Shadowgraph Research lamp made from several LEDs

Hello all,

I am a researcher at a US university, and my adviser has instructed me to come up with a solution to a lighting issue we're having. "Shadowgraph" is a technique used for high-speed imaging, gas and density-based research. For example, shadowgraphs are commonly used to view the blast front of an explosion, due to the change in air density associated with said blast front being visible in the shadowgraph images.

This technique requires a fairly significant amount of light from what is functionally a point-source, and definitely not a distributed series of point-sources. Typically, we use an arc-lamp with a series of lenses appropriate to the task at hand, giving us a relatively bright light, focused onto a finger-tip sized area. My adviser suggested I look into using the "100W LED panels" that Amazon sells, as we already have several of those laying around. (This one: https://www.amazon.com/dp/B00CZ75TWA/?tag=cpf0b6-20) One of these with an appropriate power supply is quite bright, significantly more so than an HID headlight, but it's not enough.

What I'm looking for is suggestions on how to combine several of these LED chips into a singular beam of light that functions as a point source, not as several notably distinct sources of light. I have 4 ideas so far:

1. An old attempt from a colleague of mine, strikingly similar to this old article from '06: http://spie.org/newsroom/0197-non-imaging-optics-combine-leds-into-one-bright-source I saw an older thread on this site from a decade ago discussing this same topic, and involving this article here: http://www.candlepowerforums.com/vb...mbining-multiple-LEDs-to-a-point-source/page2 My colleague assembled a simple 2-D dome shape out of acrylic sheets in 6 layers, flat at the base and covered in reflective mylar plastic. The "top" of the dome was flat and frosted, and light could be seen "seeping" out of the "sides" of the dome, likely due to not properly observing refraction math.

2. A far more expensive option of individually collecting the light from the LED chips into fiber-optic cables, perhaps requiring collimation first, then using a laser splitter the "wrong-way 'round" to merge the light beams into one, then de/focusing it to the correct diameter upon exiting the fiber-optic.

3. A far more machining-intensive design where the LED chips would not be mounted co-planar, but in a ring facing into each other, then using mirrors to aim the light "out" of the ring, perhaps with a reflector around that ring to help aim the light, then even an additional lens after that reflector.

4. I have seen the Lambda 421: https://www.sutter.com/IMAGING/lambda421.html and while I could possibly construct something "homemade" in a similar fasion, it would require collimation, the same drawback #2 has. I am at the moment unaware of any simple tools or strategies for collimating light that exist on the common market, so this certainly seems like a challenge and a drawback to me.

Any feedback you gentlemen could provide me on these ideas or your own LED combination strategies would be helpful in this pursuit.

How bright is a "fairly significant amount of light"?


Is there a maximum size requirement for the "single beam of light" or is it just important that it appears as a single beam?

Ozythemandias said:

How bright is a "fairly significant amount of light"?


Is there a maximum size requirement for the "single beam of light" or is it just important that it appears as a single beam?

Click to expand...

Our arc lamp is rated at 1000W, with an estimated efficiency of 30 lumens/W, meaning roughly 30,000 lumens. I can't give a much more precise answer than that, as the lamps aren't explicitly rated for lumens, and the efficiency rating is from a different company altogether. But perhaps one of you can get more out of their site than I could: https://www.newport.com/f/research-arc-lamp-sources-450-1000-w#page-top

My technical references say that the output light needs to be sharply focused onto the end of a silvered glass rod. I can easily go from collimated light to a focused point, so if the lamp-device puts out either a column or cone of "single beam light", that's perfect for our purposes.

I hope I was able to answer your questions adequately, thanks for your interest.

This isn't my particular branch of engineering, but the general rule of thumb says that what you want to do isn't possible. The rule of thumb goes like this: the maximum intensity will occur at the surface of the light emitting material and it isn't possible to (by purely optical means) to exceed that surface intensity.

Beam combiners throw away half of the light, for example. The other methods mentioned aren't so inefficient but no combinations of lenses or mirrors will do the trick.

For more explanation of how such a device would violate the laws of thermodynamics see https://what-if.xkcd.com/145/

parametrek said:

This isn't my particular branch of engineering, but the general rule of thumb says that what you want to do isn't possible. The rule of thumb goes like this: the maximum intensity will occur at the surface of the light emitting material and it isn't possible to (by purely optical means) to exceed that surface intensity.

Beam combiners throw away half of the light, for example. The other methods mentioned aren't so inefficient but no combinations of lenses or mirrors will do the trick.

For more explanation of how such a device would violate the laws of thermodynamics see https://what-if.xkcd.com/145/

Click to expand...

In other words, we CANNOT scale up an emission area in order to scale up intensity. I am interested in what you're implying when you say "by purely optical means".

parametrek said:

This isn't my particular branch of engineering, but the general rule of thumb says that what you want to do isn't possible. The rule of thumb goes like this: the maximum intensity will occur at the surface of the light emitting material and it isn't possible to (by purely optical means) to exceed that surface intensity.

Beam combiners throw away half of the light, for example. The other methods mentioned aren't so inefficient but no combinations of lenses or mirrors will do the trick.

For more explanation of how such a device would violate the laws of thermodynamics see https://what-if.xkcd.com/145/

Click to expand...

+1

All Xenon short-arc lamps have an extremely high luminance (coming from a small spot on the tip of the anode in the arc) and are very close to a point source. There is no LED that can match this. This means that for an LED with a very high luminance one would still need a much larger optical system.
Another problem is that the LEDs with the highest luminance are very small (Osram Black Flat Gen2, Cree XP-G2 de-domed) and have a much smaller lumens output compared to the 1000W Xenon bulb (around 700-1000 lumens compared to 30.000 from a much smaller source!).
Compared to these the 100W COB LEDs from Amazon which you linked to are extremely large, much brighter, but have a very, very low luminance. They are basically useless if one wants collimated light. No amount of optical trickery will change that.

UHP bulbs (used in projectors) are compareable to Xenon short-arcs. UHPs are basically super high pressure, low-wattage mercury short arc bulbs with a much longer lifetime (2000-10000h). They are also much smaller in size. Fitting them into a special optical system is dangerours though and requires a lot of work. Downsides are their less continous spectrum with bad color rendering and their long warm-up times. They also need to be modified in multiple ways to be used in non-projector applications.

Also compareable are laser phosphors sources used in BMW and AUDI car headlights and in high-end projectors. These have the potential to be better than most (but not all) Xenon short-arcs.

Thanks for the suggestions. We may simply end up having to move to a more powerful version of arc lamp.

I'm not sure lasers are what we want, unless we can get a laser that has several frequencies it will emit in simultaneously. That may be what the Lambda 421 was created for.

Thanks for all the replies.

What I meant with laser:
http://optics.org/news/4/11/17
A blue laser (or multiple ones) are focused onto a phosphor disc to create a concentrated point of white light which can the be focused with optics like an LED. It works because the converted light is emitted isotropically inside the phosphor.

We need more information to give you a proper response.

What exactly are you imaging? .. you said focused onto a fingertip area, but what is your depth of field for imaging? Is this something in free space that you are imaging? I have some basic understanding of the shadowgraph technique, but not enough specifics. What is your imaging system? ... is it direct projection onto an imaging device (given the sizes you mention).

As stated previously, thermodynamics "enforces" the concept of Etendue' ... essentially you can never increase surface brightness (of an emission source), but that does not mean you cannot increase power per unit area delivered. I.e. you can obviously shine 1, 2, or more lasers on a given surface and increase the power delivered per unit area. This does not violate Etendue'. If you think of your spot as the new "emission source", while the power is higher, all those lasers are going in a different direction, so the surface intensity is never higher than any given laser.

For the shadowgraph, from my understanding, you are not illuminating a surface and imaging the reflection off the surface (which could take advantage of multiple illumination points), but passing light through "something" and imaging the shadow. Essentially you want a collimated light source ... I think you used the analogy of "rod" of light ... .and it is that collimation that also gives you your focus in a shadowgraph?

Unfortunately a more powerful arc source may not improve your situation. Larger arc lamps do not necessarily have higher source brightness ... they just have larger arcs that make them harder to focus.

Laser pumped phosphor (as was suggested) can achieve source brightnesses on the level of arc lamps and optically may be easier to work with.

If you can use actual laser, possibly using a combiner to combine multiple wavelengths, then a beam expander, that will probably be the best.

combining multiple laser beams in order to make a "white" laser will have it's issues as well seeing that it's almost impossible to combine them close enough to appear as "one" beam. think of the old school Projection TVs, they always had the RGB shimmer on the outside edges. you MIGHT be able to work with an AR/KR laser which will provide 6-11 separate laser freqs in a single beam, but those lasers are getting harder and harder to find.

is "shadowgraphing" anything like a schlieren camera setup? if so, only a point light source will work due to the physics of light itself.

The shadowgraph system we use is as follows: A retroreflective screen is placed at one end of the test room, with a camera opposite. The test specimen is placed between the 2. A glass rod, roughly .5" diameter, cut at a 45* angle on one end and mirrored on the same end is mounted to the end of the camera lens, with the mirrored end pointing "out" of the camera. There is a light source (what we're discussing here) placed 90* from the camera, with the emitted light set to be reflected from that glass-rod. Obviously, the glass rod is angled so that the 45* angle will reflect the light into the test area.

From the camera's POV, light is being reflected from a small surface in front of the lens through the test area, retro-reflected back through the test area, and into the camera sensor.

WHAT we are imaging varies quite wildly, but usually has something to do with energetic materials. Our test setup will also vary from test to test. For the lab, we may have some consistency (and wiggle room) built into our general setup, but it requires us to focus the light for each test individually. So the specific output characteristics of lab-shadowgraph light aren't as important as it being a nearly-point source. This type of shadowgraph setup is relatively easy to deal with, and we don't actually need the arc-lamp (or anything nearly as powerful) for the lab tests. In fact, we'd like to NOT have to use it, as using it for extended periods of time will cause indoor sunburns!

For our field-shadowgraph, we DO make use of the arc-lamp, and rather inefficiently, I've been informed. Our lab techs purposefully remove the reflector in the lamp to ensure that reflected light doesn't become a second point source, and we usually have a lot of light spill between the emitter and any focusing devices we have between the light and the mirror. So, my earlier estimation of 30KLumens emitted (as shaky as that estimation is) means that only about 3-thousand lumens (a super rough guess of 10% effectively-directed light) are being effectively focused onto the rod-mirror. This makes the project a bit more approachable, and I've been looking into HID lights, but I'm not sure of their output ranges or IF they are able to act as point sources or nearly-point sources.

After speaking more to some of my lab-mates who I inherited this from, an optics engineer originally proposed a pinhole with tons of emitters behind it, meaning a rather inefficient but nearly-point source of light.

Shadowgraphing is related to schlieren setups, in that they both use optical trickery to observe changes in gas density. The schlieren setups in our lab make use of fairly low-level LEDs (1 LED, not a matrix of them), if that helps any.

Here's 2 papers that we use for reference around our lab:
http://infohost.nmt.edu/~mjh/Pubs/2009-Hargather-AO.pdf
http://infohost.nmt.edu/~mjh/Pubs/Settles2017.pdf

FRITZHID said:

is "shadowgraphing" anything like a schlieren camera setup? if so, only a point light source will work due to the physics of light itself.

Click to expand...

You don't need a point source, you need a collimated beam for schlieren, just like shadowgraph. It is similar.

In a lab environment, combining the beams is pretty easy, not cheap, but easy. It's a much simpler thing than a projection television where you have variable geometry due to magnetics, offsets of the tubes, optics, etc. With lasers, you are just aligning three simple lines. It is a given you would need an aperture to eliminate the outside of the beams.