Making LED light molds from silicon?

[GHA]

Anyone tried this?

Moving on a bit on yet another project on hold...
Opensource anchor light

So now there are a couple of PCB designs waiting to get made into something useful >

Original plan for the anchorlight circular one was to mount in a piece of plumbing pipe but with 2 rows of LEDs that's proving to be a complete pain so looking for alternatives - cast in epoxy and make a little mold to create the shape - so has anyone actually tried something like this? The flat PCB is to replace nay/stern light so hopefully might be a little easier to make a little mold for.

So any better ideas? Tips to mold with silicon?
Boards were about $25 for 5 of the circular ones and about $22 for 10 of the rectangular ones, regulator built in, just add cree 501 LEDs.

 

[Stemar]

I made an anchor light with a bit of drain pipe, just poked 18 LEDs through holes in the pipe, each group of three in series, with a 120Ω resister, I soldered them all together loose, and potted the whole lot in epoxy, with ally discs top and bottom. It was the brightest light in the anchorage for years until it got clattered in the club mast store.

 

[GHA]

I made an anchor light with a bit of drain pipe, just poked 18 LEDs through holes in the pipe, each group of three in series, with a 120Ω resister, I soldered them all together loose, and potted the whole lot in epoxy, with ally discs top and bottom. It was the brightest light in the anchorage for years until it got clattered in the club mast store.

Done that already, though with a constant current regulator (costing a few pence..).. DIY anchor light
Boards are so cheap to get made it seems well worth a play with some molding.
AnchorLightMaster - EasyEDA

 

[GHA]

If Big Clive does it then hope yet

 

[ltcom]

Never tried but the American video looks like a good result. Do you sell anchor lights and if so how much including postage to UK?

 

[lw395]

I used hot melt glue.
These days I'd consider 3D printing a case, then filling it with hot melt glue.
If you coat the ends of the LEDs, then you're into lens design.
If you don't and use a rigid resin, you are relying on the bond between the LED's plastic case and your resin. Let's just say I've seen that go wrong on a commercial scale.

 

[GHA]

Bish. Lack of experience showing now, this FET, Q2 is a little toasty.
https://datasheet.lcsc.com/szlcsc/1811081213_Alpha-Omega-Semicon-AO3400A_C20917.pdf
Probably a bit small for the power dissipation. To hot to keep touching but been running an hour or so and the magic smoke is still inside.

So how good is epoxy at heat dissipation? Or will this need a heatsink to stay out of the bin?

 

[lw395]

Guessing a few figures.
RN1 drops 1.25V?
Vsupply max 14.5V?
Vf of LEDs 3.3V worst?
So the FET drops 3.35V vaguely worst case.
160mA from the other thread?
536mW.
thermal resistance junction to case is about 100degC per W
(the package can lose more heat via the legs)
max junction temperature 150degC
So you need to keep the case temp down to something like 90degC. doesit burn your finger? sizzle?

Epoxy is a mediocre conductor of heat. You could epoxy a metal heat spreader to the FEt then a reasonable area of epoxy will conduct the heat away.
Or a biggish area of copper on the PCB will take the heat away via the package legs?
Thermal conductivity of air 24.35 mW/(m K)
Epoxy epoxy resins have a poor thermal conductivity (0.15–0.25 W/mK 6 to ten x air.
Which suggests if it's OK in air it will be better in epoxy.

I haven't checked any of this, but them I'm not charging.....

 

[William_H]

For an anchor light I would suggest some sections of 5050 LED strip wrapped around a section of plastic water pipe. I don't think you need a regulator for current limiting. The strip uses tiny series dropping resistors. You will only have 12v supply no charging when at anchor. Stick the circle of leds around the outside of the pipe doing several levels of circle depending on the brightness and power you want. Mount the lot in an inverted jar with a base (lid) of some sort. ol'will