Jitter - converting ps to db ?

[percy]

How can one convert periodic jitter value from ps to -db, or vice versa ? For example, how can one find what would be the level relative to 0dbfs(-db) of the spikes that flank the fundamental signal for a known time value (200ps, 500ps, etc..).

 

[poynton]

As I understand it, the values quoted for 'jitter' are completely independant of signal level and therefore stand on their own merit.

attempts at equating are meaningless.

Andy

 

[KOON3876]

Very roughly and not accurate for the theory

44.1kHz = 22.67us
if I have average 200ps jitter, some signal output from DAC will come 100ps faster / or 100ps later.

it will distort signal voltage level at each sampling time.
the distortion ratio estimation will be 200ps / 22.67us = about 9.0e-6 = about -100dB.

 

[percy]

so you can only 'measure' the spectrum amplitude of jitter, you can't compute it ?

 

[KOON3876]

simply, if all of PCM data are 0x0000, then DAC output is always 0V,
nothing happen by jitter.
terrible jitter cause .... still 0V output.

 

[lcsaszar]

... and if all of the PCM data is 0xFFFF then a jitter peak amplitude of 1/44100s divided by 2^16 causes +/-1 bit error.

 

[dmills]

Not quite: for any steady (DC) level, jitter has no effect.

You can consider jitter to be phase modulation of the recorded signal, and as phase is the time integral of frequency you can apply the analysis that applies to FM modulation to the problem.

Basically it comes down to bessel functions, and can be computed but it is not trivial and a simple conversion to db is not possible.

Matlab and Octave are both useful for visualizing this sort of thing.

Regards, Dan.

 

[jkeny]

Here ya go http://www.jittertime.com/resources/pncalc.shtml