Hi Dave - I love this post. There are lots of simple, rigorous tests that can be done to test an assumption -- such as the assumption that all 400nm LEDs are actually 400nm -- which are so important to the work we're doing, and this is a perfect example. I gave you a Barnstar for the note to mark this.

Were these LEDs of the same brand? The same part number? Or were they a variety? Which LEDs did you actually test -- can you post part #s? Thanks!

Other simple questions we should answer:

Are the spectra of different halogen (specifically Solux) bulbs the same? How close?

How about those of different RGB LEDs?

Also, do you have these as a set in Spectral Workbench, or could you share the ID #s?

Thanks for your posting. I was particularly surprised when I looked at Spekwin32 on the internet. I found that it was freeware for individuals like myself, and presumably like you.

Thanks once again,

Guillaume851

Hey Jeff! Yes they were all from the same company (superbrightled.com) here is the part number - # RL5-UV0430-400. Originally I bought 10 LEDs but yeah, I toasted one, my fault, so I had to deal with the other 9 (I also bought an LED driver for a high wattage LED I got, which the manufacturer stated didn't need a resistor for lower watt LEDs, which this particular LED driver can interface with a TTL, but indeed this driver's max is 350ma and YES it did need the resistor, LED got smoked!

Soooo, upon testing the other 9 I ran them off my Arduino boards TTL so voltage would not fluctuate at load and I could get a more reliable meter reading on each LED. I may purchase an assortment from a variety of companies and test maybe 20 and try and get a derivative average on the probability of getting a "good" LED that one could use with a high percentage of repeatable results.

also here is the link to the set I created for the testing https://spectralworkbench.org/sets/3208

Thanks for the barnstar!

Dave H

Hey Guillaume851, yeah, I have been using spekwin32 for several months now and once you start getting the hang of it it is easy to use and very effective for data editing and analysis. Yes it was free, although you do have a choice if you want to use it commercially, but that was costly and really I didn't need to do that, I do this as a private researcher so I just requested the free license, no problem. I hope this research helps because I know it certainly helps me, and I love the fact that there are so many of us here that working and collaborating together, I think that's pretty cool.

Have a good one!

Dave H

Dave - I have to ask - would it be a lot of work to do the same set of 9, but with a fluorescing sample in there as well? I'd love to see how much the variability in excitation wavelength (8nm!) affects the emission spectrum of a sample. Maybe a heavy motor oil?

Hey Jeff! No problem at all, I am retired (a lot of time but no were to go!) All the LEDs are numbered so it's just a matter of setting the experiment up again with a sample. Will do tomorrow with the 10w40 sample I have, no solvent, just straight up oil.

The variation in wavelength you're seeing probably is normal variation in the manufacturing process for LEDs. Their wavelength depends on the particular mixture of the semiconductor, which for blue is gallium, indium and nitrogen, and the natural bandwidth of an LED is typically about 30 nanometers, as you can see if you look at the plot (it's measured full width at half maximum, or 0.5 of peak output). Makers of cheap LEDs don't try to produce precise wavelengths, so you need to measure them to see what you get. That's different than fluorescent lamps which have narrow emission lines.

Laser diodes have narrower emission lines, but the cheap ones also vary in wavelength because their main applications don't require high precision. (Green laser pointers do not have the same variation.)