Differences of Phase noise of VCO and close loop PLL?

[dd2001]

pll carrier close noise

Hi, dear

Some datasheet: give synthesizer close loop phase noise, e.g. -110dBc at 1MHz, and SpectreRF can measure VCO phase noise in open loop condition, my question is:

1. What is diffrence of those two measurements? and how to measurement of Close loop phase noise?

2. Why use close loop phase noise?

3. Relationaship of those two phase noises?


Thanks.

 

[nathan]

closed-loop phase noise

Just a general statement:

think about the PLL as an amplifier that works on phase rather tahn voltage and think about the VCO as a noise generator.

When you close the VCO in the loop the effective phase noise is reduced by a factor equal to the loop gain of the PLL ...

... got the idea ?

nathan

 

[dd2001]

pll vco phase noise transfer function

nathan

Thanks. Do you mean:

1. If measure phase noise from the outputs of PLL, its value is smaller than the output of VCO's. since there is gain in PLL loop, more gain, smaller phase noise.

2. Since Phase/Freq Detector, Charge pump, Loop filter, prescaler, all of those blocks contribute noise to the output of PLL, is the phase noise of PLL still lower than VCO's in general, duot to there is loop gain?

3. Measure VCO phase noise is rather simpe, however, whole PLL loop simulation in SpectreRF is impossibale, even HSpice, Hsim, UltraSim, none of them can simulate whole PLL loop, how can we get close loop phase noise? Any method, any tools?


Thanks.

 

[Mazz]

vco phase noise pll

Some answers:
1. It's not so simple: the VCO is just one element of a PLL. When closing the loop (is it clear why we use PLL instead fo bare VCOs?) in a PLL the noise is shaped by the transfer function of the PLL. As the VCO transfer function is high pass type, there is a certain offset from the carrier at which the VCO noise is cutted out.

2. Each block in the loop followes the same rule (except that the transfer function is different if the block is in the forward path or in the backward one).

3. The way used to predict phase noise is usually with the transfer function of each block, computed with the help of a math tool. These are really custom tools and I think that it doesn't exist soething available, but maybe I'm wrong.

I hope it can help
Mazz

 

[dd2001]

pll close in phase noise

Mazz

Can we conclude that the phase noise of PLL is usually better than VCO's in general? More narrow bandwidth, more better phase noise. If my VCO does not meet specs, however, my PLL may meet specs. (since datasheet only list closed loop phase noise specs.)


thanks

 

[Mazz]

I can make an example:
VCO phase noise:
@ 1kHz from carrier = -40 dBc
@ 10kHz from carrier = -80 dBc
@ 100kHz from carrier = -100 dBc

If properly designed you can obtain:
PLL phase noise
@ 1kHz from carrier = -70 dBc
@ 10kHz from carrier = -78 dBc
@ 100kHz from carrier = -100 dBc

With a loop bandwidth of about 15/20 kHz.
There is a floor flat inside the loop bandwidth and out of the loop bandwidth the VCO noise is seen at PLL output.

Usually the spec is given in "clever" way. In our example @10KHz and @100KHz
or, even better, as integrated phase noise over 1KHz-10MHz. And you cannot enlarge too much the bandwidth for at least two reasons: the first is that you'll be limited by VCO noise (once again), the second is that you cannot forget the spurious generated by reference frequency (of fractional if the PLL is not an integer one), that will not be filtered out if the loop bandwidth is too large.

It's just an example, but maybe clarify what in practice is done.

I hope it can help.
Mazz

 

[dd2001]

Mazz, Thanks.

You make a clear picture for me. Thanks again.

 

[borodenkov]

About measuring phase noise and jitter of the whole PLL using SpectreRF - read Ken Kundert's paper at

**broken link removed**