Sampler diodes with more barrier height?

PN diodes neither turn on nor turn off in much less than a ns.

I know, but I was hoping that I didn't have to piece it all together and someone would have made a nice IC for that. I can't be the only one.

I've done it with BFS17 a lot. They can really pulse. What one cannot rely on too well is the "let go" phase, of course. I need to bang a slope of around 100psec onto a load. That requires some fast BFP-series part.

I don't have to go quite that fast, 100psec would be sufficient.

We will have to SW-correlate a lot to clean this up. Mainly because the trailing edge of my pulse will look ugly as heck. Because I need it to also contain some lower frequency energy, down to 100MHz or so.

By that time my pulse is already east of Hangtown :-)

That's interesting. From a quick scan of the BFS17 datasheet, it doesn't look as though it's anything special, at least not on the scale of the BFR505 or BFG520. What do you like about it especially?

Cheers

Phil Hobbs

This is the classic 2-diode feedback sampler:

which is pretty much what's in the HP185 scope from 1962. The analog switch, just before the integrator, can be any decent CMOS part; it will be on for a couple of us after each sample. The opamps can be any cheap jfet dual. There are no expensive parts here... the SRD might cost 75 cents.

The signal at A is a small bump that's the difference - a few per cent - between the input and the output at the sampling instant. That gets amplified and gated into the integrator as the baseline for the next shot. Loop gain can be 1, or less than 1 for "smoothing". Any drift is in the integrator, and schottky diode leakage hardly matters.

B1 and B2 represent the voltages that back-bias the sampling diodes. In the Tek full-bridge 7S14, they were literally 1.3 volt mercury batteries... which died in a few years and were hell to replace. Bandgaps would work fine.

I like BFS17s because they are pretty fast but no too fast. They are less likely than really fast parts to oscillate in nanosecond current-steering sorts of apps. But Ft is only about 2 GHz.

I'm still disappointed by the pulse response of microwave NPN transistors, even the SiGe stuff. I'm thinking that a PN junction works a lot faster in small-signal mode (where it's already been on for a while) than going from off to on in one whack. PIN diode concept, carriers or something.

So sort of the big brother of the BC817?

Cheers

Phil Hobbs

Thanks.

I'd have to run the post-sampler faster because my whole signal period is under 1/2usec.

True, with this scheme the transformer isn't so critical anymore either because it's output gets spiffed up by the SRD.

Yikes. I've seen that before, where they generated the bias for a tube with a battery. A red one with a black cat on there. I guess those must have been cheaper than resistors half a century ago. Needless to say, it leaked all over the place.

Because they had manners and didn't oscillate so easily, like John said. But the main reason was that the price was right, around 3c. That's hard to beat. Haven't used on in a few years now but I still have lots of them.

HP had the deconvolution software in their 54750A scopes & associated TDRs. There is a PSPL app note that compares industry standard samplers that talks about that.

regards, Gerhard

What is that mess used for?

Nothing that you'd ever be involved with.

I think that one used the Bracewell Transform, sort of an FFT that is its own inverse.

The classic deconvolution was an FFT-divide-reverseFFT, since convolution in the time domain is multiplication in the frequency domain. But the FFT thing has noise and divide-by-zero problems. Imagine a graphic equalized where you have to crank some of the gains to infinity, to make zero bins come out flat.

I'm all ears for the dual gate MOSFET trick. Are they using one of the gates to shield the charge injection of the other gate?

Back in the day, much effort went into complimentary switches in charge transfer circuits. All sorts of games regarding clock rates (not too fast, not to slow), etc.

The "Sports Edition" of the BC847 :-)

But then they could not claim many GHz of bandwidth in amplifier configurations. Usually that is used at serious output swings in cable distribution systems and the like. One guy got almost 200psec in transitions out of the old BFR92. I've never managed to quite get that from a BFS17.

Normally you aren't supposed to saturate them but it's tough to walk the fine line between that and not enough amplitude. They can't do very sort pulses, just a fast pull-down, unfortunately. So the other side of the pulse doesn't look like much to write home about.

It's pretty simple--you connect the input to the source and G2, storage cap to D, sample pulse to G1. (It works best with parts that bias properly with V_G2S = 0.) Works great with 3N201s. ;)

You usually wouldn't do it that way because of the resistance nonlinearity, but you don't care much about that in a sampler, assuming the sample gate is several time constants wide.

Cheers

Phil Hobbs