My application requires an oscillator continuously tunable between 1GHZ and 10GHZ (or a wider range). The output (somewhere between 1mW-15mW) has to be coupled to an ultrawide band antenna. No PLL allowed. Generous phase noise and frequency stability allowed.
I've read a lot in the web, and it appears that yttrium iron garnet sphere (YIG) tuned oscillators are probably the solution. Unfortunately, unless I throw a fortune (relative to my modest budget), the best range I can obtain is 2-20 GHz (so the upper bound is OK, but not the lower bound). On the contrary, YIG tuned filters offer wider suitable ranges at affordable prices; I've read somewhere this is because they use several YIG spheres to cover the full range. I know nothing about the construction of YIG tuned filters, but assuming they use one YIG sphere at a time, shouldn't it be possible to build the oscillator by myself, using the filter as the resonant part and microwave transistors?

  • \$\begingroup\$ Spin Torque Nano Oscillators have wide tunability and can be easily tuned with a DC current as fast as 5 GHz/ns. They also have high frequency stability. Unfortunately, as of 2017, they still must be custom made. \$\endgroup\$
    – axsvl77
    Dec 10, 2017 at 20:48

1 Answer 1


YIG tuned filters use several spheres for a high order filter.

Unless you take the filter apart and assemble the amplifier on the coupling to the spheres, the electrical length in and out of the filter will dominate the phase shift round a sustaining amplifier, meaning you will probably need additional switchable phase shift to be able to oscillate at all frequencies.

There's a lot of knowhow, aka technology, goes into the making of a YIG oscillator. It's difficult, and the people that do it for a living know more about it than you, or me. Cobbling together an oscillator out of a huge filter is rather like trying to build a cheap 4-wheeled vehicle by welding two motorbikes together. By the time you've solved the essential problems, you'll have taken a lot of time to be left with something far inferior to buying a Fiat 500.

You can get very wideband varactor tuned LC oscillators. You should be able to cover the 1 to 10GHz band with 3 or 4, for cheaper and smaller than a YTO, if you really have a generous phase noise allowance.

Depending on what you mean by 'continuous' - no missing frequencies, or the ability to sweep from one end to the other with absolutely no dropouts - you could use a 2-10GHz YTO and a divide by 2 IC to cover the bottom octave. Or maybe a 5-10GHz YTO (which ought to be smaller and cheaper) and a divide by several to cover the 1 to 5GHz range.

  • \$\begingroup\$ It's really not my field, but would it possible to use the YIG filter as a one port device and build an oscillator around it? I mean you keep filter out closed on 50 ohms (or opened or shorted whatever found best) and use frequency dependant S11 toghether with some negative resistance oscillator? \$\endgroup\$
    – carloc
    Dec 10, 2017 at 14:52
  • \$\begingroup\$ Carloc, I think this is essentially my question. The point is: is there something in the construction of (most?) YIG filters that will prevent them to act as resonators ? \$\endgroup\$
    – MikeTeX
    Dec 10, 2017 at 15:00
  • 1
    \$\begingroup\$ Have you thought about dividers, I've updated my answer. \$\endgroup\$
    – Neil_UK
    Dec 10, 2017 at 16:44
  • 1
    \$\begingroup\$ No, somebody sold (at least they did, a few years ago) a range of LC tuned oscillators that covered 1 to 10GHz in octave bands, try Sivers. \$\endgroup\$
    – Neil_UK
    Dec 10, 2017 at 16:49
  • 1
    \$\begingroup\$ Well, Sivers can take you from 3 to 13GHz in 3 oscillators, and minicircuits can take you 1 to 3GHz in 2 or 3. \$\endgroup\$
    – Neil_UK
    Dec 10, 2017 at 16:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.