It’s just a standard PA.

Awesome plots! On the microstrip, notice the ENIG slopes are not linear until 5 GHz. Pretty sure that's the frequency dependent permeability. The permeability drops quite sharply in UHF, but the effect would be hard to see in thin plating.

I had read a paper from the 40's where they measured Nickle coaxial lines, generating a nice curve of permeability vs frequency through 10 GHz. I'd like to repeat the experiment, comparing two Maury airlines, but I need to fine a used airline I can effectively trash by Nickle plating it.

FWIW there is a company making an RF friendly paraleyne; have not tried it.

https://www.imperial.ac.uk/media/imperial-college/faculty-of-engineering/electrical-and-electronic-engineering/public/optical-and-semiconductor-devices/pubs/2008_06_PIERSO.pdf

https://iopscience.iop.org/article/10.1088/0370-1301/62/6/305/pdf

Nickle naturally falls to ur=1 by 10 GHz, and is under ur=10 between 1 and 10 GHz; i.e. ur is very frequency dependent.

I've directly compared bare copper vs ENIG on the same coupon to 60 GHz and seen a good amount of dispersion between the two (of course the thin gold would cause that too), so I'm not entirely sold on ENIG as being harmless. I always use immersion silver, but that just means you need to store the bare boards properly to prevent oxidation prior to soldering.

I've done this a few times. You can plot the stability circles and then adjust the transmission line length between the amplifiers to avoid the unstable region. So you'll end up with a forbidden length. Just watch it as things become more unstable as they get cold.

Of course with that much gain, just put a 3 dB pad between the two, to accomplish the same regardless of phase. With 30 dB gain in the first stage, that will dominate the noise.

What do you mean by "large noise"? If you terminate the input with 50 Ohms, do you get the expected -174 dBm/Hz + 60 dB at the output? Is the antenna impedance making the first stage oscillate?

I plan on going to IMS this year, though paying myself as employer is too cheap.

Well you won’t understand how the rats-nest of VNA error correction works unless you learned flow graphs and Masons gain rule.

https://www.rfmentor.com/sites/default/files/NA_Error_Models_and_Cal_Methods.pdf

Yes, your last sentence is correct. A normal M/F cal uses a zero length thru. Those adapters should be phase matched. If you want to do a M/M cal you could attach a F/M adapter on port 2, do a M/F cal, then swap the adapter to M/M. Any errors introduced are the slight differences between the M/M and F/M adapters.

There are also adapter removal cals, which fully remove the adapter. The 8510 allowed this by combining two cals sets with the adapter on either port. With the eCal, most of this has been obsoleted with the exception of mixed media such as coax on port 1 and waveguide on port 2.m

There was actually a method to do two-tier calibration fixture characterization using adapter removal and reading out the error model from the VNA. This is now lost knowledge.

I believe a non-zero length thru is essentially LRL cal.

An association with IEEE will likely be stuffy and bureaucratic; you already mentioned committees and stamps of approval. I’d much rather have the politics and content of questionable taste. It also seems all the forum content was lost when IEEE re-wrote the site.

Looking at EE enrollment over the last 30 years, MTT is definitely not going to grow. Not sure why they’d even try to monetize the content, or charge a subscription. This isn’t exactly a huge industry. If you had a YouTube channel with every IMS attendee subscribing, it’d be 10k subscribers, which is laughable small.

It probably depends on how a single amplifier fails. Say you had a 1/4 wave combiner (i.e. a Wilkinson without the resistor). If one amp fails with a short circuit at its output, that places an open circuit at the summing node, so the working amplifier just sees a 100 Ohm impedance mismatch. If the amp fails with a short circuit, the working amp sees an open circuit. If you added additional quarter wave lines at the amps, a short failure would present a short to the working amp. Perhaps a short is a better failure than an open since you have over-current (thermal problem) rather than breakdown from over-voltage.

I’m sure there’s topologies that optimize degradation. I wonder if quadrature would be better.

Yes, though Z0=sqrt( (R+jwL)/(G+jwC) ) so a high R will boost the impedance; it’s not artificial. You may have to increase C to compensate. Though I would expect R to be very frequency dependent. Can you sweep it on a VNA, and see what happens at very low frequency such as 300 kHz? You may see a spike in Z0 at very low frequency, which will be exasperated by the R since you are getting sort of a divide by zero.

Use a VNA in time domain mode to gate out the reflections.

The connectors are mechanically compatible, but you should not use a 3.5 mm cal kit on a 2.92 mm connector as you will have an electrical discontinuity at the reference plane. Rather you use the adapter to get the proper interface.

No, you should not use a dielectric filled 2.92 mm on a 2.92 mm cal kit for the same reason. Sounds like a bastardized 2.92 mm.

Now SMA cal kits are rare (as they are an abomination) so they expect you to use 3.5 mm on SMA. Though never connect an SMA to a 3.5 mm cal kit, especially an HP. Use a 3.5 mm jack saver.

I'm pretty sure thermal noise is half phase noise and half amplitude noise. In other words, -177 dBm/Hz from phase noise and -177 dBm/Hz from amplitude noise gets you -174 dBm/Hz thermal noise. So technically it's a low phase noise splitter just because it has low loss.

Absolute lower limit of phase noise for a passive, room temp device is -177 dBc/Hz, and if that splitter is 0.6 dB excess loss, I'd think that would put the phase noise at -176.4 dBc/Hz, but this lists -175 dBc/Hz.

That said, I'm sure piezo effect from substrate dielectric or capacitors will cause phase noise, so maybe they use low-glass substrates or certain caps.

This is interesting. Maybe there is more to it:

https://www.researchgate.net/publication/321324476_Phase_noise_measurements_with_a_cryogenic_power-splitter_to_minimize_the_cross-spectral_collapse_effect/link/5b1be357aca272021cf46626/download

So the above paper describes how an isolation resistor in a power splitter introduces a differential thermal noise that messes up (underestimates) the DUT phase noise in a cross-correlation phase noise analyzer. They fix it by cooling the splitter to 4K but 77K may also work.

Maybe the Holzworth splitter is a purely reactive type (though not with 38 dB isolation), specifically needed for their analyzers, though the paper mentions issues with reactive types too, though I don’t understand what they mean.

It’s also neat to see the splitter isolation decrease near 4K, which makes sense as the resistor is changing.

Note I’m using the term power splitter, but maybe a power divider would behave differently.

Maybe use a TDR. This guy has some lower cost pulse generators and probes. In this video he demonstrates probing some traces:

https://youtu.be/kXpYIczdta8

Probe across the resistor to measure the impedance at the probe tips. Of course you could probably just use the probe with a VNA too in TD mode, and gate out the response just after the probe tips, but in essence you can isolate the measurement just around the resistor, limited by bandwidth or rise time.

Even if you can’t probe directly across the resistors, if you can probe the splitter outputs differentially, you’d see the resistor further down the lines. This is assuming the resistor is closer to the probe than the stuff the splitter is feeding, as the pulse would travel in both directions.

Some multi-line TRL compute a frequency weighted average, thus if you overlap the lines, you won’t get the discontinuity at the transition frequency. Yes though, it allows wider bandwidth than a single line.

I assume you are trying to do this in-situ without access to any of the ports?

I was messing around several months ago with non-destructively measuring dielectric constants of curved plastic radomes. You can cut a piece of 1/2 wavelength copper tape and lay it on the plastic, then take two small coax loop probes and magnetically couple near the center of the dipole, looking at Q and resonant frequency. It easily tells the difference between ABS and anti-static plastics.

Maybe you can couple to the output traces, as the isolation will degrade with the resistor open or shorted.

I’d think you’d need to write something in Matlab or Python to get the 3D orthogonal view like that.

Have a look a Baudline. It does sliding FFT analysis, and you can have the time domain displayed with the spectrogram. You can slide the window along in the time domain and see the spectrum change. You can also pipe it signals from GNU Radio via a FIFO to run in real-time.

https://www.baudline.com/

GNU radio would also work as you can plot time domain (with triggering), spectrum, and waterfall.

AWR VSS will also run in real-time (continuous) and you can move tuners to change signal properties.

There is a open source copy of Baudline called Inspectrum, but it is not as fancy.

Is there a limit? Traveling wave antennas are tapered transmission lines. Maybe they have to be non-TEM lines? If you had a high impedance transmission line made of perfect conductor, with no dielectric loss, then there would be no loss. Those transmission line formulas do not account for radiation. So if a transmission line is always driven at a low enough frequency for TEM, I assume that means it can never radiate?

I don't think so, but now I see you can buy 600 Ohm ladder line, so I don’t know. Skim through this, particularly section 5 where he looks at TM waves on two wires lines. The TM cutoff is lambda/2 wire separation, so it'd be lambda/4 for wire-over-ground if you treat it as an image. Really no different than microstrip.

What's more interesting is the TE mode for really large wire diameters, but that's expected as it's in microstrip too.

I'd just add a disclaimer saying it's for TEM only.

https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwiOt6Cd4PvuAhUWbs0KHVBKBKAQFjAAegQIBRAD&url=http%3A%2F%2Fwww.jpier.org%2FPIER%2Fpier40%2F01.0206091.Green.pdf&usg=AOvVaw3KI2NddabYy80mc2rx2-C8

I was also looking in the Ramo Fields and Waves book, and they had the reminder that TEM wave impedance can be derived by the static DC fields. So it reasons that the wire separation can go to infinity if the frequency goes to zero.

This, and that other post about quasi-TEM causing nulls for long lines, really makes me want to redo my whole education; dumber everyday.

The WR-15 worked well. For a 3” length, the insertion loss was 0.3 dB, compared to 0.2 dB for the 3” reference. The WR-10 did not work well.

The surface roughness models in the 3D simulators do not work well; they overestimate the loss for the rather rough surface.

I had previously stated the flange was polished. It was just machined flat. I suppose you could polish it, but the machining is probably adequate.

I don’t know how high-Q structures would perform. I’d like to make a WR-15 filter or resonator; just no time.