another wasted attenuator

What's your proposed alternative?

Your problem is that you want your attentuators to operate at power levels they weren't designed to handle. A short, very high current pulse delivers a moderate amount of energy in a very short time, which is a spectacular amount of power. The limited area that has to handle it gets blown away.

Laser trimming exploit a similar process to blow away limited areas of resistive material. The laser pulse is brief, so all the energy is concentrated in the area under the beam, and evaporates the surface of the target area, without getting the substrate hot enough to damage it.

The risk with that is that you put enough power into the water to produce a lot of steam in a very small volume, you might create a supersonic shock wave. If it blows the glass tube apart, the shards could be dangerous.

There are the huge pulses you can imagine, and the risk of even larger pulses which you have failed to imagine.

how is it not a pi?

We're making a high voltage pulse generator. It's making fast 650 volt pulses now, and we expect 1KV when we get our new transformer winding next week.

Even at low avearge power, it eats attenuators. We may have to make our own.

I can buy suitable attens, for $1500 to $5K each.

So the reason you need 6Ghz BW is just to preserve those fast edges? Because obviously it's that high BW that's making these things expensive. Esp high BW combined with high power.

We want to be able to measure, say, 1 ns rise times accurately and make pretty pictures of our lovely pulses without the atten adding ringing. Realistically, 2 GHz would be OK.

Here's the guts of a VAT-20.

Thickfilms are (literally) flakey enough, and then the laser trimming makes them much worse. If we do our own attens, they will have to be untrimmed thin films.

The atten I have in mind would be an astounding numerical problem to design. I'll use an intern.

Well that should bring the price down somewhat anyway.

Yeah, I've no idea what that might be for (if anything). At any rate, the internals of that one look much better manufactured than the first pix you posted on this topic.

They become a serious PITA as you go higher in frequency and power. I wish you the best of British luck! Please keep us informed how progress goes. I suspect it might prove a bit too challenging for the average intern, though.

The gold dot might be a manufacturing test point.

Could you use a coaxial transmission line transformer to reduce the voltage before a more conventional attenuator?

John

I suspect so. It might make the trimming much easier.

I guess, if we could trust it. That would be another whole project.

I have a kid who is nothing like average. But he goes back to school in two weeks. We're discouraging him from doing that.

Transmission line transformers are remarkably predictable and consistent. They don't dissipate much power, so the termination might still be problem, but you might be able to split the signal on the way down.

Setting them up to do step-down is tricky. You can't fiddle with the number of turns, but rather use several transformers in series to get whole number reduction ratios.

Matick R.E. "Tranmission line pulse transformers - theory and applictions" Proc.IEEE. 56 pages 47-62 (1968)

spells it out in some detail. Getting down from 1kV might take a few stages.

Would one of these do the job?

Yes, if it was still good. The price is not much below that of a new one.

Interesting that esdemc is selling a lot of "brand new" stuff on ebay.