> I'm thinking of making some custom LEDs for a project with my interns.
> =A0Think of a custom 7-segment LED, except the segments won't be laid
> out in the standard 7-seg pattern, but will have other
> shapes/sizes/orientations.
>
> I'm thinking the application PCB will have 0603 LEDs mounted on the
> top, and then we'd just cap the PCB with a custom plastic "housing"
> with the segment patterns cut into it.

> I found LED encapsulation compounds, but I'm not
> clear on what specific parameters I should be looking for (for the
> encapsulation compound). =A0How and why should I choose different
> opacities, etc.

I've not built any LEDs from scratch (as opposed to using them in
arrays for lighting)  BUT if you want your display to have maximum
possible  brightness for a given LED then you want minimum possible
opacity and diffusion. You can choose LEDs that have the correct
radiation angle to suit as supplied by the manufacturer with your
encapsulation being a protector. Or even build your own lenses if keen
(see below) but adding diffusion alone is a mistake [tm].

In applications where maximising  light output mattered I have seen
very few cases where any sort of diffusion makes sense. - but I have
seen many many people doing it nonetheless.  Lights which function
perfectly well without diffusers have them added, presumably to
achieve some effect which is by no means apparent - but which reduce
light output and make the distribution pattern less even. (No names
mentioned unless they read this and fix them :-) ).

If I was looking at 0603 LEDs I'd start looking here - chosen from
Digikey catalog after a quick look - I've not used their brand myself

          http://www.kingbrightusa.com/images/catalog/SPEC/APG1608SURKC-T.p=
df

and other APG1608 family LEDs.
These and some other colors in the family  are around 5 cents in 2000
quantity and near the top end of available brightness at 20 mA. There
may be cheaper and higher lumen/$ but those look like a good start.

120 degree radiation angle and rather rounded distribution pattern
(more spherical than lambertian on polar diagram)

If you want very bright displays per Watt you'll need a much smaller
radiation angle and are either going to need other than 0603 SMD
(probably) or to use reflectors or similar (not nice in volume) or to
start learning about making your own lenses. That may be easier than
you'd expect - and you'll get some dispersion along the way.

When looking at specs - lumens are absolute light out, candela are
centre brightness and vary ~~~~ with inverse square of radiation angle
for similar radiation pattern shapes and same lumens. (eg 15 degree
and 30 degree LED with same die inside but different lens. 30 degree
LEd will be ~~~ 25% the canela of 15 degree LED.

Take a say 15 degree radiation pattern 5mm LED. Shine on a wall etc
and note radiation pattern. Brutally hack a few mm off the top so you
get a flat topped package. Shjne again and note substantial widening
of beam. Shorten lens height  and observe.  Chamfer edges. Look at
"top hat" wide angle LEDs and at narrow angle LEDs and compare. Get
some of ther LED encapsulation compound you mentioned - as clear as
possible and add a nice vertical dollop similar to5mm 15 degree LED
shape. Observe.

NOTE: I've widened LED patterns by removing lens material - I've never
tried rebuilding them as above BUT its clear that with the right
material and some care you could get useful control of an otherwise
wide angle pattern. You'll get more losses than a LED manufacturer
does.

Manufacturers who sell narrower angle higher power (500 mW +) SMD
white phosphor LEDs tend to use soft silicone lenses rather than epoxy
or plastics. I'm not certain why but note that epoxy is prone to
degradation with high energy light (blue to UV end of spectrum) which
are present in White LEDs. Silicone compounds are far more resistant
to blue & UV as Si-Si bond is far higher energy than C-C bond in
epoxies. (Which is why eg PV panels in cheap lawn lights turn grey /
opaque after a few seasons).

Plastic of choice for long term light transmission is acrylic but it
has other issues and you are unlikely to be able or wanting to work
with it in this context.



                           Russell

-- =

http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist