That won’t make upwards facing leds look much different in that case. No amount of sanding is gonna smooth out those intensity differences.
Should still be an improvement and an easy stop-gap at least.
1 Like
Would not be that sure, sanding 3d prints can have weird results. Often it looks quite a lot worse till it looks better. Enabling ironing fir the top layer would have probably been the better move here.
Ironing would smooth it out too much, I’d think. The goal would be to make the surfaces rougher so the light can’t take a straight path through the material. Doing both the ironing and sanding could work so that you don’t have so much sanding to do.
Or you could use a diffuser that’s literally made for that purpose XD
1 Like
How about trying white filament instead? Will still have hot spots, but it won’t be too bright.
It looks really cool! Have you considered using something like sandpaper on the 3D printed logo to diffuse the light?
1 Like
Great work Luke; as others have suggested take some 800 or 400 grit sandpaper and wetsand (just get the paper wet is all you would need to do) the underside of the logo closest to the LEDs. Between 400 and 800 grit should get a nice matte look to diffuse the light. The LEDs are probably too bright overall to be super uniform.
Other thoughts:
-
You could also experiment with piece(s) of wax paper between the logo and the LEDs to get it a little more opaque.
-
If you are really feeling creative, try putting a little bit of cotton in the holes to dim the light and diffuse the light coming from the LEDs a little.
-
If Framework offered this as an optional cover in the Marketplace like the bezels I would buy it for sure.
-
Future iterations of it could dimly show the battery remaining while the laptop is in sleep mode like a clockface (100% all dim and glowing super slow; 75% 1-3 o’clock dark; 50% 1-6 o’clock dark, etc.)
A couple minor updates:
I’m back on campus, and the first week has been hectic, so I haven’t accomplished much. The machine shops are closed until next week, so I don’t have access to sandpaper quite yet (I tried sanding at home, but all we had was 100 grit, and it didn’t work super well). I also re-printed the logo as almost entirely walls and no infill, which changes the appearance pretty significantly, and I actually like it a lot better that way.
I did try cotton, but I couldn’t get it to sit in the holes very well, and it came through rather splotchy. I did end up disassembling that old LCD I had and yanking the diffusion layer, as well as two semi-mirrored sheets of something. Messing with those I can get pretty good diffusion, so I may cut them into fromework-logo shaped pieces and try and put them underneath the 3D printed piece.
I went to the EE lab today to try and find 5V I can yank, and it’s going to be harder than I thought. The fingerprint reader definitely uses 5V and some kind of control to change its light, but the signal is either more complicated than I can parse in the pins (there are 3 of them, a W, R, and G signal, and I couldn’t figure out what they were doing, I don’t have perfectly stable fingers or a 3-channel oscilloscope). It may be doing the dimming internally, because with the laptop asleep, I couldn’t read any change in the signals when the fingerprint reader was flashing. I think I could get into the sensor and take power directly from the LED’s on the board, because it is at 5V, I would be effectively just adding a bunch of them in parallel. The PCB for the fingerprint sensor is plastic-welded in the casing though, and I don’t have an extra one of those at the moment to try and disassemble, which I’m not even sure would work. The lights flickered really bad with 3.3v, so that isn’t really an option unless I redesign it to use different lights or smaller resistors or something, and if I’m resetting to change the design I’d end up changing just about everything.
The backlight for the display uses 5-21V, according to the pinout, but I didn’t want to pry into the screen at risk of damaging something; that isn’t a cheap part.
This is a really cool idea, but definitely a few steps above my paygrade. I did consider using a dedicated expansion card with some kind of USB RGB controller that I could wire up to the lights, but that would take an entire slot, and getting a controller that small seems like it would be difficult.
They also make sk9822 based 5v RGBW, so with white in addition to RGB. It’s a thought. Probably would need to use a ATtiny85 or something, and talk to it over SPI or UART. Which you’d then need a USB hub chip to put on the USB lines to the webcam or something.
Or something with a Bluetooth, so you can pair the 2 together. an ESP8266 may not be quite small enough, but something to think about.
If you’re interested in exploring options, I could offer one.
Since space is limited, rather than some purpose made RGB controller, you could go for just a general purpose microcontroller. The RP2040 is used in the FWL16 keyboard and other input modules. It’s not the smallest possible option, but it’s easy to use, and still might be small enough to stick somewhere without needing to take up an expansion card slot.
Here is a 20mm x 18mm x 3mm high RP2040 board aliexpress.us/item/3256805628494981.html
I don’t know much about RGB controller firmware, but openRGB would probably work. Someone made an QMK openRGB port for the FWL16 keyboard. I tried flashing it to a spare RP2040 board I have and it loaded up ok at least.
For the RGB LEDs themselves, I’d use some WS2812B compatible LEDs. They will only need 3 wires, power 5v positive, negative, and 1 data wire from the microcontroller. The LEDs come as small as 1.4mm x 1.4mm. And jlcpcb has them, so you could get a board assembled. jlcpcb.com/partdetail/OpscoOptoelectronics-SK6805EC14/C2909055. You can also get them on a strip for just testing out how they work, if you’ve never used them before. A 2.7mm wide WS2812B LED strip aliexpress.us/item/3256806542970058.html
2 Likes