OT- [sdiy] LED maximum peak current

[Brock Russell]

Hi Martin:

I guess I qualify as an LED expert. I have been designing solar
powered LED marine marker lights for Carmanah Technologies
for six years and outdoor LED message boards for Boulevard
Technologies for two years. I hope I've learned a few things in
this time.

>I want to build very bright LED indicators.

How bright? Serious question. You can easily build LED
indicators that can cause eye damage. Most people are
not aware that super-bright LEDs come with optical hazard
warning labels on the factory packaging. This gets lost
when you buy small quantities.

>Now, the normal spec is average current , e.g. 20mA.
>If I use transients, this can be a lot more.
>I remember my first design project as EE, it was an "inherited"
>design that should send IR telegrams. It had three integrated transitors
>on it, each 30000um wide (this IS substantial) and could
>deliver peak currents up to 1A into each of the three IR
>LEDs. The current was regulated.
>
>Now I want to do something similar, but I don't wat to do
>the current servoing in a discrete design for obvious reasons.
>I want to use the internal parasitic resistance of the
>LED.

I don't understand what you mean here. The LED has series
bulk resistance and parallel parasitic capacitance. If you are
thinking of using the bulk resistance as the current limiting
element, it can be done but tends to be electrically and
optically inefficient. External current control is always more
accurate and efficient.

>Do red (very bright) LEDs have the same current tolerance,
>i.e. 1A for a single pulse?

No. Assuming a 5mm (T1 3/4) package, manufacturers specs
range from 70 mA to 200 mA peak pulse current. I find it
useful to look at package power dissipation also, typically
125 mW to 150 mW for 5mm package. Other packages
have different specs of course.

>What pulse rate and duty cycle is needed to make the apearent
>brightness as large as possible?

DC, 100% and heat sink it like crazy with short leads and lots
of copper area on a pcb.

All super-bright LEDs (GaP, AlInGaP, etc.) have non-linear
forward current to luminous intensity curves, with the increase
in luminous intensity proportionately less that increase in
current. This is the reverse of standard LEDs and the standard
methods of implementing them as panel indicators where one
would use short duration, low duty cycle, high current pulses to
optimize intensity.
More important are the effects of forward current on junction
temperature and the inverse effect of junction temperature on
luminous intensity. This is hidden or not even mentioned in most
data sheets but the result is that super-bright LEDs are most
efficient operating at 5 to 10 mA. You want to keep the die as
cool as possible. Ambient temp of -40C works great...

Back to the question of how bright do you want it to be.
Typical indoor panel indicators are in the 2 to 50 mcd range.
You can get super-bright red LEDs that provide over 10000 mcd
at 20 mA. In darkness this would be visible from over 3 km.
Up close it can cause permanent eye damage.
It pays to think of these things as lasers. The light is very
coherent with interference patterns visible even at low currents
and painfully obvious at high currents.
What is your objective here?


I'm willing to discuss this further if you want, off list preferred.

Brock