Looking for a well-collimated light, which I think means a very small emitter. I don’t want a flashlight to use by hand, I’ll be taking it apart and using it in a light source for an instrument which is basically a spectroscope. It would be nice to have a wide and smooth output spectrum, which probably most people don’t care about, but I’m thinking “high CRI” should work.
This won’t run off batteries and I only need the light head part, but I thought an existing flashlight design is the easiest and cheapest way to get what I want. I tried buying a bare reflector, but now it looks like a fair amount of machining to make a stable and complete light source out of it, which an off the shelf flashlight would already have done. I was looking for something roughly like the 2-AA “de.power” DP-014AA-C claiming 263 lumen, a light that I got from Amazon a while ago, but I didn’t want to cut this one apart. Wanted to buy another, but Amazon says it’s unavailable. As long as I’m looking, thought I’d see if there’s something with higher CRI. I don’t actually know what the CRI rating on this one is; I have a plot of its spectrum, but I don’t know the formula for calculating CRI from that. I was not thinking of any enormous spotlight to get a small divergence angle; I was thinking a small head as I’m using another condenser anyway, to refocus the light to a small spot. I’m just after the specs and I don’t care about specific brands. Any ideas?
Edit: now that I search for high-CRI flashlight it appears the best ones (eg CRI 95) use several different LEDs to get the broad spectrum. Is there such a thing as a single-LED, high-CRI light? Or at least if they use multiple die, if they’re mounted close together in the same package, so it could be focused down to more-or-less the same spot?
Edit 2: I realize the modern lights apparently have all kinds of different power modes, blinking etc. I was thinking I would just apply external power, and it would turn on in a consistent way, but if there’s a smart controller chip it sounds like things get more complex. Am I going to need a digital interface to even turn it on and have it come up to the same brightness each time?
Once it is out all you’d have to do is provide proper heat sinking and the right amount of power to the LED, e.g. via a bench power supply.
If you are even modestly good with soldering you can reflow almost any kind of LED to a fitting MCPCB on a hotplate without the need of disassembling a flashlight - there are shops that sell all kinds of MCPCBs and LEDs.
Please elaborate on what kind of colour temperature, tint and brightness you are looking for.
What kind of power supply will you be using? I highly doubt a typical flashlight driver would be compatible with it, both electronically and mechanically, so you probably don’t actually want a flashlight head. I’d suggest finding a light that already works, and disassembling it to replace the LED with a high-CRI one.
There are plenty of single-die 95+ CRI LEDs, the ones with the best light collimation include SST20 (4000K and below are 95+), SFT40 (4000K or below), and 219F. We can help you narrow down further if you tell us more about your color temperature preferences, and what type of optical setup you’re using. (Some optics are poorly designed and produce rainbow-colored beams, defeating the point of high CRI.)
If you post a picture of the spectrogram, some of us here (myself included) can eyeball a CRI estimate. LEDs with the same CCT and CRI tend to produce very consistently similar spectra.
Small light source, high CRI, will not be powered by batteries, and flat spectrum…, and cheap? Just use a halogen bulb. The output spectrum of a LED, even a high CRI one, isn’t exactly flat (cyan dip, blue spike), while a halogen bulb output spectrum is very flat.
Thanks for the reply. I’m not that worried about the LED driver as I’ve done other projects that needed those. Mostly I wanted to get the mechanical assembly of the LED and the reflector in a nice package to get the tightest, most collimated beam out of it without too much effort.
I don’t have a color temperature in mind, but in terms of the output spectrum I was hoping for more output in the 450-500 nm range than the light I mentioned has. Its single LED has a peak at 440 nm (deep blue) and 536 nm (green) and a big dip in the blue-cyan range of 450-500 nm.
I don’t know if I can upload images but if it works, here’s its spectrum:
Right now what I’m doing to fill in the light spectrum is to mix together beams from several different LEDs at a diffuser, and maybe that’s still the only way to improve CRI. But I thought I’d check to see what’s available.
Yes, old-school tungsten seems to be the simplest thing, and in fact I’m using the head from an old AA Maglite (with its tiny halogen bulb) to get that nice smooth spectrum, although lacking in blue but I’m adding in one blue LED. The annoying thing is even with a constant current driver, the bulb output drifts a fair bit right after turnon, and its never super stable.
If you want the most collimated beam possible, have you looked into TIR lenses instead of a reflector. Those offer the least spill and most concentrated beam. Take for instance:
To find the light/setup you need we need to work backwards from the emitter given how specific your needs are, which narrows down options tremendously. The cyan dip is a universal problem for blue-pumped LEDs, so to eliminate it completely you have 2 options: an incandescent source or a violet-pumped LED. But very high CRI blue-pumped LEDs do address the cyan gap partially, so maybe it’s ok for your usage. Here are some typical spectra that the current best blue-pumped LEDs can offer.
Well-collimated beams only come from large optical elements, so it’s hard to decide what to recommend you without knowing the remainder of your optical setup. But if you are adding a condenser later on, you probably don’t want a secondary optic at all as it would not play nicely with a condenser, which I understand is optimized for point sources, i.e., the bare emitter.
The spectrum looks solidly sub-70 CRI, I’d say 60-70 with high probability.
Thanks for that link! I will have to see if those are conveniently available in small quantity. I’ve used a number of such lenses before but the ones I used did not achieve such small beamwidths, I didn’t see 10 degrees and I certainly did not see sub-5 degrees, as their chart indicates is possible with some devices.
TIR versus reflector only changes the beam profile; with all other variables fixed, the intensity (which is a good metric for how well-collimated a light beam is) depends, to a first-order approximation, only on the effective diameter of the optical element, not the type of optical element.
These charts fail to account for many aspects of the interaction between LED and optical element; I would consider claims for any specific angle to be misinformation. But knowing the optic diameter and emitter size, and assuming uniform beam intensity and perfect collection efficiency, it is possible to estimate the angle of the beam.
Thanks for that link to the page with CRI measurements. Some of those look pretty decent, if I’m understanding those plots. Are such high-CRI devices actually available, in small quantity?
Gaggione offers more information than other optical manufacturers that I know of and specifies their data for different LEDs. I only offer this idea as a lay person here as an idea. I have never seen this kind of data on a reflector as there are WAY more variables in reflectors no?
Each lens is less than 3 Euro and shipping to the states in 13$ with no minimum. This was just an example BTW, I have no idea if these fit your use case. There are many high CRI flashlights with very narrow beams. Typically these are referred to as “throwers”.
Short arc xenon is another option, it doesn’t provide an exactly flat output like the halogen bulb, but the spectrum is more full compared to the LED spectrum, and it goes from 350nm all the way to 800nm with a relatively flat spectrum.
Edit: it is 100 CRI btw, just like the halogen bulb.
It is true that there is virtually no such data on reflectors, but it is difficult to compare the difficulty of estimation. Performing this sort of estimation for any optical element other than a convex lens is in general very difficult if the LED used is not on the list, in that case one might be better off doing some back-of-the-envelope calculations to estimate the beam angle.
EDIT: if you have reliable measurements for the radius R of the optical element (in mm) and light-emitting area A of the LED (in mm^2, sometimes can be found in LED tests on this forum), a first-order estimate for the beam angle (in degrees) can be given by
\frac{360}{\pi^{3/2}} \frac{\sqrt{A}}{R}
assuming uniform beam intensity and perfect collection efficiency. For a reflector the actual beam will be narrower due to reduced collection efficiency.
