Really bright yellow LEDs?

We like the Osram led's, like LO A676-R1S2-24, Digikey 475-1041-1-ND. That's orange, 606 nm, but they have a yellow at 587. I don't know the power in mW, since they are rated in mcds and mlms, but they are blinding at 5 mA or so, easily visible at 1 uA.

What does 750 mlm/yellow mean in milliwatts?

Could you use an optical filter on a white LED? The yellow phosphor typically peaks around that range, though it's not a very sharp peak.

Best regards, Spehro Pefhany

I'm hoping to modulate the source wavelength by turning various LEDs on and off. That way I get wavelength resolution and an AC measurement more or less for free. I may need to put filters on each block of LEDs to narrow down their emission bandwidths, but I'd much rather just find a few reels of LEDs that work, and let the customer crank out sensors from those for some years. That may not be possible, of course.

The measurement needs a few ~3-10 nm bandpasses at particular wavelengths, which are determined by spectroscopic interferences.

Multispectral measurements of solid objects are full of nuisance data. I usually try to avoid them, but this one is a bit of an exception. I'm measuring hens' eggs, whose unit-to-unit variations are very much larger than any variation with time, so once it works it should work more or less forever.

Cheers

Phil Hobbs

In the green (540 THz), 1 watt gets you 683 lumens. (Which is almost certainly the largest prime number ever used for unit conversion, and the absurd nomenclature of photometry is just getting going at that point. But I digress.)

Assuming that the particular yellow hits the photopic curve at, say, 600 lm/W, 750 millilumens would be 0.75/600 W, i.e. 1.25 mW. I'd really like 100 mW at 577 nm, but I could use as many as 10 LEDs to get there.

Cheers

Phil Hobbs

Cree makes some serious illumination-type LEDs in colors. We were going to use their green one in a fixture to test free-space o/e converters, but the time-domain (pulsed) response was awful. Jonathan tested 14 different green LEDs before he found a Lite-On part that pulses well.

Probably not; what is surprising is that i understand that "white" is derived from "blue" and phosphor overlays, then "yellow" is another overlay on that. Maybe take the most efficient lower frequency blue and filter it for the yellow? Might be slightly better. Hell,one may even try red as some have strong blue bands. Use a DVD made by the pros for angle diffraction grating spectrum analyzer. Or...fuss with "orange"?

Interesting, thanks. I just need them to turn on and off at 20 kHz or so, so I'm not too worried about anything made out of direct bandgap semiconductors. It seems as though 575-580 is a lot harder to get than

Cheers

Phil Hobbs

High power LEDs only work for certain colors and that's not one of them.

Any chance you could create the right emissions from violet chips? The have WPE around 40% at intensities that are a fire hazard.

There are some part numbers dating from 2009 to 2012 listed at

Matt Roberds

These 3W (found via mouser) claim (page 17) to have an Amber bin 1 as low as 580-582.5, but it's not clear that you can easily order a single bin (some hand-waving about 4 adjacent bins...)

Then there's 570/575 greens like this surface mount 75mW:

It's a possibility. At that point I'd have to use multiple photodiodes and multiple filters, which I'd love to avoid if I can. My fond hope is that I can buy ten reels of carefully-chosen LEDs and set the customer up for life.

Thanks

Phil Hobbs

Good suggestion, thanks.

Cheers

Phil Hobbs

If you don't know them already, Roithner have a wide range of leds (and lasers).

Thanks. I sent them an email earlier this week and haven't heard back yet. A 578-nm laser would be awesome, but it'd need to be at least 100 mW.

Cheers

Phil Hobbs

The angular brightness curves invariably peak at 0 degrees. Real lensed LEDs fire wildly off-center from part to part. And the patterns usually have features, nothing like the beautiful smooth curves on the data sheets.

High power LEDs only work for certain colors and that's not one of them.

Any chance you could create the right emissions from violet chips? They have WPE around 40% at intensities that are a fire hazard.

It looks as though the right answer is probably to use white LEDs and interference filters. With three or four detectors, the duty cycle goes up and the required bandwidth goes down, so I can get by with less light. The Luxeon LXS8-PW40 is pretty amazing: 1360 lumens from an 8-mm diameter device.

With a 150-nm wide bandpass, I'll get something like 26 lumens in a 3-nm bandpass, but of course the efficiency will be lower, probably 50%. So two or three of those should be good enough.

The main issue will probably be angle-tuning of the filters, which will limit the acceptance angle of the receiver. It would be nice to be able to use a single lens on the RX side, which means that the filters have to go on the photodiodes.

An interesting question is what the optimal duty cycle should be for an on-off AC measurement--the "off" state should be stable for a long time, and with digital lock-in techniques, there's no particular reason to choose 50%. It seems as though the optimum should be near 100% if there's no systematic variation of the background, since one can average many "off" samples and get the noise in that state as low as one likes.

Cheers

Phil Hobbs

Kevin, where can I find these violet LEDs? Thanks.

The

Depends on how violet / ultraviolet you want. Both Cree and Nichia make them to less than 300 nm. Not terribly bright but at that wavelength they don't need to be, they get quite dangerous.

?-)