[SOLVED] How to calculate phase noise of a frequency modulate signal

[icemc2008]

Hello everyone,

I am a newbie in phase noise and jitter. I want to ask how to calculate the phase noise of a frequency modulate signal.

Ex: I have a sine wave of 10MHz and 0.5V - > - 0.5V amplitude. a modulation signal: a sine wave with 10kHz frequence and 1MHz amplitude(deviation).

I want to calculate the phase noise of this modulated signal. please give me a expression and example to calculate it with the data above.

please help me, thanks a lot for your kind and patience!!!

 

[biff44]

Your question does not make too much sense. If you have one ideal carrier, modulated by another ideal modulating tone, the phase NOISE is zero. There is no random noise on the signal.

Try to imagine your modulated signal in the frequency domain. You are very much modulating the carrier with a large angle modulation. Your spectrum will look like a picket fence of tones.

 

[icemc2008]

Thanks biff

Actually I want to calculate the period jitter of this modulated signal. by convert the phase noise to period jitter.
From the book <Clocking in Modern VLSI Systems> by Thucydides Xanthopoulos, in page 143 expression 5.6



I dont know how to calculate L(f) of this modulated signal: phase noise spectrum or single sideband noise distribution.

In my case, the noise souce is not the thermal noise or white noise, but the external pertubation.

So I want to find the deterministic Jitter(period jitter) of the ideal sine wave (10MHz and 0.5V - > - 0.5V amplitude) when apply a frequency modulation by sine wave (10kHz frequence and 1MHz amplitude(deviation))

I hope that I state the question clearly. And thanks for your help.

 

[biff44]

You can do that by inspection. You have a 10 MHz carrier with a 1 MHz deviation. If I can assume you mean peak deviation, then the frequency varies from 9 MHz to 11 MHz.

1/9 MHz = 111 ns
1/11 MHz = 91 ns

Looks like 20 ns, (+/- 10 ns peak), of "period" (time) jitter to me.

 

[icemc2008]

Thanks biff

It seems that I make the problem complicate!

 

[biff44]

It is a complicated thing, and unfortunately not too well explained anywhere. That is because half of the engineers worry about "phase noise" in the frequency domain, and the other half (mostly DSP engineers) worry about clock jitter in the time domain. Few actually bridge the gap, so you seldom see a book or paper explaining both sides. And to top it all off, there is a heavy dose of probability theory thrown in too!