VCO interface with a non-linear circuit

Kindly explain,why does a locked VCO o/p gets affected if its interfaced with a non-linear circuit like varactor based reftection
type analog phase modulator?

It has been observed that ,if isolation between the two circuits(VCO and modulator) is not very good(>50dB),you will get
over modulated or imbalanced side-bands as shown in the figure.

You have stated a cause & effect. How sure are you that the non-very-good-isolation is actually the root cause ? It could be some other factor in the setup. Like (for example) misalignment of the phase modulator from the central phase-shift point due to incorrect tuning...

kripacharya said:

You have stated a cause & effect. How sure are you that the non-very-good-isolation is actually the root cause ? It could be some other factor in the setup. Like (for example) misalignment of the phase modulator from the central phase-shift point due to incorrect tuning...

Click to expand...

I am very much sure about the cause ,because I have tested these modulators with commercial siggen which has very good reverse isolation and
it has performed perfectly.

And as I increase the isolation between the two circuits(VCO and modulator) the side-band imbalance starts decreasing and at one point
they are perfectly balanced.

Is about VCO frequency pulling by the load.
The load in your case is a varactor which value changes dynamically in the rhythm of the modulation, so the VCO sees a variable load, finally generating the spectrum that you show in the picture.
Instead increasing the isolation, I think better is to place a buffer between VCO and modulator.

VCO phase will be impacted by instantaneous impedance changings at the input of the modulator because there are probably differential pair that is switched by LO signal.Hence, pulling effect will shift amplitude and phase both,that's why modulator spectrum is not as good as in ideal case.It's normal and evident..
In order to preven this effect, you should use a buffer amplifier to increase the isolation.

Thanks for the suggestions.

The observation which is coming out of the discussion is that the dynamic impedance variation of varactor diode causes dynamic load change at VCO output.
This load change at VCO output tries to pull to some other frequency,forcing VCO control voltage to change dynamically. If control voltage keeps changing,results in phase modulation of VCO signal.This results in spoiling of phase modulated spectrum of modulator output.

afz23 said:

The observation which is coming out of the discussion is that the dynamic impedance variation of varactor diode causes dynamic load change at VCO output.
This load change at VCO output tries to pull to some other frequency,forcing VCO control voltage to change dynamically. If control voltage keeps changing,results in phase modulation of VCO signal.This results in spoiling of phase modulated spectrum of modulator output.

Click to expand...

good to see you have your explanation

afz23 said:

,,..as I increase the isolation between the two circuits(VCO and modulator) ...

Click to expand...

I was curious to know how you achieved this variable isolation ?

kripacharya said:

good to see you have your explanation



I was curious to know how you achieved this variable isolation ?

Click to expand...

When we come across with such a problem of variable load at VCO out put,

Possible solutions

If change in VCO load or reflection coeff. is δΓ

1) Increase PLL locking speed- depends upon loop bandwidth, may not always be possible

2) Add matching network to make the loads look similar( reduces δΓ )-is not a easy solution

3) Add isolation between VCO and load. This can be done with buffer amplifiers (reverse isolation) or attenuator-should always be possible

● Attenuator isolation calculation will be shown with an example.
If current δΓ = -20 log(0.3) = 10.5 dB
If we do not want frequency pulling to exceed by 500KHz, Required δΓ = -20 log( 500kHz / 19.9MHz ) = 32 dB

(here frequency pulling for the given VCO is 10MHz p-p @ 12dBr all phases. since 12 = -20 log(0.251)
so frequency shift δf = ρf δΓ and ρf=10 MHz / (0.251*2) = 19.9 MHz)

then required attenuation is ½ ( 32 – 10.5 ) = 10.75

As can be seen,depending upon the application,required attenuation can be variable,it can be as high as 90dB for
lower frequency pulling tolerances.

yep, had an application where I needed around 80 dB of isolation from open/short type loads. I made the requirement a little less bothersome by using non-reflective switches--the type that always present a 50 ohm load in any state (insertion loss or isolation).

There are a few tricks you could use. There are amplifiers that are designed for high reverse isolation. You would put in a pad and the high isolation amp for maybe 50 dB. Going thru a frequency translating device would help a lot too. Going thru a laser diode/photodetector would give you the 90 dB. Etc etc.

renesas μPC8178TK

biff44 said:

yep, had an application where I needed around 80 dB of isolation from open/short type loads. I made the requirement a little less bothersome by using non-reflective switches--the type that always present a 50 ohm load in any state (insertion loss or isolation).

There are a few tricks you could use. There are amplifiers that are designed for high reverse isolation. You would put in a pad and the high isolation amp for maybe 50 dB. Going thru a frequency translating device would help a lot too. Going thru a laser diode/photodetector would give you the 90 dB. Etc etc.

renesas μPC8178TK

Click to expand...

Can somebody suggest a transistor IC which can provide a reverse isolation of 40dB @4 GHz and can be used for SATCOMM.

you do not understand the concept. a transistor will never do the job.

biff44 said:

you do not understand the concept. a transistor will never do the job.

Click to expand...

biff44
renesas μPC8178TK is a bipolar analog IC (silicon low power MMIC) which can operate upto 2.4GHz, it integrates 4 BJTs monolithically.It can be used as
a good buffer because of its 40dB reverse isolation.
The same technology is being used by some manufacturers to make amplifiers upto 6GHz. Above 6GHz you donot have choice, you have to use GAAS amplifiers.

If you say concept is not understood ,can you please elaborate this point.

i am not an mmic designer, but I believe the technique is to use a cascode amplifier with the proper feedback. You are never going to get that to work with discrete packaged FETs or transistors at 6 GHz! Too many parasitics. So you are going to have to use one of the other techniques.

what is your modulation type?

biff44 said:

i am not an mmic designer, but I believe the technique is to use a cascode amplifier with the proper feedback. You are never going to get that to work with discrete packaged FETs or transistors at 6 GHz! Too many parasitics. So you are going to have to use one of the other techniques.

what is your modulation type?

Click to expand...

I am doing phase modulation(reflection type ,using hyperabrupt varactors) at 4 GHz,modulating signals in KHz range.

I have found one from Avago technologies, this can provide 33dB isolation at 4 GHz and 22 dB gain. So, you can have an isolation of > 50 dB
by using combination of attenuators and this device with out sacrificing VCO o/p power.