Unsymmetric sidebands in modulated signal

I have a linear phase modulator at 4GHz.It modulates carrier with audio frequencies. Modulator is varactor based reflection type.

I see a perfect phase modulated spectrum with symmetric sidebands at fc±fm,when carrier is fed from signal generator.

But I observe unsymmetric sidebands at fc±fm, when carrier is generated from PLL.

Problem persists even after introducing an isolator at phase modulator i/p.

You obviously need more isolation. In really critical applications, I have seen the need for up to 90 dB of reverse isolation in some situations. Try some amplifier/attenuators in series and see if it goes away. Remember, it might be harmonic pulling, where your isolator is not broad band enough to isolate it.

obviously, if your modulator is somehow leaking analog signals onto the PLL (either the rf line or the ground plane) it could screw things up.

The typical problem is that the PLL does not resond quickly, while the VCO response to the load pull is nearly instantaneous, so the PLL is always "chasing after" the pulled VCO.

thanks biff44

I tried some attenuators in addition to isolator between PLL and modulator,it does
help in improving the symmetry and confirms that there is reverse leakage from modulator towards PLL.

But putting additional attenuators means losing crucial power, can there be better
way for doing this, will a buffer amplifier help in improving the reverse isolation
in this case.

A buffer amp normally has around 15 db of reverse isolation. There are a one or two chips out there with much more reverse isolation, although as I recall only at low frequencies. So you take an amplifier with 20 db of gain, put a 20 dB pad in front of it, and you have 0 dB gain and 35 dB of reverese isolation.

Digital dividers work pretty good at reverse isolation. If you can run the VCO at twice the frequency needed and divide by two, you will have a lot of isolation.

Opto isolators are very good with reverse isolation. If you can run your vco into a laser diode, and run a short fiber to a photodetector, you will have over 100 dB of isolation.

Or you could try to figure out why your modulator is screwing up the PLL.

thanks biff44 for these great ideas.

I hope many members will be benefited by your experience.


I can try first option of amplifier and attenuator in combination straight away,I am
not sure whether isolator + amplifier + 20dB attenuator can provide me 55dB
isolation, i need to see this.

Second option can also be tried , which is to drive VCO at higher frequency and
prescaling to desired frequency, I need to change VCO for this in my circuit,can be
done easily.


third option requires a diode detector and optical cable, I am not sure about this as I donot have them right now.

Added after 6 minutes:

The modulator has a branchline coupler with more than 25dB isolation.
Two identical varactors at the two coupled ports, the reflected RF from th coupled
ports is suppose to be 25dB down at the input port, as the PLL o/p is 10dBm
the amount of energy being reflected back from modulator towards PLL is sufficient
to disturb VCO operation even after putting a isolator at PLL o/P.