UV blockng transparent LCD

[mpp2508]

Like the exposure table in this project (that was discussed on this forum somewhere sometime ago).

This project uses MIPI-HDMI and a Raspberry Pi, which can be removed. I'm currently working on it. Two SSD2828 and two PSRAM64H provide a 10 Hz refresh rate, which might be enough. It's still a prototype, but I've gotten an image on a 2K LS055R1SX04 display from a 3D printer. So, let me know if anyone's interested in this solution; it'll give me extra motivation to get the project done.

[rteodor]

Like the exposure table in this project (that was discussed on this forum somewhere sometime ago).

This project uses MIPI-HDMI and a Raspberry Pi, which can be removed. I'm currently working on it. Two SSD2828 and two PSRAM64H provide a 10 Hz refresh rate, which might be enough. It's still a prototype, but I've gotten an image on a 2K LS055R1SX04 display from a 3D printer. So, let me know if anyone's interested in this solution; it'll give me extra motivation to get the project done.

Two questions:
1. Would it scale to higher resolutions ? I have a 4K display and now I see there are 8K and even 16K displays.
2. I did not made my homework well on this subject and so I can not understand what would be the practical advantage of removing the RPi. Can you detail a bit on this ?

[mpp2508]

Two questions:
1. Would it scale to higher resolutions ? I have a 4K display and now I see there are 8K and even 16K displays.
2. I did not made my homework well on this subject and so I can not understand what would be the practical advantage of removing the RPi. Can you detail a bit on this ?

1. I can't say for sure; I need to know the specific model. If the 4K display has MIPI 8 LANE, like mine, then my setup will give 5 fps with a 25 MHz clock. It's also important to understand that the PSRAM64H operates at a maximum of 84 MHz in fast read mode. 8K and 16K are a completely different matter, and budget solutions are unlikely to be sufficient.
2. Removing the scaler (MIPI-HDMI adapter) and the Raspberry Pi itself. The scaler costs as much as the display, and the Raspberry Pi is about the same. For those with a lot of Raspberry Pis or money, this isn't an issue.

[tooki]

I think my UV light is 365nm.  It works good through the glass that I have.

UV nail curing uses inexpensive 395nm LEDs. That’s also why it passes readily through glass. 365nm is already significantly attenuated.

[coppercone2]

I see severe attenuation with acrylic with the higher frequency UV

[tooki]

I see severe attenuation with acrylic with the higher frequency UV

Which is fine, since my understanding is that UV photoresist is not optimized for shortwave UV (i.e. UV-C), but for UV-A.

[tatel]

So, let me know if anyone's interested in this solution

Yep, really interesting. Bookmarked https://hackaday.io/project/178451-qwicktrace-pcb.

Not so interested in complete solution, already have CNC mill...

[Fire Doger]

UVTools has support for gerber https://github.com/sn4k3/UVtools
Why to bother making a custom driver with Pi? Just buy a chitu mobo https://www.chitusystems.com/products/chitu-e10-lite-motherboard?variant=44530781749484
And a supported LCD.
But in the end it will probably cost you more than a whole printer... which can also be used to print staff...

I used to etch pcbs around 2013 at uni. Mirrored gerber printed on transparent sheet to avoid scattering from the sheet...
Photocopy shop outside of university did a great job on printing them...

I can't imagine myself doing it now, it's like 15€ delivered for 5x 4 Layer boards....

wont the UV slowly destroy the polarising film on the display?

i wonder if rasterizing a uv laser and drawing directly to the pcb would be better?
maybe use the raster mechanism from a laser printer and a moving bed with geared down steppers driving it.

They do, new printers even count the exposure time that the LCD has been threw to know when to replace it.
A famous 3D printer reviewer has a bullet about how easy it is to change screen. It's considered a consumable.

Formlabs tried the uv laser mechanism in forms 3 model. Ditched it in forms 4. It's doable though...

[rteodor]

Two questions:
1. Would it scale to higher resolutions ? I have a 4K display and now I see there are 8K and even 16K displays.
2. I did not made my homework well on this subject and so I can not understand what would be the practical advantage of removing the RPi. Can you detail a bit on this ?

1. I can't say for sure; I need to know the specific model. If the 4K display has MIPI 8 LANE, like mine, then my setup will give 5 fps with a 25 MHz clock. It's also important to understand that the PSRAM64H operates at a maximum of 84 MHz in fast read mode. 8K and 16K are a completely different matter, and budget solutions are unlikely to be sufficient.
2. Removing the scaler (MIPI-HDMI adapter) and the Raspberry Pi itself. The scaler costs as much as the display, and the Raspberry Pi is about the same. For those with a lot of Raspberry Pis or money, this isn't an issue.

Is the FPS relevant here, apart from the initial delay until the image is fully shown ?

By the large number of interfaces and display types I would see this better made with an FPGA. But even that solution comes with its own challenges (connectors, voltages, IP, etc.) and it may blow the cost margin.

[coppercone2]

no you can see it with like 380 or whatever, I think the resist is very close to the light I use within 10nm iirc, on a bright source it looks dim. Glass is OK though

Looks like you would need to have different exposure times if you use something like acrylic instead of glass. the plastic got warm

so if you use photoresist, i recommend buying a glass plate to use as a weight for exposure, because you can barely see a change in brightness through it, compared to the plexiglass

[Berni]

Is the FPS relevant here, apart from the initial delay until the image is fully shown ?

By the large number of interfaces and display types I would see this better made with an FPGA. But even that solution comes with its own challenges (connectors, voltages, IP, etc.) and it may blow the cost margin.

Refreshing LCD displays with a too low refresh rate does cause problems because the pixels start to fade between refresh cycles. A black pixel might slowly fade towards white or a white pixel might slowly fade towards black, the longer the time the farther it might drift. If this happens too much you could expose areas that were supposed to be black.

In general the challenge in driving these SLA printer displays is that they are very high resolution, so they use fancy high speed protocols that not everything might support, or it might not support running them at fast enough clock speeds.

Id say just buy a SLA printer and use it for both. It can be useful sometimes being able to print really tiny features and the printing vat is always separate so it is easy to remove and replace with a DIY PCB alignment fixture.

Only thing to watch out for is that your photoresist can actually be exposed at the wavelength you need. These 3D printers use a fairly long wavelength UV so that it is easy to produce the light using simple LEDs and the LCD can better handle it. Some PCB photoresists are designed to be used with the more spicy shorter UV wavelengths (hence why PCB exposure boxes absolutely need an enclosure to be safe for people around it)

[ftg]

The cheap 2K 3d printer displays do 60Hz all day every day with the compatible HDMI->MIPI-DSI bridge.

[mpp2508]

Is the FPS relevant here, apart from the initial delay until the image is fully shown ?
...

......
Id say just buy a SLA printer and use it for both. It can be useful sometimes being able to print really tiny features and the printing vat is always separate so it is easy to remove and replace with a DIY PCB alignment fixture.
......

  There are tools for every task.
  Everyone has their own unique circumstances and financial capabilities. Not everyone can or will want to buy a printer or even a minimal part of one for PCB exposure.
  A semiscaler based on the SDD2828 and PSRAM64H is being developed as the most affordable option for an exposure device with an LS055R1SX04 display.
  At the very least, it should be cheaper than a Raspberry Pi and an HDMI-to-MIPI converter or a printer motherboard. The SDD2828, PSRAM64H, and STM32F401 are the main components.
  The FPGA is likely less common among the target audience than the STM32. An FPGA-based solution won't be cheaper or simpler. Perhaps we could just shove MIPI into it to eliminate the SDD2828, but I'm far from qualified to do that.
  The SDD2828 is currently on a four-layer PCB. I'll try to make it two-layer, although the difference between them on JLC is not that big, nevertheless.

[Berni]

Well for example the Elegoo Mars 4 costs 189 EUR and comes with a 7 inch 9K LCD (8520x4320) or even a Mars 4 DLP for a similar price.

By the time you buy a LCD, the UV light source, electronics and put it all together you are likely already half way there to that price. So it is pretty hard to compete against mass produced Chinese products when it comes to price. This is mostly the reason why i stopped making PCBs at home. I can go to JLC PCB and get a proper 4 layer PCB made for 7$ while the materials alone to make 5pcs single layer PCB at home would likely cost around that much. This is why most of my boards are now 4 layer for the superior signal integrity. Not to speak of the 2$ double layer boards that are so cheap that there is no way even the Chinese are making any money off those.

Tho having a board that can spit out MIPI video at really high clock speeds is pretty useful for a lot of things where you want to drive most modern high resolution displays. Lattice has a series of FPGAs called Crosslink that can do 4K resolution MIPI at high frame rates using dedicated MIPI transceivers, the chips cost around 10 to 20$ in small quantity.

[mpp2508]

Well for example the Elegoo Mars 4 costs 189 EUR and comes with a 7 inch 9K LCD (8520x4320) or even a Mars 4 DLP for a similar price.

Those who need printing, who have a workstation for working with resins, a place to put a printer, can spend that much money, etc. That's what they'll do. Everyone else will consider other options.

By the time you buy a LCD, the UV light source, electronics and put it all together you are likely already half way there to that price. So it is pretty hard to compete against mass produced Chinese products when it comes to price.

LS055R1SX04          ~$20
Fresnel   140x90mm   $11.19
PCB                         ~$10
misc                        ~$6
ssd2828 x2                 $5.03
10W UV LED               $2.30
ESP-PSRAM64H x2     $2.03
STM32F401CCU6       $1.56
WP7B-S050VA1         $0.69
LP3102QVF              $0.564
74LVC245A  x2         $0.3
EA3036CQBR           $0.15

~$60 And need to figure out something about the case.

... Lattice has a series of FPGAs called Crosslink that can do 4K resolution MIPI at high frame rates using dedicated MIPI transceivers, the chips cost around 10 to 20$ in small quantity.

The cheapest I found was the LCSC LIFCL-40-7SG72C, $31.
It's a really cool thing, but $20 more than the SSD2828+PSRAM64H+STM32 option. You also need to know Verilog or VHDL. I don't.
Not to mention that you also need an FPGA programmer, while the STM32 can at worst be programmed via UART, not to mention DFU.

I should say that the SSD2828+PSRAM64H+STM32 project is the simplest and cheapest way to get an image on a 2K display.

I'd rather read about how to make it even cheaper and simpler.
For example, I'm currently using a 10W LED and a Frenzel lens, which increases the height of the device. How can I get a parallel light source with a small case and a small budget?

[ftg]

JLCPCB assembly can source Lontium LT6911C (C2149630) for an estimated price of $5.4919, but real prices seem to be 3.5 - 5usd, depending on the day and order size.
Assuming one has a PC with HDMI output, then some absolute cheapest mcu with usb, like CH570 can be used to control the UV light's timing.

[rteodor]

I'd rather read about how to make it even cheaper and simpler.

Can't imagine getting much cheaper than what @ftg proposed above: HDMI would already be there, just need the HDMI-MIPI adapter and the screen.

For example, I'm currently using a 10W LED and a Frenzel lens, which increases the height of the device. How can I get a parallel light source with a small case and a small budget?

While I do not know much on the optics and materials bet here are my thoughts:
1. keep the source at a higher distance
2. an array of UV LED strips would be better than a single source
3. 3D print a black block of separators over the array to absorb uncolimated light
3. maybe even 3D print an array of lenses over the LED's out of transparent material. I can not tell how well this last measure could work ... there must be someone who thought or tried it out there... I am really curious over this last one

[mpp2508]

Can't imagine getting much cheaper than what @ftg proposed above: HDMI would already be there, just need the HDMI-MIPI adapter and the screen.

I was talking about standalone solutions.
And in that case, you need to add the cost of the computer.