Updated my UV-led test method (=final), tested a Luminus SST-10 365nm led (and some others)

Good idea!

No need, he already lists the led for sale. But his picture is an older generation of LG UV-leds than I thought it would be. Still cool to test but not the latest and greatest.

At least he has 2. best flux bin and 3. lowest voltage bin
960-1060mW @500mA is impressive

Djozz, this is a great thread! I applaud your efforts to find a repeatable measurement method. :+1:

Simon asked me for your shipping address to send you a free smaple djozz

this host with triple UV led and a DD driver will outperforming many high-end UV flashlight, i guess

Yes, could be a perfect combination, but the large diameter filter glass (ZWB2) is difficult to purchase.

Nice work, djozz. Would you care to update this thread by testing the LG UV-A 365 nm? I’d like to see it fare against the 365 nm UV LEDs in your test. I’m sure it will not top the Luminus SST-10, but I think it will beat the two Nichias. I’m not sure about the LiteON. I have the LG UV-A and I’m using it. The visible stray light of the LG is tolerable, maybe not as bad as the LiteON. I’m curious to see if getting the SST-10 is worth the extra price.

I have one of those LG leds in house to test but I’m waiting for the Seoel Semiconductors version from Kaidomain that is stuck in the mail somewhere, to do a test of both.

Yes, the one from Seoel Viosys, too. The CUN66A1A. I’m curious to see both of them in your test.

The SST-10 has no white wastelight and pretty few violett visible light, so an UV filter is not really nessesary

Have you considered using a diffraction grating to measure the output at a desired wavelength? All you would need is a slit at the calculated angle, and your detector behind it. You wouldn’t need all that filter stuff.

We’ll I am sure djozz will chime in but in order to do this your detector has to be sensitive to UV wavelengths and these are quite expensive. Djozz elegant solution is to convert the UV to visible wavelengths using the uranium glass so he can use regular inexpensive light meters to measure the intensity.

UV photodiodes are inexpensive too.
(https://shop.boselec.com/collections/uv-sic-sensors/products/sg01s-18)

True if you wanted to use an ordinary lux meter, a fluorescent material is needed. But with a grating there is no need for the visible light filter, and since 365nm is a component of sunlight, you don’t need the UV filter. The meter will either detect it (which is fine) or attenuate it.

Thanks for this detector exekutive and welcome to BLF. My brief search didn’t turn up any inexpensive flux meters using it but if you know of one please let us know. Wonder if you can just replace the sensor in one of those cheap meters? I see you haven’t posted much (neither have I) but that doesn’t mean you aren’t familiar with the forum. People here are focused on flashlights and budget so spending a few hundred dollars on test equipment is generally (not always) not an option. People are always receptive to new ideas so please do help us out with your knowledge of this topic.

In djozz setup the UV filter is to keep out stray UV from the flashlight, not from sunlight

Hi Scientist.
I realize that the UV filter was to keep UV out of his light meter. I was trying to explain how it isn’t necessary. (had a little typo there sorry).

I also realize that this is a flashlight forum, and that most people aren’t interested in this — in fact they probably won’t be reading this far down. My comments were directed at djozz or anybody who is interested in/might attempt this experiment.

The detector simply drives a current proportional to the incoming light. In fact they give you an example circuit right in the datasheet (it looks like about a dollar or two in parts). Coincidentally, the setup is very affordable and well within the budget range of somebody who can afford a fancy flashlight.

Hi exekutive,

Part of why my method measuring UV leds is how it is, is because it developed out of measuring white leds first. It was very rough at first and now it is finetuned a bit but does not have the accuracy of white led testing. It seems to work, and that is always a valid reason to stick to a method.

There will be other approaches indeed, the most obvious one is leaving out the fluorescence step if you use a UV detector instead of a luxmeter and go from there.

UV detectors do introduce new sources of error: i.e. the sensitivity slope in the wavelength area of measurement can cause significant errors: in case of the detector that you linked: a led with peak at 365nm will be measured higher than a led with peak at 370nm (5nm peak difference is a common variation among “365nm” leds). If you have not also measured the spectrum to correct for that (which I usually have not done) a 10% error is easily made. I know that the ZWB2 filter in my set-up does a similar thing but I hope (not measured) that the U-glass that I use in my current method has a reasonable flat radiationpower-to-visiblelightoutput response over my usual around 365nm wavelength measuring range.

Also, I’m not good in electronics and not good in computers. So not being able to design electronic circuitry and computer interfaces, I always use standalone instruments. And as Scientist already pointed out, they can be expensive and my hobby money is very limited. I do allow myself something pricey every now and then though :slight_smile:

One thing that I learned from measuring white leds is that it is not obvious to measure total light output of a source if the beam profile differs from source to source. In that case you need some integrating device. My current method does provide some integration by projecting all radiation onto a fluorescing surface and measuring the resulting fluorescing light in my usual integrating sphere made for visible light. Not perfect but it does to quite some extend compensate for different beams.
If a UV detector is used, also an integrating device is needed and if the UV light is bouncing around in an integrating sphere I must determine if the reflectivity of the sphere’s inner surface does not vary too much for wavelength differences. Because the light undergoes multiple reflections before reaching the detector, small differences get multiplied.

So I do my measurements the way I do for several reasons:

-historic reasons, it has been developed towards this method
-convenience/lazy reasons, it works for me and it will be a major undertaking to develop something new
-financial reasons, a new method will require buying new equipment
-fear of the new, can I learn the new stuff? (electronics, computer stuff, interprete datasheets etc.), and will I ever be able to sufficiently first measure then solve all the causes of error in a new to be developed method.

My white led tests that I have been doing for 5 years, in the meantime are being done better by a few other BLF-members, who have better equipment, methods and skills than me. So I know that my possibilities and skills (and space!, my hobby is 2 meter2 from our 3-room city appartment that I share with wife and son) are limited and I welcome people to do better UV led tests too :slight_smile:

The detector is a wideband device (UVA+UVB+UVC, 221 to 358 nm). They have other devices for other wavelengths. The data sheet provides a spectral response graph, so with a bit of back-of-the-envelope calculation, or spreadsheet math, I think you could get within a couple of percent error.

The amplifier circuit has only 4 parts, which you could pop into a breadboard. No equipment required. It’s perfect for a beginner.

I think in this situation where you’re comparing products, the most important thing is consistency and reproducibility. The relative measurements are the only thing you’re interested in.

The ideal setup would use slits to sample the light, and be performed in a lightproof enclosure. You could also take several measurements while rotating the emitter through 180deg to get an angular intensity profile.

Devices which measure total light output are spherical and fairly sophisticated.

I know too well. Anyway, I thought I’d throw some ideas out there.

I think your tests are very good. Independent testing like this is very valuable to people trying to make purchasing decisions, so thank you. Never be afraid of learning new stuff! I’m always glad I did.

To compensate for a non-flat spectral response of a radiation detector you can not do some post-measurement correction calculation unless you know the spectrum of the radiation source. The only way is to use a dedicated optical filter before the detector that fixes the non-flatness of the detector response. Such filter-detector combinations are usually very expensive.

I still think that you under-estimate the importance and difficulty of integration of the radiation in total light output measurements.

Anyway, I hope that someone will do all those nice things that you propose (not meant in a cynical way :slight_smile: ), thanks for the suggestions and discussion, I hope it will be picked up by someone.

Me, I have so many nice little hobby projects on my list that for UV tests I will stick to what I do now until someone else does it better (hence the mention “final” in this threads title).

You can. That is exactly what response curves are for. By the way, this is after passing through a diffraction grating + slit , so you are starting with fairly monochromatic light.

I don’t underestimate it at all. As I mentioned, the apparatus which labs use for measuring TOTAL output is fairly sophisticated. But I think for the purposes of comparing one LED to another, measuring at 0deg from the emitter (usually peak intensity) will get you pretty good idea. As I said, you could measure at a few different angles in case the emitter has some weird profile.

In regards to the fluorescent source, I think quinine (ordinary tonic water) might make a good candidate. You could start with a fresh sample for each measurement in case there is some bleaching.