Jump to content
UltravioletPhotography

My circular UV LED project


BruceG

Recommended Posts

I've been working on this for a while now, and just recently finished making the mark II verison of my circular LED light, I used different numbers of 365nm, 370nm, 395nm and 405nm LED bulbs, which gives a pretty well balanced spectrum distribution, close of natural sunlight in the region the camerea can record. Its only problem is overheating when used under full power.

 

post-127-0-53773900-1538579680.jpg

 

post-127-0-99600200-1538579874.jpg

 

The mark I version is much bulkier, and due to the quality of of the LED bulbs I used it is less bright. Also, because of my oversight on color reprocuction, the light output is heavily weighted in 370-385 region, which makes the photos kind of monochrome. It was a bit difficult to take out individual bulbs so I just left it as it is.

 

post-127-0-79134400-1538581218.jpg

 

post-127-0-97108700-1538581422.jpg

 

post-127-0-36906600-1538581822.jpg

 

These two light are both powered by a 15V battery, it can support up to 96W of power which I made full use by connecting as many LED bulbs, in both lights, 32 of 3W LED bulbs were used. LED is a bit different from any traditional light, they do not have a fix resistance, but the value follows a curve as voltage and temperature changes, therefore its ressitance/power is highly sensitive to voltage applied, slightly higher voltage can give higher luminance, but will also cause it to overheat and also overload on the power source. Therefore when I built my mark II version I also bought a LED dimmer, which changes the voltage in a small range, which allows me to avoid overheating when I don't have to.

Link to comment
Nice work! I have built something similar in the past, but with only one wavelength of LED. With regard to the problem of controlling light output from LEDs, a transistor circuit is helpful - that may be what your LED dimmer is - since what you really want is to control current not voltage, and transistors can essentially map voltage to current. Additionally the wavelength of the LED changes the voltage drop (in fact, rule of thumb is that the voltage drop of an LED at full power is the same number as the wavelength in eV. So a 365nm LED is (1240 eV-nm)/365nm = 3.4 eV photons, and the forward voltage drop is 3.4 volts).
Link to comment

When I saw your excellent flower shots yesterday, I was going to ask you about the light.

Thank you for posting this. This is very interesting. I have thought about making something similar using 340nm, 365nm and 385nm led lights. But the cost is still too high. Hopefully in a year or so they will come down.

Do you get odd looks, wielding this light around? Its bigger than a typical ring light I was expecting.

Link to comment
That is a great build. Really like that you've incorporated multiple wavelengths into it. And the pictures you posted from using it looked really good.
Link to comment

When I saw your excellent flower shots yesterday, I was going to ask you about the light.

Thank you for posting this. This is very interesting. I have thought about making something similar using 340nm, 365nm and 385nm led lights. But the cost is still too high. Hopefully in a year or so they will come down.

I was concerned with the cost too, it definitely be too expensive if I use Nichia LEDs, however there are a few manufacturers in the market who make quality UV LEDs that are relatively cheaper, one of them is Seoul Viosys, you can check out their website to find out their different models, they also make shorter wavelength LEDs including 340 and 350 if I remember correct, however their prices are a bit beyond our ideal budget. There are many more Chinese manufactured UV LEDs, they are not branded, most of them are inferior to Nichia and Seoul Viosys in terms of luminance level per unit power, but there are also good ones that performs nearly as good.

 

Do you get odd looks, wielding this light around? Its bigger than a typical ring light I was expecting.

I had a softbox attached to my LED, so I just look like a regular macro photography guy.

Link to comment

That is a great build. Really like that you've incorporated multiple wavelengths into it. And the pictures you posted from using it looked really good.

Thank you, Jonathan.

Link to comment

Nice work! I have built something similar in the past, but with only one wavelength of LED. With regard to the problem of controlling light output from LEDs, a transistor circuit is helpful - that may be what your LED dimmer is - since what you really want is to control current not voltage, and transistors can essentially map voltage to current. Additionally the wavelength of the LED changes the voltage drop (in fact, rule of thumb is that the voltage drop of an LED at full power is the same number as the wavelength in eV. So a 365nm LED is (1240 eV-nm)/365nm = 3.4 eV photons, and the forward voltage drop is 3.4 volts).

Yes, a transistor, that's what the dimmer is. The 365nm LED was indeed labelled as 3.4V, but so were the other LEDs, 370, 395 and 405nm. What's the theory behind your formula?

Link to comment

Yes, a transistor, that's what the dimmer is. The 365nm LED was indeed labelled as 3.4V, but so were the other LEDs, 370, 395 and 405nm. What's the theory behind your formula?

 

It's just that the band gap of the semiconductor is 3.4eV (for the 365nm example), so it takes about 3.4 volts to move one electron across the band gap if quantum yield is near 1 (each electron makes one photon). I'm not sure why your others were labeled the same way. Here is how some data (taken from this quick search) stacks up against the rule of thumb:

post-94-0-83618900-1538638667.png

 

The conclusion I would draw is that the quantum yield on the blue end of the spectrum is probably not all that close to 1, so it takes more voltage to move one electron across the bandgap than the rough assumption above would predict.

Link to comment

It's just that the band gap of the semiconductor is 3.4eV (for the 365nm example), so it takes about 3.4 volts to move one electron across the band gap if quantum yield is near 1 (each electron makes one photon). I'm not sure why your others were labeled the same way. Here is how some data (taken from this quick search) stacks up against the rule of thumb:

post-94-0-83618900-1538638667.png

 

The conclusion I would draw is that the quantum yield on the blue end of the spectrum is probably not all that close to 1, so it takes more voltage to move one electron across the bandgap than the rough assumption above would predict.

I see I see, it makes sense since 3.4V is a value above the required voltage for any longer wavelength than 365nm, it could be because the manufacturer wants to standardise their specs. There are also some models of 365nm LED from Nichia and Seoul Viosys that has forward voltage rated at 3.6, 3.7 and 3.8V, but I haven't seen one that is below 3.4V.

Link to comment

If your "dimmer" isn't a switch mode converter you might consider getting some DC-DC converter modules with a constant voltage/constant current ability.

 

There are step-up, step-down and step-up/down topologies available. You should drive the LEDs from converter modules running in constant current mode.

I would also set the voltage output without LEDs to a level that begin to limit before you reach the current ratings for the LEDs.

Some of the modules have really good efficiency. They are often designed after a reference design in the Switcher datasheet and work quite well.

 

Such modules are available on eBay for surprisingly low cost.

 

By grouping each wavelength-type together you can tune the colour-balance by adjusting the drive currents for each group.

It is also a good idea to add heatsinks on the backside of the LED Modules.

 

If you like the idea and want advices how to select DC-DC modules, please message me. I will try to help.

Link to comment

If your "dimmer" isn't a switch mode converter you might consider getting some DC-DC converter modules with a constant voltage/constant current ability.

 

There are step-up, step-down and step-up/down topologies available. You should drive the LEDs from converter modules running in constant current mode.

I would also set the voltage output without LEDs to a level that begin to limit before you reach the current ratings for the LEDs.

Some of the modules have really good efficiency. They are often designed after a reference design in the Switcher datasheet and work quite well.

 

Such modules are available on eBay for surprisingly low cost.

 

By grouping each wavelength-type together you can tune the colour-balance by adjusting the drive currents for each group.

It is also a good idea to add heatsinks on the backside of the LED Modules.

 

If you like the idea and want advices how to select DC-DC modules, please message me. I will try to help.

Thanks for all the good advice, they are all very relavent and helpful to me!

Installing a constant current module is indead urgent, it will defintiely help prevent overheating. I'm right now trying to find a light weight solution for heat sink too, I still want my light to be small enough to be hand held.

Interesting idea to control each wavelength separatly too, I never thought about that, however I will have to deal with the first two problems first.

Link to comment

Color me UV-impressed !!!

Great work. Keep us posted on the progress for the heat sink and other mods.

 

Everyone is going to want one of these. You may have just started a nice little business.

Link to comment

Color me UV-impressed !!!

Great work. Keep us posted on the progress for the heat sink and other mods.

 

Everyone is going to want one of these. You may have just started a nice little business.

Thanks Andrea, Sourcing the parts and materials would take some time, meanwhile I'm planning to build another one for UV fluorescence too.

Link to comment

Here is a question for Bruce and Ulf and other electronics knowledgeable folks.

 

Is it possible to replace the 365nm LED in a flashlight/torch such as the Convoy with a 340nm LED? Because I would really enjoy having even a small amount of UV light in some band below the ubiquitous 365 nm.

Link to comment

Here is a question for Bruce and Ulf and other electronics knowledgeable folks.

 

Is it possible to replace the 365nm LED in a flashlight/torch such as the Convoy with a 340nm LED? Because I would really enjoy having even a small amount of UV light in some band below the ubiquitous 365 nm.

It might be possible, but a bit difficult, needing the correct equipment to be done right.

 

There are several things that must be the same for the replacement LED as the original one:

  • It has the same component casing to fit the pattern on the circuit-board.
  • It has to have a similar emision pattern to work well with the funnel-shaped reflector.
  • It has to have the same or bigger power handling rating to survive, as the Convoy is running its LED close to max of its LED.
  • The needed drive voltage and/or current must be the same to match the drive electronics.

I know to little about the now available 340nm LEDs to tell if that is possible.

Bruce- Do you know?

 

If those things are not fulfilled more modifications or a redesign of the electronics and circuit-board is a must.

It might also be a bit tricky to solder the SMD LED without harming it, but with correct equipment it is quite doable.

The LEDs used in Bruce's project are already mounted on heat-spreading generic PCBs.

The component itself is the small 4x4mm iem in the center.

  • The LEDs are surface-mounted with big solder-pads on the underside.
  • These circuit boards often have an aluminium core to spread the heat and improve cooling.
  • The LEDs are sensitive to any over-temperatures as that will degrade their final efficiency permanently. That temperature is not that far above the soldering temperature.
  • The solder, circuit-board and component must get hot enough to make solder melt and make a good solder connection.
  • The LEDs are also sensitive to static electricity and should be handled accordingly.

With a simple soldering iron full success is unlikely.

Link to comment

Thank you for this additional information.

I think the best plan would probably be to let hire of the flashlight builders on Candle Power Forums to make a custom UV Led torch.

Link to comment

Thank you for this additional information.

I think the best plan would probably be to let hire of the flashlight builders on Candle Power Forums to make a custom UV Led torch.

 

If you abandon the need of a very neat, compact, elegant cylindrical flashlight design it becomes much easier.

 

Using one of the LEDs with heat-spreader PCB like the ones in Bruce's images above combined with a reflector and heatsink can give a lamp head.

Combine that with a suitable constant current DC-DC module and a few 18650-LIPO-cells.

Encase and connect to taste, with or without a cable, between lamp-head and battery/driver.

This concept can be scaled to bigger units with more powerful LED modules, if available.

 

I have a slowly progressing project for a bigger 370nm lamp with a 50W LED module built with that kind of structure.

Will share about it when it is closer to completion.

Link to comment

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...