Formlabs Website Store Support

Uneven "Flashing" on First Layer Using Color Resin with Form 3

Using Form 3 machine - I’m getting about 1 mm of uneven, rippled “flashing” on my first layer using Formlabs Color resin (mixed as “Walnut” brown). It appears to be only the very first layer, as the rest of the models look perfect. I wanted to print these models directly on the build platform, but the models are so small that the 1 mm halo really affects the appearance. Trimming it off would be too labor intensive and result in a scuffed bottom. So I really need to figure out how to remove the flashing.

I checked the LPU and bottom of the tank and both are free of dust. Rigid 4000 resin prints fine with no flashing.

PXL_20210210_141911129_2

I have some better photos that show the flashing on parts after removal from the build platform.

For troubleshooting, I did try both the Beta and the Default profiles for Color resin at 100 microns. Both exhibited the same problem.

Any ideas?

I have been troubleshooting this issue on my own, as Formlabs support advised that it may be a faulty tank, which does not appear to be the case. They then said that models printed directly on the build plate may have this “skirting,” as they called it. Buy why?

My working hypothesis is that the additional laser power applied to the first layer, combined with reflections on the build plate surface, can work together to create an uneven perimeter.

To test my hypothesis, I modeled a circle with a chamfered bottom which was sliced in 0.1 mm increments and arrayed 14 times on the build plate, as illustrated below (the first 3 layers are not chamfered):

I then printed this file on several different substrates, each with the intention of testing the hypothesis that excess laser energy is being reflected around the perimeter of the first layer:

  • Direct to standard build plate
  • On shiny 3M aluminum tape
  • On green polyester tape (used for powder coating)
  • On amber-colored Kapton tape

As expected, the more reflective the surface, the more likely and worse the flashing/skirting. I also want to add that it is ONLY the first layer that is affected.

Printing directly to the build plate produced some skirting, as the surface is roughened. Yet, being aluminum, there were still some reflections.

Printing directly onto shiny aluminum tape produced the worst skirting, jagged, non-uniform, and extending > 1 mm from the first layer model perimeter. In fact, the first layer of the circles were not perfectly circular. Because the surface is shiny, more energy is reflected and distortions occur to the laser dot. Printing on shiny aluminum results in a first layer like this:

Printing on green polyester tape produced a small amount of smooth and uniform skirting. Because it is green in color, some of the UV energy was allowed to reflect off the aluminum on the back side. Unfortunately, standard resins do not adhere well to this tape, so it is not a viable option. Rigid 4000 resin does adhere well enough, at least for small models.

Printing on the amber-colored kapton tape produced < 0.1 mm of skirting–practically zero–as the excess energy reflected off the aluminum build plate is absorbed by the red color. Unfortunately, like the green polyester tape, standard resins adhere poorly to this material, making it unsuitable for use. See in the below photo, none of the models from the second column adhered to the tape.

I will next experiment with altering the aluminum build plate by oxidizing/anodizing the surface in an attempt to darken it and minimize its reflectivity.

Nice work! What I can also recommend is checking the LPU window for dirt/smears, very often this kind of print defect is a result of optics that are not perfectly clean.

In my most recent test, I applied a strip of 3M 3340 aluminum tape to the build platform, used isopropyl alcohol to remove the red ink from the surface of the tape, and then treated the tape with Birchwood Casey Aluminum Black, an acid solution that oxidizes and blackens aluminum.

This tape is normally very shiny and produces an extreme amount of skirting/flashing as the UV laser scatters and bounces off the reflective surface. However, the black oxidized layer prevented this and produced a perfect first layer with no skirting. The only problem is that the oxidized layer is very thin and is poorly bonded to the unaffected aluminum underneath, resulting in models potentially peeling off the oxidized layer and failing to adhere to the build platform. In the image, you can see one of the discs did just that.

Based on the tests so far, I would speculate that a properly anodized aluminum build plate dyed with a dark color would likely resolve the flashing issue with minimal effect on part adhesion.

Disclaimer: As an Amazon Associate, I earn from qualifying purchases.

The below photos show the problem in practice when parts are placed directly on the build platform.

For this, I used a brand new and unused build platform with the color resin I’ve been using.

As you can see, even though the amount of skirting/flashing on the first layer is minimal with a new build platform on account of the matte finish, it is still present and affects small parts a great deal.

Of note, the matte finish on the build platforms is not abrasion resistant and rubs off easily with a paper towel and isopropyl alcohol, eventually wearing down to an increasingly shiny layer of aluminum. So the skirting will only worsen as the build platform is used.

Also notice how the skirting is not a consistent width around the perimeter.

1 Like

thank you very much for this info

Today, I powder coated the exposed aluminum part of the build platform with a black, textured polyester resin. The aluminum tape on the sides is to reinforce the rubber gasket that prevents resin and alcohol from leaking into the hollow space, a frequent problem I have with the standard build platforms.

I then printed a test in which 8 of 14 small models adhered successfully to the build platform. The others probably failed during the peel of the first layer. There was virtually no “flashing” on the bottom layer of the models because the black surface absorbed the UV laser instead of reflecting/scattering it.

The prints that did adhere properly were able to be removed with a reasonable amount of force (enough to beg the question as to why the others failed), and none of the powder coating was stuck to the bottom of the model. So it seems the powder has good adhesion to the metal, but the cured resin has less adhesion to the powder.

I will continue experimenting with different powder types (polyester, polyurethane, epoxy), textures, and densities to determine if powder coating is a viable option for eliminating the laser reflections that occur on the first layer of prints.

So the good news is that a dark finish on the build platform in place of the shiny aluminum does virtually eliminate the effect of uneven, ragged perimeters on the first layer. The problem is that it is difficult to obtain the perfect level of adhesion where parts stick to the build platform but also release easily enough without damage to either the part of the surface of the build platform.

After the failure of all models to adhere to the first test, I sanded the powder on the build platform. This resulted in parts that adhered too strongly to the powder coating.

For removal of small parts, I have resorted to using a 6" bench vise to apply lateral compressive force to the parts to aid in their removal without damage. This is incredibly annoying but does a decent job if done carefully.

For the bottom surface of printed parts, I have resorted to laser ablation to remove about 0.2 mm off the bottom surface and achieve a more uniform, matte finish. This completely eliminates the uneven perimeter of the first layer. Since I do not have dedicated equipment for this, I use my CO2 laser by loading parts onto an anodized aluminum sheet with the bottom side facing up, and then de-focus the laser beam so the energy is distributed over a larger dot–I do not want to engrave deep lines with a more focused beam. Then an engraving program passes over the parts in 0.5 mm strokes, slowly melting and sanding away the surface.

The following photo shows parts that were printed on a brand new build platform with the standard aluminum surface. The parts near the top have been lased while below are the original bottom surface. Note the uneven perimeters around many of the parts and how it is removed with the laser.

The following image shows the results of laser ablation on a small part’s rounded bottom surface in practice:

Compare to the untreated surface:

1 Like

These are some pretty in depth experiments - nicely done. Smart idea to ablate using your laser.

This does seem somewhat overkill though. Are you unable to make these parts in a different manufacturing methodology if you’re making so many of them?

You could probably disassemble the build platform and redesign it so you can use some black anodized 6061. That’s obviously some amount of work but it seems like your already putting a lot of time into this anyway. Anodic coatings usually have a lower coefficient of friction though, and unsure how that will affect your first layer adhesion if at all. It’s also bound to get scratched and dinged up over time.

I agree; it’s all overkill. Had I known I would need all these extra steps, I probably would have stayed with my MSLA machines. On the LCD printers, the overexposed first layers evenly enlarge the entire first few layers, which can be compensated for in the model. The “elephant’s foot” is much less of an eyesore compared to the Form 3’s burn-in implementation.

The chamfer at the base of my parts was an attempt to compensate for the “flashing,” but it is mostly an exercise in futility. The reduced base also helps with part removal since they’re printed directly on the build platform.

I’d love to have these injection molded, but the part’s complexity would require handloading of the tool, which would make the per-part cost prohibitively expensive.

Concerning anodization, the build platform consists of a single cast piece of aluminum and then has the rubber gaskets that secure two plastic side panels. The handle is simply pressed on the metal rails and secured with a single screw, making it easy to remove, as well.

I agree that a dark-dyed anodized surface may work well, but I don’t have the experience to do it myself. I’m pretty sure the build platform can be anodized directly.

For now, I’ll just stick with minimally invasive modifications I can implement on my own. Hopefully I can find the sweet spot and achieve an efficient workflow for these parts.

Any reason not to switch back to a MSLA printer and use the Form 3 for something else? AFAIK MSLA printers are really cheap and probably wouldn’t be too bad of an investment if all they’re printing is this one part. Obviously not ideal and annoying since you bought the F3 for this, but just trying to help figure out your process and see if you can use the F3 for other things that don’t require such a tight control of first layer process.

Have you tried adjusting the default slicing parameters to try to minimize the amount of over-compression that happens on those first layers? I think by default it has a 0.75mm “compression” layer and also merges the first 0.3mm or something like that. Just wondering if reducing these might result in less flash. I regularly change 0.75mm to 0.25mm on my prints that go direct on the build platform. I still usually see flash of some sort though, but I only really print flexible/elastic on the build platform and I know from experience these are more likely to flash than other resins.

Understood on molding, that’s too bad.

Over here in Vancouver, Canada, sending parts out for anodizing is pretty straightforward. The local shops have a $60 CAD minimum charge so it’s not too bad to send it out. Not all aluminum grades can be anodized easily though so if it’s cast, it might be harder to find a shop who can anodize it well. Most shops usually only like doing 6061 and 7000 series.

It sounds like you’re already on a good path to finding a workflow that works. I know it sucks to always have to come up with work-arounds, I can relate to that 100%.

I’m not generating supports at all in PreForm (the model is designed to print without supports). What I’m seeing in practice is that the first 0.3 mm of the model is compressed, even without generating supports. So if the first 0.1 mm was a circle that transitioned immediately to a chamfer on layer 2, it would be compressed and there’s no way to mitigate it in PreForm. Consequently, I model the first 0.3 mm as a flat surface with no transition.

Thinking about it more deeply, I feel this is an engineering problem that Formlabs could solve if they wanted. Instead of applying the burn-in layer to the entire perimeter, PreForm could offset the perimeter of the first layer inside about 0.3 to 0.5 mm for the primary burn-in time (even better, make the offset user configurable), and then apply less laser power/time to the true perimeter, so as to allow good adhesion of the part to the build platform while eliminating the over-exposed perimeters.

The reason I didn’t want to use the MSLA machines for production was because the supply chain for part availability on these foreign machines can be unreliable. If I’m operating on an annual duty cycle of > 50% per machine, I can’t afford months-long delays where it’s typically faster to buy a new machine than wait for a replacement part.

Thanks for the feedback. I may revisit anodization again at a later date.

1 Like

I was referring to the parameters in the advanced settings here - but actually looking at it, not sure if it would help do anything at all. So are you just dropping your model in and hitting print? I’m curious if Preform does anything different in the background if you select “generate supports” for a raft type of “none” (usually what I do). I doubt it’s different, but worth a test.

What I wouldn’t give for a copy of their source code to play around with ideas like this. There’s a lot of room for improvement via experiments and easy quality-of-life tweaks. I wish they had a couple hackers unleashed from the constraints of their bug backlog and roadmap who’s full time job was to take novel ideas from customers (and their own team), and play with them to see if they’re feasible.