Ping JL: High-Voltage White LEDs

Our fastest one is a New Focus fiberoptic o/e converter, internal battery, unamplified, 50 ohms. Our Tek 11801 sampling scopes can display a 1 mv signal with some averaging. We can use the New Focus for free-space light, if we have a lot of light.

The LED is a small source, very bright. The phosphors are diffuse and much less bright.

Osram SHF206 is a nice, fairly large-area TO92 silicon diode, spec'd at 20 ns.

SFH2400 is a smaller surface-mount part, spec'd for 5 ns.

I think Mouser has them.

First Sensor has some nice parts, too, especially their UV enhanced photodiodes.

Gain is cheap these days, but for really sensitive applications, you need to minimize the first amplifier noise or using a larger signal (e.g. larger area sensor) to get a decent SNR. It should not matter if you connect the sensors in series or parallel as long as long as you use an amplifier with proper input impedance.

What bandwidth do they spec? Do you know what the photodiode is? These have to be pretty common devices, probably used in OC-192 by the zillions.

Is New Focus related to the company LeCroy uses for their oe converters?

Are these the people?

Looks like an interesting web site to peruse.

OK, thanks. You have to have a completely different mind set to work with this stuff:)

Thanks for the nice collection. They would probably work fine.

I was sifting through the hammamatsu web site and came across the S5973-

This has been very enlightening. Thanks for your help.

Years ago I bought one of these,

~200 MHz. I stuck a minicircuits amp on top of it. (Good enough for what I wanted.)

Inside it's just a Battery (~10V) small bias resistor and 50 ohms to gnd. These days I might try and make my own. (Cheaper.)

George H.

Check out GPD. I got some 4 GHz InGaAs diodes from them for about $35 each. (60-um diameter though)

Cheers

Phil Hobbs

Hamamat$$$u!

Appointech has fast pd's too. The fast GaAs and InGaAs stuff responds mostly in the infrared.

I recently saw a detector design that started with 5 volts and then used a double c-multiplier (cribbed from one of Phil's designs!) to wind up with under 3 volts on the pd. Followed, of course, by the usual noisy opamp TI. It was really slow.

Try the same thing with a ruby (or one of those pink grindstones).

It should have about a millisecond falloff. You'll want the room light dimmed for this...

The glow-in-the-dark extreme is strontium aluminate. You can buy the paint on ebay. It glows for about 12 hours. You can put a panel of the stuff out in the sun, take it in at sunset, and have usable illumination for several hours.

Very good info. Thanks!

Good info. Thanks. I think this is the page:

I noticed that also. I wonder if most of the fast emitters are also in the infrared. Are there any fast ones in the visible or UV?

I agree. It is particularly good for the housing of emergency torches so that you can find them in the pitch dark after a power cut. In fact you can see by the light that the case emits once dark adapted!

This one is particularly impressive and fairly compact. I think the GloTorch plastic is made by 3M with a persistent green yellow glow.

Unfortunately though a brilliant and useful gadget it doesn't sell particularly well. It makes the old ZnS stuff look incredibly dim.

Halloween plastic tat is still mostly ZnS :(

LEDs of all colors can turn on and off in nanoseconds. Apparently similar parts can have wildly different junction capacitances, so it's best to test candidates and make sure they are compatible with your driver circuit.

Small LEDs can run from under 10 to over 100 pF.

Laser diodes can be really fast. We managed to tease a sub-100 ps optical pulse out of a fiber-coupled butterfly laser, pretty good considering that the leads were plugged into pin sockets.

Yup. GaAs and InGaAs have way higher electron mobility than silicon, and the photodiodes are also very much smaller. That's because (a) InGaAs is mostly used for fibre things, and (b) since tiny stoichiometric errors translate to pretty large doping, it's hard to make really "intrinsic" regions in compound semiconductors, making the capacitance per unit area much higher.

Cheers

Phil Hobbs

LEDs are much slower than laser diodes, because their turn-off is limited by spontaneous recombination rather than stimulated emission.

Commercial datacom modules do 10 Gb/s direct modulation on VCSELs, and have for awhile.

Cheers

Phil Hobbs

That's fast. Thanks!

I come across articles, perhaps on optical heterodyning, that claim jitter mesurements down in the femtoseconds. Other articles mention very poor jitter converting electrical signals to optical and back.

Do you have any feel for what types of high speed emitters and detectors have have the lowest jitter? I'm looking for bandwidths of 5GHz to 10GHz and coupling into optical fibre.

Optics is considerably more complicated than electronics in general, because there are a lot more things you have to worry about in any given case. Optical feedback on the order of 10**-6 (in power) can cause a diode laser to mode hop at a rate of ~ 100 kHz, which will mess up nearly any optical measurement you can name.

For those sorts of bandwidths, I'd be looking at CW diode sources with two- or three-stage Faraday isolators, and good quality zero-chirp telecom modulators. You'll find them only in the 1.5 um telecom band.

Cheers

Phil Hobbs