# pll phase-noise/ rms phase error

Status
Not open for further replies.

#### afz23

##### Full Member level 3
In a Phase-modulation scheme,rms phase error is very critical,it should be less than 2deg in most cases.

This rms phase error directly relates to in-loop phase noise of pll output.

while working with PLL frequency synthesizer, we come across two formulae to calculate phase noise:

1) in-loop PLL_PhaseNoise[dBc/Hz] = Noise floor[dBc/Hz]+ 10*LOG(Comparison Frequency[Hz]) + 20*LOG(N)

noise floor in this formula is synthesizer

2) phase-noise= reference phase noise + 20*log(N)

Now, the question is which formula to consider for comparing a measured phase-noise at PLL output,secondly the VCO
phase-noise is not accounted in above formulae, does that mean that reference phase noise or synthesizer noise floor control
the final phase noise? What if I use a VCO with better close-in phase noise,will I get any improvement in final phase noise,how do
I consider vco phase-noise this in above formulae?

eg. reference phase noise =-135dBc/Hz
Noise floor =-210 dBc/Hz
Fc= 10MHz
N=44
then from 1) PN=107 dBc/Hz
2) PN=102 dBc/Hz

Where as measured phase noise is about -90 dBc/Hz at 10 KHz offset for a carrier of 4GHz.
Why the measured phase-noise is off by 12 dB ? where to look for improvement?

#### biff44

The PLL has a loop bandwidth. If you are inside of +/- the loop bandwidth from the VCO carrier, then the phase noise of the output is no better than 20 Log N + Reference phase noise. i.e. the reference phase noise as degraded by the PLL.

If you are outside of the loop bandwidth, the phase noise can be as low as the free running VCO phase noise at that offset from the carrier.

Modern communicaitons systems often have "carrier tracking loops" in the demodulator, and they can track out close to the carrier phase variations, so getting a frequency reference with super low phase noise close to the carrier is often not needed (unless you are doing very low data rate stuff, like deep space satellite communications).

Also, since one can not make a silk purse out of a sow's ear, if you have a very bad VCO, the PLL may not have enough gain to correct the phase noise to anything close to 20 Log N + the reference noise.

afz23 and tony_lth

Points: 2

Points: 2