UHF laser modulation

We've been doing a lot with GaN fets lately. Cool parts. I can imagine a simple servo loop.

How much current do you want to divert? At what frequency? Maybe a MMIC would work.

A distributed amp might do, if cost is no big deal.

Take a look at SY88022AL. It's gorgeous and fairly cheap. And crazy fast. I use it to make clean fast variable-amplitude pulses to drive distributd amps. I couldn't find any other way to do it.

It would dissipate more power than a shunt GaN fet.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Not a bad part, especially for $10. (That one is obsolete, but it has surviving relatives.) Way too low power for this use, unfortunately. I need a good amp of AC.

I'm thinking a high-side series GaN FET driven by a transformer, like this:

IbiasDC (very very quiet)

| | GND GND

with the laser bypass chosen so that the current doesn't drop too far below threshold in the OFF periods, and the Ibias bypass big enough to keep the bias supply running normally without destroying its sub-Poissonian character below 1 MHz.

This approach saves dissipation, which is important for satellite use. (There are terrestrial applications too, but its space that's paying the bills at the moment.)

Cheers

Phil Hobbs

You might consider an RF technique, like a series RC or tuned circuit or something, matched into the laser diode. From a GaN fet running at low voltage maybe.

That could be low power, which probably matters in space.

MMICS waste a lot of power, as do diffamps.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

My hope is to make a little tile with an EPC8004 (52 pF Cin) resonated with a few nanohenries of inductance, and drive it at a Q of around 3. The voltage gain of the FET will be less than 1, so Mr. Miller shouldn't be a big problem.

I just don't have a lot of experience doing 1-ns stuff with GaN FETs.

Cheers

Phil Hobbs

Take a look at EPC2037. Cin is 14 pF and reverse transfer is zero.

If you use a bit of drain voltage, some sort of RF matching thingie will make current gain.

Are you using a single laser junction or one of those bar stackup things? Namely, what sort of dynamic impedance do you expect?

I have an intern doing a small 100 amp laser driver now and there's another project here for a 1500 amp 250v driver. The bar lasers are getting up into welding territory.

It may be better to use a push-pull architecture and have the base of that at an elevated voltage where the laser diode just barely does not yet turn on. That way you don't have to muscle through its capacitance every time.

It'll be a lot of experimenting. GaN aren't the only game in town. LDMOS sometimes come in pairs but it can be hard to find them for such low power. Assuming you need pulses here and there you'd drive them through a small RF transformer so it becomes push-pull despite the common source architecture. Like an RF power amp. Instead of GND, the base of the secondary of the transformer which is connected to the laser diode could be at an elevated DC level so the laser diode remains somewhat pre-biased.

someone snipped-for-privacy@example.com wrote: <snip AI crapola>

Have you ever built an instrument based on a diode laser? I gather that you haven’t. Your average lowish power diode will start mode hopping at around 1 ppm of feedback into the cavity.

That level is difficult to avoid in a compact focused beam system.

Cheers

Phil Hobbs

Cheers

Phil Hobbs

If you modulate the laser current, does that cause problems? Will it hop in and out of laser modes? How much modulation did you have in mind?

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Ideally from full output to just below threshold. The terminal voltage changes by only about 1V out of 6.3V or so.

The idea is to make the oscillation build up from noise on each pulse, so that there’s no phase memory to allow instability.

Doing this puts a gigantic chip on the laser frequency, which makes the linewidth ~1 THz rather than the 1 GHz you’d expect from simple AM.

Cheers

Phil Hobbs

I suspected it might be a moding problem. How does AM make it better?

That's pretty clever. Seems impossibly microscopic though.

Right, the notion would be to choose the capacitor values and operating frequency such that the bias current is stable at low frequency to control the noise down there, but drops below threshold long enough that the oscillation has to build up from noise on each cycle.

Because of the hideously wide chirp, that last might not even matter—it’s an analogy to old oscillator-keyed radars, but the radar oscillator has stable circuit constants, which the laser really doesn’t.

Cheers

Phil Hobbs