For some advanced mathematics, can you correlate the forward voltage with the colour (e.g., band-gap energy of the semiconductor versus the
wavelength of operation).

Also, does the I-V characteristics depend upon ambient illumination? How about self illumination (e.g., holding a reflective mirror in front of the LED
to reflect its own light back into it).

How does temperature affect the I-V characteristics?

Anyway, just some fun things to think about.

Dave

All of the calculations of actual LED current use the actual measured LED voltage drop.

I just posted up an article about an adjustable current sink that might be helpful with respect to your LED tester idea. Let me know if I can help you with this project.
http://www.uchobby.com/index.php/2008/02/03/adjustable-current-sink/

Make sure your pot can handle the load you are putting on it.

LED’s drop 2 or more volts depending on the type. Your math was using 1V for LED drop.

I was interested in getting even more resolution at low currents (mostly out of curiosity), so I also added a couple of diodes to get a nonlinear current function of voltage with more resolution at the low end. The exponential characteristic of the LED is exactly wrong for this: we want the controlling signal to be something like log(current), not exp(current), so you need either a double log conversion or a current source + log converter.

In a darkened room, I had enough control at the low end to measure LEDs turning on somewhere around a tenth of a milliamp.

Have you considered giving the AVR control of the current? Since it’s a DC signal, you can do simple D/A by just aggressively low-pass filtering a PWM pin (or two, for more resolution). You could then have the AVR sweep through the desired current range and graph the voltage/current characteristic, or automatically read out your voltage and resistor combinations at some predefined currents (say, 5, 10, 15 mA).

Another useful attachment would be a photodiode or phototransistor, so you could match LEDs for a display application. You could use it to bin LEDs by brightness at a specific current, or to select customized resistors for each LED. You don’t need photometric accuracy if you’re matching the same kind of LED, just repeatability.

If you’d like to fit the whole under a 5V supply, you can use a small current-sense resistor and the differential ADC mode of a more recent AVR (like the 261) which has a built-in selectable gain stage.

Maybe you shouldn’t listen to my “feature creep” encouragement, though; I managed to convince myself to put my own version of this project on the back burner by adding too many features.