Suggestion for device to switch Blumlein pulse generator

The Blumlein pulse generator consists in:

----R----o---=============--RL--=============-- | | Vdc / sw Zo,Td Zo,Td | | ---------o---=============------=============--

Both transmission lines are slowly charged to Vdc through a biggish R. Then, the switch closes and generates a pulse across RL with amplitude Vdc and width 2*Td. If the switch is ideal, a square pulse is obtained. If the switch closes slowly, you get a triangular pulse resulting from the superposition of the slow edges.

Some years ago I played with this circuit using a BFR83a (?) as a switch and I remember that the pulses (~500 ps) were limited by slow switching. In the next days I will try with a BFP405, which I have by the bucket, but I perhaps there might be better alternatives. So, the question is: which device would you suggest for a fast _falling_ edge? (I guess that strict RF performance is _not_ the key point to look at in this case).

The ideal objective would be get 100 ps pulses with amplitude in the 2-5 V range (although higher values would be welcome).

Pere

Reply to
o pere o
Loading thread data ...

--
I think a mercury-wetted reed relay would work.
Reply to
John Fields

Maybe raise the charging voltage to 50V or so and use a BFG541, which will avalanche nicely at that voltage. Estimated risetime should be in the 150ps ballpark.

If you find a better way, I'd be interested to hear about it.

Jeroen Belleman

Reply to
Jeroen Belleman

A PHEMT might get you there. An NE3508 with good gate drive might do it, or a couple in parallel to reduce lead inductance.

But at those speeds, I'd prefer a different topology.

  1. Step recovery diode and shorted-transmission-line shaper. 10 or 20 volts would be feasible at 100 ps.

  1. Fast logic edge, like from an ADCMP582, RLC or txline shaper, optionally a second comparator, and an amplifier, a darlington MMIC or one of the distributed amp fiber modulator things maybe.

The SRD thing is easier and cheaper, if anything can be said to be easy at 100 ps. The comparator/amplifier thing would allow variable pulse widths.

100 ps is like the speed of sound, the "wall in the sky", where things get difficult.

John

Reply to
John Larkin

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D--

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D--

I remember a triggerable lambda diode switch circuit with really short fall time that I can't find just now, but fig. 8 here seems similar:

formatting link

Mark L. Fergerson

Reply to
alien8752

if a few hundred mV will do. Most parts of the family are $10. The generator part of a typical tdr is not much faster. I'll try them next month.

regards, Gerhard

Reply to
Gerhard Hoffmann

The ADCMP58x series is nice too, with 37 ps edges. The comparator function is nice.

We built a TDR that uses the '580 as the step generator. Seems to work fine.

John

Reply to
John Larkin

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D--

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D--

A long time ago, while in college, I worked in a laser physics laboratory, and the professor had built a high power pulsed UV laser powered by a Blumlein pulse generator. The actual pulses were induced by a spark gap (Rogowski profile electrode etc.,) One of my first projects under the professor was to create a triggered spark gap, by having a small circuit that would create a small spark just a few nanoseconds before the main discharge, basically to ionize the nitrogen gas inside the spark gap., to ensure that the Blumlein capacitors were totally discharged. Worked very well.

Reply to
Daku

From what I have read, I have come to the conclusion that a spark generates edges in the ~1 ns reange, but I got no definite reference. I would have expected this to be shorter...

Pere

Reply to
o pere o

Now I seem to remember that you already had suggested this PHEMT in another thread. Thanks, this may be worth trying.

Well, the final objective is to be able to choose the sign of the pulses (I started a thread some time ago on this). And by now it seems that a Blumlein line fired from opposing ends would do the job.

I have played with SRDs but have not found a way to be able to change polarities...

Pere

Reply to
o pere o

Following this hint I have googled a paper "A Comparison Between Mercury Wetted Relays, Ruthenium Relays, And Solid State Switches In A Thermoelectric Z Meter" where the graphs showing the comparisons in

0.0001 (s) (!!) time increments...

Although I was targeting a solid-state solution, can you provide a link to switching behavior of mercury-wetted relays?

Pere

Reply to
o pere o

Thanks to your input I have come to an Electronics Letters paper (NOVEL SCHMITT TRIGGER BASED ON LAMBDA DIODE) showing a simulated DC (!) transfer characteristic of a Schmitt trigger circuit based on a lambda diode. I will research a bit on this... but I am afraid that the switching time will be a (possibly unfortunate) combination of both transistors' times.

Pere

Reply to
o pere o

Thanks for the suggestion. Reportedly, a 2N2369 should also work well. Any idea on how precise the avalanche startup is? Or, equivalently, what jitter should be expected when working in this mode?

Pere

Reply to
o pere o

Avalanche is typically 70-120V, so a trimpot is necessary for the supply. Typical way is a series resistor bringing voltage up to just below threshold, then drive the base with a pulse transformer or something like that. There's more info in Jim Williams' AN-47 IIRC.

Tim

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

Yes, a 2n2369 will avalanche too, but with about 90V across it. It's slower too: in the 300ps ballpark, if I remember well. As for the jitter, I don't really know, as it never mattered to me. This is quite some time ago, but from fuzzy memory, when I tried to measure the risetimes using a Tek7904 with S-6 sampler plug-in, the edge looked smeared over a few tens of ps. Hard to tell how to partition that between the scope and the pulse generator.

Regards, Jeroen Belleman

Reply to
Jeroen Belleman

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D--

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D--

.

ch

g.

t

-5

The mechanism of a spark breakdown is well-defined, and - as Daku pointed out - it does pay to have charged particles floating aound in the gap before you apply the high voltage.

Spark gaps break down by an avalanche mechanism, where electrons in the gap see enough of an electric field to pick up enough energy between collisions with gas molecules to be able to ionise the gas molecules that they hit and knock out a new electron when they hit.

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.

Apparently the mean free path of an electron in air at atmospheric pressure is about a cm, which a photon would cover in about 30psec, and low-kV electrons in about 300psec.

Illuminating an electrode with hard enough UV light to knock electrons out of the metal can speed things up quite a bit.

formatting link

but it's old technology

formatting link

formatting link

and dates from a time when nanosecond rise times were hard to measure.

The scattering cross-section for electrons can be a bit odd

formatting link

-- Bill Sloman, Nijmegen

Reply to
Bill Sloman

But that floats the load. Are you OK with that?

You could generate the pulse and flip the polarity after the fact. Fun but messy to do under electronic control. It would take RF relays or some such.

Or use two SRDs, one for each polarity.

John

Reply to
John Larkin

--
Here's a few hits:

http://cds.linear.com/docs/Application%20Note/an122f.pdf
 Click to see the full signature
Reply to
John Fields

In this case yes, as it will feed a small bowtie antenna.

Electronically, perhaps a diode mixer ring would do the job. But I have not been able to find information on how the diode would contribute to degrade the pulse edges.

That could also be ok!

Pere

Reply to
o pere o

formatting link

Well, that's a wealth of information! I just browsed through those links and they look promising -it will take some time to go through them, however.

Thanks for your offer, John. I have sent you an email ;)

Pere

Reply to
o pere o

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.