Rectify 500MHz Small Signal

We've done this sort of thing already in the early 90's :-)

On a chip Jim has a ton of opportunities that us discrete designers don't have. Such as very good matching between neighboring components. Provided they are oriented in the same direction but that's easy.

You can then run DC or an otherwise calibrated RF signal through a reference chain and error-correct the output. If the loop Jim needs doesn't have to be super-fast one could even servo.

However, 1mV accuracy at 500mV is no small feat at 500MHz.

Ok, maybe I am totally off here now, but could you mix the VCO signal with a saturated (bang-bang) image of itself and also 90 deg of itself, then I/Q demodulate these two channels? That way you'd be close to DC, or to whatever loop speed you have.

TDs are insanely expensive nowadays, ballpark $100. I used to get them for a couple bucks from Allied. The fabrication process is insane, and nobody ever modernized it.

There are some more modern planar germanium back diodes, essentially low Ip tunnel diodes, but they're RF detectors, useless for switching. Pity, I used to like tunnel diodes.

John

Now functioning, need to make the loop-gain track the set-point and it'll be a winner ;-) ...Jim Thompson

I posted it so the next time someone shows up with a 40 year old design and wants a tunnel diode, people will know where to send them. :)

They even have LM3909 LED flashers for $6 each:

:)

Try PiN diodes then.

Jim Thompson a écrit :

Uh... Why don't you get it simple? Have your rectified level constant and a VCA in a feedback loop.

For what? Certainly not switching, amplifying, oscillating, detection, or mixing.

There was a single-TD circuit that was an RF amp, a local oscillator, a mixer, and an IF amp. One TD and a couple of passives would make an FM transmitter.

The beefier td's would make a voltage step with a 22 picosecond rise time... in 1964.

John

Good, but not fast. PIN diodes specialize in having a lot of stored charge, so that the signal current can be quite a bit larger than the DC current without causing excessive distortion.

Cheers

Phil Hobbs

PINs stop behaving like PINs at low frequencies, too. So they don't make useful wideband switches.

But I meant active switching when I was referring to a TD. A TD would

John

Sure. TDs are neat. I have a bunch of low-current ones, but they have some totally absurd capacitance and so aren't that fast. The really quick ones have peak currents of something like 50-100 mA. Kind of like a turbocharged rock crusher.

Cheers

Phil Hobbs

Got to watch the carrier lifetime. The lower the bottom of your spectrum and the higher the RF current, the longer its carrier lifetime must be. I found PIN diodes to be great and most of all cheap variable attenuators as well as switched. Designed in tons of them.

I've drooled over SRDs all my life and every time I wanted to buy one I either couldn't have one or it was outlandishly expensive. Guess avalanching is the only game in town and if you want avalanche-rated then a bone-simple BJT can easily shoot up to twenty bucks.

Forgot to mention, I was also never able to lay my hand on a tunnel diode. The hobbyist books had schematics with them in there but that was all bogus, just like UJTs were. Unobtanium. Can you buy a TD in large qties at a reasonable price somewhere? I mean, not the $100 ones.

We use them at 10GHz bands. Or maybe it is only used with a trap filter.

Old radios (transceivers).

When I was a kid I repaired a lot of transceivers. None had TDs in there.

SRDs aren't hard to get. MA/Com has distributor parts, under a buck. M-Pulse and Metelics are good about samples. If you want a few, send me a SASE.

Oh, here it is...

MA44769-287 PENSTOCK

Price 58 cents in small quantities.

They also have MA44767-287, 600 ps risetime, a little easier to drive because it stores more charge.

These make nice edge generators and frequency multipliers. I have a rubidium clock that generates the 6.3846826128 GHz frequency from a 10 MHz rock with an absurdly small number of cheap parts.

John