Measuring jitter

[oshifis]

I am looking for ideas to construct a simple jitter measuring set. The frequency I want to measure would be the master clock of a CD player, say 16.9344 MHz. Even better would be to measure the jitter of the DAC latch pulse.

Basically a signal frequency modulated with noise is to be demodulated. The only difference is that the modulation is very low. I found the following FM demodulation methods:

- The dual tuned circuit phase discriminator, where the output amplitude is proportional to the frequency deviation from the center frequency.

- The PLL method. The amplitude of the analog error signal will be proportional to the frequency deviation.

- The scaling demodulator. Here the FM signal is shaped to square wave and constant pulse width. After low-pass filtering we get the analog signal.

For me the PLL method looks most easily realizable. Has anybody experience in this topic?

Laszlo

[oshifis]

Further thinking a very narrow band phase discriminator could be realized by using a crystal as a filter, tuned slightly away from the nominal frequency of the signal to be measured. An LP filter must follow.

[Guido Tent]

Originally Posted by oshifis

I am looking for ideas to construct a simple jitter measuring set. The frequency I want to measure would be the master clock of a CD player, say 16.9344 MHz. Even better would be to measure the jitter of the DAC latch pulse.

Basically a signal frequency modulated with noise is to be demodulated. The only difference is that the modulation is very low. I found the following FM demodulation methods:

- The dual tuned circuit phase discriminator, where the output amplitude is proportional to the frequency deviation from the center frequency.

- The PLL method. The amplitude of the analog error signal will be proportional to the frequency deviation.

- The scaling demodulator. Here the FM signal is shaped to square wave and constant pulse width. After low-pass filtering we get the analog signal.

For me the PLL method looks most easily realizable. Has anybody experience in this topic?

Laszlo

Yes, we did construct such setups. They are not simple, and to start with, you need a very good VCXO (on par with the clock to be measured) and low noise design.
We now cab measure down to levels below 100fs.

best

Guido