More LED's

[Veeduber]

Little while back, on the subject of LED's, I mentioned that if they weren't
bright enough simply use more of them. That produced a couples of queries from
homebuilders wanting to know how many and how-to.

The how-to is easy, especially if you're a ham radio operator, pick your teeth
with a soldering iron and have been passing notes to your buds in Morse code
since the fourth grade.

Circuit board is an insulator such as phenolic or fiberglas with a layer of
copper glued to one or both sides. Single-sided stuff is inexpensive and
commonly available. To mount an LED by the bulb you simply drill a 5mm hole
(.196" givertake), poke the LED thru the hole and give it a dot of crazy glue.
That leaves the LED with its head sticking out one side and its legs out the
other. One leg is for power, the other for ground. Bend the ground leg over
and solder it to the copper then collect the other legs singly or in sets,
according to how you want to wire it, solder them to a dropping resister and
connecting the other of the resister end to power. Let there be light... and
there usually is.

Want them to flash? Then you gotta pony up another dollar fifty. That'll buy
you a NE555 timer chip, a capacitor, three quarter-watt resisters and a generic
PNP transister. Crazy-glue the timer chip to the circuit board 'dead bug,'
meaning it's legs are sticking up, solder the thing together in the proper
manner and your LED will blink sixty times a minute with whatever duty cycle
you've selected. If we're talking a nominal 12vdc and up to 45 LED's, a 1:10
duty cycle will give you a nice flasher that should last about 100,000 hours...
mebbe a little less.

That's one way.

Trouble is, with the LED just poking through the circuit board it's staring
straight ahead; most of its light is going to be in a cone that's only about 20
degrees wide. Even a hundred LED's won't put out very much light if viewed
from the side. To be seen from the side the LED has to be pointing to the
side... which means it can't be pointing anywhere else at the same time. So
you use a buncha them, pointing ever whichaway.

Do you need FORTY-FIVE of those suckers? At about half a buck each the red
ones aren't too expensive, their production having been subsidized by their use
as automotive tail lights. And ultra-bright white LED's are down to about two
bucks because they are coming into common use for flashlights and such. But
those green jobbies cost the earth... nearly three bucks each. So you're
looking at two hundred and fifty bucks worth of LED's... which is about a big
handful.

On the up-side, it's a one-time cost. Build them right into the airframe, GLUE
on the molded Lexan cover, forget about it.

Okay, but why FORTY-FIVE of those suckers?

Got an orange? Okay, a grapefruit then. Cut it in half. Now you got a
hemi-sphere. That's your tail light's coverage.

Take the other half of the grapefruit and slice IT in half. Now you got two
quada-spheres, or whatever. They are your wing tip nav-light's coverage.

Remember the part about the 20 degree angle? Take one of the wing tip chunks
of grapefruit and divide it into twenty-degree slices. (Don't cut it, use a
marking pin on the rind; you can eat it for breakfast tomorrow. Yeah, I
know... but she already thinks you're crazy anyway.) 180 degrees, you need
about 9 LED's. 90 degrees, about five. (yeah, it overlaps a bit) Nine times
five is 45.

So how do you do that? How do you arrange 45 LED's so their output covers a
quarter of a sphere?

Take a look at an LED. The tip is about the size of a grain of popcorn. Now
imagine your wing-tip nav light is little EAR of popcorn with nine kernels from
top to bottom that tapers smoothly to the end FIVE kernels away. Now all we
have to do is replace the kernels with LED's :-)

Looks something like the eye of a bug.

To make something like that I used the comptuer to make a drawing showing rows
of holes for the legs of the LED's. Glued it to the non-copper side of the
circuit board. Started drilling. (#80 carbide burr; about 12,000 rpm. SOP
for anyone who fiddles with electronics.)

(The pattern is about 3-3/16" long, 1-3/16" wide at its fattest point, which
happens to be 1-3/16" back of the nose, and is a smooth tear-drop shape. The
holes are logically placed so that their LED can be bent to the proper angle.)

Over on the copper side of the board the hole for the grounded leg is left
straight. The leg will be poked through it then bent over and soldered to the
copper cladding. But the hole for the power leg gets deburred with a regular
1/8" diameter drill bit having an angle of 116 degrees. That gives me a
shallow NON-CONDUCTIVE hole through which the hot leg of the LED protrudes.
From that point on the wiring is exactly the same as for the first example,
with a dropping resister (about 4 cents each) attached to each leg, or to pairs
or even sets of three... depending on how you want to wire it.

The LED's get poked into their holes to a depth determined by the ANGLE to
which they have to be bent, which I worked out on the comptuer before I started
drilling. Since the pattern is symmetrical this isn't as tough as it sounds.
Basically, the LED's in the middle stick up about 3/4" and stand up straight.
The ones on the outside only stand up about a quarter of an inch and get bent
over at 90 degrees. Between the middle row and the outer row you've got three
LED's, each standing about an eighth of an inch higher, each bent at a
different angle.

Not a very good description but you should get the idea.

The computer also provides the profile of the Lexan cover, which starts life as
two poster-board templates to which you glue some crunchy urethane foam then
sand it into a smooth surface. Coat that with something that will harden up
nice -- I used sheet-rock mud -- then use it to make a female mold and from
that pull a solid plug of either Portland cement or Plaster of Paris. The
cement shrinks a bit more than the plaster but both are saturated and have to
be cured for about ten days before you can use them to make your cover, which
you do by gluing flannel to the plug, greasing it up and popping it into the
oven with a hunka Lexan balanced on top. When the Lexan gets rubbery, protect
your hand and mold the plastic to the plug. Do that until you have a nice
symmetrical shape. Now do it again for the other wing. If you know about
thermoplastics and canopies and who shot John you know you can use the original
mold to flare the skirt so as to form a flange. Trim that, apply sealant and
attach it with glue, rivets or whatever... you won't be removing it.

Did you want fries with that? Or mebbe a strobe? Then order a batch of item #
06040 from Harbor Freight. You can throw away the case; all you want is the
guts. It has its own plastic shield. Trim a piece of aluminum so the clear
plastic is a nice fit then glue the guts to the aluminum... or whatever... and
mount them ALL OVER your airplane... wing tips, top & bottom of the fuselage,
top of the vertical stabilizer, ass-end of the rudder... wherever you want a
bright little spark of light.

The little strobe costs about eight bucks and runs on two AA cells. Forget
that and pony up another buck per unit for an LM340T-5 voltage regulator and a
couple of caps, an electrolytic 470uF/16v for the input and a disk ceramic .1uF
for the output. Screw the LM340T-5 to the piece of aluminum to which you've
mounted the guts and solder it all up.

18ga. wire is over-kill for either the stobe or the flashing LED's

The strobes aren't very bright but then, neither are they very large nor
expensive... use a buncha them. It won't keep the Big Boys from running your
ass down but it might scare the crows out of your way.

----------------------------------------------------------

Too expensive, right? I think so too. But the cost of LED's continues to drop
and ultrabrights will probably be down to a nickle each by the time the bird is
ready to fly, if ever.

-R.S.Hoover

[Richard Lamb]

all to replace a perfectly good light bulb...

[Veeduber]

>
>all to replace a perfectly good light bulb...

---------------------------------------------------

Yeah.... sorta silly. Which is the same thing they said when that fragile,
expensive Edison lamp was used to replace a perfectly good gas light, which
provided a nice white light, if it was the mantle type. Or even kerosene. Or
earlier candles.

But the tricky bit with nav lights --- or traffic lights --- is the fact you
can't wait until they burn out. The whole idea of having them is for them to
be there ALL THE TIME.

So you make them as reliable as possible then replace them BEFORE they can burn
out.

To make a filament type lamp more reliable you either increase the
cross-section of the filament or reduce its operating temperature. The former
dictates more current, the latter less light output.

The output of aviation nav lights for light planes has never been carved in
stone but the trend has been for brighter lights. Back in the Good Ol' Days,
whenever that was, a three candle-power lamp was considered good enough.
Nowadays the lamps are running 50 candlepower. To get 50-cp of light from a
rough-duty filament, you gotta feed the thing quite a bit of power.

Regular Grimes wing lights use that flat-sided lamp with the built-in
reflector. Nowadays they list for about twenty bucks although most folks sell
them for less. That's a good, reliable lamp. But you still replace them every
hundred hours or whatever, because the whole idea is that you don't want to be
doping around at night without nav lights. Makes you get lost or something.
So you don't wait for them to burn out and THEN replace them, you replace those
perfectly good (but 100 hour old) lamps with spiffy new lamps. And if your
uncle Sam is paying the bill, you never even think about it.

But if you DO ever come to think about it, mebbe a set of nav lights that NEVER
has to be replaced isn't such a bad idea after all, even if the set costs you
$250 bucks. Because the last time I checked, Grimes wing light fixture goes
for about $75. Tail light, too. So you're already up to $225 and those
suckers weigh a bit more than a handful of LED's, use ten dollar light bulbs
and suck three times the power as that handful of LED's.

-------------------------------------------------

But I'm with you. ...all to replace a perfectly good bulb. And you still
run the risk of having some BUF run you down if you're silly enough to try
threading the VFR corridor over LA, day or night, even when you're blinking and
flashing and painted International Orange and have ten thousand dollars of
FAA-mandated equipment screwed to the panel.

Them kerosene lamps is sounding better every day.

-R.S.Hoover

[Tim Ward]

"Veeduber" > wrote in message
...
<snip>
> Them kerosene lamps is sounding better every day.
>
> -R.S.Hoover

Well, I suppose we could run them on jet fuel...

Tim Ward

[Richard Lamb]

Veeduber wrote:
>
> >
> But I'm with you. ...all to replace a perfectly good bulb. And you still
> run the risk of having some BUF run you down if you're silly enough to try
> threading the VFR corridor over LA, day or night, even when you're blinking and
> flashing and painted International Orange and have ten thousand dollars of
> FAA-mandated equipment screwed to the panel.
>
> Them kerosene lamps is sounding better every day.
>
> -R.S.Hoover


I'm not worried about the Buffs.
But the t-38's running on the Charlie Brown route go right over our
traffic pattern...

[Veeduber]

>'m not worried about the Buffs.
>But the t-38's running on the Charlie Brown route go right over our
>traffic pattern...

-------------------------------------------------------

Flying behind a VW engine, a Cherokee Six is a BUF :-)

[Ryan Young]

(Veeduber) wrote in message >...
> Do you need FORTY-FIVE of those suckers? At about half a buck each the red
> ones aren't too expensive, their production having been subsidized by their use
> as automotive tail lights. And ultra-bright white LED's are down to about two
> bucks because they are coming into common use for flashlights and such. But
> those green jobbies cost the earth... nearly three bucks each. So you're
> looking at two hundred and fifty bucks worth of LED's... which is about a big
> handful.

How bright are you thinking they need to be, each?

Newark has some greenies for .39$ each, but they only spec out 150
millicandle/80 mcd
http://www.newark.com/NewarkWebCommerce/newark/en_US/support/catalog/productDetail.jsp?id=46F5023

Perhaps something like this is more what you had in mind:
http://www.superbrightleds.com/leds/g2_specs.htm 8000 mcd $1.24 each
in the sort of quantities you mention.

[Richard Lamb]

Veeduber wrote:
>
> >'m not worried about the Buffs.
> >But the t-38's running on the Charlie Brown route go right over our
> >traffic pattern...
>
> -------------------------------------------------------
>
> Flying behind a VW engine, a Cherokee Six is a BUF :-)

Touche'!

[Jeff Peterson]

just a couple of suggestions...

-if you grind the domed front off an LED the light pattern becomes
broad instead of narrow. that seems easier than trying to arrange
many narrow angle patterns to overlap.

-before you grind off the front, spray paint the entire plastic bit
white. then, after grinding, all the light has to come out the ground
off end.

-the place to get cheap LEDs by the 100s is ebay. maybe 20 cents each
for ultrabrights.

cheers,

Jeff

[Ryan Young]

(Jeff Peterson) wrote in message >...
> just a couple of suggestions...
>
> -if you grind the domed front off an LED the light pattern becomes
> broad instead of narrow. that seems easier than trying to arrange
> many narrow angle patterns to overlap.

You could do this, but it would be wrong. Spreading the beam in this
manner makes it more diffuse. Plus, all the scratches you'll induce
in the lense will make light transmission less efficient.

>
> -before you grind off the front, spray paint the entire plastic bit
> white. then, after grinding, all the light has to come out the ground
> off end.

More like get absorbed by the paint before it ever gets out of the
packaging. Another idea I wouldn't pursue.

>
> -the place to get cheap LEDs by the 100s is ebay. maybe 20 cents each
> for ultrabrights.

Depends on what's meant by ultrabright. Check the MCD numbers. More
is better. 12000 mcd green LEDs run about $1.20-$1.50 ea, but yes,
there are some 10,000 mcd LEDs running about 20 cents, and that's
probably plenty bright enough. Meself, I wouldn't futz with anything
dimmer.

So, one good tip out of three, IMHO. Better than the RAH average.

[Ernest Christley]

Veeduber wrote:

> Too expensive, right? I think so too. But the cost of LED's continues to drop
> and ultrabrights will probably be down to a nickle each by the time the bird is
> ready to fly, if ever.
>
> -R.S.Hoover

Did you count on using less LED's as you went up and down. The
superbright LED's come in packages that will spread the light through a
45 degree angle. Combine that with the need to have less light output
in the up and down directions, and you come out needing only about 10 LEDs.
--
http://www.ernest.isa-geek.org/
"Ignorance is mankinds normal state,
alleviated by information and experience."
Veeduber

[Jeff Peterson]

Hi Ryan,



(Ryan Young) wrote in message >...
> (Jeff Peterson) wrote in message >...
> > just a couple of suggestions...
> >
> > -if you grind the domed front off an LED the light pattern becomes
> > broad instead of narrow. that seems easier than trying to arrange
> > many narrow angle patterns to overlap.
>
> You could do this, but it would be wrong. Spreading the beam in this
> manner makes it more diffuse. Plus, all the scratches you'll induce
> in the lense will make light transmission less efficient.
In order to get the wide angle coverage you need, you can either use a
large number of LEDs, each with a narrow beam or you can use the same
number of LEDs each with a broad light beam. When you grind the front
off the LEDs the forward-directed brightness (officially called the
luminous intensity ) will be decreased
because the light is no longer directed forward but is spread out
instead. The total amount of light emitted (officially called the
luminous flux ) is not affected by grindind off the
front. This means you need just as many ground-off LEDs as you do
intact ones. The advantage of grinding them off is that you dont need
to go to all the trouble of pointing them every which way. As far as
scratches go, yes, they do diffuse the light output, and that's just
what you want.


>[i]
> >
> > -before you grind off the front, spray paint the entire plastic bit
> > white. then, after grinding, all the light has to come out the ground
> > off end.
>
> More like get absorbed by the paint before it ever gets out of the
> packaging. Another idea I wouldn't pursue.
>

white paint is pretty reflective, and not too absorbing. I think white
paint would do more good than harm. if you are really fussy you could
silver the outside. or dont. I think this only makes a 10 percent
difference.


[i]
> >
> > -the place to get cheap LEDs by the 100s is ebay. maybe 20 cents each
> > for ultrabrights.
>
> Depends on what's meant by ultrabright. Check the MCD numbers. More
> is better. 12000 mcd green LEDs run about $1.20-$1.50 ea, but yes,
> there are some 10,000 mcd LEDs running about 20 cents, and that's
> probably plenty bright enough. Meself, I wouldn't futz with anything
> dimmer.
what you really want is a high mcd number AND and a wide beam. that
gives you high luminous flux (aka total light output). generally LED
vendors are charging a premium for the very narrow beam units...the
ones with the highest mcd values. so i would avoid these, and try to
get the most light output for the dollar.

the auction below offers 100 each red LEDs, 5000 mcd, with a veiwing
angle of 15 degrees for $21.00 including shipping

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=26207&item=3816545720

by the way, i wish the ebay vendors would quote luminous flux (as
lumileds does).

instead they quote intensity. worse, they use a variety of
definitions of the beam angle. some give the full angle of the beam,
others give the half angle. some give a number and dont say which.
also they dont say if that angle is the half power point of the beam
pattern. so its a bit hard to get the flux from their numbers.

i noticed this coincidence: all these devices seem to use about the
same current, 20 ma. so very likely they have similar flux. i think
the more expensive ones are the ones that have the LED chip located
very close to the focus of the hemispherical lens....these have the
narrow beam that gives them high mcd numbers. if you are grinding the
dome off, it doesnt make sense to pay extra for this.


>
> So, one good tip out of three, IMHO. Better than the RAH average.
If I counted such things, the number would be three... ;)
-Jeff

[Jeff Peterson]

>
> Did you count on using less LED's as you went up and down. The
> superbright LED's come in packages that will spread the light through a
> 45 degree angle. Combine that with the need to have less light output
> in the up and down directions, and you come out needing only about 10 LEDs.

ok, good idea. but, the more light the better, when you want to be
seen and avoided, so perhpas a you want a mix of narrow angle LEDs
covering the horizon and broad beam LEDs as well.

-Jeff

[Jay]

(Jeff Peterson) wrote in message >...

> In order to get the wide angle coverage you need, you can either use a
> large number of LEDs, each with a narrow beam or you can use the same
> number of LEDs each with a broad light beam. When you grind the front
> off the LEDs the forward-directed brightness (officially called the
> luminous intensity [in milli-candella, aka mcd] ) will be decreased
> because the light is no longer directed forward but is spread out
> instead. The total amount of light emitted (officially called the
> luminous flux [in lumens, aka lm]) is not affected by grindind off the
> front. This means you need just as many ground-off LEDs as you do
> intact ones. The advantage of grinding them off is that you dont need
> to go to all the trouble of pointing them every which way. As far as
> scratches go, yes, they do diffuse the light output, and that's just
> what you want.

You save having to do as much pointing but you have to grind and
POLISH each one. The scratches cause a loss in the total flux. Look
at what you have to do to prepare the ends of fiber for installation
of connectors, and thats even for the longer wavelengths in IR.

> i noticed this coincidence: all these devices seem to use about the
> same current, 20 ma. so very likely they have similar flux. i think
> the more expensive ones are the ones that have the LED chip located
> very close to the focus of the hemispherical lens....these have the
> narrow beam that gives them high mcd numbers. if you are grinding the
> dome off, it doesnt make sense to pay extra for this.

They use the same current because its the same little chunck of Si.
The cost of molding the lens doesn't vary much depending on the angle.
What you're paying all the money for as I understand it is:
1) Pick of the litter, the semi-conductor die vary in efficiency and
are binned. You pay more for the really good one, and less for the
others.
2) New process technologies- All these really bright LEDs are the
result of recent changes in semiconductor processing, so there is less
competition because less guys can make them.

[Jeff Peterson]

(Jay) wrote in message >...
> (Jeff Peterson) wrote in message >...
>
> > In order to get the wide angle coverage you need, you can either use a
> > large number of LEDs, each with a narrow beam or you can use the same
> > number of LEDs each with a broad light beam. When you grind the front
> > off the LEDs the forward-directed brightness (officially called the
> > luminous intensity [in milli-candella, aka mcd] ) will be decreased
> > because the light is no longer directed forward but is spread out
> > instead. The total amount of light emitted (officially called the
> > luminous flux [in lumens, aka lm]) is not affected by grindind off the
> > front. This means you need just as many ground-off LEDs as you do
> > intact ones. The advantage of grinding them off is that you dont need
> > to go to all the trouble of pointing them every which way. As far as
> > scratches go, yes, they do diffuse the light output, and that's just
> > what you want.
>
> You save having to do as much pointing but you have to grind and
> POLISH each one. The scratches cause a loss in the total flux. Look
> at what you have to do to prepare the ends of fiber for installation
> of connectors, and thats even for the longer wavelengths in IR.
you polish optical fiber so one end will press tightly up against the
next and transmit the light with little reflection. effectively you
get a direct glass to glass connection, at 1.3 microns, if the ends
are polished well. in our application the LED plastic case terminates
and the light continues in air. there will be about a 4 percent
reflection here. unless we were so fussy as to anti-reflection-coat
the plastic we are stuck with this 4% reflection.
cant fix this by polishing. a scratched up surface diffuses the
transmitted light to a wide pattern but doesnt reduce the flux.

by the way, painting the body white helps increase output by giving
that light reflected back into the plasitc a second cahnce to get out.


>
> > i noticed this coincidence: all these devices seem to use about the
> > same current, 20 ma. so very likely they have similar flux. i think
> > the more expensive ones are the ones that have the LED chip located
> > very close to the focus of the hemispherical lens....these have the
> > narrow beam that gives them high mcd numbers. if you are grinding the
> > dome off, it doesnt make sense to pay extra for this.
>
> They use the same current because its the same little chunck of Si.
> The cost of molding the lens doesn't vary much depending on the angle.
> What you're paying all the money for as I understand it is:
> 1) Pick of the litter, the semi-conductor die vary in efficiency and
> are binned. You pay more for the really good one, and less for the
> others.
i am no expert on this, but i think the selection is to find the ones
with the narrow (and therefore bright) beams. we want braod beams so
would should
buy the cheaper LEDs


>

[Jay]

(Jeff Peterson) wrote in message
> i am no expert on this, but i think the selection is to find the ones
> with the narrow (and therefore bright) beams. we want braod beams so
> would should
> buy the cheaper LEDs

I've looked into this a bit and used to work at a company that made
extensive use of LEDs for detection and color sensing. The beam width
is dictated by the lens that is formed on the epoxy encapsulant, take
a look at a wide and narrow beam LED side by side and you can see the
difference.

I think it does make a lot of sense however to use a mix of different
lenses. It would be an intersting excercise to ask Excel's Solver how
many of which variety of LED to use provided with the right economic
and light inputs.

p.s. Congrats to CMU on that oragami folding robot.

[Snoopy]

Is this something like what your talking about?

http://www.killacycle.com/Lights.htm



"Veeduber" > wrote in message
...
> Little while back, on the subject of LED's, I mentioned that if they
weren't
> bright enough simply use more of them. That produced a couples of queries
from
> homebuilders wanting to know how many and how-to.
>
> The how-to is easy, especially if you're a ham radio operator, pick your
teeth
> with a soldering iron and have been passing notes to your buds in Morse
code
> since the fourth grade.
>
> Circuit board is an insulator such as phenolic or fiberglas with a layer
of
> copper glued to one or both sides. Single-sided stuff is inexpensive and
> commonly available. To mount an LED by the bulb you simply drill a 5mm
hole
> (.196" givertake), poke the LED thru the hole and give it a dot of crazy
glue.
> That leaves the LED with its head sticking out one side and its legs out
the
> other. One leg is for power, the other for ground. Bend the ground leg
over
> and solder it to the copper then collect the other legs singly or in sets,
> according to how you want to wire it, solder them to a dropping resister
and
> connecting the other of the resister end to power. Let there be light...
and
> there usually is.
>
> Want them to flash? Then you gotta pony up another dollar fifty. That'll
buy
> you a NE555 timer chip, a capacitor, three quarter-watt resisters and a
generic
> PNP transister. Crazy-glue the timer chip to the circuit board 'dead
bug,'
> meaning it's legs are sticking up, solder the thing together in the proper
> manner and your LED will blink sixty times a minute with whatever duty
cycle
> you've selected. If we're talking a nominal 12vdc and up to 45 LED's, a
1:10
> duty cycle will give you a nice flasher that should last about 100,000
hours...
> mebbe a little less.
>
> That's one way.
>
> Trouble is, with the LED just poking through the circuit board it's
staring
> straight ahead; most of its light is going to be in a cone that's only
about 20
> degrees wide. Even a hundred LED's won't put out very much light if
viewed
> from the side. To be seen from the side the LED has to be pointing to the
> side... which means it can't be pointing anywhere else at the same time.
So
> you use a buncha them, pointing ever whichaway.
>
> Do you need FORTY-FIVE of those suckers? At about half a buck each the
red
> ones aren't too expensive, their production having been subsidized by
their use
> as automotive tail lights. And ultra-bright white LED's are down to about
two
> bucks because they are coming into common use for flashlights and such.
But
> those green jobbies cost the earth... nearly three bucks each. So you're
> looking at two hundred and fifty bucks worth of LED's... which is about a
big
> handful.
>
> On the up-side, it's a one-time cost. Build them right into the airframe,
GLUE
> on the molded Lexan cover, forget about it.
>
> Okay, but why FORTY-FIVE of those suckers?
>
> Got an orange? Okay, a grapefruit then. Cut it in half. Now you got a
> hemi-sphere. That's your tail light's coverage.
>
> Take the other half of the grapefruit and slice IT in half. Now you got
two
> quada-spheres, or whatever. They are your wing tip nav-light's coverage.
>
> Remember the part about the 20 degree angle? Take one of the wing tip
chunks
> of grapefruit and divide it into twenty-degree slices. (Don't cut it, use
a
> marking pin on the rind; you can eat it for breakfast tomorrow. Yeah, I
> know... but she already thinks you're crazy anyway.) 180 degrees, you
need
> about 9 LED's. 90 degrees, about five. (yeah, it overlaps a bit) Nine
times
> five is 45.
>
> So how do you do that? How do you arrange 45 LED's so their output covers
a
> quarter of a sphere?
>
> Take a look at an LED. The tip is about the size of a grain of popcorn.
Now
> imagine your wing-tip nav light is little EAR of popcorn with nine kernels
from
> top to bottom that tapers smoothly to the end FIVE kernels away. Now all
we
> have to do is replace the kernels with LED's :-)
>
> Looks something like the eye of a bug.
>
> To make something like that I used the comptuer to make a drawing showing
rows
> of holes for the legs of the LED's. Glued it to the non-copper side of
the
> circuit board. Started drilling. (#80 carbide burr; about 12,000 rpm.
SOP
> for anyone who fiddles with electronics.)
>
> (The pattern is about 3-3/16" long, 1-3/16" wide at its fattest point,
which
> happens to be 1-3/16" back of the nose, and is a smooth tear-drop shape.
The
> holes are logically placed so that their LED can be bent to the proper
angle.)
>
> Over on the copper side of the board the hole for the grounded leg is left
> straight. The leg will be poked through it then bent over and soldered to
the
> copper cladding. But the hole for the power leg gets deburred with a
regular
> 1/8" diameter drill bit having an angle of 116 degrees. That gives me a
> shallow NON-CONDUCTIVE hole through which the hot leg of the LED
protrudes.
> From that point on the wiring is exactly the same as for the first
example,
> with a dropping resister (about 4 cents each) attached to each leg, or to
pairs
> or even sets of three... depending on how you want to wire it.
>
> The LED's get poked into their holes to a depth determined by the ANGLE to
> which they have to be bent, which I worked out on the comptuer before I
started
> drilling. Since the pattern is symmetrical this isn't as tough as it
sounds.
> Basically, the LED's in the middle stick up about 3/4" and stand up
straight.
> The ones on the outside only stand up about a quarter of an inch and get
bent
> over at 90 degrees. Between the middle row and the outer row you've got
three
> LED's, each standing about an eighth of an inch higher, each bent at a
> different angle.
>
> Not a very good description but you should get the idea.
>
> The computer also provides the profile of the Lexan cover, which starts
life as
> two poster-board templates to which you glue some crunchy urethane foam
then
> sand it into a smooth surface. Coat that with something that will harden
up
> nice -- I used sheet-rock mud -- then use it to make a female mold and
from
> that pull a solid plug of either Portland cement or Plaster of Paris. The
> cement shrinks a bit more than the plaster but both are saturated and have
to
> be cured for about ten days before you can use them to make your cover,
which
> you do by gluing flannel to the plug, greasing it up and popping it into
the
> oven with a hunka Lexan balanced on top. When the Lexan gets rubbery,
protect
> your hand and mold the plastic to the plug. Do that until you have a nice
> symmetrical shape. Now do it again for the other wing. If you know about
> thermoplastics and canopies and who shot John you know you can use the
original
> mold to flare the skirt so as to form a flange. Trim that, apply sealant
and
> attach it with glue, rivets or whatever... you won't be removing it.
>
> Did you want fries with that? Or mebbe a strobe? Then order a batch of
item #
> 06040 from Harbor Freight. You can throw away the case; all you want is
the
> guts. It has its own plastic shield. Trim a piece of aluminum so the
clear
> plastic is a nice fit then glue the guts to the aluminum... or whatever...
and
> mount them ALL OVER your airplane... wing tips, top & bottom of the
fuselage,
> top of the vertical stabilizer, ass-end of the rudder... wherever you want
a
> bright little spark of light.
>
> The little strobe costs about eight bucks and runs on two AA cells.
Forget
> that and pony up another buck per unit for an LM340T-5 voltage regulator
and a
> couple of caps, an electrolytic 470uF/16v for the input and a disk ceramic
..1uF
> for the output. Screw the LM340T-5 to the piece of aluminum to which
you've
> mounted the guts and solder it all up.
>
> 18ga. wire is over-kill for either the stobe or the flashing LED's
>
> The strobes aren't very bright but then, neither are they very large nor
> expensive... use a buncha them. It won't keep the Big Boys from running
your
> ass down but it might scare the crows out of your way.
>
> ----------------------------------------------------------
>
> Too expensive, right? I think so too. But the cost of LED's continues to
drop
> and ultrabrights will probably be down to a nickle each by the time the
bird is
> ready to fly, if ever.
>
> -R.S.Hoover

[Blueskies]

Notice, no white rear light there...

--
Dan D.
http://www.ameritech.net/users/ddevillers/start.html


..
"Snoopy" <Snoopy is > wrote in message ...
> Is this something like what your talking about?
>
> http://www.killacycle.com/Lights.htm
>
>
>
> "Veeduber" > wrote in message
> ...
> > Little while back, on the subject of LED's, I mentioned that if they
> weren't
> > bright enough simply use more of them. That produced a couples of queries
> from
> > homebuilders wanting to know how many and how-to.
> >
> > The how-to is easy, especially if you're a ham radio operator, pick your
> teeth
> > with a soldering iron and have been passing notes to your buds in Morse
> code
> > since the fourth grade.
> >
> > Circuit board is an insulator such as phenolic or fiberglas with a layer
> of
> > copper glued to one or both sides. Single-sided stuff is inexpensive and
> > commonly available. To mount an LED by the bulb you simply drill a 5mm
> hole
> > (.196" givertake), poke the LED thru the hole and give it a dot of crazy
> glue.
> > That leaves the LED with its head sticking out one side and its legs out
> the
> > other. One leg is for power, the other for ground. Bend the ground leg
> over
> > and solder it to the copper then collect the other legs singly or in sets,
> > according to how you want to wire it, solder them to a dropping resister
> and
> > connecting the other of the resister end to power. Let there be light...
> and
> > there usually is.
> >
> > Want them to flash? Then you gotta pony up another dollar fifty. That'll
> buy
> > you a NE555 timer chip, a capacitor, three quarter-watt resisters and a
> generic
> > PNP transister. Crazy-glue the timer chip to the circuit board 'dead
> bug,'
> > meaning it's legs are sticking up, solder the thing together in the proper
> > manner and your LED will blink sixty times a minute with whatever duty
> cycle
> > you've selected. If we're talking a nominal 12vdc and up to 45 LED's, a
> 1:10
> > duty cycle will give you a nice flasher that should last about 100,000
> hours...
> > mebbe a little less.
> >
> > That's one way.
> >
> > Trouble is, with the LED just poking through the circuit board it's
> staring
> > straight ahead; most of its light is going to be in a cone that's only
> about 20
> > degrees wide. Even a hundred LED's won't put out very much light if
> viewed
> > from the side. To be seen from the side the LED has to be pointing to the
> > side... which means it can't be pointing anywhere else at the same time.
> So
> > you use a buncha them, pointing ever whichaway.
> >
> > Do you need FORTY-FIVE of those suckers? At about half a buck each the
> red
> > ones aren't too expensive, their production having been subsidized by
> their use
> > as automotive tail lights. And ultra-bright white LED's are down to about
> two
> > bucks because they are coming into common use for flashlights and such.
> But
> > those green jobbies cost the earth... nearly three bucks each. So you're
> > looking at two hundred and fifty bucks worth of LED's... which is about a
> big
> > handful.
> >
> > On the up-side, it's a one-time cost. Build them right into the airframe,
> GLUE
> > on the molded Lexan cover, forget about it.
> >
> > Okay, but why FORTY-FIVE of those suckers?
> >
> > Got an orange? Okay, a grapefruit then. Cut it in half. Now you got a
> > hemi-sphere. That's your tail light's coverage.
> >
> > Take the other half of the grapefruit and slice IT in half. Now you got
> two
> > quada-spheres, or whatever. They are your wing tip nav-light's coverage.
> >
> > Remember the part about the 20 degree angle? Take one of the wing tip
> chunks
> > of grapefruit and divide it into twenty-degree slices. (Don't cut it, use
> a
> > marking pin on the rind; you can eat it for breakfast tomorrow. Yeah, I
> > know... but she already thinks you're crazy anyway.) 180 degrees, you
> need
> > about 9 LED's. 90 degrees, about five. (yeah, it overlaps a bit) Nine
> times
> > five is 45.
> >
> > So how do you do that? How do you arrange 45 LED's so their output covers
> a
> > quarter of a sphere?
> >
> > Take a look at an LED. The tip is about the size of a grain of popcorn.
> Now
> > imagine your wing-tip nav light is little EAR of popcorn with nine kernels
> from
> > top to bottom that tapers smoothly to the end FIVE kernels away. Now all
> we
> > have to do is replace the kernels with LED's :-)
> >
> > Looks something like the eye of a bug.
> >
> > To make something like that I used the comptuer to make a drawing showing
> rows
> > of holes for the legs of the LED's. Glued it to the non-copper side of
> the
> > circuit board. Started drilling. (#80 carbide burr; about 12,000 rpm.
> SOP
> > for anyone who fiddles with electronics.)
> >
> > (The pattern is about 3-3/16" long, 1-3/16" wide at its fattest point,
> which
> > happens to be 1-3/16" back of the nose, and is a smooth tear-drop shape.
> The
> > holes are logically placed so that their LED can be bent to the proper
> angle.)
> >
> > Over on the copper side of the board the hole for the grounded leg is left
> > straight. The leg will be poked through it then bent over and soldered to
> the
> > copper cladding. But the hole for the power leg gets deburred with a
> regular
> > 1/8" diameter drill bit having an angle of 116 degrees. That gives me a
> > shallow NON-CONDUCTIVE hole through which the hot leg of the LED
> protrudes.
> > From that point on the wiring is exactly the same as for the first
> example,
> > with a dropping resister (about 4 cents each) attached to each leg, or to
> pairs
> > or even sets of three... depending on how you want to wire it.
> >
> > The LED's get poked into their holes to a depth determined by the ANGLE to
> > which they have to be bent, which I worked out on the comptuer before I
> started
> > drilling. Since the pattern is symmetrical this isn't as tough as it
> sounds.
> > Basically, the LED's in the middle stick up about 3/4" and stand up
> straight.
> > The ones on the outside only stand up about a quarter of an inch and get
> bent
> > over at 90 degrees. Between the middle row and the outer row you've got
> three
> > LED's, each standing about an eighth of an inch higher, each bent at a
> > different angle.
> >
> > Not a very good description but you should get the idea.
> >
> > The computer also provides the profile of the Lexan cover, which starts
> life as
> > two poster-board templates to which you glue some crunchy urethane foam
> then
> > sand it into a smooth surface. Coat that with something that will harden
> up
> > nice -- I used sheet-rock mud -- then use it to make a female mold and
> from
> > that pull a solid plug of either Portland cement or Plaster of Paris. The
> > cement shrinks a bit more than the plaster but both are saturated and have
> to
> > be cured for about ten days before you can use them to make your cover,
> which
> > you do by gluing flannel to the plug, greasing it up and popping it into
> the
> > oven with a hunka Lexan balanced on top. When the Lexan gets rubbery,
> protect
> > your hand and mold the plastic to the plug. Do that until you have a nice
> > symmetrical shape. Now do it again for the other wing. If you know about
> > thermoplastics and canopies and who shot John you know you can use the
> original
> > mold to flare the skirt so as to form a flange. Trim that, apply sealant
> and
> > attach it with glue, rivets or whatever... you won't be removing it.
> >
> > Did you want fries with that? Or mebbe a strobe? Then order a batch of
> item #
> > 06040 from Harbor Freight. You can throw away the case; all you want is
> the
> > guts. It has its own plastic shield. Trim a piece of aluminum so the
> clear
> > plastic is a nice fit then glue the guts to the aluminum... or whatever...
> and
> > mount them ALL OVER your airplane... wing tips, top & bottom of the
> fuselage,
> > top of the vertical stabilizer, ass-end of the rudder... wherever you want
> a
> > bright little spark of light.
> >
> > The little strobe costs about eight bucks and runs on two AA cells.
> Forget
> > that and pony up another buck per unit for an LM340T-5 voltage regulator
> and a
> > couple of caps, an electrolytic 470uF/16v for the input and a disk ceramic
> .1uF
> > for the output. Screw the LM340T-5 to the piece of aluminum to which
> you've
> > mounted the guts and solder it all up.
> >
> > 18ga. wire is over-kill for either the stobe or the flashing LED's
> >
> > The strobes aren't very bright but then, neither are they very large nor
> > expensive... use a buncha them. It won't keep the Big Boys from running
> your
> > ass down but it might scare the crows out of your way.
> >
> > ----------------------------------------------------------
> >
> > Too expensive, right? I think so too. But the cost of LED's continues to
> drop
> > and ultrabrights will probably be down to a nickle each by the time the
> bird is
> > ready to fly, if ever.
> >
> > -R.S.Hoover
>
>

[baltobernie]

"Jay" > wrote in message
om...
> (Jeff Peterson) wrote in message
>...
>
> > In order to get the wide angle coverage you need, you can either use a
> > large number of LEDs, each with a narrow beam or you can use the same
> > number of LEDs each with a broad light beam. When you grind the front
> > off the LEDs the forward-directed brightness (officially called the
> > luminous intensity [in milli-candella, aka mcd] ) will be decreased
> > because the light is no longer directed forward but is spread out
> > instead. The total amount of light emitted (officially called the
> > luminous flux [in lumens, aka lm]) is not affected by grindind off the
> > front. This means you need just as many ground-off LEDs as you do
> > intact ones. The advantage of grinding them off is that you dont need
> > to go to all the trouble of pointing them every which way. As far as
> > scratches go, yes, they do diffuse the light output, and that's just
> > what you want.
>
> You save having to do as much pointing but you have to grind and
> POLISH each one. The scratches cause a loss in the total flux. Look
> at what you have to do to prepare the ends of fiber for installation
> of connectors, and thats even for the longer wavelengths in IR.
>
> > i noticed this coincidence: all these devices seem to use about the
> > same current, 20 ma. so very likely they have similar flux. i think
> > the more expensive ones are the ones that have the LED chip located
> > very close to the focus of the hemispherical lens....these have the
> > narrow beam that gives them high mcd numbers. if you are grinding the
> > dome off, it doesnt make sense to pay extra for this.
>
> They use the same current because its the same little chunck of Si.
> The cost of molding the lens doesn't vary much depending on the angle.


This is correct. The flux, or total light output, does not change. We
either spread it out, or concentrate it.


> What you're paying all the money for as I understand it is:
> 1) Pick of the litter, the semi-conductor die vary in efficiency and
> are binned. You pay more for the really good one, and less for the
> others.


Not quite the whole story. In ascending order of output AND price:
1. Gallium Phosphide
2. Gallium Aluminum Arsenide
3. Allium Indium Gallium Phosphide (AlInGaP, pronounced like the
abbreviation)
4. Indium Gallium Nitride (InGaP, ditto)

All of the major die manufacturers bin their die. I won't tell you the
ratios of brightest to dimmest, quantity-wise, but reputable LED
manufacturers are interested in producing product yielding the most
efficient use of the wafer. Specifying bins (bright OR dim ones) dictates a
significant price penalty to the user. Naturally, over a product lifetime,
the entire wafer goes up in intensity, but the primary factor in retail
price is the wafer material, shown above.


> 2) New process technologies- All these really bright LEDs are the
> result of recent changes in semiconductor processing, so there is less
> competition because less guys can make them.

Yes, and add that material #4 is the subject of a patent dispute, only
recently negotiated (or subverted).

Regarding position lights, commercially-available 3mm (aka T-1) LEDs are
possible at 500 mcd over a 50-degree angle. There are some VERY high
intensity LED packages not available to the retail user (yet), e.g. the red
and blue lightheads and in unmarked cruisers. Yes, these are LEDs, not
strobes.

Bernie
in the LED business for 20 years
remove my age to reply directly

[Richard]

For an excellent four channel dimming system that handles LEDs see:

http://www.fdatasystems.com/LC_40.htm

R


"baltobernie" > wrote in message >...
> "Jay" > wrote in message
> om...
> > (Jeff Peterson) wrote in message
> >...
> >
> > > In order to get the wide angle coverage you need, you can either use a
> > > large number of LEDs, each with a narrow beam or you can use the same
> > > number of LEDs each with a broad light beam. When you grind the front
> > > off the LEDs the forward-directed brightness (officially called the
> > > luminous intensity [in milli-candella, aka mcd] ) will be decreased
> > > because the light is no longer directed forward but is spread out
> > > instead. The total amount of light emitted (officially called the
> > > luminous flux [in lumens, aka lm]) is not affected by grindind off the
> > > front. This means you need just as many ground-off LEDs as you do
> > > intact ones. The advantage of grinding them off is that you dont need
> > > to go to all the trouble of pointing them every which way. As far as
> > > scratches go, yes, they do diffuse the light output, and that's just
> > > what you want.
> >
> > You save having to do as much pointing but you have to grind and
> > POLISH each one. The scratches cause a loss in the total flux. Look
> > at what you have to do to prepare the ends of fiber for installation
> > of connectors, and thats even for the longer wavelengths in IR.
> >
> > > i noticed this coincidence: all these devices seem to use about the
> > > same current, 20 ma. so very likely they have similar flux. i think
> > > the more expensive ones are the ones that have the LED chip located
> > > very close to the focus of the hemispherical lens....these have the
> > > narrow beam that gives them high mcd numbers. if you are grinding the
> > > dome off, it doesnt make sense to pay extra for this.
> >
> > They use the same current because its the same little chunck of Si.
> > The cost of molding the lens doesn't vary much depending on the angle.
>
>
> This is correct. The flux, or total light output, does not change. We
> either spread it out, or concentrate it.
>
>
> > What you're paying all the money for as I understand it is:
> > 1) Pick of the litter, the semi-conductor die vary in efficiency and
> > are binned. You pay more for the really good one, and less for the
> > others.
>
>
> Not quite the whole story. In ascending order of output AND price:
> 1. Gallium Phosphide
> 2. Gallium Aluminum Arsenide
> 3. Allium Indium Gallium Phosphide (AlInGaP, pronounced like the
> abbreviation)
> 4. Indium Gallium Nitride (InGaP, ditto)
>
> All of the major die manufacturers bin their die. I won't tell you the
> ratios of brightest to dimmest, quantity-wise, but reputable LED
> manufacturers are interested in producing product yielding the most
> efficient use of the wafer. Specifying bins (bright OR dim ones) dictates a
> significant price penalty to the user. Naturally, over a product lifetime,
> the entire wafer goes up in intensity, but the primary factor in retail
> price is the wafer material, shown above.
>
>
> > 2) New process technologies- All these really bright LEDs are the
> > result of recent changes in semiconductor processing, so there is less
> > competition because less guys can make them.
>
> Yes, and add that material #4 is the subject of a patent dispute, only
> recently negotiated (or subverted).
>
> Regarding position lights, commercially-available 3mm (aka T-1) LEDs are
> possible at 500 mcd over a 50-degree angle. There are some VERY high
> intensity LED packages not available to the retail user (yet), e.g. the red
> and blue lightheads and in unmarked cruisers. Yes, these are LEDs, not
> strobes.
>
> Bernie
> in the LED business for 20 years
> remove my age to reply directly