help me: integrated phase noise in PLL
A real world oscillator will tend to oscillate at one frequency, but jump around slightly due to various perturbations. If you measure the jumping around with a frequency discriminator, the output would be frequency noise. If you measure the jumping around with a phase detector, the output would be phase noise.
Assuming you have a broadband phase detector with an analog low frequency output. You would look at the phase detectors output voltage as it jumps around, and try to measure the jumps. How you measure the jumping voltage determines what you would call it.
If you crafted a analog bandpass filter that passed 99.5 to 100.5 Hz, and rejected everything else, then the noise voltage you got at the filter output would be phase noise in a 1 Hz bandwidth. Since it is a measure of noise at 100 Hz from the carrier, I would call that "close-in phase noise". If you redesign the bandpass filter to be at 49.5 to 50.5 Hz, I would also call that "close-in", although you will likely get a different reading.
If, instead, you took the phase detector output voltage, passed it thru a lowpass filter with a 10 MHz cuttoff frequency, and then integrated it over say 1 second, then the integrator output would be the "integrated phase noise".
Of course, you would have to somehow calibrate the phase detector output voltages to an actual phase modulation deviation, with some sort of calibration method, for the numbers to mean anything.
Practically, you can not easily craft a 1 Hz bandwidth bandpass filter, so measurements are often made in larger bandwidths (10 Hz, 100 Hz, etc) and the answer is mathematically corrected to phase noise/Hz.
What you call "close-in" depends on the system. If I were transmitting data at 100 bps, close in might be at 0.1 Hz from the carrier. If I were transmitting 10 Gbps data, close-in might mean something below 100 KHz.
Rich
biff44,
Thank you so much for your detailed introduction, it really helps me. Regards.