Step Recovery Diode

[PETERg0rsq]

I am repairing an old sampling volt meter, and a "Step Recovery Diode" has failed.

Searching around shows not only is the part obsolete (no suprise there) but also the technology is obsolete!

This is used as a pulse shaper, and takes a random pulse and reduces it to a 300pS pulse by discharging a small inductor as the diode bias reverses.

Can anyone suggest a source foe a Step recovery diode, preferably a glass package, or an alternative component to replace one?

Space is a premium, as it is inside a probe.

Thanks

Peter

[GMB]

That sounds a bit like a tunnel diode. Another name for it?

I had some fail in my Tektronix 1S1 unit but was amazed to be able to buy them as spare parts! I haven't seen tunnel diodes for sale anywhere these days.

I'm getting the impression that their may be more obsolete semiconductors than valves!

[PETERg0rsq]

GMB

Alas they are not the same thing!

I have found tunnel diodes available, and replaced a few in the front ends of HP frequency counters.

Step Recovery Diodes have a property that stores a charge when they are forward biased, and when they are reverse biased this stored charge keeps the diode conducting very briefly, then as soon as the stored charge depletes the diode switches into an OFF condition in a matter of pico-Seconds.

This very rapid switching is then used, either to generate harmonics, or in my case, to discharge a small indictor of about 10nH and produce a sharp pulse about 300pS wide.

The diode I need to replace is an A4S112!

I am experimenting in LTSpice with different diodes, and find some strange results discharging a small inductor this way!

A 1N4007 diode produces multiple 1kV spikes, and a 1N4148 seems to have similar properties to what I need, but can I trust LTSpice at these extreems!

[Radio Wrangler]

No, you can't trust LT spice in this respect, Peter. The spice core has no coverage of diode stored charge mechanisms. LT spice has been mucked about with (quite successfully) in a number of respects to siout SMPS simulation, and I wondered if something had been done to diodes to make their efficiency calculations more efficient, but looking at the parameter strings doesn't show any unexpected inclusions.

I think you'll find the same SRD in the 1GHz sampling scope plug-ins for the 140 and 180 series mainframes. These often go for very little money as orphans. Even a working unit in a scope frame is cheap because relatively few people know what they are or have a use for them.

Even the later 8508A instrument is obsolete and out of support now. Oddly Agilent say that the ENA network analyser is a replacement for it. They seem to not realise that a network analyser needs to source the signal it is used to measure. The vector voltmeter can use signals which are already there... quite a different capability.

You should have a better chance of finding a scrapper of a scope plug-in than another Vector voltmeter.

David

[kalee20]

All PN diodes show reverse recovery after passing a brief reverse curent pulse.

Diodes like 1N4148 have a very short recovery time, achieved by doping with gold which created charge recombination centres. Unfortunately this considerably increases leakage.

Fast switching diodes for switch-mode power supplies have carefully controlled junction profiles - some give a short recovery time but start blocking abruptly; others pass reverse current for longer, but recovery is a more gentle, soggy affair (which avoids ringing oscillations in stray capacitance and inductance).

I'd have thought the best chance you'd have is with a diode that's not a fast recovery type. Maybe 1N4003 is not quick enough? Have you thought of a small Zener diode? Or a junction FET, with drain and source linked? Or a transistor, using base/emitter junction? None of these will have any funny techniques to control reverse recovery, so although you'll have to play around, you may well be lucky.

[PJC58-Hythe]

There are a few places in the States that do them...

http://www.advancedsemiconductor.com...ve_diodes.html

http://www.mpulsemw.com/index.html

Didn't have an exhaustive look but I'm sure a little more digging might find you an equivalent?

Peter

[PJC58-Hythe]

These might be able to help...

http://www.melcom.co.uk/melcom_principals.php?view=17

[WME_bill]

Step recovery Diodes. If it is any help. I have several brand new HP diodes, type 5082-0112, which appears to be identical to the TRW A4S112. I bought them perhaps 20 years ago for a comb generator project which never proceeded. Still in their HP plastic envelope. Send me a PM if of any interest.

Step Recovery diodes were used to generate very fast pulses. Usually in Sampling Scopes to make the <1ns pulse to the Schottky or GaAs sampling bridge. I attach an extract from the HP microwave catalogue of 1991 dealing with Step Recovery diodes with specifications. Used in Sampling scopes, Sampling or Vector voltmeters

Tunnel diodes were used to generate a trigger pulse, as they could switch from 65mv/5ma to 350mv or so in <1ns. The usual ones used are either 5ma 1N3716 or 10ma 1N3718. I also have some of these (2ma/5ma/10ma/22ma), some new, as I acquired the whole remaining stock of a distributor some 30 years ago when he was closing that product line down. He wanted full price also. So if anyone is desperate, really desperate, send me a PM. But I generally use them to keep my HP140/HP180/Solartron CD1740/Tek564 scopes, and Philips PM3400, HP180-1810 sampling scopes going. Tunnel diodes are usually Ge, and seem to fail easily from age. Perhaps like me.

Also from a GE 1991 catalogue upon Tunnel diodes, on which I have added a proper tester ex Wireless World March 1965, (as an ohm-meter test is the best way of ruining them), and also on GE Backward diodes, occasionally used by Tek as a UHF detector.
wme_bill. m0wpn

[PETERg0rsq]

Gents

Many thanks for your responses.

The three links supplied I had not come across, so will make enquiries, might just get lucky and get a diode at a decent price.

I hadnt considered looking at salvaging from other equipment, its a bit against my nature. If I got a working plugin, I would then be looking for a mainframe to fit it to

I do have a HP 8508A, but I am not going to scrap it to repair a little old Racal 9301A!

I have seen suggestions of using other devices, diodes, or transistors as diode gates, but no mention of specific devices that work, I am sure I am not the first to find problems sourcing SRD's, so I guess I will have to make a little test jig and make some experiments.
I will start with some fast rectifier diodes.
Kalee are you suggesting 1n4003 and 1N4148 might make suitable candidates? What parameters should I be particularily looking for in other possible diodes?

Regards

peter

[PETERg0rsq]

Bill

You posted as I was writing my last reply!
I will contact you by PM shortly.

I did a lot of research a short while ago on tunnel diodes, as I had a HP5252A that I purchase about 30 years ago, and never got around to fixing!

One of the tunnel diodes had failed, and I made a test jig similar to the one in the pdf you enclosed, as I also read that testing with an Ohm meter could kill one. I also limited my PSU to 50mA to be on the safe side!

I found a German seller who had some Ge diodes of similar characteristics, but in a different package, and these worked fine, didnt even need to re-biase the diode.

I think tunnel diodes are a fascinating device, and in their day were going to revolutionise electronics (acording to the sales blurb), but alas the transistor improved and made tunnel diodes obsolete (almost).

Regards

peter

[WME_bill]

Diode Recovery Measurements.
I attach an article from a Philips booklet aimed for schools which describes a set-up for measuring diode recovery time. This purposely deals with low frequency and large delays. But I did spend a fascinating day playing around with a fast rise pulse generator and a sampling scope. Rather like the elaborate set up shown in the HP Step Recovery details I posted.
I forget the details now, but diode with a small stored charge worked best, I seem to remember OA47, 1N914/1N4148 and some random computer switching diodes seemed to give sharpest cutoff. So they might prove an alternative for less taxing uses. I didn't try a small signal schottky diode (like BAT10, BA280, MDB101) as they were hardly around then. Power diodes would be very slow, though perhaps a modern computer power supply schottky diode might be interesting. I must repeat my experiments.
The pulse generated for the telephone line testers using Time Domain Reflectometry like the Cossor CME110 of 1972 (which Anchor at Nottingham was selling some years ago, and still turn up), or the later BiccoTest T213 was a 10V pulse of 10ns rise, and uses a switching transistor (as 2N2369) operating in avalanche mode which is passed that through a 1N916, so the diode delay must be significantly less.
These units could provide the basis for quite a nice little scope. wme_bill

[kalee20]

Originally Posted by PETERg0rsq

Kalee are you suggesting 1n4003 and 1N4148 might make suitable candidates? What parameters should I be particularily looking for in other possible diodes?

I'd have thought not a 1N4148; possibly a 1N4007; more likely another non-fast-recovery type. Hence I suggested a Zener diode.

Other than that, I'd not have much idea, except that you're going to need a fast scope and a decent current probe!

[woodchips]

Another source of step recovery diodes are the frequency extender plug ins for the HP 5245 counter. If you are lucky might find a tunnel diode in them as well. Bob

[John_BS]

If you can live with a pulse of around 250-300pS, the avalanche technique is quite repeatable and inexpensive. 2N2369 or 2369A seems to work well in this role.

John

[PETERg0rsq]

John

I did look at using a transistor in Avalanch mode, but the high bias voltages mean it is impractical for my application, as the sampling head is in a small probe. I would need to provide an additional supply to the probe of around 100V dc, as well as a few additional components!

I have been experimenting with an assortment of diodes, but even my 350Mhz scope (with a 1Ghz resistive probe) is struggling. All I seem to get is "ringing" due to stray capacitance in the test jig!

Trying to measure the discharge of a 20nH inductor is asking a bit much I think!

Time for a re-think!


Peter

[Radio Wrangler]

SRDs were a bit of a devil to test. Back in their heyday, the fastest scopes had a sampling aperture which was set by the pulse width of um, err, well, their SRD.

The effective way of checking one was to look at the spectrum output on a multi-GHz spectrum analyser.

David

[PETERg0rsq]

David

Thanks for that, I hadnt thought of looking at the harmonic content on a spectrum analyser, might be worth a look before I fit the diode i have now aquired.

I have made some tests with the best pulse generator I have, (a 15nS rise time and 19nS fall time) and some screen dumps from my scope are here (200Mhz bandwidth 2Ghz sample DSO).

I tested the SRD then variouse other diodes from my odd diode box. Nothing really compares with the SRD though!

Test circuit was 5v pulses through a 1nF capacitor to give + and - elements to the pulse.
I had to guess at the circuit loading, but in my application I think 1K input and output loading was reasonable, so I tried this.

I tested the following diodes:- 1n4007, 1n4140, and unknown diode network, possibly a varactor, BZY88, BYX36, AO202.

I have yet to try a high frequency transistor emitter-base junction, or a jfet.

I tested many other diodes, but these produced nice clean results with no conduction when reverse biased, similar to the 1n4140.


Peter

[Radio Wrangler]

Hi Peter,

The SRDs in the vector voltmeters gave sampling apertures close to 100 picoseconds. So they could shift a bit.

David