LED driver to illuminate 300+ LEDs efficiently help

[Active8]

On 12 Oct 2003 21:19:53 -0700, Spehro Pefhany, said...
[snip]

Voltage drop of the LED goes up with current...

that's the non-linearity i mentioned earlier. like a reverse Ebers-
Moll... I controlling V as opposed to Vbe controlling Ib via Ic/Hfe.
groovy.

this increases
power, which could be considered to have a proportional to I
component and a proportional to I^2 component. Sometimes
even if the voltage is free, the derating to keep the die
temperature in spec will kill any small improvement.

It will use even less energy if you leave it off entirely, but that
will tend to reduce the brightness.

really? ;-)

If the average current is the same, and the supply voltage fixed, the
resistor will run cooler

sorry. i should have just stated that. i was under the impression that
john wasn't seeing that, but upon revisiting his post, i see he was
talking about the internal resistance of the diode.

and the LED hotter for pulsed operation.

ok thanks. i think i'll reread that article. it did mention some good
excuses for pulsing an LED AFAIK for now. i'm wondering how many led
pulsers i've run across that were for naught and how many were good
designs. glad i haven't wasted any time with 'em, though

brs,
mike

 

[Active8]

I just tried this with my signal generator:

GEN ---->|----WWW--0V

With a DC voltage of 5V, a small red LED is about twice as bright as it is
with a 60Hz square wave. Too bad, think of the energy we could save...

Regards,
Bob Monsen

yeah and as john said:

Persistence of vision is not a peak follower effect.

so all that (or most of it) pulsing crap is... crap!

thanks for doing the test for me.

brs,
mike

 

[Hal Murray]

I'm building an LED clock with 5" display numbers. The LED manufacturer

(lite-on) seems to suggest a 10% duty cycle at 1kHz. (At least, thats what
they test max current at.)

Its a bit of a pain (well, not too painful) to flash them, so I'm quite
interested in the group take on this datasheet.

If I don't have to do it, I can save a pin on my PIC, and use a cheaper
part.

My reading...

To first order, the brightness of LEDs is proportional to the
average current.

The specs are telling you that you can run it at a 10% duty cycle (at 1 KHz)
if you want, not that it will be brighter if you do that.

The peak current (above conditions) is 180 mA. That's an average of
only 18 mA. The next line is the spec sheet says you can run them
at at (up to) 60 mA steady. That will be much brighter than an
average of 18 mA.

 

[Da Man]

I wrote a tiny program for my STK 500 microprocessor programmer that flashed
the 8 bright little LED's with increasing order of duty cycle, from 1/8 to
full duty, and it was pretty much linear, and did not matter on frequency,
from a few hundred kilohertz down to around the spot flicker starts.

 

[R.Lewis]

"Jeroen Vriesman" <[email protected]> schreef in bericht

[snip]

Also, a 50% duty cycle at 2*X lumen looks brighter than continious X

lumen.

Is that really true? I've read/heard that many times, but I also
seem to recall that this is an urban legend.

[snip]

The key word here is "looks". An eye's response is not linear. The
best way to answer your question is to perform a few experiments. Get some LEDs
and power some with DC and some with various duty cycles of current. Compare
the visual effects. Measure the power applied to the LEDs. Try different
colors of LEDs. Try different amounts of background ambient light.

You could quickly become an LED expert and settle the question once and
for all.

Remember, a peek at the answer is worth a thousand guesses. Or ten
thousand news group "theories".

Perhaps it would help the pulsing led protagonists if they realised that the
eyes response is perfectly linear in its response to (the lumen value of)
light. The CIE have expended extensive efforts over many years to ensure
that this is so.
The eyes response to radiometric quantities is, however, non linear but
irrelevant in this context.

 

[Active8]

I wrote a tiny program for my STK 500 microprocessor programmer that flashed
the 8 bright little LED's with increasing order of duty cycle, from 1/8 to
full duty, and it was pretty much linear, and did not matter on frequency,
from a few hundred kilohertz down to around the spot flicker starts.

funny. i was planning on doing that to test the programmer i'm building.
too bad i'm using ho hum, off the shelf, who knows what they are or
where they came from and how long they've been in the junk box leds
i don't expect anything exceptional but it will help with the LCD
contrast code while i'm waiting for the LCD to arrive <== need
emoticon for light bulb.

brs,
mike

 

[Robert Monsen]

I'm building an LED clock with 5" display numbers. The LED manufacturer
(lite-on) seems to suggest a 10% duty cycle at 1kHz. (At least, thats what
they test max current at.)

Its a bit of a pain (well, not too painful) to flash them, so I'm quite
interested in the group take on this datasheet.

[snip]
The peak current (above conditions) is 180 mA. That's an average of
only 18 mA. The next line is the spec sheet says you can run them
at at (up to) 60 mA steady. That will be much brighter than an
average of 18 mA.

Thanks, that sounds like the answer I wanted to hear. No pulsing! Yow!

Regards,
Bob Monsen

 

[Don Klipstein]

My reading...

To first order, the brightness of LEDs is proportional to the
average current.

I want to add to this the LED's efficiency/efficacy as a function of
current. LEDs normally are nonlinear, and many to maybe most of
chemistries that have been in production since the late 1980's or further
back have efficiency/"luminous efficacy" maximized at higher currents in
the 10's of milliamps (for a "usual size chip"). These can benefit from
pulsing if the average current per illuminated chip has to be a few
millimaps or less.

Some additional data/comments in a web page of mine,
http://www.misty.com/~don/ledp.html

- Don Klipstein ([email protected], http://www.misty.com/~don/index.html)

 

[Don Klipstein]

"Jeroen Vriesman" <[email protected]> schreef in bericht

[snip]

Also, a 50% duty cycle at 2*X lumen looks brighter than continious X
lumen.

Is that really true? I've read/heard that many times, but I also
seem to recall that this is an urban legend.

[snip]

The key word here is "looks". An eye's response is not linear. The
best way to answer your question is to perform a few experiments. Get some LEDs
and power some with DC and some with various duty cycles of current. Compare
the visual effects. Measure the power applied to the LEDs. Try different
colors of LEDs. Try different amounts of background ambient light.

You could quickly become an LED expert and settle the question once and
for all.

Remember, a peek at the answer is worth a thousand guesses. Or ten
thousand news group "theories".

Perhaps it would help the pulsing led protagonists if they realised that the
eyes response is perfectly linear in its response to (the lumen value of)
light. The CIE have expended extensive efforts over many years to ensure
that this is so.
The eyes response to radiometric quantities is, however, non linear but
irrelevant in this context.

I want to add that human vision has some sort of partial
time-integration process. As ar as I can tell from kilotons of data hat I
have read and generated on my own, human vision is linear prior to this
partial time-integration process and all significant nonlinearities occur
after this process. To make an LED appear brighter than its average
brightness, it has to flash or visibly flicker.

I did experiments in this area, noted in
http://www.misty.com~don/ledp.html

Some LEDs, especially ones of chemistry that had nonlinearities favoring
higher currents, were popular and in wide use at the times when this
business of pulsing them to take advantage of some quirk of human vision
took off (in the early 1980's). The relevant nonlinearity was in the
usually-concerned LEDs and not as often believed in human vision.

- Don Klipstein ([email protected],
http://www.misty.com/~don/index.html
http://www.misty.com/~don/ledp.html)

 

[Tony Williams]

I want to add to this the LED's efficiency/efficacy as a function
of current. LEDs normally are nonlinear, and many to maybe most
of chemistries that have been in production since the late
1980's or further back have efficiency/"luminous efficacy"
maximized at higher currents in the 10's of milliamps (for a
"usual size chip"). These can benefit from pulsing if the
average current per illuminated chip has to be a few millimaps
or less.

7-segment LED displays were most often scanned. It required less
drive lines and produced a much higher brightness for the same
power consumption. The HP Optoelectronics books were always a good
source of data and sums to get best brightness/watt out of LEDs.
I suspect many cheap noname LEDs and displays are still made with
old technologies and could still benefit from pulsed operation.