Yttrium iron garnet

YIG oscillators were quite the thing back in the day, but I'm guessing they've been completely superseded by now to get to ever higher frequencies. Seems we've gone from -

R/C to L/C to Xtal to YIG since about 1900. Did I miss any development(s) out pre-YIG?

Tuning forks, SAWs, BAWs, mechanical ceramic resonators, dielectric ceramic resonators, coaxial ceramic resonators, sapphire, cavities, atomic things.

This misses Jan Panteltje's thread "Small magnetic tunable filter for 6G and beyond" which is about Yig being used today.

As usual, John Larkin hasn't noticed that most of his list can't be rapidly tuned to a different frequency.

That article makes it seem like YIG is some revolutionary, new, emerging technology!

Use of YIG filters as a replacement for varactor tuning could turn out to be significant. 2022 Microwave Journal article:

Of course it does. University researchers always want to create that impression.

They may need to add a Hall plate to their stack to keep track of the actual magnetic field where it matters.

The VIDA oscillators still look like giant expensive power hogs. They don't specify modulation bandwidth on the data sheets that I see, but it must be terrible.

One can't modulate a hundreds-of-mA electromagnet very fast.

An LC osc with a varicap is a more sensible VCO. Narrowband, one can varicap a coaxial ceramic resonator, or a PCB ring oscillator, or something. Cheap and fast.

Of course, it's inherently difficult to modulate a high-Q resonator fast, even without an electromagnet in the way.

In fact they tweak quite a compact permanent magnet stack. How fast they can do it isn't discussed (or at least if they did I didn't notice it).

Varicaps are horribly non-linear. Narrow-band is always easier than wide-band, but the YIG tuning scheme is good for at least a factor of two frequency range and once you've got that you can use counters to go down from there until you run out of dividers

You don't need an electromagnet to get the fields required. You may want a non-conducting permanent magnet to proved the bulk of the field - or it might be enough to split your magnetic path into lots of parallel wire magnets insulated from one another. There are ferrite permanent magnets which aren't all that electrically conductive.

This reads more as if you don't want it to work - it's the sort of contribution that gets people chucked out of brain-storming sessions.

And far, far noisier than the best YIGs.

It isn’t, actually, at least over a restricted range. If the resonator obeys a differential equation, you can modulate its resonance much faster than f_0/Q by changing L or C.

Resonators with significant time delay, such as a long piece of coax, aren’t as friendly that way.

Cheers

Phil Hobbs

Coaxial ceramic resonators have Qs in the thousands, and low tempcos.

If you can find one at the exact frequency you need. YIGs have a huge tuning range.

IIRC you also said that they’re piezoelectric.

I’m not saying that YIG is the answer to everything, but for some things it’s amazing and (AFAIK) unique.

Sure improves spectrum analyzers!

Cheers

Phil Hobbs

The CCRs are high-K, usually shorted, transmission lines, not piezoelectric. Prop delay is a tiny fraction of c. You can TDR them as such. Z is usually in the 10 ohm ballpark.

No argument, but they will always be big and expensive slow-tuning power hogs, which is fine in a spectrum analyzer.

RF synthesizer chips are pretty amazing these days too. They make a pretty good first LO too, but they are small and cheap.

I wonder if the latest SAs use YIGs.

Except that they don't have to be, as

Not all that cheap, and there's quite a bit higher harmonic content to filter out.

Keysight's do.

I gather many spectrum analyzers these days mix successive slices of the spectrum down to where an ADC can acquire the whole slice, and the remaining processing is all software FTs.

No need for YIG oscillators, and the LO synthesizer needs only coarse steps.

Jeroen Belleman

Sure, that’s the software-defined radio (SDR) approach. Works great for many things, but good close-in phase noise is not one of them.

Cheers

Phil Hobbs

Yes, digital IF.

The price of a several-GHZ sa has dropped by over 10:1 in the last 20 years, mostly from using a lot of digital stuff.

You can get an impressive 8 GHz RF synth chip for about $5 now.