Lithium niobate

We sell a little Xtal ~10mm on a side, that we use to show the Pockels effect. (Electro-optic effect) High voltage is usualy required to do anything useful. What do you want to do? (We use ours in a Sagnac interferometer.) (Sorry for all the name dropping.)

George H.

Do you have a real optical table setup? I think the limits are set by the laser phase noise.. I'm sure Phil could write (another) book on this kind of thing.

Fiber optics doesn't need an optical table. You can just let the parts flop around anywhere! Just don't make any sudden moves that could rip the fibers out of things.

I want to make really short optical pulses. We can get down to about

I don't have a real customer for this yet. It's one of those build it and they will (maybe) come things.

It's amazing how easily a FOG can be made with off-the-shelf parts, but I think the devil is in the details (temperature, magnetic sensitivity etc. etc.).

A good telecom modulator is a thing of great beauty. 30-40 GHz bandwidth, nice clean ~12-15 ps rise times, zilch phase modulation. The ones I've used were DC coupled, with a slow bias input as well as the

Cheers

Phil Hobbs

OK, I really don't know very much about them. I know they are used in Q-switched lasers, and sometimes in Pockels cells. I used a pockels cell at vanderbilt, but I don't know what xtal was in the thing. The pockels cell did nothing as fast as 100ps. (You set the high voltage pulse length with a piece of coax.. so that says nano seconds to me.)

I don't know what telecom would use them for... nor how they could drive them with 5 volts... I thought it took kilo volts.

For fast pulses the laser jocks use mode locked Ti-sapphire lasers. I've watched a 'jock' tune one up... but never got to play myself.

George H.

ed text -

They are used to put the high speed NRZ data onto fibers. One problem with modulating a laser directly is that thay aren't as fast, and they chirp, which causes dispersion in long fiber runs. An EOM can modulate a CW laser output. The mach-zender is an interferance effect so doesn't need a lot of voltage; all we need to do is shift the light by a half wavelength, to get destructive/constructive interferance. Numbers like 3-9 volts p-p will usually do.

These EOMs are sometimes used in analog mode, like picking up wideband RF at the top of a tower and sending it over fiber to a receiver some distance away.

The modelocked lasers make absurdly narrow pulses, femtoseconds, but free-run at their own rate, numbers like 20-50 MHz maybe. EOMs are used for "pulse picking", rate reduction, in those systems. I think those units need kilovolts for some reason. I had a chance to bid on a driver, but I bailed... envision buckets of blown mosfets learning how to do that.

on

ng.

com

do

nac

Awesome! (got it) I think we've got x-cut LiNbO. We could use another mach-zender experiement.

George H.

ed text -

Telecom modulators are transverse Pockels cells, i.e. the bias voltage is applied at 90 degrees to the light propagation direction. Since they're single mode, the crystal is very skinny in that axis, so you can have a big E field with a smaller applied voltage, and the birefringence builds up along the length of the cell. That relies on the off-diagonal elements in the electro-optic tensor, but they can be pretty big if you cut the crystal right.

(I haven't actually done any calculations of that sort since I got out of Steve Harris's nonlinear optics class in 1985, but it's cool stuff and not particularly difficult if you start with the measured material properties. Calculating what those properties should be from first principles is a whole 'nother kettle of fish.)

They get the bandwidth by using a travelling wave geometry for the electrical signal as well, and tuning the propagation speed to match the speed of light in the niobate. Travelling wave is neat, because you don't have to drive all that capacitance at once--it just looks like 50 ohms to the driver.

The other cool thing is that folks have figured out how to tweak the crystal (cutting it at some weird angle, I expect) to get rid of the phase shift with birefringence. That means that in the propagating mode (the one that makes it through the polarizers) there's very little phase modulation. If that weren't so, the resulting FM would make the signal scribble all over neighbouring channels. (That's why it's called 'zero chirp'.)

They really are pretty amazing devices.

Cheers

Phil Hobbs

on

g.

om

do

nac

t -

Thanks for that,

We put a tranverse pockel in a polarized sagnac, and do a difference detection with another PBSC at the output. Free space HeNe laser so the xtal has to be big enough for students to steer it through.

George H.