ADALM2000: Diode I-V curve

Hello:

It is difficult to tell just looking at your graph, but it looks like you might have a significant offset in your measured data below where the diode "turns on". For example the current should be zero at zero volts. Because of the exponential in the equation small offset like this can have a large impact on the results of the equation fit. Take a close look at your raw data values below 0.4 V and see if there is a consistent non-zero value (DC). Add or subtract this DC offset to the raw data and re fit to the adjusted values.

Doug

Hi Doug, thank you for your answer.

The DC offset is already taken care of by the p[3] term. By the way, the fit function above is wrong and instead I used p[1] * e^(V/p[2]-1) + p[3].

Here is a measurement with the yellow LED:

The diffusion current I_s is about 8nA for the 1N914 and 7e-12nA for the yellow LED. In both cases, the thermal voltage kT/q is about 50mV.

I have read that the Schockley equation does not work for Si and GaAs diodes. Is this the case?

I'm not sure what you are doing exactly but I've gotten excellent results fitting measured voltage / current data obtained using the M1k hardware (AKA ADALM1000). See some example results here:

https://wiki.analog.com/university/courses/alm1k/intro/data-analysis-1#curve_fitting

I assume kT/q is a known constant and the results of the fitting is the saturation current and the non-ideality factor n

Seems to me that if your aim is to measure / confirm the value of kT/q you need to know Is and N which are the real unknowns for any given diode. Is is a very small number and would be extremely hard to measure using M2k.

Hmm, I was not aware of the non-ideality factor n, which does not appear in eq. 5.2 in https://wiki.analog.com/university/courses/electronics/text/chapter-5. For the diode measurements, that value seems to be slightly above 2, but as far as I read, it can be at most 2.

As I implied above the M2k is not the best instrument to use for precision DC measurements. So I would question the data before questioning the science.

Thank you.

One more thing: in your measurements, I cannot figure out what value for n you obtained.

The error in the determination of the non-ideality factor in my plot comes from the measurement of the voltage across the diode. Since I get n=2.06, one must conclude that the ADALM2000 voltage input has an error of at least about 3%. Is this possible?

That M1k curve fitting tutorial was meant to show a user how to use the scripting function for data analysis in general and not necessarily teach the diode equation. The results are printed in the User spaces in the XY screen shot and are rather hard to read actually. The value for N was 1.32 I think.

Yes, without doing additional averaging / filtering (which Scopy does not do) the M2k measurement could be off by that much. The M1k will be at least 16 X better for DC measurements.

Now I see it: in the last equation you use n=1.9.

Thank you for all the clarifications.