Multi-conductor transmission line impedance

No, that's (I think) the supply voltage into the output stage. The laser voltage is much lower.

Of course now that John showed this pic we are all squirming in our seats, dying to know what they are going to do with a 200psec light pulse. That application must be some really hot stuff.

This will be the seed laser of a multi-stage pumped-fiber laser that will put out scads of power for some application that the customer isn't too explicit about. Tattoo removal? Bowling trophies? Asteroid destruction? We don't know.

be

heal

er.

uess

-)

Mosquito gun?

Cheers

Phil Hobbs

Ok, I meant with catalog parts from Digikey and total BOM cost of under two bucks for the modulator :-)

One issue with IC solutions for this domain is that product lifetimes are often short. Sometimes only a few years and then they go unobtanium.

When I liven in New Orleans, I wanted to make a ceiling fan that also swept a low-power laser beam around. If it detected a mosquito, it would zap it with a big pulse. I think someone has done it now.

I moved to San Francisco, which rarely has mosquitoes, so never pursued the idea.

(Send this already but it didn't show up--apologies if it appears twice.)

The problem with John's one is that the capacitance is so huge. Since the pulse is relatively small, only 32 mW peak, I have no idea why the customer specified such a monster. I just measured a 50 mW CW diode (780 nm) at 120 pF at zero bias, 260 pF at full output. I was expecting something in the 200 mW class for that much capacitance. Maybe it's a weird wavelength, or they're derating the laser a lot for reliability.

Trying to get a nice pulse shape with a big nonlinear capacitance and an even more nonlinear load admittance is hard, but on the other hand, a nanohenry or so of bond wire capacitance is just about right for the job.

TO-can lasers are miserable for this, but (as John complained) not as bad as butterfly packages.

Cheers

Phil Hobbs

Of course, a modulator requires a CW laser, and throwing away 99.99% of its power. I suppose you could make, say, a 2 ns laser pulse and slice some picoseconds out of that... if you didn't mind some of the wider pulse leaking through... plus the cost of the EOM and *its* driver.

Why is the world so complicated?

$2 is ambitious. You might could get there with avalanche transistors, probably hand-selected.

Welcome to the 21st century!

That jibes with what I've found out recently, 700pF would be suitable for a 200mW laser diode. At zero bias though. Do you know how that is when you get closer to Vf, to where is begins to lase?

I assume you meant inductance there :-)

Well, what else is there? The coaxial package which is in essence an elongated TO-can with a fiber coming out the far end is, RF-wise, the best I've encountere so far. Oh, and I also hat butterfly packages. From an RF point of view they pretty much managed to pick the worst case possible for a "standard". They probably forgot to invite an EE to the table.

Ask the IRS. They can tell you, they are able to make an ever so simple process complicated :-)

Hand-selection is usually not desired and I don't like avalanching because of the PRF phase noise that causes. Some day I'll have to try one with an SRD. Anyhow, I haven't tallied up mine yet but I am fairly sure it is under two bucks in production qties. I mean, if Trader Joe's can do it with wine we ought to as well.

It was already that way in the 20th century. There is a surprising percentage of EEs who are highly uncomfortable with transistor-level board design and then they use chips for just about anything. Right now I am trying to persuade one not to do that but it seems I am not getting anywhere at this time :-(

It seems to about double from the zero-bias value. Probably it's worse in InGaAs diodes, because the built-in voltage in the junction is lower due to the lower band gap. Threshold current is typically on the order of a quarter of the operating bias. John's building a seed laser for an amplifier string, so he needs the laser to be really really off in between pulses--Poff/Pon

Sorry, my stuff is so fast that even the inductors look capacitive. ;)

Telecom mostly uses external modulators. Direct modulation chirps the laser, which makes it scribble all over neighbouring channels. More recent lasers are much better for this, but AFAIK still not good enough. Besides, a modulator is a nice well behaved 50-ohm load. The fast ones are all travelling-wave devices--really gorgeous, zero chirp, phase matching temperature compensated, you name it. About $5k in onesies, unfortunately, but last time I heard, an OEM can get them for about $1k, and for less in volume.

I'm doing something fun today, measuring the time & frequency response of photodiodes as a function of voltage, wavelength, and position on the die, so I can pick the right ones for a 100 MHz laser noise canceller. Fun.

I discovered that a normal BPW34 photodiode leaks about 50 nA at 100 V of reverse bias. (It's rated for 60V abs max, but the capacitance doesn't bottom out until about 80, and it's important to run PDs really really fully depleted to get the best response.)

Cheers

Phil Hobbs

Ugh. I was hoping the increas would be a bit less. Mine are usually in the telco IR ranges.

Then he'll have to discharge the capacitance to probably less than 1/2 Vf to be sure there isn't even anything coming out below the lasing threshold.

But your new HP-3585 only goes to 40MHz :-)

Congrats on that auction success. Under $400 is a steal even with a busted reference. I have hardly ever seen them go much under $4k, for a good one from a used equipment place with a 30-day warranty or so.

In my line of work those kinds of Dollar amounts are often totally out of league, it's mostly production-ready stuff. $10 parts here and there could be justified if there really, really is no other way. But only then.

I'll have to finish up the module spec and do the CAD stuff for a design I just finished. That's where the fun phase is more or less over but, as John Wayne used to say, a man's gotta do what a man's gotta do.

The highlight of today will be a visit to the Alzheimer's place with our therapy dogs. The smile on those people's faces is worth more than any fun EE stuff could be.

Ah, one of those pedal-to-the-metal circuits. If it's bottoming out at

The customer loaned us the laser, which costs somewhere over $2K, and Jonathan was reluctant to drive it really hard from the prototype driver board, so didn't crank it up too much. This laser is around 700 pF, but other candidate parts are more like 250, which should make them easier to drive. We'd probably get more power out of a lower capacitance part, since a lot of our milliamps are being wasted on the capacitance.

When we think about adding series L here and there to crisp things up, we usually discover that we have too much L already!

Hmmm... 1 amp in 100 ps into 1 nH is... 10 volts!

We TDRd a few butterfly packages, and they were surprisingly good, a lot of C and not much L. We want to blow one up so we can crack the lid and see how they did the wire bonds. But at $2-3K each, everyone is waiting for someone else to do it.

The driver should be a hybrid, *inside* the butterfly package. But that would put me out of the driver board business, so forget I ever mentioned it.

Old telecom EOMs show up on ebay. I've been meaning to play with some.

These don't have very good extinction. One the NIF project, we used

At ballpark 2 ns, you can do it with SOT23 mosfets, for a few cents each. Another 12 cents will buy you a TinyLogic gate driver. At 200 ps, it gets trickier.

SRDs are a nice solution if you don't need amplitude or pulse width tunability. MA/com makes SOT23 SRDs that you can get from distributor stock for, like 70 cents, the only SRDs that are distributor items, as far as I know.

Sadly, the IC boys can put stuff on chips that we can't get, or use, as discrete parts. Take a look ADN2871. Cool chip, if we can find a use for it.

Can't you keep it pre-biased a bit? Maybe a few hundred millivolts below where you are absolutely certain that it's dark enough in non-lase mode? Every millivolt that doesn't have to be traveled through upon a rising edge eases the pain.

That is the same when pre-loading oneself with calories before a major hike or bike ride, looking at the belly that's already there.

Shhhhht! I think there's a lot of folks around here whose biz would suffer if that was available. But, it's like with all those fancy driver chips. After the first unobtanium situation nobody trust such boutique solutions any longer. Has to be catalog parts.

But be

nd heal

sses

I'd

that

etter.

I guess

? :-)

an

it

oard

f an

.highlandtechnology.com=A0 jlarkin at highlandtechnology dot com

Bill Gates has funded the development of such a gun for use in Africa to kill the malaria carrying mosquitoes.

nanohenry? capacitance?

hang your head in shame! :)