Fast impulse from fast-rise-time square wave (microwave)

[ezalys]

Suppose I have a very high bandwidth (harmonics up to/beyond 12.5 GHz) square wave (from say a ONET8501P). What's the most reliable way to derive a train of narrow impulses from this square wave? It seems to me that differentiating and rectifying might be a good way to go... but I probably shouldn't trust a single capacitor to be a good differentiator up to high frequencies. Any recommendations from the microwave gurus out there?

[rhb]

I'm certainly not a microwave guru, but if you need  a signal source for testing I suggest you check out Leo Bodnar's fast rise time 10 MHz square wave modules.

http://www.leobodnar.com/shop/index.php?main_page=index&cPath=124&zenid=58b771673c16e5a14713fc19019d82e8

[ezalys]

I ultimately want to build a sampler from a circuit like that (modified for differential out)... hence why I need to derive sharp pulses from it.

[rhb]

I assume you have a *very* fast scope.  My BNC version was measured by Leo at 36 pS.  Nominally that's a 14 GHz spike.  My fastest scope is a LeCroy that goes to 1.5 GHz.  This portends an interesting thread if you tell us what worked.  I'm hoping to hack my LeCroy to 3 GHz.

[ezalys]

I have access to a modern 30 GHz sampling scope. I also have the sampling module from a frequency counter, which can be hacked into a high quality sampler with a bandwidth of 12 GHz. I also am not attached to retaining the full 12 GHz bandwidth of the square wave... if I can obtain a pulse train where the teeth fall 3 dB or so at 4 GHz I'd be very happy. 6 and I'd be ecstatic. Whatever I make will be on Isola FR408 so I'm not trying to be a hero.

[awallin]

seems like a comb-generator (impulse train) is slightly different from a sharp square wave generator (?)
e.g. http://www.vk2zay.net/article/41
or http://www.ke5fx.com/cg.htm

[tomato]

Step recovery diode.

[RoGeorge]

A tunnel diode might help, W2AEW have an intro with more resources linked in the video description:





Another technique I would look into is the chirped pulse technique used in optics. In theory, it should work for microwaves, too.





Do you have any specification numbers for the pulses you want to achieve?

[tggzzz]

Another technique I would look into is the chirped pulse technique used in optics. In theory, it should work for microwaves, too.

It certainly works for radars. In the 70s there were radars with 1MW peak and 10kW mean outputs.

[tggzzz]

You didn't mention voltages, so...

Take the output, split into two, invert one, have slightly different delays on each output, recombine them. If you have a differential output, the splitting is unnecessary.

Differential delay could be via wire length or a digital delay ic such as
https://www.digikey.co.uk/products/en/integrated-circuits-ics/clock-timing-delay-lines/688?k=programmable%20delay

The details are left as an exercise for the student.

[MrW0lf]

I have used shorted stub method (made from connectors lying around) just for playing purposes. Leo's pulser on 500MHz scope:

[jungle vegetable]

Both small-value SMD capacitors and inductors (just a few mm of wire across the microstrip or connector) work quite well for this application (and, specifically, with ONET8501P). Inductors made from wire are easier since some tuning is required to obtain maximum pulse amplitude. Pulses, obviously, will be of alternating polarity, so a frequency doubler (or a passive mixer in general) might be required afterwards

[jungle vegetable]

Here are some pulses made by shorting the SMA connector with a 1mm hex key that was lying around (100MHz square wave is from the ONET8501P, one division is 625ps, dots/samples are 5ps apart)

[Marco]

If you are making a PCB any way, just put a shorted stub on the transmission line.

[ezalys]

Why does a stub work? How long should the stub be?

[Marco]

Why does a stub work? How long should the stub be?

Good question, I thought I understood why ...  I only have a very approximate understanding though on reflection The principle is easy, the current pulse splits at the stub, then when the negative reflection comes from the grounded end it cuts the signal to the other branch to zero. So instead of a square wave you get pulses at the edges.

That said, how exactly it works with impedances I have no idea. When I put it in a simulator it tells me the stub impedance should be 25 Ohm to make it work perfectly, with other impedances you get multiple reflections and a screwed up trailing edge. Is the simulator screwing up? I have no idea.

PS. twice the length of the stub divided by (lightspeed*Velocity-factor) gives the length of the pulse, ignoring rise time.

PPS. if you wonder why the 36 mA current source, the ONET8501P can have a 900 mV signal, which with CML, it's internal 50 Ohm resistors and assuming it's driving a 50 Ohm terminated transmission line means it's switching 36 mA.

[mark03]

Another technique I would look into is the chirped pulse technique used in optics. In theory, it should work for microwaves, too.

It certainly works for radars. In the 70s there were radars with 1MW peak and 10kW mean outputs.

AFAIK with radar you never have the pulse-compressed (non-chirped) signal at high power---there's no reason you'd want to.  The chirp is collapsed as part of the receive/imaging path.

[jungle vegetable]

Btw, LTspice is quite excellent at simulating transmission lines (it even has lossy ones, but they aren't really needed if you do short runs of FR4 or coax). Shorted stub effectively acts as a filter to block the DC component of the square wave (could also be approximated as a parallel LC notch filter)

[Marco]

Shorted stub effectively acts as a filter to block the DC component of the square wave (could also be approximated as a parallel LC notch filter)

That's the theory, but the moment you start contemplating the time domain impact of impedance mismatches it gets a bit headache inducing. From the point of view of the feed in transmission line, the stub is in parallel with the output transmission line after all ... so there's an impedance mismatch. That said, I think the simulator (simetrix in this case) has it right and the stub should be 25 Ohm impedance. Or you could put two 50 Ohm impedance stubs either side of the transmission line so you can use the same trace widths everywhere.

[jungle vegetable]

That's the theory, but the moment you start contemplating the time domain impact of impedance mismatches it gets a bit headache inducing. From the point of view of the feed in transmission line, the stub is in parallel with the output transmission line after all ... so there's an impedance mismatch

Sure, that is only an approximation and to get accurate result you need more than one LC circuit (it think SPICE simulators do something along these lines anyway). Interestingly enough, stub impedance controls the falling edge decay -- impedance above 25 ohm generates (stepped) exponential decay while below 25 ohm results in decaying oscillations

[tggzzz]

Another technique I would look into is the chirped pulse technique used in optics. In theory, it should work for microwaves, too.

It certainly works for radars. In the 70s there were radars with 1MW peak and 10kW mean outputs.

AFAIK with radar you never have the pulse-compressed (non-chirped) signal at high power---there's no reason you'd want to.  The chirp is collapsed as part of the receive/imaging path.

You only have low power non-chirped signals in RoGeorge's graphic.

But I agree with your statement. I believe the radar's chirped Tx signal (2 in the graphic) was synthesised rather than stretched from a pulse (1).

[aandrew]

Marco, what software are you using for your simulation shown in the schematic and graphing output?

[mark03]

Another technique I would look into is the chirped pulse technique used in optics. In theory, it should work for microwaves, too.

It certainly works for radars. In the 70s there were radars with 1MW peak and 10kW mean outputs.

AFAIK with radar you never have the pulse-compressed (non-chirped) signal at high power---there's no reason you'd want to.  The chirp is collapsed as part of the receive/imaging path.

You only have low power non-chirped signals in RoGeorge's graphic.

But I agree with your statement. I believe the radar's chirped Tx signal (2 in the graphic) was synthesised rather than stretched from a pulse (1).

Are you sure about this?  I'm not familiar with the laser technique, but I heard the inventor interviewed on public radio and I thought they said the whole thing was designed for the purpose of achieving higher peak-power laser pulses.  I think they were saying it couldn't be done before this invention because of some peak-power limitation intrinsic to the lasing apparatus.

Back to the original goal, it would be cool if there were some sort of dispersive transmission line which you could drive with a chirp and get a real bang out the other end.  I imagine obtaining the ultra-wideband chirp (DC-ish to 10s of GHz) in the first place is the hard part.  Radar chirps are nowhere near that broadband AFAIK.

[ejeffrey]

Are you sure about this?  I'm not familiar with the laser technique, but I heard the inventor interviewed on public radio and I thought they said the whole thing was designed for the purpose of achieving higher peak-power laser pulses.  I think they were saying it couldn't be done before this invention because of some peak-power limitation intrinsic to the lasing apparatus.

With lasers people definitely recombine the amplified chirp into a short pulse with high peak power.  The reason for this is to strongly drive non-linear interactions that depend on peak signal strength. For instance, if you want to do optical comb generation with a non-linear fiber, you want high peak power to drive high harmonic generation.  Also, there are experiments people do on atoms where they need a high peak electric field compared to the electric field of the nucleus. 

Radar systems are (ideally) linear, so it would make more sense to transmit the chirp then apply the phase shift to bring it back to a pulse on receive.  This avoids non-linear interactions while preserving the good time resolution of a fast pulse.  If there are microwave applications where the high peak power pulse is desirable, you certainly could shift the chirp back into a pulse.

[jungle vegetable]

Back to the original goal, it would be cool if there were some sort of dispersive transmission line which you could drive with a chirp

There are NLTLs made from transmission lines and varactors: http://www.its.caltech.edu/~hajimiri/pdf/non-linear.transmission.pdf

Some specific NLTLs manufactured on GaAs MMIC processes achieve sub-picosecond risetimes