Fiber-optical switch/multiplexer

Joerg · Sep 16, 2008

Ok, this is a borderline electrical and optical issue so I posted to
both groups. I am looking for optical "toggle" switches that are
controlled electronically (typically via a highish voltage). Needs to
tolerate 500W laser pulses in the IR range, 5-10uJ, average laser power,
is less than 100mW, needs >20dB isolation for the off-channel, must
switch within very few microseconds (that excludes a lot of them),
should be <$500 a pop. A larger 1: X multiplexer would be even nicer.

Many of the classical switches can't stomach the pulse power, others are
too slow. Basically it is something like this:

http://www.bostonati.com/Products_Nanona.html

Any ideas who else might make something similar?

bernado · Sep 16, 2008

George Herold · Sep 16, 2008

Ok, this is a borderline electrical and optical issue so I posted to
both groups. I am looking for optical "toggle" switches that are
controlled electronically (typically via a highish voltage). Needs to
tolerate 500W laser pulses in the IR range, 5-10uJ, average laser power,
is less than 100mW, needs >20dB isolation for the off-channel, must
switch within very few microseconds (that excludes a lot of them),
should be <$500 a pop. A larger 1: X multiplexer would be even nicer.

Many of the classical switches can't stomach the pulse power, others are
too slow. Basically it is something like this:

http://www.bostonati.com/Products_Nanona.html

Any ideas who else might make something similar?

We (meaning someone else aligned it) used Pockels Cells for pulse
picking of high power IR pulses. But this does not switch the pulse
to two different outputs. And the price tag is way out of line!! How
about AOM's (acoustic optic modulaltors). I've used these to switch
30 Watt CW lasers in under a microsecond. Not sure of the current
price tag or if they can handle high pulse powers. Or do you want
something that is fiber coupled?

George

Joerg · Sep 16, 2008

George said:
We (meaning someone else aligned it) used Pockels Cells for pulse
picking of high power IR pulses. But this does not switch the pulse
to two different outputs. And the price tag is way out of line!! How
about AOM's (acoustic optic modulaltors). I've used these to switch
30 Watt CW lasers in under a microsecond. Not sure of the current
price tag or if they can handle high pulse powers. Or do you want
something that is fiber coupled?

Like these?

http://www.crystaltechnology.com/products/Modulators.html

I am not an optics expert, just designing their electronics, but the
optics guru at this client to me that Lithium Niobate switches would
degrade or burn up at 500W peak power.

And yes, we must fiber-couple all this stuff.

George Herold · Sep 16, 2008

Like these?

http://www.crystaltechnology.com/products/Modulators.html

I am not an optics expert, just designing their electronics, but the
optics guru at this client to me that Lithium Niobate switches would
degrade or burn up at 500W peak power.

And yes, we must fiber-couple all this stuff.

- Hide quoted text -

- Show quoted text -

Like these?

http://www.crystaltechnology.com/products/Modulators.html

Yup, those look like AOM's. I noticed one with a peak power of 250 W/
mm^2. Now you need to know things like the aperture into the AOM and
what is your laser beam size. But these are not a fiber solution.

George

Joerg · Sep 18, 2008

Phil said:
AODs are medium-latency devices due to the acoustic propagation
delay--that few-us number isn't impossible, but it's easier if you have
a bit of pipelining happening....how far in advance do you know your
pulse pattern?

The pattern is completely know upon system start so if the latencies are
well defined and don't flop about it may be possible.

You can use one fibre in, a collimating lens, an AOD, another lens, and
a few fibres in a row for the output. There are people who'd be happy
to build that for you (OZ Optics springs to mind), but it wouldn't meet
the cost target unless you can amortize over, say, 8 outputs.

We will usually have more than a dozen outputs which must be served in
round robin fashion, just plain sequentially. We've been through pretty
much all domestic sources and the only viable one so far seems to be
BATI. But theirs is also still a bit in the early stages.

To do 1x8 that with ordinary fibre, you'd need about 100 resolvable
spots, since the fibre core is about 12 times smaller than the cladding
(Resolvable spots = scan range/spot diameter).

Matsushita makes cells like that--I've often used their EFLD-340, which
has 340 spots in the visible and about 180 at 1 um.

Right now we are at 1060nm but long term will change to longer and more
standard (meaning less $$) wavelengths.

LCDs have very high damage thresholds but are way too slow.

Another approach would be to use an electrooptical scanner, e.g. a wedge
of lithium niobate, to replace the AOD. The key is going to be to use
bulk optics for the power level, and amortize the cost over many channels.

The optical guy at the client told me that most lithium niobate
solutions wouldn't be able to handle the pulse power. The mechanical
solution we have right now works but it is iffy to push in to higher
rpm. It is already running at jet engine rpm.

George Herold · Sep 19, 2008

AODs are medium-latency devices due to the acoustic propagation
delay--that few-us number isn't impossible, but it's easier if you have
a bit of pipelining happening....how far in advance do you know your
pulse pattern?

You can use one fibre in, a collimating lens, an AOD, another lens, and
a few fibres in a row for the output. There are people who'd be happy
to build that for you (OZ Optics springs to mind), but it wouldn't meet
the cost target unless you can amortize over, say, 8 outputs.

To do 1x8 that with ordinary fibre, you'd need about 100 resolvable
spots, since the fibre core is about 12 times smaller than the cladding
(Resolvable spots = scan range/spot diameter).

Matsushita makes cells like that--I've often used their EFLD-340, which
has 340 spots in the visible and about 180 at 1 um.

LCDs have very high damage thresholds but are way too slow.

Another approach would be to use an electrooptical scanner, e.g. a wedge
of lithium niobate, to replace the AOD. The key is going to be to use
bulk optics for the power level, and amortize the cost over many channels..

Cheers,

Phil Hobbs

Cheers,

Phil Hobbs- Hide quoted text -

- Show quoted text -

Another approach would be to use an electrooptical scanner, e.g. a
wedge
of lithium niobate,

Hmm how does this work? (mostly I found patents on the web, or
journal articles that I couldn't access.) I've measured the Electro-
optic effect in Lithium Niobate, (change in the index of refraction
for one of the polarization directions with an applied E-field.) But
this is a pretty small effect. Does the scanner use the same
principle? It must be a real small change in angle.

George

Joerg · Sep 19, 2008

Phil said:
They're very very stable--better than 5 ps p-p by my actual
measurements--so no problem there. (They're probably better than that
even--my measurement was <0.1 degree at 60 MHz.)

You can do that with about 150 resolvable spots. You'll need a decent
lens on the RX side.

AO cells become more expensive at longer wavelengths, because the
diffraction efficiency goes as the phase shift across the crystal
thickness, which goes as the optical frequency (it's even a bit steeper
than that). Thus you lose efficiency the further you go into the IR.
Brimrose is the major supplier out there.

If you can use 820 nm instead, it'll get significantly cheaper.

I think he was talking about niobate waveguides, which are only ~10 um
across, vs. a bulk-optics modulator, which would be more like 1 cm.

Yes, that may be true. I will discuss it some more with him. But right
now I am trying to get to a simple design for a half bridge to swing
12V/1A with almost no losses. But my favorite motor drivers have all
gone lalaland and the MIC4451 ain't big enough. P/N channel follower has
too much crossover. Dang.

Joerg · Sep 19, 2008

Jim said:
I'm about to date myself... are there still Kerr cells ?

John Kerr discovered that at a time when the first singing wires were
strung in your neck of the woods. A few years later there was the famous
O.K. Corral shootout

Are you this old?

Joerg · Sep 20, 2008

Michael said:
What backup? It didn't have the capacity to back up. It just drifted
with the currents. ;-)

It did have a controller and needed no backup. But it was not an earthly
one ...

Joerg · Sep 20, 2008

Phil said:
Sure, but they need ~10 kV and contain nasty things like carbon
disulphide or nitrobenzene.

Yikes! And here I was fretting over having to generate fast 200V pulses.

Hey, anyone know a MOSFET driver with around 100-200mohm internal Rdson?
But not the big Infineon TO220 style, that loses too much efficiency at
low power. External p/n channel has too much shoot-through for this
circuit. Noise issues, walking on egg shells, the usual.

JÃ¼rgen Appel · Sep 21, 2008

Joerg wrote:

[Kerr Cells = 10 kV]

Yikes! And here I was fretting over having to generate fast 200V pulses.

200V is more like the half-wave voltage that is needed for 90Â° polarization
rotation in an average-sized electrooptical modulator.

Moore's Lobby Podcast

Menu

Categories

Platforms

Content

Connect With Us

Network

Fiber-optical switch/multiplexer

Fiber-optical switch/multiplexer

Joerg

bernado

George Herold

Joerg

George Herold

Joerg

George Herold

Joerg

Joerg

Joerg

Joerg

JÃ¼rgen Appel

Similar threads