airbus material spec and is used for flight parts on the A350. The certified grade is traceable back to the raw resin lot, with certificates of analysis for each batch. Nobody is pulling that off a hobby-store shelf.

If your part has to pass flame, smoke and toxicity requirements, or your customer asks for a material cert, the conversation is over before it starts.

2. Cost: the printer is the cheap part

Everybody compares sticker price. $2,000 vs. six figures looks like a slam dunk.

Now add up what a part actually costs: the hours someone spends babysitting prints, the failed builds, the parts that "looked fine" but delaminated in the field, the engineer who spent two days tuning a profile instead of doing engineering. A cheap printer that makes a part you can't trust is the most expensive printer in the building.

And size still matters. Our Stratasys Fortus 900mc builds up to 36" x 24" x 36". A big fixture or housing comes out in one piece instead of five glued sections.

The honest math: for one-off prototypes, desktop wins on cost every time. For parts that have to be right, repeat, and ship with paperwork, the cost per good part is what matters.

3. Quality control: can you print the same part next year?

This is the one nobody on Reddit talks about. Printing one great part is easy now. Printing the 100th part the same as the first, and the same as the one you'll make two years from now, is the actual job.

Industrial FDM is built around repeatability: controlled build environments, locked process settings, calibrated machines, and inspection against a known tolerance. Ours is +/- .005" plus .002" per inch after the first inch, and we can tell you why a part meets it.

Ask your desktop printer for a first article inspection report. I'll wait.

4. Security: the part that's getting spicy

Here's where it gets controversial. Bambu printers connect to the company's cloud by default, and some of those servers are in China. In April 2026, the UK Ministry of Defence ordered a security review after Bambu printers turned up in a military exercise. Josef Prusa, who builds competing printers, publicly said "never again" to Bambu in May.

To be fair: Bambu offers a LAN-only mode, launched a Trust Center with a security white paper, and runs a bug bounty program. They've taken it seriously. For a hobbyist printing a phone stand, this is a non-issue.

But if you're printing parts for defense, aerospace or medical customers, or anything covered by an NDA, the question isn't "is Bambu secure?" The question is "can I prove to my customer's auditor where my CAD files went?" If the answer is "probably," you have a problem.

So where's the line?

Use the desktop printer for:

  • Fit checks, form studies and quick design iterations
  • One-off shop aids nobody depends on
  • Learning, testing ideas and failing fast

Go industrial when:

  • The part goes on a product, an aircraft or a production line
  • Someone will ask for a material cert or inspection report
  • You need the same part again in six months
  • The part is bigger than a desktop build plate
  • The CAD file is something your customer would not want leaving the building

The smartest shops I know run both. Bambu on the engineer's desk, industrial FDM for the parts that count. The mistake is asking one to do the other's job.

We run Markforged for strong, fiber-reinforced shop-floor parts and Stratasys for bigger parts and engineering plastics like Ultem. See our FDM capabilities or request a quote.

Now I want to hear it: would you trust a Bambu-printed part in production? Tell me why I'm wrong.


About the Author

Tim Ruffner has spent almost 20 years in additive manufacturing and works at 3D Parts Unlimited. Connect with him on LinkedIn.

Sources

Tuesday, November 3, 2026

Your Next Fixture Doesn’t Need to Wait for the Machine Shop

Your next fixture probably doesn't need to wait for the machine shop. I've watched a lot of lines wait on a fixture. The part is ready, the operators are ready, and everyone is waiting on a holder, a nest or a drill guide. The machine shop has it in the queue behind paying production work, and it'll be a while.

That's where FDM earns its keep. A fixture printed from engineered plastic can often be on the bench within days. If the first version is wrong, you change the model and print another one without paying for new aluminum or waiting in the queue again.

Here's how I'd think about it.

What FDM fixtures do well

1. Assembly and holding fixtures. Nests that hold a part in the same spot every time, so the operator isn't guessing. Shape the pocket to the part and you're done, with no complex machining.

2. Drill guides and locating jigs. The jig puts the hole where it belongs. For a guide that gets used all day, press in a steel bushing where the drill runs and let the plastic do the locating.

3. Check gauges and go/no-go fixtures. If the part drops in, it passes. A quick gauge at the station catches problems before they reach inspection.

4. Soft jaws and non-marring holders. Finished, painted or polished parts need to be held without being scratched. Markforged's Onyx is non-marring, and you can print jaws shaped to the part.

5. End-of-arm tooling and grippers. Lighter tooling on a robot or pick-and-place arm is easier on the arm. FDM makes it practical to build a custom gripper finger for each part.

Picking the material

We run two kinds of FDM: Markforged for strong, shop-floor parts and Stratasys for bigger parts and a wider range of plastics. Here's the full list, but this is how I'd pick:

Onyx is the everyday choice. It's Nylon 6 with chopped carbon fiber, so it's stiff, has a good surface finish and won't mar your parts. Most holding fixtures can start here.

Onyx with continuous carbon fiber is for parts that have to stay stiff under load. Markforged rates it at 6 times stronger and 18 times stiffer than plain Onyx. With the right design, it can take the place of some aluminum fixtures.

Onyx with Kevlar is the pick for parts that get hit or loaded over and over, like grippers and end-of-arm tooling. It's tough rather than stiff.

Onyx with fiberglass adds strength at a lower cost than carbon fiber, for parts that need more than Onyx but not the stiffest option.

ABS-ESD7 is static-dissipative, so it's a good fit for fixtures that hold circuit boards or electronic assemblies.

Ultem 9085 and Ultem 1010 handle heat and harsher environments. Tell us what the fixture will see, and we'll help you pick.

Size matters too. Our Stratasys Fortus 900mc builds up to 36" x 24" x 36", so a big fixture doesn't have to be split into pieces. The Markforged machines are smaller, at 13" x 10.6" x 7.9", but that covers most hand-held fixtures and grippers.

Design it for printing, not for machining

The biggest mistake I see is sending a fixture that was designed to be machined and asking us to print it as is. It'll work, but you leave a lot on the table. A few habits help:

Put metal where it wears. Threaded inserts for bolts that come in and out. Bushings for drills. Dowel pins for precision locating. The plastic holds everything in place, and the metal takes the wear.

Use what printing makes easy. Add pockets for magnets, finger cutouts so operators can grab the part, lightening pockets to save material and print time, and the part number printed right into the fixture so nobody grabs the wrong one.

Think about orientation. FDM parts are strongest along the layers and weakest between them. If you know where the load goes, tell us. Orientation is something we look at on every quote.

Know the tolerance. Our standard is +/- .005" plus .002" per inch after the first inch. If a feature needs to be tighter, call it out. Often the answer is to print it slightly undersize and ream it, or to locate on a pressed-in pin.

When not to print it

I'd rather tell you this now than have you find out on the floor. A printed fixture isn't the right answer when it has to hold a tight tolerance across a long span, take heavy clamping force on thin walls, or wear constantly without any metal inserts. Sometimes the best answer is a hybrid: a printed body on an aluminum base plate. Sometimes it's just aluminum. We'll say so.

Getting one made

Send the model of the fixture if you have one. If you don't, send the model of the part it has to hold and tell us what the fixture needs to do. It also helps to know:

• how many you need, one per station or one per line
• what it's exposed to: heat, coolant, chemicals, or static-sensitive parts
• which features are critical

You can request a quote here, or message me directly.

What's the fixture in your shop that took the longest to get made? I'd like to hear about it.




About the Author

Tim Ruffner has spent almost 20 years in additive manufacturing and works at 3D Parts Unlimited. Connect with him on LinkedIn.

Tuesday, October 27, 2026

Markforged Continuous Fiber Reinforcement in 2026: When FDM Actually Replaces Machined Aluminum

Has Markforged Continuous Fiber Replaced Machined Aluminum in 2026?

Not every part, and not every application. But for one category most engineers don't realize qualifies, the honest math has moved: brackets, end-of-arm tooling, inspection fixtures, and low-volume structural parts that fit on the print bed.

If you ruled out FDM because it was "prototype only" in 2020, the platform has changed. Markforged's continuous fiber reinforcement (CFR) lays unbroken strands of carbon fiber inside an Onyx part, right where the loads run. For brackets, EOAT, and fixtures under about 100 units, it now often beats machined aluminum on weight, lead time, and total cost.

The honest math

Continuous carbon fiber reinforced parts reach a flexural strength of about 540 MPa and flexural stiffness of about 60 GPa. That puts them in the same range as 6061-T6 aluminum, which has a modulus of roughly 69 GPa.

The difference is density. Onyx and carbon fiber parts come in at roughly 1.2–1.4 g/cm³, versus 2.7 g/cm³ for aluminum. That means a printed part typically weighs about half as much as the same aluminum part, and often less once you strip out material that only existed for machining. Markforged reports a strength-to-weight ratio about 50% higher than 6061 aluminum in flexure.

Then there's time. A print-ready file becomes a part in 24–48 hours, with no fixturing, no setup, and no waiting in a loaded CNC queue that can stretch to weeks.

Where continuous fiber wins

  • Weight-sensitive structural brackets in aerospace, drones, and robotics, where every gram costs flight time or payload.
  • End-of-arm tooling (EOAT). Lighter grippers and jaws let a robot move faster, carry more, or run on a smaller motor.
  • Inspection fixtures and CMM holders. Carbon fiber's low thermal expansion helps fixtures stay true in a shop that warms and cools.
  • Low-volume structural parts (1–50 units), where machining setup and fixturing costs never get amortized.
  • Part consolidation. Assemblies of several machined pieces can often be printed as one part.
  • Legacy replacement parts. Obsolete brackets for older equipment can be scanned or modeled and reprinted on demand.
  • Load-path fiber routing. You put fiber exactly where the part needs strength, instead of carving it out of a solid block.

Where aluminum still wins

  • Large parts. Anything bigger than the FX20's 525 × 400 × 400 mm build volume.
  • High service temperatures. Polymer parts have limits; above roughly 150°C, aluminum is the safer choice.
  • High volumes. Production runs of a few hundred or more of a simple geometry, where machining setup cost is spread thin.
  • Tight mating surfaces. Precision faces held to better than ±0.1 mm across the part.
  • Isotropic loading. Printed parts are strongest along the fiber and weakest between layers. If the load direction is unpredictable, or the part sees high-cycle fatigue or heavy bolted joints, aluminum doesn't care which way you push it.

X7 vs FX20: which printer for which job

PrinterBuild volumeBest for
Markforged X7330 × 270 × 200 mmThe production workhorse: brackets, EOAT, fixtures, and most everyday tooling
Markforged FX20525 × 400 × 400 mmLarger parts, aerospace tooling, and ULTEM 9085 with continuous fiber in a chamber heated up to 200°C

Quick decision checklist

Continuous fiber is worth a serious look if most of these are true:

  • The part fits inside the X7 or FX20 build volume.
  • You need fewer than about 100 units.
  • Weight matters, or lead time matters more than unit cost.
  • Service temperature stays under about 150°C.
  • Loads follow a predictable path you can route fiber along.
  • No mating face needs tolerance tighter than ±0.1 mm.

If several are false, machined aluminum is probably still the right call.

The aluminum default is slipping

Continuous fiber hasn't replaced machined aluminum across the board, and it won't. But for brackets, EOAT, and fixtures under 100 units, Markforged usually wins the honest math. If your shop still defaults to aluminum for every one of those parts, it's worth running the numbers again.

Want the full technical breakdown, including the Markforged vs Stratasys vs UltiMaker decision framework? Read the full comparison here.

Sources