X band cavity design and microwave safety

[ChristofferB]

Hi!

I'm slowly assembling an electron paramagnetic resonance (EPR) spectrometer, which involves coupling X-band microwave energy into a resomnator inside a strong magnetic field.

Setup is going to be something like  [Gunn diode oscillator]---->[magic tee/circulator]---->[detector]
                                                                                                            l
                                                                                                            l
                                                                                                           V
                                                                                                         [iris]
                                                                                                            l
                                                                                                            l
                                                                                                           V         
                                                                                                [cavity/resonator]              

So basically the resonator is the only X-band part needing homebrewing.                               

The iris is commonly just a nylon screw with a small metal (ferrite?) slug at the end that can be moved in and out in front of a smaller-than-waveguide hole.
The resonant mode on commercial solutions usually TE102. There needs to be a hole through the cavity to allow for the insertion of a 6mm sample tube.


Here's the actual question:

A) Could one just cut a lenght of copper waveguide with flange, to the desired lenght of the cavity (and what would that be?), solder the other end shut with copper foil, drill a hole through the center of the H plane and be done with it?

B) Can the iris just be copper shim stock inserted between two flanges, the non-cavity end having a ferrite tipped screw?

C) How serios should microwave safety be taken with an x-band gunn diode? From what I can read, it's a fairly low powered device in a band that doesn't really resonate with human tissue.



This is what a commercial X-band EPR probe looks like, for reference:



Thank you in advance!

[radar_macgyver]

With copper/brass waveguide tubing, yes you can just solder a plate across the ends and have a nice cavity. You can model it in something like openEMS to verify the dimensions.

Commercial Gunn modules have the diode in its own cavity with an iris for coupling. The iris has a tuning screw some distance away to control the coupling factor, but they are just regular stainless screws.

As for the safety aspect, Gunn oscillators are in the 10-20 dBm range, so won't pose a general heating risk. However, avoid staring into open waveguide ends or horns to avoid exposure to your corneas. Heating does not correspond to resonance - pretty much any band will cause heating in soft tissue. Corneas are high-risk because they lack blood flow.

Edit:
Just curious - why Gunn oscillators, when low-power amplifiers are cheap and cheerful, as are low-cost synthesized sources?

[ChristofferB]

Thank you very much, that's good info!

I'll look into openEMS, I assume a waveguide cavity should be a whole number of half wavelenghts long?

Gunn oscillators simply because that's what I have - a little Toshiba X-RX15 module. It doesn't have any visible tuning screws by the way.

Also most of these EPR spectrometers uses either Gunn oscillators or Klystrons.

Aiming for something like this:

[radar_macgyver]

Yup, half wavelength if you want a TE₀₁₀ cavity. If you're lazy like me and didn't want to derive the formulas I forgot since grad school, here is a reference (see section 3.5).

I did a bit of reading on EPR, gave me the same headache as trying to wrap my head around MRI It looks like frequency stability of the source is less important since you sweep the magnetic field and observe the resonant peaks, right?

The diagram shows an isolator at the output of the source, this is done in low-cost Gunn sources by controlling the coupling factor from the Gunn cavity with a screw (and is therefore not a true isolator). Additional screws are provided to control the cavity size and hence the diode's output frequency. The one I have here at home also includes a varactor for electronic tuning. If you look on E-bay you may be able to score a circulator from an old marine X-band radar, maybe even the detector modules they use.

[ChristofferB]

Good reference! That one goes in the folder!

--Even lazier cavity design: Couldn't I just make a waveguide 'slide whistle' - have a close fitting copper plate (backed by some sponge or coax braid) on a stick and move it in and out of a lenght of waveguide to find resonant lenghts?  Basically a super simplified version of a cavity wavemeter


Thanks!

[radar_macgyver]

You could do this with a regular cavity, plus a couple of tuning screws, which would give you much better control over the cavity size.

[ChristofferB]

The problem is the cavity is very constrained in width, as it needs to be placed between the poles of a powerful magnet, and that space also needs to accommodate some sweep and modulation coils. Which is why using the H plane of some WR90 would be ideal.

[ChristofferB]

Another safety concern I've been considering: Is waveguide ever made from beryllium copper? I know some flexible waveguides are.

Machining, drilling and filing beryllium copper is a BAD idea, but this fear may be unfounded, as I can't find any references of standard waveguides being made of anything but copper, brass, aluminium or, rarely, stainless.

[radar_macgyver]

The only time I've seen warnings about Beryllium are on BeO windows (on high-power tubes) and BeO insulators used on (older) high-power semiconductors. Both are rare. Never seen anything about BeCu stuff. Maybe someone with an EMC testing background can chip in?

[coppercone2]

I heard its nasty and i heard from people that work no spark jobs using beryllium copper hammers that their a big hazard if you decide to clean it up with a sander. They use them because their non sparking, but when you use copper like a steel hammer head its going to deform no matter what kind of alloy you use.

With a normal copper hammer you can clean it up on a belt sander or whatever with little risk, but if you do it with a beryllium copper hammer its a bad thing. https://www.amazon.co.uk/Ega-Master-Beryllium-Copper-Hammer-Sparking/dp/B07MSFNBLB

But in general brass dust is bad too. Yes normal brass if pretty bad if you sand it (often wet sanded). You don't hear much about sanding brass because it polishes up really nicely with brasso or whatever.

As a general case, anything you read about 'metal fume fever' applies to the base metal powder, since basically under normal conditions all metal fume exposures are actually particle exposures (it condenses, you don't actually get any metal in gaseous form), based on what OSHA says. So if there is any air float component, the toxicity is pretty much the same.


The only EMC BeCu thing I heard about is RF fingers, in addition to the connectors pins that everyone knows about.

The only waveguide type I could think of would be the corrugated kind.
https://patents.google.com/patent/US4429290A/en

This catalog claims its available to order:
https://www.microtech-inc.com/Products/WRD-350-Waveguide-Flexible-Double-Ridged/p3s25/

I don't think it makes sense to make a BeCu wave guide that's not a weird flexible one because the safety controls for manufacturing are expensive and wave guide is the type of component someone might actually modify in a lab, so you don't want to sell something you know can poison someone. Unless its ancient stuff where no one cared. The safety behind BeCu fabrication is taken seriously in the modern world, which makes it expensive.


Obviously be careful sanding brass because that would be my go to for making a waveguide, you can do very nice sanding and cutting and soldering work on brass and make a precision assembly without manufacturing difficulties (you can use a chisel even, or is this trick for bronze? never tried anyway). I recommend you do as much work as possible with tools that do not make fine dust or fumes, then do as much finishing work as possible wet/ventilated. If you are new to shop work, you will find that copper is irritating to work with, while brass is very pleasant.

[ChristofferB]

Thanks for the insight! I agree in that it wouldnt make much sense for it to be BeCu but almost all 2nd hand x band waveguide parts I come across are 1950s and onwards vintage radar/marine/aircraft parts, so you never know.

The waveguide piece I have looks to be brass. I dont plan on grinding it with power tools, I think I can get close enough by hacksawing to rough lenght, hand file to within 0.2 mm or so, then wet sanding flat on a glass plate to 1000 grit,, then solder closed with copper sheet.

I didnt know metal fume fever was a risk without heating/grinding! But yeah it makes sense that it's not actually "fumes". Vapor pressure at room temp is through the floor for most metals.

[coppercone2]

i think its that the amount of product you get that you can inhale is small compared to the amount you vaporize during welding

I don't think I would be too scared of brass, unless you plan on doing it for a career. I meant don't get medieval just throwing squares of metal on a belt sander to make the shape you want (with a fast sander its tempting, you can do it with steel). (the metal eating sanders seem like very popular shop projects on youtube, in lieu of using jeweler saw, plasma cutter, coping saw, etc). I see some equipment knife makers made where it hardly even makes sense to do anything but just sand the whole thing down when it has a long fast sharp belt on it. (maybe you call this a belt grinder not a sander). Seriously those 72 inch sanders are pretty crazy in terms of how fast they devour metal.

Your work plan is pretty good. All I can think of is that if you had a precise saw (adjusted band saw or cold cut metal saw with vise) you could make a closer cut to reduce filing  (its probobly a negligible amount of work to file a precise hack saw cut anyway). The hard part will be precision soldering, as not to have a fillet (I have seen very skilled solder workers do this some how, but they have a extremely good feel for how solder flows, along the lines of stained glass work).

[ChristofferB]

An inside fillet is an acceptable deviation from the resonator shape, I think. I plan on oversizing the copper end plate, standing the waveguide on it, then blowtorch some electronics solder all the way around (higher flux:solder ratio than thick plumbers stuff). I REALLY should silver solder something like this, but that's not going to happen.

The best solution would to solder on a brand new WR-90 flange in the correct lenght and then just bolting on an end plate, that way the end will be perfectly plane.

[ChristofferB]

Oh, and to answer my own question, I found a paper "Improved rectangular cavity for EPR in TE102 mode" in which the cavity is WR90 dimensions and then 4.15 cm long - which they call one wavelenght for some reason. I don't get that, but 4.15 cm might be a good lenght.

[coppercone2]

you might not be too happy with silver solder because its a high temperature process and without nitrogen purge, the cleaning is more difficult then solder cleaning is, not to mention when the temperatures get high enough (red heating) you get might still get RF surface finish problems because of copper 'separating out'. I think it would require internal repolishing. Not sure what brass will do.

i also calculated a while back that the effect of fillet on impedance in the waveguide is small until frequencies are high. But I still don't like it, makes the project feel super half assed.

[ChristofferB]

I agree, I don't have all the oxy-fuel torches and stuff to do silver soldering, but I'm pretty sure that's "the" correct way of attaching waveguide flanges to waveguides. I think there's a US signal corps waveguide 'plumbing' handbook, and I think that's where I've read it. Soft solder should be decent enough, if the waveguide isn't supporting anything heavy.

[ChristofferB]

I had the fortune to be allowed to examine a commercial EPR spectrometer cavity today at my university. Taking off the side caps, it has a potted coil on each side, these are the modulation coils, but look at the white substrate below it. The cavity is silver plated on the inside, and I think the sides must be made from either some ceramic pcb or Rogers substrate. The point is the sides that are normal to the magnetic field must be thin walled to reduce eddy currents from the 100 khz modulation coils. The silver plated ceramic/teflon sides are just edge soldered onto the rest of the cavity. It might be worth making a cavity like that; cut/grind off the edge and solder some silver plated pcb onto there. The hole in which the test tube is placed is quite wide bore (under the black screws top and bottom), I didn't have calipers with me but I'd say 10-12 mm.

[ChristofferB]

I've received some X-band parts! Among those this strange homebuilt cavity thing. No idea what he used it for, but it shows a very simple way of fabricating an iris.

It's difficult to see, but the waveguide has been cut about 12mm deep all the way through the H plane symmetrically on either side. In the slots, brass shims has been soldered to form an iris. A tapped hole with a brass screw forms the 'shutter' or coupling adjust.

Not sure I can reuse this particular piece of waveguide, as it has 3 consecutive iris' 1,5 cm apart but who knows. What do you think? And does it matter if the iris narrows the E or the H plane?

After the 3 irises, the cavity has a plunger, setting the cavity lenght from 83 mm to 96 mm, or from 5.2 to 6 half wavelenghts at 9.4 GHz. I'm wondering if it's reusable as a first cavity. Would save me some trouble.

[ChristofferB]

Also, a more concrete question: when a magic teeis drawn as it is in the schematic above, which port goes to what? port naming and numbering is NOT consistent at all. And most references I find uses magic tees for either combining or splitting signals, not as a reflection isolator.


 Thanks for your interest!

[coppercone2]

there are arrows

Also that wave guide is in serious need of cleaning. I would put it on the buffing wheel before working on it (contamination) (plug the holes). I have one ancient big high power filter that looked like that. Not a lovely patina. However when I cut one of them apart when I was very curious as to how the insides looked like (it was capped with coaxial feeds), the inside was nice looking. 

Also how old is that university equipment? It looks like the plating is failing on it. I have not worked with bare waveguide, only sections with coaxial feeds, but I wonder if you should grease sections before mating them with deoxit grease or something. The old pieces I have looked like brass bells.

I think I will try to electroclean the inside of this open piece of WR90 i have. Also on the interior of mine, there is no solder spill through, the solder joint looks flush.

[ChristofferB]

There most definitely are not arrows on the actual tee, unfortunately. I think the two colinear arms should be signal source and reflection detector respectively, but whether the cavity should be at the sum or difference port I can't really wrap my head around.

I agree, waveguide looks abysmal,  insides looks much better though. And the university probe is pretty new, no more than 30 years I'd guess. It's not the plating, it's solder I think.

[coppercone2]

btw what exciting uses does this machine have, I have been looking for a engaging microwave project for a long time, they don't really excite me

[ChristofferB]

It's called an EPR spectrometer - "electron paramagnetic resonance" - you sweep a magnetic field (around 3500 gauss for X-band) through the sample in the cavity, and once you hit the resonance field strenght of an unpaired electron (paramagnetic ions, organic radicals etc) the electron is kicked up in a higher spin state, absorbing some energy. This causes a dip in energy (kinda like hitting resonance on a grid dip meter) and you can plot that as a spectrum.

Different spin systems will give different sets of spectral peaks, and these peaks split up (instrument resolution willing) into very fine peaks called hyperfine coupling - stemming from electron being spin-coupled to other atom nuclei (especially hydrogen).

It's a more obscure technique than the related MR/ NMR, but decent resolution can be achieved without a superconducting magnet so that's my motivation.

as well as, like you, finding an interesting microwave project that's NOT a 144 mhz down/upconverter transciever. I'm a bit bored with HAM stuff tbh.

[coppercone2]

so do you sweep frequency?

I.e. waveguide has a stable pass band

oops, 3500 gauss is not that much magnetic field. thought you said 3500 tesla

So its like S parameters with the sample in h field flux?

I made one good antenna and I figured out how many evangalists were on the air and gave up with reception lol


What interesting materials parameters can you study?

[ChristofferB]

Sweeping frequency is tough to do with good repeatability. Sweeping a DC electromagnets current with a slow (minutes to hours) ramp gen is the option people has gone for. More recently, short bursts of RF energy and then fourier transforming the decaying echo. EPR is still usually CW swept field.

It has been used to fingerprint authentic dyes and pigments in ceramic glazes and artwork to detect forgeries, medical imaging, and investigating chemical reactions. Reactions involving radical mechanisms can be monitored and the shape of the radical molecule can be deduced from the spectrum. It is pretty neat actually.