Raspberry Pi 4 Experimental Resin 3D Printer -- Episode 414

Hello! I wrote some ideas about a large format MSLA printer in the youtube comments and was asked to post em here. I've thought about three ways, each with their advantages and disadvantages:

1. Easiest and simplest to implement would be to just expose each layer in sections instead of all at once. It can be done with the current rig from the video, you would just need to separate the large bitmap into several  sections, then expose each section and peel it without advancing the Z axis. It may be possible to split the bitmap in various ways like alternating checkerboard patterns to optimize them for easy separation from the FEP film. Main downside from this method is that although it's simple, the resolution is still relatively low, and i doubt we'll be seeing any improvements in the near future for off the shelf LCD monitors.

2. Use one of the standard small 1440p smartphone LCD screens from Sharp which printers like Anycubic's Photon use, but put it on a gantry that will allow it to move on the X or X/Y axes. To cure a layer you would split the bitmap into sections like in the first method, but physically move the LCD and backlight into the correct positions and expose them one by one. Though slow, this would keep the nice ~50x50micron resolution no matter how large the build volume is.

3. An approach similar to Formlabs's "Low Force Stereolithography", but using the LCD screen instead of a laser scanline. The whole unit would need to be placed in a gantry and for each layer it would sweep the layer while creating a small bulge in the FEP film. The idea is that the resin would cure only in the area where the bulge contacts the buildplate and then quickly get peeled off as the contact point moved away. Formlabs has a very nice video on how their Form3L works here:

https://www.youtube.com/watch?v=dLIRAkXa6t8

Not sure how feasible it is to do this with an LCD, it would require some way to synchronise the displayed bitmap with the motion of the LCD gantry such that the image remains "still" with respect to the buildplate. I'm guessing this would require an HDMI-to-MIPI converter board since the bitmap would be animated.

So that's basically what i've come up with. I'd love to hear any feedback on these ideas, and perhaps hopefully we'll get to see something interesting come of them

Hi thanks for posting yxour ideas! welcome to the community.

I go through this one by one to give some feedback of mine, but I hope others join in on these topics too:

I see all these Ideas keep the FEP film-> that was one thing I wanted to get rid off , for the reasons of cost (strips of FEP are not crazy expensive but wide sheets are) and the problem wit hpeeling forces at scale that make it more likely to rip/tear or just not peel that well. this holds true even for the thickest FEP I could find(the wider the build volume_> the bigger the lift-> the higher the force ->the more sever  distortion).

I tried to combine lcd masking with techniques more used in DLP machines.

1. I not sure if i understand that idea correctly please chime in if i get it wrong.
Every section gets built ,peeled(the lift in general is between one and five millimeters depending on resin and machine) the next one exposed and so forth until a layer is completed.
I see a simple mechanical problem here, in which as one section is exposed the gantry cannot go back to it the exact position for the next section without pressure to the Foil.
But on the other hand if this would be a Non-FEP but liquid system no such problem would arise. the pixel density of the whole screen is the same as if it would be exposed at once, speed is slower( divided by amount of sections). I think you are aiming to get the peeling forces lower, but the sheet would still have to be in one piece, which means the same stretching problem over wide areas would occur, but the forces would be lower per peel.

2.I could see this idea viable for very high resolution machines! two gantrys are needed :
Z for liftuing the part and X/Y for the screen underneath the build volume or in a top down configuration(which I think would work better) vice versa.
I'm intrigued to give this one  a shot, not for a big printer like in the video but for a reasonable small but very high resolution one.
the biggest challenge here is the firmware& pefectly aligned mechanics! to get everything in snc I'd tend to use a beaglebone for its PRUs and later spin off a board based on it. another variant is to use a TMC5161 driver and load off a lot of the movement calculations by setting it dynamically to the loading times of the image..but I'm currently not confident that I could pull off a machine like that in a timeframe reasonable for a video.

3.I also like this idea, not sure if formlabs or anyone else has a patent or similar on this technique. it basically works like a photocopier. peeling forces are highly reduced.
what bothers me is there is stil FEP involved and the firmware would be very tricky. to get the screen in sync with the movement i think a beaglebone would be a good option, the PRUs could act as synchronized stepper drivers that get triggered on the image bbeing displayed. (synchrionizing the movement to the image , not the image to the movement.)

Hope this sparks soem more discussion!

Hi thanks for posting yxour ideas! welcome to the community.

I go through this one by one to give some feedback of mine, but I hope others join in on these topics too:

I see all these Ideas keep the FEP film-> that was one thing I wanted to get rid off , for the reasons of cost (strips of FEP are not crazy expensive but wide sheets are) and the problem wit hpeeling forces at scale that make it more likely to rip/tear or just not peel that well. this holds true even for the thickest FEP I could find(the wider the build volume_> the bigger the lift-> the higher the force ->the more sever  distortion).

I tried to combine lcd masking with techniques more used in DLP machines.

1. I not sure if i understand that idea correctly please chime in if i get it wrong.
Every section gets built ,peeled(the lift in general is between one and five millimeters depending on resin and machine) the next one exposed and so forth until a layer is completed.
I see a simple mechanical problem here, in which as one section is exposed the gantry cannot go back to it the exact position for the next section without pressure to the Foil.
But on the other hand if this would be a Non-FEP but liquid system no such problem would arise. the pixel density of the whole screen is the same as if it would be exposed at once, speed is slower( divided by amount of sections). I think you are aiming to get the peeling forces lower, but the sheet would still have to be in one piece, which means the same stretching problem over wide areas would occur, but the forces would be lower per peel.

2.I could see this idea viable for very high resolution machines! two gantrys are needed :
Z for liftuing the part and X/Y for the screen underneath the build volume or in a top down configuration(which I think would work better) vice versa.
I'm intrigued to give this one  a shot, not for a big printer like in the video but for a reasonable small but very high resolution one.
the biggest challenge here is the firmware& pefectly aligned mechanics! to get everything in snc I'd tend to use a beaglebone for its PRUs and later spin off a board based on it. another variant is to use a TMC5161 driver and load off a lot of the movement calculations by setting it dynamically to the loading times of the image..but I'm currently not confident that I could pull off a machine like that in a timeframe reasonable for a video.

3.I also like this idea, not sure if formlabs or anyone else has a patent or similar on this technique. it basically works like a photocopier. peeling forces are highly reduced.
what bothers me is there is stil FEP involved and the firmware would be very tricky. to get the screen in sync with the movement i think a beaglebone would be a good option, the PRUs could act as synchronized stepper drivers that get triggered on the image bbeing displayed. (synchrionizing the movement to the image , not the image to the movement.)

Hope this sparks soem more discussion!