Ping JL: High-Voltage White LEDs

I'm at home today, but I can dig out the part numbers tomorrow.

I usually tape the Digikey or whatever bag label to the back of the coin envelope, or stick it inside, to remember what's what. I hope I did that properly.

OK, thanks. That would be a big help. I tried searching for high voltage LEDs and came up with nothing. Any pointers would be very much appreciated.

Digikey lists hundreds

George H.

Here are the Digikey tags:

Thanks very much. I was wondering how these could have such a high forward

"

Sure enough, the forward voltage is listed as 21.5 to 24.5 V.

Near the top of the document, they include the phrase "Multi Junction Technology." I take that to mean there are a number of ordinary LEDs in series, since it doesn't seem reasonable they would be in parallel.

I figured that had to be the solution, so that solves my problem.

Thanks!

Thanks. I was specifically looking for the ones JL used.

You can look at a lot of these parts and see the multiple LED chips in series.

A lot of these multichip parts look like welding torches. They are blinding.

Nobody seems to make series strings from different-colored chips, which would be cool.

I was impressed with your statement that they are visible at 1uA. I have an application where it would be nice to use a LED as a photodiode as well as a light source. I'm interested to see if the multichip parts would be more effective than a plain LED. Time to get some. Thanks for your review on the light output.

I'm starting to see what appear to be large RGB outdoor displays that show pictures as well as text. No problem reading them in direct sunlight, so I assume they must be LEDs. I'll have to pay more attention next time and check with binoculars to see if I can figure out how they work.

A "white" LED consists of a deep blue emitter and some fluorescent material to convert that blue light to red and yellow.

I very much doubt that this process would work in reverse, converting white light to blue :-).

Of course, if you get a deep blue emitter without fluorescent material, this could work well for blue light.

RGB LEDs might work well, if the different colours are made with different fluorescent materials to generate the different colours.

Series-connected photodiodes don't work especially well. Because a photodiode is a pretty good current source, you get only the photocurrent of the least-illuminated or least-sensitive diode, with a lot of weird dynamics due to the nonlinear capacitances of the others.

Cheers

Phil Hobbs

I was going to say that!

Thanks for the reply. I had already figured out that since the photodiode acts as a current source, putting them in series would not increase the output. I did not know about the effect of the nonlinear capacitances, which would definitely screw things up. Thanks for the additional information.

However, there is a more fundamental problem. The white led is actually a blue or violet led with a phosphor coating to convert the led output to

clearly in Fig 1. Color Spectrum, on page 4 of

"

The spectrum shows a peak at about 440nm, which is blue/violet according to Wikipedia:

The led as a photodiode responds only to wavelengths equal to or shorter than the predominant wavelength it emits:

This means the led would require a blue or violet light to produce any output. In addition, the phosphor coating would block the incoming light and convert any blue to white, so very little would reach the led. My conclusion is this whole idea is doomed to failure.

Overall, it was a complete success. I found a bunch of reasons why something won't work!

Well, don't give up: try it!

You will (probably) get some photodiode effect from a series HV LED.

The photovoltaic output of the series diodes will ADD!

I just tried a single Cree T1-3/4 white LED, connected to a (very) high impedance voltmeter. I got 2.2 volts from an LED flashlight and the same from an incandescent mini-maglite, both up close.

The HV LEDs tend to have parallel ESD zener diodes, which could mess up using them as photodiodes.

I don't know about blue-on-blue LEDs, but Forest Mims used the reds (or greens or yellows?) for both send and receive.

LEDs' shinning on each other generate rather small photocurrents, I wonder if you could heat or cool one to move the spectrum a little.

Get a RGB string of LED's and just use the reds as the sensors.

(why do people give up so easily?)

George H.

Thanks, but I have to switch them: 100us on, 1khz rate or faster. Photovoltaic won't work in this application.

The original concept was white on white. Due to the low photocurrent of the series junctions, blue/violet requirement, and the phosphor downconversion, the photocurrent output would be so low it's not even worth trying. A regular photodiode would be much better, and it would accept a much broader range of wavelengths.

The low cost and high output of the multijunction leds would be ideal in combination with a plain photodiode. For example, the Seoul SAWFS72A- T2/U1-GA is only $0.23CAD in quantity:

"

I assume the multijunction led would have no problems switching at 100us on, with a 1kHz rep rate. That may turn out to be optimistic but it's worth trying.

No problem. The Cree white LEDs go on/off in nanoseconds. The down-conversion phosphors seem to be very fast.

That's good news. Thanks.

Just out of curiosity, how do you measure the response time?

Just pulse them and aim them at a fast photodiode. We have a PD+Scope combination that's good to about 25 GHz.

I bought a bunch of assorted colors of glow-in-the-dark paint on ebay, and lit them up with a pulsed uv (360 nm) LED. I was surprised that they were all fast.

I used my trusty PH200 photodetector, and the phosphors seemed to be a lot faster than the pd, a couple hundred ns indicated. The light from the phosphors here wouldn't couple into our really fast detectors.