VCO Pulling ,How it effects circuit performance?

I have two VCOs at C-Band from two different manufacturers,they have pulling parameters as follows:

1.-28dB, RL, 23MHzp-p

2.@1.67:1 VSWR ,47MHz

Which is better and why? Please explain.

VSWR = 1.67:1 means -12dB Return Loss.

To compare apple to apple have to have the p-p value at the same Return Loss.

But I would say that nr.1 is better because have less frequency variation at -28dB RL (VSWR=1.08:1).
The 2nd one has higher frequency variation for a VSWR not such bad.

-28dB Return Loss and 23 MHz shift...
The oscillator looks so bad...

yeah the 1.67 vswr one is much better.

Sometimes you need good performance, sometimes you do not care. what is your specific application?

I used the second VCO(1.67 vswr 47MHz pulling),as shown in the figure1(PLL and modulator both operate at same frequency). The circuit resulted in a severe VCO pulling problem,i.e. over-modulation of carrier and side-band imbalance.The phase modulator has been tested independently and it works just fine.
My question is, although the i/p ReturnLoss of modulator is better than 20dB and I donot have a PA,why VCO is getting pulled?

Right now,its just a open(unshielded) test boards with PLL on one board and modulator on another board.Will shielding/enclosure of PLL circuit help in this case.

I used a buffer as shown in figure2, it could solve the problem of over-modulation ,but side-band imbalance is unsolved,
Experts please guide me.

I have heard that isolation of > 80 db is required for a high grade VCO for the HF bands. So I reckon that a good dose of decoupling and shielding will be required.
Frank

well there are a boatload of ways to get isolation.

1) run the VCO at twice your desired frequency, and use a digital divide by 2. That might approach 80 db isolation. You can use the divided output to also drive your PLL input.
2) run the VCO thru a laser diode, then thru a photodetector. That should give you >100 dB isolation
3) run the VCO at 1/2 the desired frequency, and use a frequency doubler (like a push push diode one from minicircuits). that should give you maybe 50 dB of isolation.
4) use a double junction ferrite isolator and a buffer amp. Probably good for 70 dB isolation.
5) use a VCO and a local oscillator thru a mixer to generate the output frequency. Because of the frequency translation, the impedance mismatches at the modulator input will be very buffered from the actual VCO output port.

Thanks for your responses.

Can we measure vco pulling practically?what would be the test set-up for this measurement?
Can we simulate vco pulling in ADS,what vco parameters need to be accounted for simulation?
Please share your expertise.

- - - Updated - - -

chuckey said:

I have heard that isolation of > 80 db is required for a high grade VCO for the HF bands. So I reckon that a good dose of decoupling and shielding will be required.
Frank

Click to expand...

Dear chuckey,when you say de-coupling and sheilding is required,do you mean sheiding and decoupling of vco circuit?My vco biasing circuit have 1uF ,10kpF and 22pF decoupling capacitor,isn't it alright.Right now my PLL circuit is not shielded,do you suggest that independent metal housing for PLL circuit will help in reducing vco pulling to good extent?

The 1µF does not do anything for 47 MHZ, I doubt if 22 pF is very useful. The sort of thing I would expect to see is the Vcc being fed through a couple of feedthrough 1nF capacitors in metal screens being coupled by a RF chokes. This sort of thing applies to the VCO voltage control lead, as this must be a very touchy area, but excessive capacitors to earth could upset the locking-up time. It is unlikely that a series RFC would affect this.
Metal shielding would normally be used but to try shielding out, make a cardboard box and covers for your circuit, then cover the box with aluminium foil with nuts and bolts through the joints to try and make good contact between the sheets and use multipoint earthing. If this makes an improvement, it would indicate that making a proper metal box would be worthwhile.
Frank

chuckey said:

The sort of thing I would expect to see is the Vcc being fed through a couple of feedthrough 1nF capacitors in metal screens being coupled by a RF chokes. This sort of thing applies to the VCO voltage control lead, as this must be a very touchy area, but excessive capacitors to earth could upset the locking-up time. It is unlikely that a series RFC would affect this. Frank

Click to expand...

Thanks chukey for your suggestion,do you mean RFC and feedthrus connected to vco as shown below:

What's the suggested RFC value(uH) here?
I hope RFC and capacitor at VCO control doesnot affect loop filter parameters,this need to be verified.

- - - Updated - - -

chuckey said:

Metal shielding would normally be used but to try shielding out, make a cardboard box and covers for your circuit, then cover the box with aluminium foil with nuts and bolts through the joints to try and make good contact between the sheets and use multipoint earthing. If this makes an improvement, it would indicate that making a proper metal box would be worthwhile.
Frank

Click to expand...

thanks chukey,that's a great idea,I would certainly try this,it will give me an idea ,to what extent, the radiated coupling is pulling VCO,that too very fast using just a cardboard box and metal foil,without fabricating an actual box.

Thats the sort of thing, I would expect to see another RFC between the two feedthrough capacitors. You could try ferrite beads if you have any, or wrap a few turns around any piece of ferrite. Value of RFCs ? >50µH.
Frank

Dear chuckey, What's the thumb rule to select ,feedthroughs and RFCs values for a particular application? I have seen designers
using highest values say 1µF feed throughs along with de-coupling polarized capacitors 0.1µF for low frequencies,1nF for moderate frequencies and 10-22pF for higher frequencies de-coupling?

How do you select these values for your design? You recommend 1nF feed-throughs for VCO,what will happen ,if I could use only 1µF feed-throughs or miniaturized 100pF feed-throughs due to non-availability of 1nF.

For decoupling, its the impedance of the capacitor at the working frequency that is important. If you only look at the capacitive reactance, theory says that a 1 MF is good for any frequency about 50 KHZ, the problem is that with the 1 MF, you get some series inductance which rises with frequency so that, at say 2 MHZ, the "capacitor" starts behaving like a low value inductor. You get this effect with all capacitors. The best RF decoupling capacitors are ceramic disc caps with virtually zero length leads. the value is not that importent (100 PF -> 10 NF) but its construction is, i.e. it must have very low internal inductance. Years ago I read an article where someone was cutting the capacitor leads to a certain length, so the lead inductance resonated with the capacitor to provide a really low impedance to the frequency of interest.
In answer to your question, for your Vcc, I would solder a 1NF across the feed through to a solder tag under its mounting thread and a 10 MF to a convenient earth. The VFO voltage control lead might need something like this but without the 10 MF. As an experiment, you could try putting an aerial (connect a piece of wire 3" long) on the control line and see if the performance gets even worse - just to see how touchy it is.
Frank