Step Recovery

PIN diodes have more capacitance than you usually want in an SRD. John Larkin has posted some pretty interesting stuff on using rectifiers as high-power SRDs--search back in this group. (One good search term is 'Grekhov'.)

Cheers,

Phil Hobbs

There are a few distributor-stock SRDs, the M/Acom MA44767 series, SOT-23's...

Small varicaps often snap, but pin diodes usually don't... they are usually grown to be sort of anti snap diodes.

Metelics and M-pulse have catalog parts, and are usually pretty good about samples.

John

Reposted from abse:

In the first place, that's one hell of a small snap to be able to do anything with. In the second place, if that is a VOLTAGE waveform, you've got it bassackwards. THe SRD remains a closed switch (min volts, max current) until all the stored charge is swept out of the junction in the reverse direction and then snaps to an open (max volts, min current) for the remainder of the reverse cycle.

The trick is to select both the SRD and the forward bias to make the sucker snap at the exact peak of the reverse voltage waveform.

Jim

--
"It is the mark of an educated mind to be able to entertain a thought 
without accepting it."
        --Aristotle


"Andrew Holme"  wrote in message 
news:V3s3l.44649$Ci.29689@newsfe10.ams2...
> Just been hanging various junk box diodes between a pair of SMA sockets, 
> one to my signal generator, the other to my \'scope, looking for step 
> recovery.

"RST Engineering (jw)" wrote

It's the current. Here's the test circuit:

R1=50 Diode ___ .----|___|------->|-----. | | /+\\ .-. V1 ( ) | | R2=50 \\-/ | | | '-' | | | | === ===

V1 = EMF of signal generator (sinusoid) R1 = Output impedance of signal generator R2 = 50-ohm coax to 'scope (50-ohm input)

. . And the frequency of interest???

Jim

-- "It is the mark of an educated mind to be able to entertain a thought without accepting it." --Aristotle

It is a voltage waveform, and he doesn't. This is the classic series-diode waveform that looks exactly like Boff's original discovery pics... on an HP scope, even.

John

As has been pointed out in threads past, PIN diodes intended for rapid band switching can make decent Step Recovery Diodes. Both proper SRD's and band switching PIN diodes are designed with very abrupt transitions from low to high doping, and band switching diodes for UHF and above will be designed for low "off" capacitance.

An easily available example in SOT-23 surface mount package is the MMBV3401, eg.:

while this has been used to multiply by 16x to around 1GHz with some efficiency:

I don't know what the transition time is.

Avago/Agilent/HP have a nice app note:

in which they show how to use their HSMP-3820 and 3822 PIN diodes as SRD multipliers, with some usefulness to >5GHz.

SOT-23 packages have enough parasitic inductance that this might be dominating the useful risetime that you can get from these PIN diodes

Depending on what you're doing with your SRD, you might care about recovery time variation with temperature. HP worked around this in their sampling oscilloscopes beginning in the 60's by using a brief forward bias pulse to build up stored charge in the SRD before applying the reverse bias pulse that leads to the snap. If the forward pulse is much shorter than the carrier lifetime (at the highest temperature) then the lifetime variation with temperature has little impact on the recovery time.

James

Sorry about the gmail address. I know that many people will have killfiled gmail for good reason.

No problem. That's what white-listing is for.

...Jim Thompson

--
| James E.Thompson, P.E.                           |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
 I love to cook with wine     Sometimes I even put it in the food

- - snip - -

So refreshing ... to hear from someone who appears to know what they're talking about.

Chris

That very well may be. By the time I got around to using them in '67, the details had mostly been worked out and the theory was decent. However, I started out using them as S-band local oscillator multipliers (x10) in shunt mode and was used to seeing the waveform on the sampling scope as voltage as I described.

You guys just gotta remember, you are a LOT older than I am and probably called DeForest, Lee {;-)

Just for giggles, suppose you have two of them driven by a common UHF source and vary the bias on one of them. What is the difference in the output waveform, neglecting the small change in amplitude because you are snapping at different points on the waveform. Oops, gave the answer away there, I did, I did.

THen went to making some decent (5-watt) S and C band multipliers and to get the power dissipation down, the only practical topology was shunt mode also. It made the input and output coupling a lot simpler and more efficient than series mode (which we tried, just to see what the difference would be).

BTW, did Boff get credit for the work? Seems to me that Moll was mixed in there somewhere also. THe same Moll from the Ebers-Moll model?

Jim

Yes. They originally called it a "Boff diode" but I guess the alternate connotation made them change it to "SRD" or "snap diode." I think the story is that HP had just invented the 185 sampling scope [1] which used an avalanche transistor sampler to get to 500 MHz bandwidth. In another lab, Frank Boff was playing with using diodes as harmonic multipliers, and found one diode that gave much higher than theoretical multiplier efficiency. They wheeled over the new scope and observed almost exactly the waveform that Andrew posted. Not long after, the 185 was using SRDs to get to 1 GHz, and a year or two later they were up to 4 GHz.

He co-published the discovery with Moll.

There are HP Journal articles.

John

[1] I have one. Some day I may try to get it to work.

Hey, weren't you going to send me a handfull once upon a time? Your server must keep dropping my e-mails or something, I never seem to get a reply.

It's my understanding that snap diodes are just hyperabrupt varactors with an even more aggressive doping profile. So as the charge leaves, it kind of leaves in a plane wave, all at once, sluurrp, and it's gone. Then inductance says hello and...*whang*.

The best edges I've ever found from diodes in my collection seem to be schottky rectifiers... and that'd be all from the capacitance, which does have a funny response (usually huge, a couple nF around 0V, for the bigger ones), but it's not a snap per se. I must have bad luck with my parts, haven't even seen a 1kV+ diode that snaps. Well, that's not entirely fair, there's that one diode that I suspect is a varactor, it went reasonably well, but about as the OP says, 3ns or so (which is still close to my observation limits, but not exceeding them, so...).

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

"John Larkin" wrote

Thanks for that fascinating link. It was an advert for the 185A/187B, which I stumbled upon in a 1961 issue of Wireless World, that got me playing with this recently. That 'scope must've been a sensation in its day.

I think the 2ns I measured for the rise time from my mystery PIN diode is limited by the 1.75ns rise time specification of my HP1725A and may actually be slightly faster. We have better 'scopes at work and I'll try it there.

I have some test equipment here containing SRDs including a Marconi spectrum analyser and a RACAL true rms microwattmeter and the service schematics make interesting reading with their use of these exotic devices.

I've also been playing with tunnel diodes and an avalanche transistor pulse generator circuit. I get sub nanosecond rises from the latter, and I've had the former oscillating at above 1 GHz; but I haven't tried generating fast steps using tunnel diodes yet.

I also found quite an interesting paper "50 Years of RF and Microwave Sampling" by Mark Kahrs which mentions you, John, in the acknowledgments.

Lumatron had already been selling sampling scopes, but they were pretty awful. I have one, and some schematics... piece of junk, mostly.

In a series circuit, the ciode capacitance is working with 100 ohms to make a time constant, and that's another limit on speed. But that's academic if you use a slow scope. Putting the srd across the transmission line makes the diode see 25 ohms, much faster.

If you're really interested in fast stuff, there's lots of sampling scopes on ebay. I'm partial to the Tek 11801 series.

I played some with higher powered PIN diodes, hoping to find a high-voltage SRD, but all the ones I tried had really slow, soft recoveries. Maybe they just have a wide, flat-doped intrinsic region. SRDs need hyperbolic doping or something.

Couldn't be simpler: signal generator, series resistor, TD.

Mark is even more of a sampling freak than I am. But I have more hardware.

John

Ok, OK. Send me a self-addressed envelope and I'll scrounge a few up.

John Larkin Eighteen Otis St San Francisco 94103

Look up Grekhov Diode. It's pretty much a regular rectifier used as a HV SRD. The trick is to forward bias it (like, 50 volts forward!) for a few 10's of ns, then reverse it hard.

Grekhov is sometimes spelled Grekov or Grehkov. AKA "drift step-recovery diode."

John

"John Larkin" wrote in = message news: snipped-for-privacy@4ax.com...

you've

the=20

However, I=20

shunt=20

voltage as=20

probably=20

source=20

output=20

snapping=20

there, I=20

to get=20

also.=20

than=20

be).

mixed in=20

Here is some information that I don't think has been mentioned yet.

Dec 1964 HP Journal: Microwave Harmonic Generation and Nanosecond Pulse = Generation=20 with the Step-recovery Diode

Ap Note 918 Pulse and Waveform Generation with Step Recovery Diodes=20

Ap Note 920 Harmonic Generation Using Step Recovery Diodes and SRD = Modules

Articles:

Moll, Krakauer, and Shen, "P-N Junction Charge-Storage Diodes," Proc. of the IRE, Jan. 1962, pp. 43-53.

Krakauer, "Harmonic Generation, Rectification, and Lifetime Evaluation with the Stepo Recovery Diode," Proc. of the IRE July 1962, pp. =

1665-1676.

Moll and Hamilton, "Physical Modeling of the Step Recovery Diode for Pulse and Harmonic Generation Circuits," Proc. of the IEEE, vol 57, no. 7 (July 1969), pp. 1250-1259.

Some more recent work: A 12 psec GaAs Doulbe Heterostructure SRD

--=20 Regards, Howard snipped-for-privacy@ix.netcom.com

Generation

Good grief, their cheapest SRD was $18.75, in 1964! The M/A-com SOT-23 parts are 50-75 cents, which is, what, 8 cents or so in '64 dollars.

John

shunt

as

source

get

also.

Generation

Why is it that M/A-com does not sell through Digikey? That would be the perfect avenue for making one's parts known to the public.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

I suspect the deal is that much of what M/A-com sells is so nichey that they don't really suffer from *not* selling through DigiKey -- people working in niches realize many of the parts they want won't necessarily be readily available, and hence don't mind having to go through "traditional" distributors.

(Look at something like FPGAs -- a decade or so ago when they weren't nearly so "mainstream" and DigiKey didn't carry many at all; now that they are very much mainstream DigiKey has lots of 'em...)

That being said, any new part I might want to use I do first punch into DigiKey (and Mouser) to check the "basic" availability -- and then look for possible substitutes if it's not so hot. (On the other hand, I know you're big on also finding a 2nd source, which I don't worry about so much given the small quantities of widgets we make at work here...)

I'm always surprised that some engineers figure it's 100% logistics/purchasing's job to check into part availability when nothing more than a 10 second DigiKey search is so... fruitful... these days.

The most common suppliers we use who aren't typically in DigiKey (or are poorly represented there) are the traditional military/MMIC guys... M/A-com, Hittite, MiniCircuits, Skyworks, Avago, etc...

---Joel