Reliable source for 3W 365nm UV LEDs?

Hi,

Does anyone have a link for a reliable seller for some reasonable quality 3W 365nm UV LEDs? Preferrably on a star PCB because I don’t really want to reflow solder emitters myself. I would prefer to spend less than €5 per LED if possible.

EDIT: I need a 120 degrees dispersion or at least 90 degrees.

Something with not too much visible light spill would be nice but I will be using these with ZWB2 filters so some spill is okay.

I don’t mind buying from Amazon, eBay or AliExpress but apparently there are many sellers that sell stuff purporting to be 365nm LEDs whereas in fact what you will get is a 395nm LED or even just a purple colored one.

I’m located in the Netherlands so a local or European source would be fine too.

I need these fairly quickly, say in the next 2-3 weeks as they are meant for a Halloween rig I’ve been working on. To be more specific they are going to be used to light up a skeleton that has been painted with white UV fluorescent paint to do a Pepper’s Ghost illusion, which is a reflection of something in a sheet of glass to make something appear and disappear as a semi-transparent ghost.

I would really like to use UV for this because the hidden chamber and the ‘display’ area are rather small and very close together and having bright white LEDs would probably cause too much spill into the ‘display’ area and ruin the effect.

I already have 6 UV LEDs that I bought from a local shop months ago and the single one that I tested back then was absolutely fine. The specs are as follows:

3W UV 365-370Nm Epistar power LEDs on black 20mm star PCB

Specifications:
Power: 3W
Emitting Color: Ultraviolet 365-370nm
Recommended current: 700mA
Forward Voltage: 3.2-3.6V
Beam angle: approximately 120-140 degrees
With double gold wire (99.99%) and copper frame
Quantity: 10 pieces

So the one I tested was just fine and plenty of UV came through my ZWB2 filters and it had a blueish-white spill to them.

The second one was fine as well but then the 3rd to 6th LEDs from the same package turned out to be very different: instead of a blueish-white visible light spill it is has a distinct purple light spill and much less UV gets through the ZWB2 filters. The paint that I used to paint my skeleton is formulated specifically to respond to 365nm, not 395nm. Below is a picture showing the difference between these two:

The one on the left is the first one which appears to be a proper 365nm LED. I think the one on the right is actually a 395nm LED because of the purple glow and the fact that much less UV seems to be getting through the ZWB2 filter.

It’s strange because all 6 LEDs came on star PCBs that were still joined together, so they got mixed up when they were reflow soldered, either by accident or deliberately.

I did find some alternative sellers of UV LEDs but like I said I have no way of knowing if I will be receiving genuine 365nm LEDs or fakes:

https://nl.aliexpress.com/item/32319049384.html

https://nl.aliexpress.com/item/33010543252.html (why is this one so expensive?)

TL;DR Would like some links to reliable sellers of 3W 365nm UV LEDs on star PCB with 120 degress dispersion, or comments regarding the sellers above.

Thanks

By the way here are some pictures of the fixtures that I designed for this application, the one below has been 3d printed by a friend of mine and he has 5 more ready and working on another 8 of them:

The fixture hold one star PCB mounted LED which sits on top of a 3W rated hearsink and there’s retainer ring for the ZWB2 filter. There’s a pot magnet in the bottom to attach the whole thing to the door frame.

Airflow is a bit restricted but there is an intake/outlet pair and I’m only driving these at 350mA instead of their rated 600mA. With that they don’t seem to be getting warm at all and I hope I won’t have to drive them harder (which will probably not increase the intensity that much anyway).

They will not be on continuously either but rather for just 15-30 seconds with a cool down period of 20 seconds or longer, depending on the number of trick-and-treaters.

After some searching I have found a number of options for UV LEDs, some of which I’m sure I’ve seen mentioned here before. Here’s a little table I built (as a screenshot because I cannot seem to insert it as an actual HTML table):

I wonder where my current EPIled UV LEDs rank in terms of radial intensity. The first two and very cheap ones from Ali and eBay in this table seem to have very low efficiency and output.

Does anyone know if kadomain.com ship to Europe (Netherlands) and how long it might take to get it here? I sent them a message but haven’t heard back yet. EDIT: Just found the information on their website, they do ship to Europe. I do find it strange that their website is not SSL encrypted, that is quite uncommon these days.

I’m a bit surprised by how much ledrise.eu charge for putting an emitter on a PCB. The price difference for the Seoul Viosys between kaidomainand and led-tech.de is also rather staggering.

From what I can tell my best options appear to be the LG, the Seoul Viosys or possibly some of the other kaidomain offerings.

Basically I want to brightly light up my UV paint coated skeleton with 6 UV LEDs for even lighting in a small 90x90cm (35x35 inches) room (the toilet in our hallway) to have it reflect in some foil in the hallway.

The brighter I can light it up without visible light , the better the reflection will be visible. The hallway will be darkened and the background against which the reflection will be visible is going to be a print of a fairly dark wooden dungeon door. The walls will be decorated with medieval style grey bricks.

I have not tested this yet. I was going to shortly, until I found out that 4 of my 6 EPIled UV LEDs are 395nm rather than 365nm and they don’t perform that well. And now that I have seen that the efficiency of these LEDs varies wildly I wonder if even the two 365nm EPIleds that I have are really that good. The EPIled specs don’t mention radial intensity so I have no comparison for other LEDs.

Any suggestions appreciated.

PS: I will be driving these LEDs at a maximum of 700mA because that is what my LED driver can output. I’m using a Picobuck or Femtobuck LED driver that connects to an Arduino.

Update: I went ahead and ordered some of both the LG and Viosys CUN66A1B LEDs. Let me know if that was a bad idea :grin: .

I’m now trying to do similar research as you did. It’s a shame no one replied but I guess UV emitter experts are few sbd dar between.

How did the LG emitters compare with the cheap ones you started with?

I did originally order 8x Seoul Viosys UV 365nm Z5 Series CUN66A1B and 8x W LG UV 365nm Ultraviolet UV LED Emitter but after I had ordered they came back to me and told me that they only had 3 of the LGs in stock, so I changed my order to 15x CUN66A1B. As such I never had a chance to try the LG’s.

The Viosys ones are absolutely fine, they do put out a good amount of UV and not too much visible light, but I did end up using the ZWB2 filters with them to reduce visible light spill as much as possible. It really worked quite well:

The UV LEds come into action at 0:34, causing ‘Skelly Pepper’, or rather her ghost, to appear.

It’s difficult to get good footage of this as it’s rather dark and the color spectrum of the display used for the girl’s portrait isn’t a great match with the sensors in my camera so it looks quite a bit better to the naked eye but you get the idea.

that’s pretty amazing! Well done!
I know what you mean about trying to capture UV on video, it definitely didn’t do it justice in my experiments.

Unfortunately those emitters are outside of my budget, as I need 1000 of them. I have come across some knock off’s by a manufacturer called XLW. But at $0.10 a piece I’m very suspicious.

It’s so hard to find info on these Chinese emitters unless you go with the top brands.

Considering that different UV emitters have different outputs, a precise number of emitters is probably not very helpful for planning purposes. It might be easier if you tell us how much radiometric UV output you need, and let us figure out what is a reasonably cheap configuration.

thanks for reply.

I have a 365nm Convoy C8. I believe the radiometric UV of that emitter is 1800 mW?
I have no way of measuring this though but it’s stated online as the “Radiant Flux typ”.

I beleive the C8 uses the seoul z5 CUN66A1G which is a 5W emitter?

I don’t need it to be as powerful, maybe half as much but I’m not sure how that translates to a radiometric UV measurement… did I mention I’m a noob with both flashlights and UV?

Welcome to the hobby!

A good way to get decent output estimates is looking at official datasheets; here’s the one for the CUN66A1G. On page 3, it specifies 1.85W radiant output at 1A current, and page 5 gives the relative output as a function of current, so you can estimate the output at any drive current. These specifications are for the bare emitter; there will be optical losses at the reflector and at the filter, so maybe a factor of x0.7 is appropriate to get the out-the-front output.

Those who claim the CUN66A1G to be a 5W emitter refer not to the radiometric output, but the electrical input, which will always be more.

What is the purpose for your getting something more than the Convoy C8? Knowing this can help us narrow down what to recommend.

This is very helpful thank you. I was actually looking for an emitter that is LESS powerful than the c8. I need something that is just powerful enough to make dust highly visible in ambient daylight.

I was following you for a while on the radiant flux, but then I tried to compare the CUN66A1G against the SST-10-UV and found there are min and max flux values, as well as relevant flux as you mentioned…

So the CUN66A1G luminous/radiant flux is 1850mW and the SST-10 is 900mW if I’m reading these datasheets correctly?

I’m not 100% sure I understand the relative flux. Is it that at 1A the relevant flux for the CUN is 100%, so over 1A you’re overdriving it? How does that translates to perceived brightness? If I drive it at 1.2A will I then see 20% more brightness? so 1.85 * 1.2?

This is the data sheet for the SST-10, (looking at the 365nm). https://download.luminus.com/datasheets/Luminus_SST-10-UV_Datasheet.pdf

There is also the little brother of the CUN66A1G, The CUN66A1B:

Cheapest UV LED Simon has as far as I see

This is correct, at the specified drive current.

All LEDs can be overdriven to some extent for some additional output. However, the increase in output is not linearly proportional to input power, but experiences the famous “diminishing returns” phenomenon. For whatever drive current you want to query, read the corresponding percentage from the relative output graph, and multiply that to the output at the binning current.

Are you scanning large areas or looking at stuff up close? Do you want a wide but unfocused beam or a narrow but very intense beam? The design of the optical system is arguably more important that the power itself. It is possible that what you want isn’t less power, but less focus.

I think I’m coming to grips slowly with how to compare these emitters. Thanks so much for taking the time to explain.

I’m designing for use up close, subject will be 5-20 cm from the lightsource. Because it’s close range and going though a small 18mm ZWB2, a wide, even beam would be ideal.

What type of optics should I consider? I didn’t know there was much to choose from apart from the viewing angle which I believe should be 120 and the reflector which I think should be and orange peel (if they exist at 18mm).

PS - how do I find the binning current in the datasheets?

20cm is a very nontrivial distance to require visibility in daylight. Given that you want a wide, even beam, I think you will need as much power as you can get, so emitters in the 3W power class are out of the question.

Your only option is a reflector–convex lenses and TIRs are opaque to UV. If you are using a reflector, the viewing angle of the bare LED does not matter. A 18mm orange peel reflector does exist (look up Convoy T3 reflector), but its opening is small enough to restrict you to 3535 emitters. I’d recommend instead a Convoy S2+ (20mm reflector), which makes no real difference in beam profile but gives you more options.

Look for keywords such as “test current” (in SST-10 datasheet), or specification in the title (“Electro - Optical characteristic at 1A” for the CUN66A1G).

Based on the above considerations, you’d want an emitter at least as powerful as the CUN66A1G. The SST10 won’t do because it is less powerful and has very high forward voltage, so even the binning current might be impossible to achieve.

I’ve had good luck with this quad-die 5050 emitter, which can handle at least 3A and has a very low forward voltage for a UV emitter, meaning it can sustain maximum output for a while. I would get this in a Convoy S2+ host and 5A driver, then set the mode group to 50% maximum.

There’s a huge range of what ambient daylight encompasses. You are very unlikely to get a UV that that will incite sufficient fluorescent response to make something visibly fluorescent in daylight at 20cm distance. You really need low levels of ambient light to get a decent result.

All that being said those 5050 quads do pretty well in an S2 or S6 sized light, but heat management is an issue. The Vf’s aren’t really great- you’re going to get voltage sag as the battery discharges.

I should have been more clear but by ambient daylight I meant indoors during the day, as I don’t want to restrict use to nighttime. Of course that depends from house to house and I won’t be able to cater to all situations.

the T2 reflector is 17.9mm, that’s probably perfect if I can get the price down on a large order. I wonder if anyone else besides Simon is making these with similar specs as currently it’s not affordable to me at scale.

I think a 3535 may actually be a sensible option as I’m a little bit worried about heat and managing that. I’d rather be less effective at brighter ambient light levels then have to manage a lot of heat.

thanks for the heads up the the SST. From what I read high forward voltage also means excess heat so I will avoid that one. Will dive back into some datasheets tomorrow.

This might be useful:

Ah yes this was quite useful, thanks.

If only they listed the emitters. Sometimes it’s quite hard to find out, especially for the more budget options.