Experimenting with waveguides using the LiteVNA

[coppercone2]

the how to web pages with formulas but not calculators say that the major diameter or longest diameter is used for the equation, not the area. But the calc I found shows area. maybe tonight I can try spread sheet comparison

[joeqsmith]

If you run the numbers for the table provided, they calculated the area from the longest surface.   They show a WR90 horn with a width of 2.66" or 67.56mm.  Close to the 80mm example I used.  At 8.2GHz, their table shows 0.25 meters.   Calculating it by hand, I get 0.24972.   

https://www.atmmicrowave.com/waveguide/horn-antenna-standard-gain-wide-band/

[coppercone2]

That makes sense. Maybe I will take a crack at the equation from that video if its appropriate, been meaning to learn some of the harder math behind these things, I put enough effort into trying to make them. not the worst math I run into anyway

[joeqsmith]

Replacing the copper foil tape with my homemade iris.  Still pretty crude but better than the tape.   Data again with VNA uncaled.  Iris1, sweeping from 8.5-11G.  Notch is really narrow around 9.1G.   As we zoom in you can clearly see the cavity resonate.     

[joeqsmith]

Looking at the drift.     

[coppercone2]

that made me remember that I found a optical iris once, they use greasy o-rings to adjust a square aperture, its a very smooth mechanism that is simple.. just friction on damping grease.. it basically drags a plate in a groove in a indiana jones way, like if you were pulling a long spaghetti over a news paper.. octopus style..so you just rotate a wheel and it rotates a o-ring that tends to drag the plate in one direction and its sticky enough to stay in place.


What is the drift from anyway? DO you think its something to do with mechanical state or mainly the instrument and temperature? Hard to imagine the screw being responsible given how stable a micrometer is.


what is the sweep duration? That would be interesting to know the time constants associated with stabilization.

[joeqsmith]

Reminds me of some idiot telling me to use some gold something or other contact gunk on my RF connectors.  Just try it, they would say.      Guessing a ham.

Drift wise, hard to say without running more tests.  The PDRO is tied to the GPS.  The LiteVNA has no reference.  Most of the bits are copper or brass with a fairly high TC.   Currently the houses temperature is fairly stable.  No A/C or furnace.   Then there is the whole paper clip, me moving things around on the desk....   

Easy enough to get rid of the paper clips and use the stands.  If I normalize the data, we can get a better view of the drift in the notch area.   Shown with the first half hour of warmup.    Now that it has warmed up,  I'll let it run for a longer time and use 100 avgs to clean up the noise.   Sun is coming up and the house will start to warm up so I expect to see some amount of thermal drift. 

[joeqsmith]

I set the LiteVNAs IFBW to 800Hz and let it log for a bit over 2 hours.   You can see I didn't have it tuned up quite as tight so we see it swing both directions. 

I think to narrow it down, I would just run it with the old Agilent.  That system is locked to the GPS.   I could tune the cavity below 9GHz and run it direct to remove the extender.   

If you start seeing me post gold plated parts, it may come to that but for now, good enough.   

[coppercone2]

too blingy for me

 I love gold. The look of it, the taste of it, the smell of it, the texture. - I love gold so much.....

[joeqsmith]

too blingy for me

Better Q. 

[coppercone2]

Gotta start calling it g factor when it comes to gold plated waveguides

[joeqsmith]

The vintage HP adjustable short has a lot of hysteresis.   The new design will use a Starrett micrometer to set the plungers depth.   I honed the plunger and tried it out to make sure that I have enough adjustment before I make the mount for the micrometer. 

[joeqsmith]

Made up an end cap for the micrometer and tried it out.   

[joeqsmith]

Before I take it all back apart,  I wanted to repeat the drift test.  So once again, I tweaked tight and let it run for about an hour.   Again, about a 20dB shift.

[joeqsmith]

While I was confident this is the LiteVNA,  I wanted to prove it.  So, I left everything running and tuned the cavity again and again normalized the data.  I let it run for several minutes than blew the heat gun over the LiteVNA.  Nothing major.  It still has the plastic protection over the LCD and that didn't deform. I just wanted to if a slight change in temperature would effect it.  Sure enough. 

[joeqsmith]

Counter is referenced to the GPS.  LiteVNA directly connected to counter and set to 1GHZ.   Then repeated the heat gun test.   It's too bad the LiteVNA didn't support an external reference but $120.....   

[joeqsmith]

Using the magic T and other parts from the LRL training kit, removed the coaxial coupler to directly measure S21.  Similar to results we saw when using this same cavity with the T and directional coupler.   This cavity does not have near the Q and of course is not sensitive to the frequency drift of the LiteVNA. 

[joeqsmith]

Adding the stub tuner, similar to what I plan to use with my cavity to improve the match, increasing the Q.    Even though the Lite was running for several hours, the drift effects the dip right away.   

My lab partner seems less enthused.

[coppercone2]

I wonder if you put a isolator on there if it would still drift on load, doggy won't care unless you microwave some hot dogs

but useful for power electronics he might smell a early fail way before anything happens on the equipments

[joeqsmith]

I don't think it would have any effect.   In this case we are working with a very low reflections, well, that's the intent anyway.   We are also work at very low levels and the VNA has no problem driving a short or open load.  We even use these to calibrate them.    With such a high Q, the notch is very sharp and any change in frequency will move us towards the skirts.   We could make a little temperature controlled box for the LiteVNA, or I have one that I have made a fair number of modifications to.  I could maybe remove the oscillator and run it from a stable source.   

[joeqsmith]

Directly connecting the LRL setup to the PNA which is referenced to the GPS.  Noise is much lower and I narrowed the span.  Of course the p-p noise is higher but there is no discernible drift. 

[joeqsmith]

I finished up the cavity.  Just in case the LiteVNA doesn't work out and I need to run it on the old Agilent, I opened up the iris to allow it to be tuned below 9GHz.   

Plots showing S21.  Started out tuning it without the sample vial (yellow).  Then inserted the vial (violet).  Resonance frequency drops.  With the micrometer, I can trim the size of the cavitiy to compensate for it and of course have to trim the iris as well.  Left it offset a bit to make it easier to see (red).   

I would like to find some shorter vials but not having any luck.  30-40mm length with a 6mm OD.  Quartz. 

[coppercone2]

very interesting. When you test substances watch out for hydrates I assume that will throw the machine off the most.

I assume water messes with it big time, and many materials form hydrates.

With quartz you can of course pour molten salt in there probobly.

One that might show results could be magnesium sulfate. Bake a portion of magnesium sulfate at approrpiate temperature to decompose the hydrate, and grind into powder, seal in tube. Grind out of box magnesium sulfate into powder of the same consistency and put in tube.

[joeqsmith]

Not too concerned with temperatures as I plan to conduct everything at room temps.   The quartz has much better impurity.   

[coppercone2]

I mean the samples could be hydroscopic and I believe water effects that quite a bit. For inside the tubes. If you bake some salts at the right temperature they will release chemically bonded water and take up a new crystaline form. I could see this messing with someone that is running samples in a microwave field. Some have a few phases of hydration so you can compare hexa to dual to pure say epsom salt

I think its hepta-hydrated (water of hydration), but you can get dua-hydrated and then non hydrated once its hot enough. Its the chemistry thing with the brackets and H2O like chemical*7[h2o]

look at the hydrates section of this page, they are meta stable and it appears dry even if its hydrated (odd property, its not like paper that it gets mushy)
https://en.wikipedia.org/wiki/Magnesium_sulfate

If you read the whole page you will want to attach a refrigerator and a pressure chamber to your sample... so much fun from a box of epsom salt, there is so many configurations. Ice is like that too I think. Just don't make ice-9 (ima call it α-ice9 because we found a boring ice-9 already in some lab), at least not on this continent please...