Building a blue light transilluminator (for viewing DNA)

Hi All,
So, This is not exactly a flashlight… but it is an LED light system. I’m a biologist, so you all know more than I regarding LEDs and how to use them, so…
(Mods, please move this if it’s in the wrong place. This was my best guess)

It’s been a while since I have built anything fun, so I decided to try my hand at making a good blue light DNA transilluminator. Commercial versions of these are available, but small/cheap ones run over $500, and ‘good’ ones run closer to $2000. These systems are effectively just LEDs with the proper wavelength mounted under a diffuser and a glass sheet, then a filter over the top. All scientific equipment is ridiculously marked up, but these seem particularly expensive for a box with lights in it.

Yes, I could just buy one. I am doing this more for fun than any real need. I thought this forum might have fun thinking, designing and testing this contraption!

The concept:
DNA is separated by size within an agarose gel matrix and stained with a fluorescent marker. To visualize the location and size of DNA fragments within the gel, it is placed on top of the transilluminator, and light of a specific wavelength is passed through the gel. The fluorescent marker is excited by shorter wavelengths, and emits a longer wavelength. The gel is viewed through a long pass filter allowing you to see the emission wavelengths of the marker while blocking the shorter wavelengths used for excitation. (The somewhat obsolete versions of these use UV (301nm) as an excitation wavelength end the markers emit green (603nm) light. These don’t need an additional filter for viewing (because we can’t see UV light) however, since UV is extremely damaging to DNA, using to visualize important samples is problematic. Blue light is non-damaging and the applicable markers are safer to handle).

Blue light transilluminators use a marker with an excitation range of about 450nm-550nm with a peak at about 497nm and an emission range of about 490-670nm with a peak around 520nm (
The “long pass filter” that is normally used is a piece of acrylic (color code #2422) which is very efficient at blocking everything below about 540nm (it’s just a piece of plexiglass, but it works really well for this purpose and costs about $10 sq/ft).

So, (finally) here’s what I am trying to build:

1) An LED panel able to provide bright light peaking around 490nm. What are your suggestions for specific LEDs to use in this panel? The illuminated surface will be about 8.25”x8.25” inches, so the panel should be slightly smaller than that. Are there inexpensive blank PCBs that big on which to solder a bunch of LEDs? (I can’t find anything suitable for sale without spending a lot on a custom PCB). Are there other options for building this panel? Shoot me suggestions!

2) A short pass filter to clean up the raw LED light allowing for nearly 0 transmittance through the long pass filter used above it. This is necessary unless the LED outputs very clean light with nothing above 540nm but most LEDs in this range output some higher wavelengths, even though their peaks are lower.

3) I already have the long pass acrylic filter. This component is sufficient, so I don’t really need suggestions for this (

I am decent at building electronics and soldering, but I really need suggestions on specific LEDs and/or filters to use for this purpose. My first attempt uses a cheap “ice blue” LED strip from amazon ($15) and a piece of acrylic #2114 ($10) as a short pass filter/diffuser. This setup results in a LOT of very, very blue light with great diffusion. When the long pass filter is placed on top, the blue is completely blocked (it looks black and dark -i.e. perfect). However, I think this setup produces closer to a 450nm peak. This MIGHT work (I have not tested it with actual DNA marker), but it’s not close to the optimal excitation wavelength, so it will not be very efficient, and will not work well for viewing very small amounts of DNA.

How do I get this rig to output closer to 470nm-490nm while still being completely blocked by the long pass filter and still be relatively inexpensive?

I’m in the process of building a spectrometer to test this contraption… Maybe that should have its own thread?

Again, let me reiterate, that this project is just for fun. If you aren’t here for a fun thinking/design challenge, please be nice anyway :slight_smile:

I’ll post some pictures soon to show what I have so far and as this project goes forward.

Let’s see what you’ve got!


1 Thank

Very interesting even if it’s over my head. Looking forward to seeing everyone responding!

I’m just about to go to sleep, so even though I read all the words, I need a distilled version of that. Is there a TLDR version of that? Like, “I need a bunch of 470nm LEDs?” or similar?

470nm is generally “sapphire blue”, but different weirdo blues like royal, ice, etc., are also out there.

Not sure I can be of much help but I took a quick look around at “cyan” emitters in that wavelength spread. Looks like most of them are pretty low power and cannot handle much current or heat, so the design will need to accommodate that carefully with the driver, heat sinking, and perhaps a cooling fan. Osram has one that looks like perhaps the best contender as it has a wide viewing angle and is a little more robust (this is the 490nm peak…they may have other versions available as well…and datasheet is linked on this page for convenience): GC VJLPE1.13-KULR-36-1-700-R18 OSRAM Opto (ams OSRAM) | Optoelectronics | DigiKey

I don’t know about premade mcpcb’s but I think the first step is to try and decide how many emitters will be needed to give the output and spread that you want…may need more running them in open “mule” fashion like this with the added layers of light-reducing covers/filters, compared to a reflector/optic like we use in flashlights (I am assuming in this case that it would be desired to omit any optics that cover the emitter, since that would add an additional factor to light reduction and likely affect wavelength). Once you know that then you can see if any boards match your needs (need to pay attention to series/parallel of course, compatibility of the smd footprint for soldering, number of mounts, etc……might be simpler to just use single-emitter stars and place them around the platform as needed?)….and then choose your power source and configuration.

These are quite tiny chips…1mm or so. I see some other emitters in that range from Lumileds, DC Components, and Broadcom but I know nothing about any of those…nor this particular Osram really, but we know from many of Osram’s other emitters that they are generally pretty well made and consistent.

Here’s another possibility on Aliexpress….again, know nothing about the emitter (and although I’ve never used them, I’ve heard that this particular store is a decent one): There are lots of similar stores on Aliexpress that deal with “general led” stuff as opposed to flashlight-centric parts, and they’re more likely to have the range of mcpcb’s and perhaps drivers that you’ll want (maybe).

Another supplier there that may be worth considering is Tangsinuo. They specialize in filters and lenses, seem like a pretty professional outfit (that cannot be said for the majority of Aliexpress sellers….). If they don’t have a listing for something you want, it’d be well worth messaging them to see if they have it/can get it. For larger pieces it might still be price competitive with what you can get domestically even if they need a higher shipping cost for protection.

Also, you may find this other recent thread of interest…biologist exploring biofluorescence in the deep blue 450nm range: Help Needed: Cree XT-E Build for Biofluoroscopy

1 Thank

A quick Google search for “480nm led strip” came up with this

Oh yeah, and you do know you can get used/non-working transilluminators on eBay for $100 or less sometimes. They are mostly UV but a good way to get a large sized filter and box.

1 Thank

Oh, hey, if an assload of 5mm LEDs are an option and you don’t want to go the chip-resistor route, there’s a way to do that very easily with 0.1” perfboard. Lemme know…

Thanks for the help so far! I wasn’t actually sure anyone was going to respond to this post at all, haha. Thank you Lightbringer, Correllux and Scientist for your suggestions.

First, I’ll address the fact that my original post was long and complicated… So the abbreviated version:

I need a light source / filter combination that produces a large illuminated area (at least 8”x8”) with bright light between 450nm and about 520nm, and specifically does NOT produce light above 540nm. This can be accomplished by either LEDs alone or by filters placed over LEDs.

Digging deeper:
I have realized that this project REQUIRES either a ton of trial and error, or a spectrometer. I opt for the spectrometer, so I’m building one. (Similar to: GitHub - leswright1977/PySpectrometer: Raspberry Pi Spectrometer) I’m making one of these because a commercial spectrometer is $1500 and I don’t think I need that kind of accuracy anyway.

I already have the spectroscope ($20), but it’s not hooked to a camera/software yet. I will definitely post some pics of various spectra once I can take pictures with this!

According to my Spectrascope, my “Ice blue” LED strip has a spike around 450nm and broad emission all the way up to over 700nm. So it’s basically “white with some blue mixed in”.

I’m getting another strip tomorrow that is marketed as “490nm”. I think this is the same product that Scientist recommended! We will see how close it comes!

The good news is that, my filters seem to be doing a great job so far! The blue acrylic #2114 shows a bright blue band, and faintly into green… So possibly exactly what I want. Other blue acryilic, like #2424, fails to block much green, and even lets some red through, so that won’t work. (Again, spectra will be posted for these when I am able to record it, sorry it’s all text for now).

@Lightbringer - I think my options are narrowed down to either a) cheap led strip lights or b) “an assload of 5mm LEDs”. If you have an efficient method for assembling a board this size, I wanna hear it! Sounds like a lot of soldering, but I’m game.

Wellp, when I made a dome-light for my car with 5mm LEDs, I decided on 3 LEDs in series, and a ballast resistor for each string. The entire light was multiple “clumps” of 6 LEDs, multiple clumps in parallel.

So you stick the LEDs in for each clump With a snip, you trim down the legs and fold them over to hold the LED in place and solder them together.

All the negative ends get folded over sideways to form a negative bus. All the positives on the other side a positive bus. Now, the 2 resistors go in the gaps between the LEDs, one for the “left string” in the clump, and the other in the “right string” in the clump.

Point is, other than connecting to power, there’s zero wiring, just snipping, folding, and soldering. And everything’s pretty much flush on the solder side.

If you can figure out by the word description, great, as I’m not sure how to “draw” in ascii, but I think you can figure out what I mean. If not, lemme know.

Anyhoo, if you want a huge array of over 100 LEDs, it’s just a matter of “scaling up”.

This doesn’t sound too difficult, however time consuming it might be to solder a few hundred LEDs. However, I can’t find perfboard that big (8x8” to 9x9”).
Anyone have any idea where to source this?

Another good option is what was mentioned by Correllux is to stick a bunch of single emitter stars to a panel and connect them all. This could get expensive, but would be easy, and could result in a really good board with heat sinking (if they were stuck to an aluminum panel, or even a huge finned heat sink).

Either of these options leaves me with a HUGE number of options for LEDs (either 5mm or SMD). If I go with either of these routes, I will have to spend some time buying and testing samples before I build the whole thing.

Honestly, if the LED strip light panel works (something like this: - YouTube), that is my first option. Easy to build, easy to repair, cheap, and low heat production. I’ll see what the spectrum on my “490nm” LED strip actually looks like today…

Here is my first pics of some of the emission or pass-through spectra I have been talking about. Please excuse the very poor quality of these, I took them with my phone through my tiny spectroscope. There is some blue halo effect around them, but these are artifacts - Ignore any color that isn't in the right place :P

Also note that there is no wavelength nm scale here - I will eventually get a camera and software that can measure and quantify these spectra, but this is what I have right now... I tried to align them, but it's not perfect.

Note that the "Ice Blue" LEDs are in fact white LEDs with a bright band in the blue region. They emit greens and reds as well as blue. The bottom two spectra are the LEDs with the filters I plan to use.

Also note that there is very little overlap in the pass-through wavelengths. When you put these filters together and look directly at the sun there is a slight green pass-through (~550nm? not pictured), but I doubt there is enough to be seen from LEDs, especially if the LEDs emit little green light as-is. I think this filter set will work just fine once I pick LEDs and build a panel.

This sounds interesting, I’m going to keep an eye on this thread :slight_smile:

Is assload an SI unit?

The filters look good. Hope you can find the right leds. I too will be following.

Naw, it’s a scalar.

490nm assload:

3 Thanks

Haha. I laughed at this much harder than I probably should have. Well done.

Any development or success with this cool project?

A blue Cree XP-E2 is a readily available emitter which peaks around between 465nm-480nm, and if you look at page 15 of the datasheet here: it shows output stops right around 540nm.
(Wavelength spec is on page 11).

You can get 3 of them ready mounted on a 20mm triple board from KaiDomain here: .

@Sirtalis is there any conclusion to your project? I’m in the process of making this exact same thing. I used cheap blue led strips side mounted in the box instead of under the gel. Works ok but I’ve yet to find the exact blue that works perfectly.

Welcome to BLF ckafer! Thanks for bumping this thread, what an interesting project!

Cree advertise their XEG LEDs as having the most colours to choose from however they’re a fiddly little non-standard footprint, and boards are pretty hard to come by (lots of them are also phosphor converted, so are less “clean” than monochromatic LEDs).

When you say you’ve yet to find the perfect blue- what have you tried, and how has it let you down? Maybe some others reading this thread can chime in with suggestions for you :slightly_smiling_face:.

1 Thank

Have a pleasant time at this friendly forum, ckafer!