[Laser] Notes on Luxeon I, III's and Rebels and Phlatlights

[C. Turner]

Hello again,

I would concur with what Chris said about the Luxeon III/Rebel LEDs and 
Phlatlight modules.

One obvious advantage of the Luxeon Rebels is that they are much cheaper 
than the Luxeon III's.  I strongly suspect (but haven't empirically 
measured) that their apparent "current density" is on par with the 
Luxeon III's - which also would imply that at their maximum rated 
current their "brightness per square millimeter" is probably on par with 
the Luxeon III's.

If this is the case, that would mean that the far-field flux of a Rebel 
would be about the same as that of a Luxeon III - but the beamwidth 
would be lower.  The obvious advantage is with a smaller device is that 
the current consumption is lower for a given amount of far-field flux, 
and another advantage is that with the smaller dies, the capacitance 
will also be lower - a point that has implications for high-frequency 
modulation (e.g. video, data, etc.)

One sticking point with the Luxeon devices is mounting them.  The "Star" 
mounted devices are the easiest to use as one simply bolts them to a 
heat sink - and all of the lines (Luxeon I, III, and rebel) are 
available (from one source or another) already mounted on slugs - often 
with electrical connections.

In our past work with Luxeon III's, I've always used the "emitter" 
rather than the stars.  Originally, this was because the "star" version 
was unavailable, but I found the emitter to be easier to work with since 
I simply epoxied them to the heat sink using "J.B. Weld" - a 
metal-filled epoxy that has proven to have excellent heat transfer 
capabilities.  By avoiding an intermediate interface (such as the "star" 
structure) the thermal resistance of the "emitter glued to the heat 
sink" was lower than a "star bolted to the heat sink" which meant a 
lower die temperature which also meant more light output at a given 
current at a given ambient temperature.

The Luxeon Rebels, on the other hand, are a different story.  Being 
intended to be reflow-mounted on a PCB substrate, they are difficult to 
use as a raw emitter and unlike the Luxeon I's and III's, you about have 
to use a substrate.  I've had some success epoxying them to a heat sink 
with the wires protruding through the heat sink, but mounting them this 
way is tedious and not always successful.

Instead, I would recommend getting them mounted on substrates, or 
mounting them yourselves.  Fortunately, this is pretty easy to do and 
pre-mounted Rebels are available from a number of sources, such as 
"luxeonstar.com".  These folks also have the "raw" mounts so that if you 
have the capability of doing so, you can attach the raw Rebel emitter to 
pre-fabricated metal slugs that already have etched (and insulating) 
wiring on them.

***

I, too, have experimented with side-emitting Luxeons.  For a 
"half-serious" project, I obtained several side-emitting Luxeon III's to 
test with a "half-million candlepower" spotlight.  As expected, they 
worked pretty well and seemed to be a decent match to the reflector, but 
they didn't have any advantage (divergence-wise) over the original 
halogen lamp.

Now, it's easy to think that a spotlight like this has a nice, 
tightly-controlled beam (it is far better-controlled than, say, a 
Maglight) but as Chris said, it's really pretty awful when compared with 
a moderate-sized Fresnel lens:  Years ago I built a "cheap" enclosure 
using foam-core posterboard, vinyl "page magnifier" Fresnels and a very 
poorly-matched Luxeon III emitter array (described here:  
http://modulatedlight.com/optical_comms/Optical_enclosure_cheap_version.html 
) - and that half-hearted attempt absolutely runs circles around the 
reflector system in terms of far-field flux!  This "cheap" cardboard 
unit was assembled only to give something for the "first" enclosure to 
talk to and is several dB lower in output and effective sensitivity than 
my other units - but it still is capable of spanning a 173km optical 
path with 2-way voice!

The side-emitting LED was intended to make a very short-range (and not 
very good) light communicator - and it would probably be usable over a 
kilometer or two.  The biggest problem, however, is getting rid of the 
heat!  If you have a Luxeon out there in "space" at the focus of the 
reflector, it's not convenient to put an adequately-large heat sink 
behind it without blocking part of the reflector.  What I ended up doing 
was to epoxy the LED to a copper rod about 8-10mm diameter and having 
that protrude out behind the reflector and putting the radiating fins 
back there.  While this works, it's really a pain!

***

Chris also mentioned the Phlatlight LEDs which are in a class by 
themselves in terms of luminous output.  The main limitation of these 
devices would appear to be the fact that it is simply not practical to 
extract more heat from the die than is currently done.  The higher-power 
devices in this line are constructed on copper slugs and achieve very 
high current densities, and unless they were fabricated on materials of 
higher thermal conductivity (silver and diamond come to mind...) along 
with active cooling it is unlikely that one could push the technology 
much harder in terms of current density.  Their unique physical 
structure also reduces the problem that most LEDs have, and that is the 
fact that a large number of the photons being emitted by an LED on the 
quantum level can't escape - that is, they can't get out of the LED die 
itself before being reabsorbed elsewhere within the die and do something 
useful!

In the year or so since they've been operational I've been able to get 
time to do only limited testing of the units that I've constructed, but 
they do work very well.  One problem (and it's not a terrible problem to 
have!) is that fact that run at full power, full-duplex communications 
is unlikely to be possible on a weak-signal path owing to "desense" from 
scattering (Rayleigh, Mie, etc.) of the transmit beam and it's 
contamination of the "receive" beam:  At full power, my detectors 
experience a dramatic increase in noise floor and may, in fact, suffer 
from a degree of overload.

The current units that I have are based on CBT-54's (no longer available 
- but I would have used CBT-40's - which are on exactly the same-sized 
substrates - at the time I built it had they been available) and operate 
with a peak LED current of 19-20 amps using a current-feedback type 
modulator with power supplied by a buck-type regulator - which means 
that there's only about 3-4 amps on average pulled from the 12 volt 
battery when at full power, using a single LED.  With a simple 
adjustment, the same modulator could be used to drive 2 or 3 of these 
LEDs in series from a 12 volt supply.

With these units (I have a working pair) we have managed one-way voice 
on a 23km NLOS (Non-Line-Of-Sight) path (but 2-way MCW) using a mountain 
as a reflector, achieving better results (that is, higher S/N ratio) 
than was obtained when we did our 278km LOS (Line-Of-Sight) contact.  
I'm sure that if we were to try this outside the area of a major 
metropolitan city that we could have easily managed 2-way voice as we 
were competing with a roar of hum from city lights!  When using these 
units across the Salt Lake Valley on a relatively short LOS path (a 
distance of only about 20km) their brilliance is eye-catching, badly 
overloading the optical receiver at full current - but then again we 
need less than 1/100th of that current (a few 10's of milliamps) to 
achieve solid, 2-way voice communications on that same path!

73,

Clint