Suggestion for device to switch Blumlein pulse generator

Can you be more explicit about what you want to do? What is the load? What sort of drive do you want?

John

Something like this might work:

John

To use mercury relays to make fast edges, a reed capsule is located in a metal tube or other structure such as to make a matched impedance transmission line. The Tek 109 pulse generator did that... I have one around here somewhere.

John

I want to be able to transmit BPSK-modulated (or, perhaps here more precisely, PRK:Phase-Reversal keying) short pulses with a suitable wideband antenna. The antenna looks resistive (100R...200R) during ~300 ps and may be symmetrical (dipole-like) or asymmetrical (monopole)

I have been able to generate unipolar pulses with the setup of HPs ancient AN 918. Doubling the structure with an inverted antenna and coupling the 1's to one pulser and the 0's to the other kind of works but I am looking for a better alternative :)

I'm afraid I don't understand the schematic you posted. I assume that both diodes are forward biased with VC > 0 > VA, but I don't see how you reverse bias them with the switches?

Pere

Sorry, I forgot to add that an output pulse of say ~3..5 V is ok, but if I can achieve 10 V or more, even better!

Pere

What sort of pulse rate do you have in mind?

I got a couple of these on ebay...

They sort of transmit pulses, between a pair of them, very distorted and ringey.

John

We are transmitting at a puls rate between 10 M and 100 MPulses/s.

Those spiral antennas look nice! Did you TDR them? With bowties and cones we get quite nice resistive impedances:

And, regarding your schematic with 2 SRDs, now it became apparent that B+ and B- are positive and negative bias and that you fire the SRDs by pulses on A and C. I did not realize this yesterday, probably because I was expecting capacitive coupling to the input signals A and B, which I would expect to perform better. Any thoughts on this?

Pere

Oh. Doesn't sound like mercury-wetted relays is a good idea.

I don't recall, but I will make a TDR mini-radar and see what happens.

With bowties and

We worked with some people who were developing a ground-penetrating radar (almost everybody tries that, sooner or later) for mine removal. Their antenna was a big cube of styrafoam carved into sort of a curvy horn, with two sort-of triangular pieces of copper foil glued to the sides. It was cheap to make and pretty broadband. We made the SRD transmit pulser for them, 60 volts in 100 ps or some such.

No, the idea is that A and C are diode forward bias supplies. You'd forward bias the diode you want to use and way back-bias the other. Oops, it needs an inductor from the diode junction to ground, for a DC bias path. Given that, imagine you don't want to use the lower SRD, so make C negative, to turn it off.

Now make A negative to bias the upper diode on, a few mA maybe. When you want to make a negative pulse, turn on the B- switch until the upper diode snaps. The shorted transmission line will make a square pulse.

There's probably a better topology, but something like that should work. It's not well suited to fast reversal of pulse polarity, but I didn't know you wanted to do that.

An SRD has to be forward biased for several ns before you can snap it, but 100 MHz pulse rates should be possible, 10M certainly. The trick would be in steering the bias, the drives, and the snap currents. Maybe drive one side of the antenna from one pulser and the other from the opposite polarity pulser. Balun or something maybe. My idea was to back-bias the unused diode to get it out of the circuit while you bias and snap the other.

John

For someone who wants a >1 MHz repeat rate, gas discharge (and even many semiconductors) has a more important problem, that of turn-off time. The electrons move out to the electrodes a lot faster than the also-charged positive ions they leave behind. Proportional counters have always needed several microseconds of dead time before they return to quiescent state.

Avalanche phenomena often take a while. In the old thyratrons, this was the deionization time (analogous to recombination time in semiconductors) of the gas, which depends on pressure and type. Xenon and mercury were the most typical thyratrons, such as the venerable miniature 2D21, or any big power thyratron. Big ions like those simply move slowly, and take a while for their energy states to decay, so typical DI times were around

1ms.

The fastest thyratrons use hydrogen (or possibly helium), where I suppose the higher direct energy levels (13.6eV for a single electron, versus numerous levels for many electrons in the cloud following a xenon ion) and high mobility ion (a single proton) allow DI times in the microseconds or less (not bad for a device with a path length of centimeters).

thyristors and avalanche transistors depend on recombination to become nonconductive, which is typically in the 10s of microseconds for devices I've measured. The typical waveform you'll see on, for instance, a PN2369 or 2N3904 avalanche generator, is an average relaxation oscillator waveform, with a sharp fall, and it stays at the bottom (dropping about

10V) for a few microseconds before rising again. It's funny to think of a 2N3904 taking that long to turn off, but an ampere of peak current is a mess of charge carriers to clear from the junction.

Tim

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

"whit3rd"  wrote in message 
news:33385557.1468.1320000386357.JavaMail.geo-discussion-forums@yqhd1...
> On Wednesday, October 26, 2011 6:18:57 AM UTC-7, Bill Sloman wrote:
>
>> The mechanism of a spark breakdown is well-defined...
>> The electrons have to move through the electric field in order to
>> accelerate, so there's a time delay there, and you've got to have a
>> lot of collisions before you've got a useful number of charge
>> carriers, so there's more delay there.
>
> For someone who wants a >1 MHz repeat rate, gas discharge (and
> even many semiconductors) has a more  important problem,
> that of turn-off time.  The electrons move out to the electrodes
> a lot faster than the also-charged positive ions they leave behind.
> Proportional counters have always needed several microseconds
> of dead time before they return to quiescent state.