Prints not sticking to the build plate, layer separation, rough surface, elephant foot: resin viscosity – the common denominator

Share via

When you scroll through the various Facebook group about resin printing, you see quite often questions about the following topics:

  • “my prints are not sticking to the build plate”
  • “my layers separate”
  • “my prints have a rough surface”
  • “I have a large elephant foot/squished bottom layers”
An example of all the problems shared in the FB groups. It is really not hard to find them.

In the first two cases, people often advise “increase your bottom layers!” and “increase your bottom exposure”, “lube FEP”, “sand your build plate!”.

But I think such advice is wrong and the best advice for all four cases should be “Introduce a light-off time”. Why? Let me walk you through a series of experiments and observations. It will be a long read, but bear with me – it is an actually simple puzzle just with multiple factors. And as we will see at the end, the same advice also applies to solving the rough surface case and also (partially) the elephant foot. We will also learn, that printing at layers thinner than 50 µm does not make much sense and it can actually degrade the print quality and precision.

Note that I have previously touched on this topic in my blog post Improving surface finish of hollowed SLA 3D prints: one aspect of blooming.

Experiment 1: What is the maximal layer height?

Let’s get two pieces of glass. We will tape 1 mm (1000 µm) thick spacers on the side of one of them. Now we will make a small droplet of our resin of choice, cover it with the second glass. We put this sandwich on the screen of our printer and crank up the tank clean function to 60 seconds (basically we want to expose our sandwich from one side). Once cured, we disassemble the sandwich and wipe out the uncured resin. We are left with a thin disk which thickness we can measure with a micrometer.

For illustration, there are thicknesses of the resins I commonly use. Please, consider these numbers as rather illustrations – I only made two samples. If we would like to obtain proper numbers, we would have to repeat the experiment multiple times.

  • Siraya Tech Fast White: 540 µm
  • Siraya Tech Navy Gray: 320 µm
  • Siraya Tech Sculpt: 200 µm

So – what did we measure? The maximal layer thickness we are able to achieve. The resins formula is designed such that it absorbs UV light within several hundreds of micrometers. Why? To avoid cross-layer curing when printing overhangs and bridges. You can see that the number varies on the resin, but is pretty low. Even if we crank up the exposure much higher, we wouldn’t cure a thicker layer (there is a linear and non-linear region regarding the thickness for a given exposure, but this is too much detail for our purposes at the moment).

That means that we cannot print layers thicker than this value, as the resin does not cure and thus, does not stick to the previous layer. Actually, it will be a little less as we need to stick the new layer to the previous one.

If you are wondering why we covered the top of the resin with glass: the resin does not cure when exposed to oxygen. So we want to eliminate this factor from our measurements. BTW: The curing inhibition caused by oxygen is the reason why everything contaminated by resin becomes sticky – the resin is left in a thin layer, exposed to oxygen, and never cures. Measuring, how thick is the inhibited layer is a topic on its own for another blog post. Also, the cover glass allows us to retain proper drop thickness via capillary forces.

Now we can repeat the experiment, but instead of curing the resin properly (with long exposure), we use standard exposure time. We obtain new values:

  • Siraya Tech Fast White: 160 µm
  • Siraya Tech Navy Gray: 90 µm
  • Siraya Tech Sculpt: 75 µm

With about 50% of the normal exposure, we get similar numbers, but the resin disc is quite fragile:

  • Siraya Tech Fast White: 130 µm
  • Siraya Tech Navy Gray: 80 µm
  • Siraya Tech Sculpt: 70 µm

These numbers tell us the maximum layer thickness for regular exposure we use during printing. If we make a thicker layer, the layers won’t properly stick together and we see a layer separation or prints not sticking to the build plate. Note that this number is also dependent on the temperature – the warmer the resin is, the quicker it cures. So in cold environments, the thickness will be probably lower for lower ambient/resin temperatures.

Note that the resins that are marketed as “high detail” usually block the UV light more to prevent cross-layer curing (as we can see in the example of Sculp here). That also means they are much more likely to experience not sticking or layer separation. We can also conclude that printing on much lower layer thicknesses, e.g., less than 25 µm, does not make much sense as the resin will easily cure-through and cover all the details. Therefore, we lose precision and we only increased the printing time.

This gives us a lead – when the print is not sticking to the build plate, is it possible the resin is not curing all the way through? Also, we can notice that meanwhile, the bottom face of the resin discs is nice and hard, the top is quite soft. So the actual maximal layer thickness can be slightly lower than the values above. They are just the upper estimate on the maximal layer thickness.

Experiment 2: How thick layers can my machine produce?

For the next two experiments, we have to get a little technical. We need a way of measuring the build plate position in the Z-direction. Therefore, I took my Elegoo Saturn and modified it a little bit:

What you see is a replaced build plate arm (machined out of aluminum) and a linear scale with a resolution of 1µm attached to the printer (it is attached via printed components as it does not transmit any significant load). The setup is not suitable for regular printing, as removing the build plate is complicated, but it serves well for our experiment. The linear scale is connected to the microcontroller, so we can read the actual position of the build plate. I also connected the stepper driver signals (step and direction) from the printer’s motherboard to the microcontroller so I can read out the “intended build plate position”. With this setup, let’s get printing will be printing the following model (basically 50×50×3 mm cube that shrinks to half in the middle).

The test model for measurements

With this setup, we can measure how much the build plate lags behind the intended position (maybe better said the position of the motor). I measured that the build plate lags behind the motor quite a lot – up to a few millimeters during peeling. That is expected and you can see by your own eyes how the Z-axis column bends. However, it also lags when sinking the build plate into the resin. The actual position of the build plate lags behind the expected and it slowly “catches up”. When the printer stops moving, the base layer is about 260 µm for Fast and 340 µm for Sculpt. However, if we stop the printer and let it sit, it slowly converges to the desired layer thickness. After 30 seconds of rest, we get 80 µm layer for Fast and about 100 µm for Sculp. The layer never reached the desired thickness of 50 µm for the base layers. This layer thickness is actually higher than what we can cure with regular times! What would happen if it was even higher? The layer would not stick to the build plate as it wouldn’t be cured through.

For completeness, I will just add that once we print over 2 mm of the model, the real build plate position tracks the desired one more precisely and it can settle to the desired position within 10 µm. The lagging also practically disappears.

Experiment 3: Why are my layers thicker than intended?

In the previous experiment, we showed that on Elegoo Saturn, the layer thickness is much higher in the base than we anticipated. Luckily, the setup I showed above has one trick in the sleeve – there are tensometers in the build plate arm, so we can actually measure the forces that push or pull the build plate arm.

We can expect the following forces to be present:

  • some friction of the axis itself, that should be negligible,
  • peeling forces of the cured resin when lifting,
  • some forces when we sink the build plate into resin.

Let’s see what printing looks like:

The printing. Note that there are two tensiometers, so the force applied to the build plate is 2x shown here.

We see that the forces are quite large – both sinking and lifting can yield forces up to 120 N with Siraya Tech Fast (that is 12 kg equivalent), even more with thicker resins (I measured up to 200 N – that is roughly 20 kg equivalent). We can see that the forces get lower and lower with increasing print height. What happening? My interpretation is that when we sink, the build plate has to push away the viscous liquid (resin) and form a really thin layer. This creates a lot of resistance that pushes the build plate up. Similarly, in order to lift the build plate, we have to fill in the void space below the build plate with resin. The resin has to flow in via the narrow opening (the current print build), so it is larger in the bottom layers and negligible in the higher layers as there is a much large space where the resin can flow in.

My test print has a narrowing after 1.5 mm, so I can actually measure the peeling forces caused by the cured resin sticking to FEP – by measuring the difference between the lifting force of the 2500mm2 crosssection and the 625mm2 one. In my setup they are in the order of units of newtons, so they are pretty much negligible for this area. You can see the change in the timepoint 16220. Also, note that there is practically no difference in the peel force between bottom exposure and the normal exposure. This basically shows that the common saying “If you cure too much, the print sticks to the FEP and not to the build plate” is wrong.

The thicker resin, the bigger resistance. I have actually measured that really thin resins like Siraya Tech Simple can have about half the resistance compared to the thick ones, e.g., Siraya Tech Blu or Sculpt.

Compared to our expectation, we can see that:

  • the peeling forces of the cured resin are pretty much negligable, but
  • the sinking and lifting forces are huge when the build plate is close to the bottom of the resin tank.

The obvious question is: what does this do with the printer? I modeled the printer and simulated it in Autodesk Fusion. The conclusion? The construction is really flexible and it can be easily deflected by the forces we measured. See the results of a simulation simulating the printer trying to peel the bottom layer (note that the visualization of the simulation is exaggerated to better see what is happening).

You can actually see the bending with your eyes. Using an indicator, I have measured that in the worst-case conditions the motor has to move the build plate up by 5 mm before it actually starts moving – this is how huge the forces are and how flexible the construction is. This is also the reason why Saturn needs much bigger lifting distances compared to e.g., Mars. How flexible it is you can also see with your own eyes – just take a side look at the Z-axis column from the side when will next print.

The interesting observation is that the Z-axis column is not the weak spot, but the base plate is. Therefore, I think the reinforcements added to the post-preorder batches of Saturn don’t have practically no effect.

Since the construction deflects and acts as a spring, we have an explanation for the thicker layers than anticipated – the resin causes resistance that bends the construction. It gets preloaded and still pushes the resin away in order to reach the desired build plate position.

Note that Elegoo Saturn uses a rubber motor mount to prevent the motor vibration to spread and also to compensate for axial misalignment of the motor and the nut in the build-plate arm. This partially adds to the overall flexibility. I have replaced the rubber motor mount with a rigid one and I will share the results for this replacement with you soon.

Closer look at the sink-expose-peel cycyle

My microcontroller also captures when the UV light is on. When plotting this, we find out something shocking – the build plate is still moving down when the resin is curing! And it stops moving before reaching the final destination. That means that the build plate has reached the maximal cured resin layer thickness and therefore it does not stop:

Build plate moving down means squeezing out the resin. Build plate moving down during UV light on? That means partially cured resin squeezing out. That also means blooming – the rough surface and dimensional inaccuracy. Also, in the worst case, it means the build plate never reaches below the maximal layer thickness and you get a print that did not stick or features layer separation.

Also, since the resistance when sinking is the highest in the base layers, I tend to believe it might be one of (or maybe the major) factor contributing to the elephant foot/squished initial layers on the prints.

What conclusions can we make from the experiments?

We should have pretty much all the observations for mitigating most of the problems during resin printing. Let’s revisit them and explain how they are all caused by resin viscosity:

How can I solve my prints not sticking to the build plate?

When your prints do not stick to the build plate it means the layer was not properly cured through so it can actually stick. There can be several sources of this:

  • Wrongly levelled build plate – if it is tilted, the layer is thicker on one side than on the other. Once it reaches the critical thickness, the resin does not cure-through. However, having build plate properly levelled isn’t a complicated and I guess most people do it right.
  • The build plate is not completely flat. There are people sanding their build plate because they think the build plate is not flat. I measured the two build plates of my Saturn and both of them are flat within 30 µm. That is enough for the first layer to stick (as we have measured that most resins cure properly withing 200 µm, so 50 µm (layer thickness) ± 30 µm (the build plate deviation)). Sanding it without proper equipment and measurement probably will make it more curved than it is. One aspect is that sanding creates rough surface and the resin sticks better to rough surface in my observation. Nevertheless, for most of the printers, this is not an issue.
  • The printer construction is flexible and the forces are huge. Therefore, the build plate does not settle in the position and the printer is printing thicker layer. I think this is the main reason why people’s print don’t stick to the build plate.

The solution is simple – introduce the light-off delay so the printer has enough time to squeeze out the resin and to properly settle in position. Personally, I use 40 seconds of rest time for the first 0.5 mm. This gives enough time for the printer to settle in position. Why 0.5mm? This is a large enough gap to make the suction resistive forces from the build plate negligible. With this setup, I can use 2 base layers with a 10-second exposure on my machine since the layers are indeed 50 µm thick and they don’t need much exposure to properly cure completely. Unfortunately, with the current dumb Chitu firmware and slicer, you cannot automate this process based on a layer crosssection and distance, and I manually change the light-off delay during the print. With this setup, my prints stick perfectly. It also has one advantage: there is no elephant foot!

Also, our measurements explain why Siraya Tech Fast is so popular “and the best resin with zero failures” and Sculp “nice, but extremely hard to print”. Fast cures-thought a lot, therefore it sticks even when the build plate is not positioned properly, but Sculp cures through only in thin layers.

How I can remove the elephant foot?

Back in the day, I wrote a tool for reducing the elephant foot on the prints. Now, this functionality is built-in Chitubox and other slicers. However, I stopped using the compensation completely. The compensation does not work properly as it just shrinks the footprint, but it does not reflect actually how the resin grows. Also, when you print tiny objects, small features get completely lost. Compare the shape of the tooth on a gear:

In my previous blog posts, I claimed that the elephant foot is caused by exposure bleeding during the high light-off times and also by the light reflecting against the shiny build plate. I was wrong.

Based on the experiments above I think there are different mechanics in place. Elephant foot is caused by the partially cured resin being squeezed out. Why it cannot be the light bleeding outside the perimeters of the model profiles? If it was the light bleeding, then the elephant foot could be at most 300 µm large as we cannot cure a thicker portion of the resin. But it is often much more (at least 500 µm in the base layers).

Let me point out, that the above is a simplification; there is definitely some light bleeding out as the LCD does not block the light completely between neighboring pixels and some light can reflect from the build plate. But the effect of these phenomenons is much smaller compared to the amount of resin being squeezed out.

And, indeed. Having the first 0.5 mm printed with 40 seconds of rest time and 10 seconds base exposure on my Saturn practically eliminates the elephant foot on my models. See the comparison (the gear with elephant foot has actually turner on the 0.2 mm compensation(!), the gear without the foot is just printed with no compensation, but with light-off). The results are astonishing:

How can I prevent blooming?

There is another phenomenon I briefly touched on in a previous blog post: blooming. Blooming appears as a rough texture on the models often with a visible line that correlates to with large cross-section. Those are observations from my blog post and the Photonsters experiments. The Photonster assumes that when the resin is cured in large cross-section areas, the curing reaction catalyzes itself (just the presence of cured resin makes it faster or the heat makes it faster) and it just “blooms” outside perimeters of the models. They propose to solve this problem by applying patterns to the large cross-section areas like this:

It seems that it helps, however, after these experiments above, I think the pattern might only slow down the curing and therefore, provide time to settle or provide “escape channels for the resin when squeezing it out. I can replicate the same positive results just by letting the printer sit in place and settle in position. Therefore, it is likely to me that the effects of self-catalyzation are at least not that significant and the biggest problem is actually again, the resin pushing against the build plate. So again, the solution is to have a proper light-off delay before there is a large cross-section being cured.

Also, note that according to my observations, you should include all areas surrounded by perimeters in the estimation of a cross-section. It seems that the air trapped in the forming cavity is pushing also against the resin – see the experiments.

How can I prevent layer separation?

A similar effect to the thicker bottom layers can happen anywhere in-middle of the print. There just have to be a bigger cross-section or you use a thicker resin and your build plate does not settle in position and therefore, you get a ticker layer. In that case, your layers separate as the new layers are not properly cured.

The solution is again to use a light-off delay. At least as the first help.

People’s printing profiles: Another evidence suggesting problems with resin squeezing

If my experiments above don’t persuade you that most of the problems are actually caused by the resistive forces when the printer is squeezing the viscous resin into a tight film, there’s another hint I might be right.

If there are no forces that the ratio between normal exposure and bottom exposures should be about the same across different printers. However, if you carefully examine most of the Google documents with exposure settings on the internet, you find that for larger printers people use higher base layers exposures (relative to the standard exposure). This is due to the fact that the larger build plate yields bigger resistance and therefore, people have thicker base layers on larger printers and, therefore, they have to use longer exposure times.

Also, most people dislike printing directly on the build plate as they often have failures. This is caused by the fact that the model on support is separated from the build plate by a region of thin supports. Therefore, the problematic region near the build plate is “handled” by the supports that are often over-cured just to make them stick.

Conclusion

I hoped the blog post helped you to understand the basics mechanics of printing a single layer on a resin printer and with this knowledge, you will be able to better diagnose problems with resin printing and you stop doing random stuff “just to make your models stick”. Note that I am not saying that people can’t have their build plates improperly leveled, they cannot print in a too cold environment. They can, but I think most of the problems above can be solved in an easier way. Introducing the delay doesn’t cost anything, it is simple to perform and OK, the test print will take longer to complete. But that is a little price to check for one, but highly probable, cause of the print failures, don’t you think?

There is also one case where the light-off time does not help. And this is insufficient peeling lift when the layer does not peel completely. It peels finally after a couple of layers are printed, but then, you have “a hole” in your model and thus, the first layer after that will be actually pretty thick and you will experience layer separation or half of the model sticking to the resin vat.

I also that after reading this you will no longer be scared of printing your models directly on a build plate and you won’t automatically put everything on supports as it usually sacrifices print quality in the form or support attachment point marks.

Also, as bigger and bigger resin printers are coming (Elegoo Jupiter, Phrozen Sonic Mega) I think the problems with layer thickness will be more severe. There is no wonder that Prozen put holes in the build plate to mitigate the resistance during sinking the build plate.

Most of the problems highlighted above could be solved with more rigid machines (e.g., by using 3 Z-axis columns), but I guess we would ultimately reach the stress limit of LCD. The forces that resist the build plate from sinking also apply to the LCD. This is why there is thick glass below the LCD on most printers. This glass provides the support structure so the LCD can withstand the pressure of the resin being squeezed into a thin layer. I also have several ideas that would remove the forces completely, but they need some extra work and are not ready to be published.

However, most of the problems could be solved purely in a software manner. We could automatically compute light-off delays, we could probably solve elephant foot by multiple exposure patterns for a single layer. We could also measure the forces and actual build plate position and make special sinking/peeling cycles with feedback. Unfortunately, most of the resin printers are currently locked in the primitive Chitu ecosystems… But this is a topic for another blog post. If you don’t want to miss any news, be sure to follow me on my social media. Links are below.

Support my work!

If you like my work (these blog posts, my software and CAD models) and you would like to see more posts on various topics coming, consider supporting me in various ways:

If you are interested in knowing what I am up to and recent sneak-peaks, consider following me on social media (Twitter, Instagram, Facebook).

My store offers

Share via
Back To Top