tiny 1200v SiC diode

Here's the data sheet.It's pretty specific about the voltage dependent reverse capacitance aka stored charge.

Most diode data sheets have c-v graphs.

I want to use it in a DC-restore circuit, to basically make negative pulses from a high voltage positive pulse generator.

The capacitance will delay the first edge a bit, but C will drop off fast so net rise time shouldn't change much.

It's just a diode.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Only those for relatively expensive diodes.

People who read application notes can remember back to "soft-recovery diodes" versus "snap-off/step-recovery diodes".

The world's generic diode is probably 1N914. They cost 1 cent in volume.

A few people still make SRDs. They are mostly used as RF frequency multipliers these days. I don't use them any more, except for off-label DSRDs. There are better ways to make fast edges.

The one I posted is a silicon carbide schottky, which presumably doesn't have reverse recovery charge. I'll see when I get some.

The coolest app note was Boff's discovery of diode step recovery, in the HP Journal ca 1960. They should have been named Boff Diodes.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

SRDs are still pretty cool for fast bipolar sampling. You get equal, opposite, and perfectly aligned current pulses, so you can drive a diode bridge symmetrically.

I’m a fan of naming clever circuits after their inventors. It helps prevent miscarriages of justice, as in the Tayloe mixer, which was published years earlier by Ed Oxner of Siliconix.

Cheers

Phil Hobbs

Yeah, there's still nothing like them.

Whose did you buy?

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

So far I’ve been sticking to the low cost end, using BAT15s as one- and two-diode samplers. You don’t need matched diodes for that, whereas you sorta do for a bridge. Otherwise one side of the bridge will hog the current and so trash the sampling efficiency.

It would be interesting to see if it’s possible to do a fast sampler with unmatched diodes. You need a bunch of extra parts to get both sides of the bridge to switch together, and that slows everything down.

Cheers

Phil Hobbs

But whose SRDs?

BAT15 is nice, 0.23 pF at zero volts.

I wonder if a phemt could be used as a spst sampler.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Apart from commercial samplers, I haven’t used an SRD since 1981.

One of my first tasks at my first engineering job was to rebias a coax-in, waveguide-out frequency multiplier. It used a small, ceramic-packaged SRD, which was self-biased via the capacitance of the mount and a shunt resistor.

The original model used a MA/Com unit with a 100k-ish blob of thick-film resistive paste, and the replacement was from Gigahertz Devices.

I used a soft graphite pencil to scribble on it, and iterated to find the right resistance, which was a bit higher iirc.

Yeah, and it has pretty low video resistance, so the sampling efficiency is reasonable.

I’ve used a bunch of SAV-541+’s to make a multipoint sampler on a single transmission line. They were good for 100 ps or thereabouts, and the sampling efficiency was excellent.

My best sampler to date is a ~60 ps TDR that talks over USB serial. It’s unipolar, and needs 10 or 20 pulses per delay point, but it’s super cheap. I used a sampling loop, an idea that I got from the Tek Concepts book.

I’d love to have an excuse to try to beat the SD-24, but I’m not holding my breath.

Cheers

Phil Hobbs

They might be cheap now, but they were a bit more up-market when they were introduced.

The 1N914 part number is a US military specification. It turned out that they were cheap to mass produce in volume, and that there's a very large market for that kind of cheap, tightly specified diode.

But that sort of detail is above your pay grade.

And the amount of stored charge isn't all that tightly specified. If you want the sharp edge to appear when you want it to you have to do a bit of extra work. I looked at the them for the half-nanosecond wide unblanking pulse for Cambridge Instrument's stroboscopic electron microscope, and went for broad-band transistors instead.

Of course it does, and it is fairly tightly specified as I high-lighted in my response above "It's pretty specific about the voltage dependent >>>> reverse capacitance aka stored charge."

I've never seen it, which is odd because I did have library access to the HP Journal, and read quite a few of the articles. Libraries do tend to lose exampled of important journal article - psychopaths do tend to steal them.

This one?

piglet

No. January 1960.

It was important enough that Wm Hewlwtt wrote the introduction.

What's cool is that Boff saw unexplained RF multiplication.

HP had a new sampling scope that used an avalanche transistor to make the sampling impulse. Boff and pals used that to discover the step recovery effect.

Soon HP was using SRDs in their sampling heads.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

It certainly looks like the right article. I'm happy to have had the opportunity to read it, though I should have been able to find it for myself if I'd known about it. John Larkin should have been able to post such a link, but his aim seems to be to look good rather than to be seen as being helpful.

Thanks, see it now.

Here's the link that you need:

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

It's not immediately useful.

on page three there's useful text

"The effect was described in detail in the paper, “A New High-speed Effect in Solid-state Diodes,” given a t the 1960 International Solid-state Circuits Conference by A. F. Boff of

-hp-, Dr. John Moll of Stanford Uni- versity and R. Shen of Harvard University. "

That's enough to get you to

Dogs and women tend to like me, which is what matters.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Heh.

Many years later, ca 2006, Jason Breithbart (who founded Holzworth Instrumentation) implemented step-response like behavior using a Non Linear Transmission Line (NLTL) with the bias voltage derived from the NLTL itself, this being the only voltage source that was remotely quiet enough to be quieter than SRDs et al. The actual bias method is not mentioned in his thesis, but emerged later.

Joe

Well, some dogs and some women like engineers. Thank goodness.

The fastest sampling scopes use NLTLs to generate the sampling edges. But brute-force ADCs are used in fast megabuck scopes now, not samplers.

It would be cool to build a small USB box that uses the old Tek SD-series sampling heads. There are lots around.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics