Small magnetic tunable filter for 6G and beyond

YIG filter and resonators have always been a bit exotic. Maybe this will make them common-place. And more compact, hopefully! The YIG was tiny, sure, but the magnet wasn't.

Jeroen Belleman

Yig filters and oscillators have been around for ages. It seems to me that tuning them with a magnetic field is messy.

YIG-tuned VFOs are the champs for low close-in phase noise. My HP 8566B’s noise floor at 1kHz is a good 30 dB better than any SDR-style analyzer.

If they manage to get them down to Digikey-level practicality without screwing that up, it would be huge.

I wonder if you could use a mag amp sort of structure, with a rare earth magnet biasing some cleverly designed bits of saturable ferrite, plus some small coils changing the effective gap in the magnetic circuit.

Fun to think about.

Cheers

Phil Hobbs

offset

is a good 30 dB better than any SDR-style analyzer.

How can one keep a magnetic field stable to parts per billion?

Seems like ambient 60 Hz fields and temperature changes and tiny noises in the coil current would dominate. It's hard to regulate a current to parts per million.

Qs are low too.

Does your HP have a big ovenized mu-metal box inside?

Normally unnecessary for a YTO, I think.

A well-degenerated BJT with a 2- or 3-pole lowpass on the base makes a deep sub-Poissonian current source. One of our laser drivers has a noise floor below -190 dBc/Hz at 400 mA, about 24 dB below shot noise.

You do have to handle the low baseband somehow, of course. For the laser driver it’s an op amp and voltage reference, and for the YTO it’s a PLL.

Cheers

Phil Hobbs

Is the yig phase-locked to an XO? I guess that would make a nice jump-tunable first-mixer oscillator. Something else would have to sweep the IF.

The 8566B has a complicated frequency plan that I've never gone into in full detail. Sure works though!

Cheers

Phil Hobbs

Electron microscopes and magnetic deflection mass spectrometers regulate magnetic fields pretty precisely - the Cambridge Instruments EBMF 10.5 that I worked on used two othogonal magnetic fields to put the electron beam precisely were it was wanted to 15-bit precision at 10MHz.

I got dragged in when the Johnson noise in the existing scanning amplifiers started making the lines it drew look a bit blobby, which I fixed by taking the low noise FETs out of the front end - we didn't need the low input impedance they offered - and relying on the low noise transistors with which they had been cascoded.

Admittedly we only had the 10MHz step rate inside a 12-bit sub-field, and stepped between those sub-fields involved 1msec of settling time - the 18-bit DAC that looked after that was bit slow.

It's certainly not easy to regulate magnetic fields to parts per million, but it can be done.

The big mass spectrometer - that I worked on for a couple of months at one point - used a Hall plate to regulate its magnetic field to that sort of precision.

I had a potentially patentable idea to make it work a bit better, but when we looked into it the idea had been spelled out elsewhere though the engineers who had put the machine together hadn't heard about it.