Microwave frequency response like it's 1971

[ChristofferB]

As I've been raving about over multiple posts, I'm building an X-band electron paramagnetic resonance spectrometer but I wanted to first be able to do frequency sweeps over a sample in a magnetic field.

Luckily, I have a Gunn diode oscillator that takes 5-35V in as tuning voltage and gives a frequency from 9.2 GHz and up to ?? probably 10-11.

Another lucky find is an electromechanical sweep gen, that just drives a very precise pot through a gearbox and a synchro motor.
link to info here: https://www.eevblog.com/forum/projects/amazing-old-school-electromechanical-ramp-generator/

So I made a 36V source from 9V batteries, and had the sweep gen sweep over the tuning range, with the sweep also going to the X axis of an XY plotter.

A waveguide mounted 1N23 crystal detector is used to detect the microwave intensity, and is connected directly to Y.


inbetween I have a little mica knife-edge attenuator and cavity wavemeter combo module from an old HP instrument, and as you can see, the cavity resonance is very easy to observe!

[coppercone2]

you might benefit from stabilizing the battery voltage with a LDO depending on the current draw of the device if you use a mechanical sweep generator

I only say this because I have no idea about the current consumption of a gunn oscillator
if its really low it should not matter

i did wonder how an old waveguide system looked like without pcb in everything and coaxial adapters everywhere and now I know

kinda wonder what would happen if you put some ultra precision starrett digital micrometer heads on it and stuff

beware the voltage drop on constant current sources if you do batteries, because I had something with a bunch of AA and discovered a lt3080 current source has a big voltage drop, so I Had to add another 8 pack.

if your equipment is that big you would like the battery brackets I figured out
to make you use loctite prism with sensitizer to glue black cheap polyethylene battery packs to a strip of acrylic (scratch both sides), then glue the acrylic to an aluminum bracket with either acrylic epoxy or more prism + sensitizer. You put screw studs on the bottom with thumb nuts on the battery pack and make sure its aligned so the bottom of the chassis pretty much keeps the batteries in to protect it from mechanical shock. Then you put a layer of thick strong tape on the bottom to act as a insulator for the batteries. If you use studs and thumb screws you can remove the bracket easily. Attach it with silicone wire and put some zip ties on it. Easy to make a 36V source with AA's for those who do not care for converters and small savings at the expense of power systems design ease being outsourced to the shop. If you use nimh you can charge it in series for a long time period I think. The bond between acrylic and polyethylene is so strong you can rip the PE battery case.



or if you have the money get the metal battery holders with rivet holes on them

[ChristofferB]

It is actually a starrett micrometer that drives the cavity. Originally it had a frequency cal chart but mine didnt come with that, and they're unique to each instrument..

Not too worried about the battery at this stage. The Gunn modules runs off of a 5V supply aside the tuning, so I think the tuning voltage current draw is fairly small.

But maybe the linearity would improve a bit, if the supply is loading down at the end points.

[coppercone2]

It is actually a starrett micrometer that drives the cavity. Originally it had a frequency cal chart but mine didnt come with that, and they're unique to each instrument..

Not too worried about the battery at this stage. The Gunn modules runs off of a 5V supply aside the tuning, so I think the tuning voltage current draw is fairly small.

But maybe the linearity would improve a bit, if the supply is loading down at the end points.

i wonder if a modern starrett micrometer head can be programmed to show GHz (look up table)

or maybe another relevant part that is mechanically adjusted (knife thing).

i think it would be amusing to have

[ChristofferB]

The knife thing is really bizzare. There is a slot cut in the waveguide, and it is then inserted into that via a worm reduction (no micrometer). I think the knife is made of metallized mica or mylar, and it's very efficient as attenuator,  when completely inserted I dont see any signal at all.

Dont know what that type of attenuator are called but it is new to me.

[coppercone2]

well you can defiantly improve that part

[cj]

Christoffer,

It looks like you are using a Toshiba SRX24 module as oscillator is this correct?

CJ

[ChristofferB]

Well spotted from that angle! Almost! It's an SRX15. They seem to be almost identical.

Unfortunately I have no further info on it, this was the first definite proof it even worked.

[cj]

AFAIK the SRX series modules are marine radar receiver front ends.
They contain a one stage RF amplifier, mixer and FET oscillator which can be tuned from about 9.2GHz to about 9.5GHz.

I suspect the detector signal you see is very small as the signal you are detecting is the leakage from the oscillator to the input waveguide.
If that signal is sufficient for you experiments that's fine.

CJ

[ChristofferB]

This is very good info but I actually thought all gunn modules were kinda transciever-like.

[cj]

Gunnplexers are used as transceivers and have a Gunn-diode oscillator with build in circulator and detector/mixer-diode. These are the modules used in automatic door opening systems.
Older marine radar receivers didn't have any RF pre-amplifier just a mixer and the Gunn-oscillator (5mW to 10mW).
These Gunn-oscillator could be mechanically tuned over about 1GHz (often less than that) and electronically tuned over about 150MHz.

The CR503 detector you use (I have an identical one) should put out about 800mV @ 9.2GHz with 1mW input this will give some idea what the signal strength is you are measuring.

[ChristofferB]

That's a very good benchmark number to have, thank you!

I'm not too worried about weak signals, as the output of the detector is fed to a lock-in amplifier with crazy sensitivity (100 nV FSD!) on the lowest range.

I'm happy for the info, though. I was wondering if the IF pin on the gunn module was an input or an output, sounds like an output from what you describe.

Do you know what the 'monitor' pin is?

[cj]

Detectors are nonlinear and stop working below a certain signal level. If at some point you want to measure amounts/concentrations of a particular chemical from the spectra you'll have to take this in to account.

The IF (Intermediate Frequency)pin is indeed the output.

I suspect the monitor pin is used to monitor the bias current of the mixer diodes (don't have datasheets for these modules).
In old radars two mixer diodes (1N23s or 1N415s) are often used and the bias current of each diode could be monitored.
The SRX module uses 4 mixer diodes and only one monitor pin so it looks like they monitor the average bias current.
The bias current is the result of the amount of RF from the oscillator.

I get my information from tearing down a few of these modules. The modules were given to me by a friend who repaired and maintained marine radars.

[ChristofferB]

Well it is way more than what info I could scrounge, so I really appreciate it!

I got mine from a HAM who just knew it output roughly 9.2 ghz when supplied with 5v bias.

[ChristofferB]

Slightly larger setup,

Here, the gunn module is connected via attenuator/wavemeter and a -30db coupler/detector to a magic tee, feeding into a home-made cavity, and on the reflection side to the crystal detector.

The idea is here to monitor the microwave power reflected out of the cavity.

The linearity below 20V is pretty promising for a frequency sweep. What I think I'm seeing is the cavity is very low-Q, the first valley being the top half of a very wide resonant region. Maybe a coupling screw infront of the iris could change this.

The sharp dip is once again just the cavity wavemeter, I think it's handy to leave in as frequency marker.

[ChristofferB]

Drilling a 7 mm hole  through both e-planes of the sample cavity did not impact the plot any.

However inserting a reference sample (7mm test tube of concentrated copper (II)sulphate) completely kills the signal.

Makes sense, it is probably an almost dead short. Commercial epr cavities has a coupling screw in font of the iris to the cavity, this may be why.

[cj]

Normally a dead short would reflect all power so you should see a maximum not a minimum.
How does your setup react if you replace the cavity with a short or terminator?
I'd expect a maximum signal with the short and a minimum with the terminator.

The function of tuning screw depends on distance to the iris.
It could be to match the cavity to the waveguide but the position of the screw would be strongly frequency dependent.
It could also be to vary the effective iris size (screw very close or in the iris) which determines the coupling factor.