LED tail strobe

anonymous coward wrote:
Have you considered using a constant current regulator, instead of a
resistor? I believe there is an example circuit given in the LM337/317
datasheet showing how to build one using only the LM337 (normally used as
a voltage regulator) & one resistor.

It would need to be bolted to a heatsink, like the Luxeon Star LEDs, but
IIRC the LM337 and cousins also shut down if they overheat.

AC


Yes. I considered it. I opted for the simplicity of a single current
limiting resistor and the constant voltage regulator that is already
there. Number one rule of fault management. If it ain't there, there's
no way to break it. Regulators not only add an additional active
component with its list of failure modes, it also adds severl solder
connections and more heat, making it even more difficult to rig the
system into a 1/4" piece of plexiglass.


--
http://www.ernest.isa-geek.org/
"Ignorance is mankinds normal state,
alleviated by information and experience."
Veeduber

Jim Weir wrote in message
And as somebody else has pointed out, running a series string close to the
bottom limit of Vcc and hoping that a single resistor will provide constant
current to these devices will cook them when Vcc rises to the charging voltage.

No designer in his right mind would use a circuit in this manner.

Jim

The more the supply varies the less effective the technique is. I
picked the simple resistor design with certain assumptions. When it
comes down to it, the voltage doesn't vary all that much on a properly
operating electrical system. That big 'ole battery is like an anchor
on the supply bus. Over voltage condition blows the breaker.

What if the alternator quits? Well how far down do you want to go on
the discharge? You could get a range of zero to 14 volts if you
consider a discharged battery, thats a pretty tough range to design
for isn't it? I can think of a lot of systems that will malfunction
when supplied from a discharged battery. Even normal lights will go
out of spec on a discharged battery.

On Wed, 05 May 2004 02:46:06 +0000, Ernest Christley wrote:

anonymous coward wrote:
Have you considered using a constant current regulator, instead of a
resistor? I believe there is an example circuit given in the LM337/317
datasheet showing how to build one using only the LM337 (normally used as
a voltage regulator) & one resistor.

It would need to be bolted to a heatsink, like the Luxeon Star LEDs, but
IIRC the LM337 and cousins also shut down if they overheat.

AC


Yes. I considered it. I opted for the simplicity of a single current
limiting resistor and the constant voltage regulator that is already
there. Number one rule of fault management. If it ain't there, there's
no way to break it. Regulators not only add an additional active
component with its list of failure modes, it also adds severl solder
connections and more heat,

A regulator will produce no more heat than a resistor passing the same
current with the same voltage drop. The ones I'm thinking of are 'in-line'
devices, so their (negligible) supply current does not have to be factored
in as an extra source of heat.

I buy what you're saying about complexity, but you would probably only
need a single current regulator + one resistor for each parallel bank of
series LEDs (eugh, but I can't think how better to put it).

BTW, do you need to use zener diodes / transorbs with aircraft power
systems as you do in cars, to avoid problems with voltage spikes?

AC

making it even more difficult to rig the system into a 1/4" piece of plexiglass.

anonymous coward wrote:

BTW, do you need to use zener diodes / transorbs with aircraft power
systems as you do in cars, to avoid problems with voltage spikes?

AC


In one form or another, I'd say the answer is a qualified yes. Again, I
didn't add any sort of regulation to the LED array. As I remember it,
and it has been quite a while since I looked at it, the larger LEDs can
absorb rather large transients themselves. Their construction is not
that far removed from zeners or transorbs after all.

As I understand it, the limiting factor of the LED's ability to suck
down transients is their ability to dump the internal heat is the reason
why 'overdriving' them with higher but pulsating current works. You can
drive more current through them, just not for very long. In my opinion
(vs me sitting down and running actual numbers which will only happen
when I'm ready to build my own prodution models), if you derate the LED
to about 80%, it'll be able to take whatever a barely functioning
regulator will ever throw at it. Anything more is overdesign, adds
complexity, and one more solder joint is just one more chance for me to
screw something up.

--
http://www.ernest.isa-geek.org/
"Ignorance is mankinds normal state,
alleviated by information and experience."
Veeduber

I posted new photos of the LED tail strobe, now with the circuit
completed, at
http://w1.lancair.net/pix/led-strobe

i also posted the schematic.

the particular component choices give a 1 second period and
50% duty cycle. tinker with values to get your favorite flash pattern

i used monolithic ceramic capacitors because they are small, but
they do have a bit of variattion of C with temp. might be
better to use something more stable. as the circuit warms it flashes
faster.

The circuit works quite well, but after some soul-searching I have
tenatively decided NOT to put the system on the LNC 360.

reasons:
-its heavy...235 grams. Installing the strobe will require more than 3
times its mass added to the rudder counterweight. perhaps a kilogram
total. all this at the maximum rearward moment arm. yuck.

-it gets quite warm. after a minute or two the temp is maybe 40C at
the
base of the LEDs and thats with a fan pushing air past at about 10
MPH. in still air it overheats. if i forget to turn it off on
landing,
I could possibly slag my rudder.

-the system satisfys the FAA Regs, but with no margin...I prefer a
safety factor 2.

-its a complicated system. lots of machining. tight fits. odd angles.
no finesse.

it just doesnt seem like a bulletproof system to me, so i may just
chalk
this one up to experience and develop plan B.

Plan B: get a bunch of 5 mm diameter superbright white leds and
mill the hemispherical ends off the plastic cases. this will give
them a
lambertian pattern. no need to polish the ends, might even sandblast
them
to make the pattern even broader.

glue these into holes drilled in the curved trailing
edge of the lancair wingtip. i will need many more of these LEDS than
with
luxeons, but that will help distribute the thermal dissipation.

the weight might actually be lower this way than with luxeons since
the
aluminum heat sink will no longer be needed. and the weight will be
near the center of lift, and will not require a counterweight.

or maybe i will come up with plan C...

-Jeff

A linear regulator will produce the SAME amount of heat as a resistor
at the same current. W=VI If the volts and the amps are the same
the power is the same. However since the car mfg's are installing
LED's in the center brake light they have paid for the development of
simple constant current switching power supply IC's to drive the
LED's. The power dissipated in the LED is the same however the power
that was dissipated in the current limit ballast resistor has been
greatly reduced so that the TOTAL system power and heat generated is
about 50% of what it was to get the same light out of the LED using
resistors.
You have to add a small inductor and since it is a switcher you may
have radio interference if your layout and EMI filter is not good.
John

On Tue, 11 May 2004 09:01:23 +0100, anonymous coward
wrote:

On Wed, 05 May 2004 02:46:06 +0000, Ernest Christley wrote:

anonymous coward wrote:
Have you considered using a constant current regulator, instead of a
resistor? I believe there is an example circuit given in the LM337/317
datasheet showing how to build one using only the LM337 (normally used as
a voltage regulator) & one resistor.

It would need to be bolted to a heatsink, like the Luxeon Star LEDs, but
IIRC the LM337 and cousins also shut down if they overheat.

AC


Yes. I considered it. I opted for the simplicity of a single current
limiting resistor and the constant voltage regulator that is already
there. Number one rule of fault management. If it ain't there, there's
no way to break it. Regulators not only add an additional active
component with its list of failure modes, it also adds severl solder
connections and more heat,

A regulator will produce no more heat than a resistor passing the same
current with the same voltage drop. The ones I'm thinking of are 'in-line'
devices, so their (negligible) supply current does not have to be factored
in as an extra source of heat.

I buy what you're saying about complexity, but you would probably only
need a single current regulator + one resistor for each parallel bank of
series LEDs (eugh, but I can't think how better to put it).

BTW, do you need to use zener diodes / transorbs with aircraft power
systems as you do in cars, to avoid problems with voltage spikes?

AC

making it even more difficult to rig the system into a 1/4" piece of plexiglass.

On Tue, 11 May 2004 17:04:22 +0000, Ernest Christley wrote:

anonymous coward wrote:

BTW, do you need to use zener diodes / transorbs with aircraft power
systems as you do in cars, to avoid problems with voltage spikes?

AC


In one form or another, I'd say the answer is a qualified yes. Again, I
didn't add any sort of regulation to the LED array. As I remember it,
and it has been quite a while since I looked at it, the larger LEDs can
absorb rather large transients themselves. Their construction is not
that far removed from zeners or transorbs after all.

Do you know where I might find any links to material about this? I'm
building a computer controlled LED flasher device (not for a tail strobe
- something unrelated) and naturally I would like to make the flashes as
bright as possible. Luxeon reckon that when pulse-width modulating their
LEDs, the current should never exceed 500-550 mA (for the 1W versions).
Given that their normal current is only 350 mA this isn't much of an
increase. Also, this is meant to be at pulse-width-modulation frequencies
of 100Hz or more. The pulse duration I need is 1/10 of a second.

On the other hand, your experience echoes my experience. My Luxeons aren't
heat-sunk, yet due to software faults I've unintentionally passed 1.5A
through some of them for several seconds. I won't be surprised if their
life expectancy is greatly reduced but they have lasted the 'development'
phase of my device surprisingly well.

As I understand it, the limiting factor of the LED's ability to suck
down transients is their ability to dump the internal heat is the reason
why 'overdriving' them with higher but pulsating current works.

What worries me is how quickly the die can dump heat to the aluminium
casing.

AC

anonymous coward wrote:
On Tue, 11 May 2004 17:04:22 +0000, Ernest Christley wrote:

In one form or another, I'd say the answer is a qualified yes. Again, I
didn't add any sort of regulation to the LED array. As I remember it,
and it has been quite a while since I looked at it, the larger LEDs can
absorb rather large transients themselves. Their construction is not
that far removed from zeners or transorbs after all.


Do you know where I might find any links to material about this? I'm
building a computer controlled LED flasher device (not for a tail strobe
- something unrelated) and naturally I would like to make the flashes as
bright as possible. Luxeon reckon that when pulse-width modulating their
LEDs, the current should never exceed 500-550 mA (for the 1W versions).
Given that their normal current is only 350 mA this isn't much of an
increase. Also, this is meant to be at pulse-width-modulation frequencies
of 100Hz or more. The pulse duration I need is 1/10 of a second.

On the other hand, your experience echoes my experience. My Luxeons aren't
heat-sunk, yet due to software faults I've unintentionally passed 1.5A
through some of them for several seconds. I won't be surprised if their
life expectancy is greatly reduced but they have lasted the 'development'
phase of my device surprisingly well.


Superbrightled.com (I think) has some of this info. The rest is just
general knowledge of semi-conductors. If Luxeon gave you a Imax for
pulsed current, they should have also given you a pulse duration and
duty cycle. That is, how long and how often the current flows. Again,
the limiting factor is how quickly can you get the heat from the CENTER
of the device. The surface temperature is really only a side effect.
It's the temperature of that little piece of doped glass in the center
of that large chunk of plastic that is critical, and it can melt while
the exterior is still cool. So you can briefly drive a lot of current,
but then you have to stop and let the casing suck the heat out.


As I understand it, the limiting factor of the LED's ability to suck
down transients is their ability to dump the internal heat is the reason
why 'overdriving' them with higher but pulsating current works.


What worries me is how quickly the die can dump heat to the aluminium
casing.

AC

You are well informed to be worried. All of these PC overclockers
thinking they can crank up the juice if they just add a bigger heat sink
are fooling themselves. The heat energy has a somewhat tortuous path to
traverse through the IC packaging before it can even be transferred to
the heatsink. To be most effective, a heat sink has to be as 'close' to
the heat source as possible. Close being defined as the least amount of
insulation between the two, ie. a 1/2" gap filled with copper plate is
probably better than a 1/10" air gap. The base of the LED is NOT the
heat source. (Though the leads are some nice metal heat conductors
connected directly to the glass in the middle 8*)

--
http://www.ernest.isa-geek.org/
"Ignorance is mankinds normal state,
alleviated by information and experience."
Veeduber

From what I've read, the bond wire that connects the semiconductor die
to the leadframe (the part you solder) can be the limiting factor on
pulse current. It has a lower thermal time constant that the chip
itself. You can smoke that little gold wire if you try to shove too
many e-'s through it at once.

I'm not sure what you're doing with your alternative application, but
in general, you pulse them to multiplex them for moving signs or to
lower the observed brightness (via PWM) for brightness controls.



anonymous coward wrote in message e...
On Tue, 11 May 2004 17:04:22 +0000, Ernest Christley wrote:

anonymous coward wrote:

BTW, do you need to use zener diodes / transorbs with aircraft power
systems as you do in cars, to avoid problems with voltage spikes?

AC


In one form or another, I'd say the answer is a qualified yes. Again, I
didn't add any sort of regulation to the LED array. As I remember it,
and it has been quite a while since I looked at it, the larger LEDs can
absorb rather large transients themselves. Their construction is not
that far removed from zeners or transorbs after all.

Do you know where I might find any links to material about this? I'm
building a computer controlled LED flasher device (not for a tail strobe
- something unrelated) and naturally I would like to make the flashes as
bright as possible. Luxeon reckon that when pulse-width modulating their
LEDs, the current should never exceed 500-550 mA (for the 1W versions).
Given that their normal current is only 350 mA this isn't much of an
increase. Also, this is meant to be at pulse-width-modulation frequencies
of 100Hz or more. The pulse duration I need is 1/10 of a second.

On the other hand, your experience echoes my experience. My Luxeons aren't
heat-sunk, yet due to software faults I've unintentionally passed 1.5A
through some of them for several seconds. I won't be surprised if their
life expectancy is greatly reduced but they have lasted the 'development'
phase of my device surprisingly well.

As I understand it, the limiting factor of the LED's ability to suck
down transients is their ability to dump the internal heat is the reason
why 'overdriving' them with higher but pulsating current works.

What worries me is how quickly the die can dump heat to the aluminium
casing.

AC

Luxeon has a patented mechanical design that allows them to dissipate the power better, thus getting more lumens. It is
not like the superbrights...

See: http://www.lumileds.com/pdfs/protected/AB23.PDF



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


..
"Ernest Christley" wrote in message news
anonymous coward wrote:
On Tue, 11 May 2004 17:04:22 +0000, Ernest Christley wrote:

In one form or another, I'd say the answer is a qualified yes. Again, I
didn't add any sort of regulation to the LED array. As I remember it,
and it has been quite a while since I looked at it, the larger LEDs can
absorb rather large transients themselves. Their construction is not
that far removed from zeners or transorbs after all.


Do you know where I might find any links to material about this? I'm
building a computer controlled LED flasher device (not for a tail strobe
- something unrelated) and naturally I would like to make the flashes as
bright as possible. Luxeon reckon that when pulse-width modulating their
LEDs, the current should never exceed 500-550 mA (for the 1W versions).
Given that their normal current is only 350 mA this isn't much of an
increase. Also, this is meant to be at pulse-width-modulation frequencies
of 100Hz or more. The pulse duration I need is 1/10 of a second.

On the other hand, your experience echoes my experience. My Luxeons aren't
heat-sunk, yet due to software faults I've unintentionally passed 1.5A
through some of them for several seconds. I won't be surprised if their
life expectancy is greatly reduced but they have lasted the 'development'
phase of my device surprisingly well.


Superbrightled.com (I think) has some of this info. The rest is just
general knowledge of semi-conductors. If Luxeon gave you a Imax for
pulsed current, they should have also given you a pulse duration and
duty cycle. That is, how long and how often the current flows. Again,
the limiting factor is how quickly can you get the heat from the CENTER
of the device. The surface temperature is really only a side effect.
It's the temperature of that little piece of doped glass in the center
of that large chunk of plastic that is critical, and it can melt while
the exterior is still cool. So you can briefly drive a lot of current,
but then you have to stop and let the casing suck the heat out.


As I understand it, the limiting factor of the LED's ability to suck
down transients is their ability to dump the internal heat is the reason
why 'overdriving' them with higher but pulsating current works.


What worries me is how quickly the die can dump heat to the aluminium
casing.

AC

You are well informed to be worried. All of these PC overclockers
thinking they can crank up the juice if they just add a bigger heat sink
are fooling themselves. The heat energy has a somewhat tortuous path to
traverse through the IC packaging before it can even be transferred to
the heatsink. To be most effective, a heat sink has to be as 'close' to
the heat source as possible. Close being defined as the least amount of
insulation between the two, ie. a 1/2" gap filled with copper plate is
probably better than a 1/10" air gap. The base of the LED is NOT the
heat source. (Though the leads are some nice metal heat conductors
connected directly to the glass in the middle 8*)

--
http://www.ernest.isa-geek.org/
"Ignorance is mankinds normal state,
alleviated by information and experience."
Veeduber