Better diffusion

[shfaxx]

Printed another white top to do a side by side comparison.  Again it's not as apparent in the pix, but the flat bottom gives a better uniformity to the panel.

Left = Standard bottom
Right = Flat bottom

[CdRsKuLL]

I did design it with a flat bottom to start with but struggled with the wiring. Are you using the V2 bottom as a comparison?


Thanks,

[shfaxx]

I did design it with a flat bottom to start with but struggled with the wiring. Are you using the V2 bottom as a comparison?


Thanks,

Yes the comparison is only using the v2 Base.  I was thinking a bit about how we could make a flat bottom. Couple options:

1 - Simply drop all the wire routing down under the flat bottom.  This will make the entire tile a little thicker.
1.1 - In order to keep the thickness down, would a different smaller connector be better?  Something like this maybe?
     https://www.amazon.com/HKBAYI-50Pair-50sets-Connector-WS2812B/dp/B00NBSH4CA
2 - Instead of hiding the wiring under the tile, would it be possible to keep it all inside around the edges?

[AgileNebula]

I did design it with a flat bottom to start with but struggled with the wiring. Are you using the V2 bottom as a comparison?


Thanks,

Yes the comparison is only using the v2 Base.  I was thinking a bit about how we could make a flat bottom. Couple options:

1 - Simply drop all the wire routing down under the flat bottom.  This will make the entire tile a little thicker.
1.1 - In order to keep the thickness down, would a different smaller connector be better?  Something like this maybe?
     https://www.amazon.com/HKBAYI-50Pair-50sets-Connector-WS2812B/dp/B00NBSH4CA
2 - Instead of hiding the wiring under the tile, would it be possible to keep it all inside around the edges?

I have been looking at this problem as well.  I think your proposed solution 1 or 2 is the best option.  The connectors in 1.1 are actually larger than the servo connectors because of the latch.  This would also make separating the tiles difficult. 

For option 1, the tiles are a little thinner than the power module, I suspect if you increase the height to be the same you could find enough space underneath to run the wires.

I am going to initially try option 2 as it means not having to print a new bottom tile. 

[shfaxx]

I have been looking at this problem as well.  I think your proposed solution 1 or 2 is the best option.  The connectors in 1.1 are actually larger than the servo connectors because of the latch.  This would also make separating the tiles difficult. 

For option 1, the tiles are a little thinner than the power module, I suspect if you increase the height to be the same you could find enough space underneath to run the wires.

I am going to initially try option 2 as it means not having to print a new bottom tile.

Ahhh ok yeah I don't have ANY of the connectors at this point to gauge how large/small anything is, def good point about the clips too...   

Curious to see how you make out with option 2!

Dunno if these would be any better / smaller:
https://www.amazon.com/Micro-Connector-150mm-Female-daier/dp/B01DUC1PW6/ref=sr_1_3_a_it?ie=UTF8&qid=1546785421&sr=8-3&keywords=micro+3+pin+connector

[shfaxx]

Did a flat bottom panel with 6 LEDs.  While it def helps fill the panel, the hotspots are kinda distracting.  Off to go test the double LEDs in the corners.

[shfaxx]

Playing around trying to get rid of those hotspots.  While this may not be very practical, if you separate the front and bottom plates by about 1/2", it cools things down quite a bit.

EDIT:  Check my build log, I've updated the top panel to be the same thickness as the power box which has helped the diffusion.

[Stevewii]

Have you guys look at this https://www.eevblog.com/forum/projects/nanoleaf-aurora-teardown/

It looks they are using a flat backplane with 6 leds (or 12 if you count the white ones also) instead of 1..
Perhaps thats the solution?

[shfaxx]

Have you guys look at this https://www.eevblog.com/forum/projects/nanoleaf-aurora-teardown/

It looks they are using a flat backplane with 6 leds (or 12 if you count the white ones also) instead of 1..
Perhaps thats the solution?

Yeah I saw it, it appears they are edge lighting a piece of plexi sandwiched in between the panel with 12 LEDs per corner (36 per panel).     I've looked up some ways to do edge lighting.  You have to have the plexi edge basically touching the LEDs.  The edge needs to be polished smooth / clear (You can supposedly torch it clear).
I was thinking of getting a piece of plexi to test with.  Was going to sand the entire surface and cut it so it lays INSIDE our moodlite panel (Laying on top the pyramid).  Then put the standard top over it.    Seems like a huge PITA since I don't have a laser cutter, etc.   Maybe some day I'll give it a go, or if someone here has some of this available.

[AgileNebula]

I have been looking at this, I think the number of leds helps but the real trick, as shfaxx mentioned, is the acrylic panel w/ holes in the middle.  By edge lighting the piece of acrylic each time the light hit one of the holes it scatters.  The causes the entire piece of acrylic to illuminate.  In conjunction with the diffusion film on top this results in the very even lighting. 

I tested this with a thin piece of acrylic I had lying around and the effect does work.  It is just very time consuming to do.
Here are some pictures showing what I mean.  This is with just 1 led in each corner.

This is my acrylic panel installed, note the position of the holes near the center.


Here it is illuminated.  Again, note that those holes are easily visible.  It definitely works but would be a pain to reproduce.


To do this right, you would need a thicker piece of acrylic than the one I used.

[shfaxx]

Instead of holes, would be interested to see if you just sanded the entire piece of acrylic.  Also more LEDs always better.

[AgileNebula]

Top tile - sanded piece of acrylic.
Bottom tiles - normal v2 w/o acrylic.



There seems to be no difference to me between the sanded acrylic and the standard tile.  Maybe there would be a bigger difference with more leds but I am not so sure.  To me, the hole pattern seems to make a more noticeable difference.

[shfaxx]

Wow, yeah don't see much dif from the pix strange.  Def agree that different colors / patterns seems to look better.   The best diffusion I've found so far is just to lift the front plate a few mm.   Ahh well, back to printing.

ps - These things are tough to photograph aren't they...

[NotionalLabs]

The approach Shfaxx and AgileNebula raised is the one I'm pursuing in my experiments too; the acrylic sheet is called a Light Guide Panel and the design of these usually takes a fair bit of engineering. From my research and testing there are a number of design elements you need to make an LGP that provides totally even lighting. but generally speaking I believe you can fabricate one with a laser engraver that should be accessible to most hackerspaces or sign shops.

The trick is to balance and compensate for the attenuation of the light as it bounces around in the acrylic, and maximizing the light emitted into the diffuser layer for brightness. If you sand the whole panel, you're still likely going to get bright spots near the corners as the LGP "diffusion spots" will be spread evenly across it's surface. This also means there's less light to bounce around in the LGP as it's been scattered near the already hot-corners.

I'm experimenting with modeling some mathematically derived patterns that can be easily raster-etched using a laser cutter, but in the meantime if you have some spare acrylic, perhaps you could try sanding it in a Y-shape corresponding to the dark spots in the centre and sides of the panel. Flame-polishing the edges will help reduce insertion loss too.

Jim

[AgileNebula]

I had just enough acrylic left to make one more panel.  I taped off the corners and sanded the middle portion of the tile.  You can definitely notice a difference but it is mainly right at the point where the sanding starts.  Ideally the piece of acrylic would be the same height as the led, the piece I had was only 1.5mm thick vs 5mm for the leds.  I am trying to find a small piece of thicker acrylic to test with.