From: Winfield Hill
Subject: Re: What is maximum current for LED?
Date: 2 Oct 2002 04:28:57 -0700
Organization: Rowland Institute
X-Newsreader: Direct Read News 2.98
Boris Nogoodnik wrote...
> ... there is one human vision factor. If you use pulses
> instead of constant current, thy still will look as bright as on
> constant current, but will dissipate much less power. So here
> you can make them look much brighter by exceeding the maximum
> current in pulses, but still keeping the average current well
> under the specs.
To my understanding, the effect one can obtain this way is not
large, under 3x, at least for foveal viewing, and only occurs
for pulsing at a repetition rate that's painfully obvious, say
15 to 25Hz. Alternately, if the LED's primary purpose is to
attract attention, peripheral as opposed to foveal viewing has
a much greater sensitivity to light variation, which means that
a greatly reduced level of flashing light can be noticed out of
the corner of one's eye, prompting further closeup examination.
"Nelson Win" asked,
>> ... I want to squeeze the brightest light I can from the
>> LEDs without prematurely burning them out. Does anyone have any
>> experience here to recommend a maximum current? The LEDs will
>> flash on and off at a frequency of a few hertz, 50% DC, for a
>> duration of a couple or hours at a time. Any advice appreciated.
Hi Nelson. We do have some relevant experience with our great
LED Blaster project. This was a successful attempt jointly with
the Red Cross to purify blood, using a special dye and massive
doses of red light at a certain wavelength. We drove 800 high-
output red LEDs at 125 to 150mA each!!, running them so hot that
their wavelength changed a bit. Just enough, as it turned out,
to optimally activate the dye. A standard bag of blood could be
completely LED-Blaster processed in just over 20 minutes, IIRC.
The LEDs were selected AND190HRP parts in a common T-3 package,
but we clamped the LED's cup-leads to a massive copper current-
distribution bar to improve heat removal, 20 LEDs per bar, and
cooled the set of 40 bars with a massive 8" fan. Both the LED
current and fan speed were adjusted to fine tune the wavelength.
I've forgotten just how much red light we got, but I think it
was about 10 watts at 655nm. My technician made a professional-
looking sign, like those for lasers, saying DANGER! LED LIGHT!
In fact, glancing directly into one of the banks of LEDs, even
at partial power, left one blinded for several minutes.
Despite the abuse, the LED lifetimes were surprisingly long,
although every now and then one would explode. Right - a loud
bang, plastic LED pieces everywhere! (The Red Cross researcher
didn't like that feature, so although it did the job well, much
faster than alternate light sources, eventually the LED Blaster
was placed into retirement. My argument that we should be given
time to create a proper surface-mount version was ignored. The
rather stormy relationship with the Red Cross didn't help.) For
our LEDs, explosion was the primary failure mode, as opposed to
declining light output. Somehow, the whole thing struck me as
very funny! Anyway, despite the hard work, we had good fun.
Now, with respect to your 50% duty flashing at a few hertz, my
transient-thermal-response measurements may be relevant. For
the common-package LEDs we were using in the copper-block setup,
the thermal mass was so low that the maximum temperatures would
be reached in a few ms, so it's likely your slow flashing won't
allow using higher power. In fact, thinking of the mistreatment
we gave our LEDs at 125mA, I suspect we would have had to lower
the current to ease the impact of thermal cycling. By contrast,
Luxeon's attractive metal-block-substrate LED emitters certainly
have a much longer thermal time constant. That's where I would
go for a lot of LED light these days.