Urgent - 3B+ spitting out parts with a dimensional inaccuracy ranging in the hundreds of micrometers

We bought a 3B+ a few weeks ago and immediately ran into issues. The high temp v2 resin we bought from a 3rd party supplier, but otherwise seemed fine.
Parts that we printed and post processed could deviate by hundreds of micrometers, and cylindrical surfaces ended up what i can only describe as “warping” to the extent that we couldn’t use the parts at all. See my previous post for details.
I have updated the firmware, ran fit tuning, printed the parts in a million different orientations and still we end up with the same result.
Since then we also have tried printing the same parts in rigid 10k and ran into the same issues.

If we do not sort out this problem within the next week, we will be returning the printer for a new one. If that one does not work properly either, we will return everything. The tolerances from the parts we are are getting are atrocious, and unusable. We previously purchased parts from an online print company and they worked well for us - I am well aware that the machines in these types of print farms can run into the 100k range, but the difference in quality should not be this dramatic, especially from a company that advertises itself as a solution for the semi-professional to professional niche. I hope this company is not just a marketing company masquerading as a 3d printer manufacturer.

Also, I have encountered a software error, being that preform says that we’ve used only 18ml of resin, but in reality it being closer to 200ml.

Sorry you going through this, In my experience designing for and using all types of 3D printing technologies, none of them are 100% accurate. Sure you can re-define your FDM printer to meet your need “Somewhat” but good luck with upkeep for that. Each resin and other materials react differently to shrink factors, solvents, heat, cure, etc. resulting in dimensional changes from your 3D model. I usually design, 3D print prototype, measure, adjust and reprint. You can try to adjust for shrink factor using model scale but be mindful not all parts of your object will be affected equally so prototype inspecting is crucial. its not the printers fault this is how this technology works. at least until we come with the star trek replicators :laughing: here is some read for this. Dimensional accuracy of 3D printed parts | Hubs


Try opening a support ticket instead of posting on the forum. Good luck!


Consider, the differences in quality between a car that costs a few $1000 and one that’s a few $100K will definitely be “dramatic” so it’s probably not unreasonable to expect the same for 3D printers. And the cost of tightening tolerances goes up non-linearly. Getting from 1mm to .1mm accuracy is much easier than getting from 0.1 to .01. If you want really tight tolerances out of your printer, you have to spend the money to get there…

I have been 3D printing for well over a decade, starting with FDM, which has terrible tolerance. I had high hopes for SLA and bought a Form1. And while tolerances tightened up significantly, they never turned out perfect. Besides errors in the printing process, parts can change shape or warp during cleaning and post-cure. I leave parts on the supports/base for curing when I am worried about this (not all shapes are prone to warping). FDM requires a huge effort to tune for decent accuracy. I never bothered. I printed the surfaces I cared about slightly oversized to ensure they’d be over the desired dimension, and then I’d take those surfaces down to the dimensions needed with some CNC machining. Much quicker than machining the entire part from scratch (assuming it’s even machinable with just my 4 axes), cosmetic aspects look great, and the machined surfaces guarantee a flush fit to whatever the part mates to. I do the same thing with SLA parts that need to meet tight tolerance stackups. Though many of the things I print don’t need that tight a tolerance so I just design them for looser tolerances.

As for the resin used gauge, it’s kind of a joke. If you search the forum you’ll find plenty of people complaining about it, including me. Not sure why FL is incapable of improving the accuracy, but most people just ignore it. It an be wildly inaccurate. I keep a spare cartridge of resin on-hand, print until the printer complains about a fill error, then swap in the new cartridge.

1 Like

Hello @praktikant ,

I am sorry to hear that you have been having some dimensional inaccuracy on your new 3B+, but please be rest assured that we will work with you on this matter. In this case, since there will be a lot of information we will need from you (logs, .form file, measurements, software and firmware versions, etc), it is best to open a case with our Support Team who can assist further. Anything of an urgent matter is much better handled by our Support team than through the Forum as we do not monitor the forums as closely as we do our Support tickets.

With that being said, the Forum is great for collaboration with other users and ongoing dialogues; you can certainly find help from others in these threads, but reaching out to us directly will get you more streamlined support. We look forward to hearing from you!

1 Like

First of all, thanks to everyone for replying.
I would like to mention that I have been in contact with support for at least 3 weeks at this point and sent them multiple form files, pictures, and measurements. I have updated firmware, preform, and fit tuning. Unfortunately, the parts we’re getting simply dont live up to the standards we need, unlike the ones we bought.

You say hundreds of microns, but what’s the actual tolerance band you are expecting?

Tolerances in 3d printing, specifically SLA, are very dependant on material type, geometry and orientation. As a global generalization I often expect +/-0.25mm.

Sometimes I see much better than this, sometimes it’s a little worse.

You may also want to see what your parts measure when they are in their green state. Post curing with heat can often introduce further warping and inaccuracies depending on geometry.

Hi, thanks for replying.
Please trust me when I say I have done multiple tests involving measuring parts in their green and post cured states.
+/- 0.25mm is unworkable for us. The Tolerance report from formlabs themselves provides a very attractive looking tolerance band, with a standard deviation of 0.03mm for small features of between 1 and 4 mm. However, the parts we ended up manufacturing ourselves are nowhere near that band of accuracy.

From this forum post I think I’ve learned that industrial SLA /DLP machines and desktop SLA machines are not even remotely comparable in the quality of parts they can provide, and that we should go back to using parts we bought off the internet. Still, upwards of 6000 euros for a starter kit ought to get you a better level of accuracy than a hobbyist machine

A general tolerance band of +/-0.03mm is more or less unattainable IMO. This is better than most average CNC tolerances that don’t have specific callouts for fits (a typical global tolerance for CNC in my world is +/-0.1mm). This is likely pure marketing from Formlabs.

I would be surprised even if 3D Systems SLA machines can claim a global tolerance of +/-0.3mm. If you look online at Protolabs or other similar vendors, they will quote much larger tolerance bands because of the reasons I mentioned in my last post.

Yes, 3D Systems machines will definitely hit higher tolerances because they work in an inherently different way and don’t have any peel forces on the part.

For the engineering work that I do, the Formlabs machines have been great. When higher tolerances are required, we’ll either CNC or send it out.

Also, another thing I have observed is that although absolute tolerances may be off, the machines are fairly consistent. So if these are production parts, it’s pretty easy to scale your part accordingly. Not ideal I know, but something I have done before in the past because some models/materials will print differently than others.

FWIW, for larger prints, tolerances are usually stated in a delta per inch, so the larger the print, the larger the “allowable” amount of error.