Nav Lights using Luxeon LEDs

A flickering light will appear to be less bright than if it were
steady. This is because the eye takes the average between the on and
the off cycles. There is an opposing theory that says this averaging is
not perfectly linear (ie the on-cycle is not given the same weighting
as the off-cycle), but for most practical purposes we can take it to
be linear. So, at 50% duty cycle you will get 50% brightness, 50% power
consumption and 50% thermal heating. Everything scales by the same
ratio. You don't gain anything.

In highly nonlinear devices, where the light vs current is not a
straight line, such as laser diodes, PWM can actually increase the
brightness. But LED's are fairly linear devices so this effect does not
come into play.

In addition, pulsing a high-current LED may introduce other problems. 1
Amp turning on and off will generate all kinds of noise you may not
want in an aircraft.



wrote:
Hi Andrew,

The way i understand it is that you get the same brightness with PWM but
in bursts. Your still pulsing at 500ma which gives the same brightness.

The eye/brain won't see the pulses, but the led is only on "X" percent
of full duty, but appears to be constant.
Kind of like florescent light which flashes at 60 hertz.

Are you increasing efficiency? Maybe not, but the duty cycle of the LED
is less which may allow the use of a heat sink without the use of a fan.

Is my understanding correct?

It's a very nice arrangement anyways!

Thanks for the ideas.

Andrew Sarangan wrote:
No, it won't. PWM can be used as a dimmer, but it does not change the
LED efficiency. Running an LED at 1Amp half-duty PWM is the same as
running DC at 500mA. Your eye integrates the signal. The only time when
PWM can increase efficiency is in nonlinear devices. LED's power-out vs
current-in is very linear. PWM is used where a digital signal is used
to control the LED brightness. It is often easier to turn the LED on
and off at a fixed current, rather than changing the current.




wrote:

Won't pulse modulation take care of the heat problem.

I have done the efficiency analysis, and had concluded that a switching
regulator did not perform much better than a straight DC current
regulator. Firstly, the LM2576 is a voltage converter. You would still
need to follow this with a constant current source to drive the LED. A
more efficient approach is to use a current-mode converter, such as the
ADP1864, which has a seperate current-sense pin. In either case, the
benefits are not that overwhelming.

The green LEDs need 700mA at 10V. There is an additional 0.5V drop
across the wiring (cockpit to wingtip) that is unrecoverable. The
conversion efficiency of ADP1864 chip is around 85% for 700mA at 10V.
The overall efficiency of the system will be 82%, which is almost
exactly the same as the simple DC current source. So there is no
advantage in this case.

The red LEDs need 1.4A at 6V. Doing the same analysis, the overall
efficiency with the switching circuit will be 81%. The straight DC
source has an efficiency of only 50%, so there is some advantage here.
However, in real numbers, this difference amounts to only 250mA. I
didn't think that was enough to justifying installing a switching
system. The potential of a noise problem made the decision even easier.




wrote:
Andrew,

Your powersupply is way too inefficient... use a National Semiconductor
simple switcher LM2576 and you'll be much better off. Don't worry, it
won't emit a bunch of RF with the toroidal inductor it uses...

Dean Wilkinson

Hello

It was my impression that mounting the LED directly to the reflector
(flashing aluminum) will not provide enough heat sinking. A large
surface area is not the only determining factor because the thermal
conductivity of the metal will limit how far the heat can spread. The
thinner the metal, the worse the thermal conductivity. Flashing
aluminum is only 1/100" thick so its conductivity is not very high. One
could construct fins on the back side to increase the surface area, but
commercial heat sinks do that much better.




wrote:
Great write up, and well documented and good composite stuff along with
the electronics.

I'll throw in my 2 cents, on top of an already good design.

1) Sometimes less is more, especially in aircraft. If one LED hits
your numbers, go that way. With a single LED, it increases the
benefits if a switching arrangement versus the linear one deployed.
The hockey pucks mentioned elsewhere might be an easy way to go. As
far as extra margin for degradation over time, this is really an issue
for applications that run 24/7, but for an aircraft and then only the
hours its used at night, these babies aren't going to see over 1000
hours in their life.
2) I hate fans/Heat sinking- airplanes and heat sinks are made out of
the same stuff. Use your reflector as the heat sink by epoxying the
LED to the reflector.

So if you went that way you'd have an off the shelf circuit, one LED on
each side, and no fans.

("karel" wrote)
PS Andrew, would you mind answering _below_ ?
It really makes sense, you know...


Besides, June is Top Posting Month at rec.aviation*.


Montblack :-)

Andrew Sarangan wrote:
Hello

It was my impression that mounting the LED directly to the reflector
(flashing aluminum) will not provide enough heat sinking. A large
surface area is not the only determining factor because the thermal
conductivity of the metal will limit how far the heat can spread. The
thinner the metal, the worse the thermal conductivity. Flashing
aluminum is only 1/100" thick so its conductivity is not very high. One
could construct fins on the back side to increase the surface area, but
commercial heat sinks do that much better.


I think you have a beautiful design with good tradeoffs, Andrew.
However, if one were inclined to go fanless, there are heatsinks now
that are much lighter will provide much greater surface area. It uses a
much greater number of fins, and each is very thin.

--
This is by far the hardest lesson about freedom. It goes against
instinct, and morality, to just sit back and watch people make
mistakes. We want to help them, which means control them and their
decisions, but in doing so we actually hurt them (and ourselves)."

On Sun, 05 Mar 2006 05:16:51 GMT, Ernest Christley
wrote:

Andrew Sarangan wrote:
Hello

It was my impression that mounting the LED directly to the reflector
(flashing aluminum) will not provide enough heat sinking. A large
surface area is not the only determining factor because the thermal
conductivity of the metal will limit how far the heat can spread. The
thinner the metal, the worse the thermal conductivity. Flashing
aluminum is only 1/100" thick so its conductivity is not very high. One
could construct fins on the back side to increase the surface area, but
commercial heat sinks do that much better.


I think you have a beautiful design with good tradeoffs, Andrew.
However, if one were inclined to go fanless, there are heatsinks now
that are much lighter will provide much greater surface area. It uses a
much greater number of fins, and each is very thin.

Ernest, it takes cross sectional area to conduct heat; if you look at
the TRULY designed heatsinks to maximize heat transfer to air, you
will note a tapered base with tapered fins, larger and thicker fins
close, smaller and thinner further away.

I am weak in describing this, but I think you will get the picture;
think about holding a piece of aluminum foil 2 inches long in a candle
flame, then a piece of 1/4" aluminum rod in the same flame. Your
fingers won't get warm with the foil, but will burn with the rod.

Hi Andrew,

Here is what you do:

Put a 0.1 ohm 1 Watt resistor in series with your LEDs between them and
ground.

Use an LM2904 op amp to amplify the voltage across the 0.1 ohm resistor
such that the LM2904 outputs 1.2 VDC when the desired current is
flowing through the resistor. The gain resistors you need is something
you should be able to figure out.

The output of the LM2905 op amp goes to the Feedback input of the
LM2576.

Best Regards,

Dean Wilkinson

wrote:
Hi Andrew,

Here is what you do:

Put a 0.1 ohm 1 Watt resistor in series with your LEDs between them and
ground.

Use an LM2904 op amp to amplify the voltage across the 0.1 ohm resistor
such that the LM2904 outputs 1.2 VDC when the desired current is
flowing through the resistor. The gain resistors you need is something
you should be able to figure out.

The output of the LM2905 op amp goes to the Feedback input of the
LM2576.

Best Regards,

Dean Wilkinson


Thanks for that explanation. It makes sense now. I am still a bit
hesitant to put a switching regulator next to my VHF antennas, but I
will give it a try.

For some reason I thought you'd mentioned that it was .1". If its that
thin (.01") you might need to either use thicker stock or epoxy (with
special high heat transfer epoxy) that heat sink behind the LED.