2
\$\begingroup\$

I would like to measure, and ultimately control, the relative phase (and amplitude, but that's easier) of two microwave signals, one at \$f_0 \approx 1 \text{ GHz}\$ and one at \$f_n = n f_0\$, where \$n\$ is odd. By relative phase of these signals, I mean the phase difference between voltage maxima. So if the both have maxima at the same time, then the phase difference is 0. (Note that they will then have minima at the same time as well, because of the requirement that \$n\$ is odd.) A phase shift of \$\pi\$ will result in one being a minimum when the other is a maximum.

I'm not sure mixing them gets me anywhere, I just end up with another two signals of which I don't know the phases. I tried looking at phase detectors, but couldn't find anything for two different frequencies.

So how do I do this?

Power at \$f_0\$ will be around 0 dBm, and it's a \$50 \text{ }\Omega\$ system. Ideally I would like \$n\$ to go up to 19. I don't need my measurement bandwidth to be high at all.

\$\endgroup\$
10
  • 1
    \$\begingroup\$ Can you divide down the higher frequency signal by n so that you end up with two signals at the same frequency? Or is n unknown and variable? \$\endgroup\$
    – user57037
    Commented Feb 14, 2022 at 2:22
  • 1
    \$\begingroup\$ If the frequency were lower you could simply sample Fn with F0, but that would be hard at many GHz. I would sequentially divide by 2, then 3, then 4... until you find the divisor and then compare phase. \$\endgroup\$
    – user1850479
    Commented Feb 14, 2022 at 2:38
  • 1
    \$\begingroup\$ Divide the lower frequency by 2, and the higher one by 2n. But knowing the delay added by your dividers will be tricky. Do you have a budget for a fancy oscilloscope? \$\endgroup\$
    – The Photon
    Commented Feb 14, 2022 at 3:31
  • 1
    \$\begingroup\$ @NLambert, that's not enough to measure the phase difference between your 19 GHz signal and the output of a by-19 or by-38 divider. If you just want to keep the phase between your two signals constant, you don't need it. But if you want to actually control the phase to some precise value, you will need it. \$\endgroup\$
    – The Photon
    Commented Feb 14, 2022 at 5:55
  • 1
    \$\begingroup\$ Or divide your 1 GHz signal down by 16 or 32 and use that to drive a sampler as suggested by user1850479. Finding a sampler that can handle 19 GHz will require you look at some specialty vendors, but such things are out there. \$\endgroup\$
    – The Photon
    Commented Feb 14, 2022 at 5:58

2 Answers 2

Reset to default
1
\$\begingroup\$

Since these are microwaves...
I would use two "identical" samplers working at the same frequency (1 GHz - x kHz).
Something as this.

Two signals are therefore recovered at much lower frequencies on which a phase measurement is easier to make.
All that remains is to synchronize the oscillator used so that the 1 GHz frequency --> becomes \$ x * kHz\$, ... and the other will be \$n*x* kHz\$ for example.

enter image description here

To show the relationship with output sampled frequency, I add this picture with some more information.

enter image description here

\$\endgroup\$
0
\$\begingroup\$

I think Phase LockedLoop was designed for exactly your problem.

Here is some guidance... https://www.microwavejournal.com/articles/28830-phase-locked-loops-enable-phase-alignment-and-control

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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