experimental interface receiver of fig.4. This circuit allows the instantaneous frequency of the recovered clock to be measured by monitoring the control voltage on the varicap diode. In order to recover the jitter signal on the clock, we must convert the instantaneous frequency signal to a timing error. Consider a frequency deviation Δ f in a recovered clock of nominal frequency fo; over a time period dt, the timing error tj can be written: Equation 12: tj = (Δ f / f0) dt If the measured frequency deviation is sinusoidal, the corresponding jitter frequency is also sinusoidal and of the same frequency, and Δ f can be written: Equation 13: Δ f = kV sin(2π fjt) where: k =PLL varicap control voltage sensitivity (Hz/V); |
V= peak measured voltage on varicap control pin; and fj = jitter frequency. Integrating across one quarter-cycle of. fj yields the peak jitter tj at this frequency: Equation 14:
tj = ∫0
1/(4fj)
Δf / f0 dt The varicap sensitivity k can be determined by decoupling the PLL feedback loop and monitoring clock frequency under direct control of a DC voltage applied to the control pin. Hence, by measuring the control voltage on die varicap using an analog/digital convener (ADC), and scaling and integrating this signal to follow the law of Equation 14, we have an indirect measure of interface jitter. (Note that although the control voltage in the PLL provides a useful indicator of interface jitter, it will not reveal jitter in the PLL VCO itself above the closed,loop cutoff frequency of the loop, or jitter sources between PLL and convener). Fig.19a-e shows jitter spectra when a CD player's digital output is connected to the experimental interface receiver. The interface signal carries the same audio data in both channels, corresponding to a 1kHz tone at 0dBFS, -20dBFS, -40dBFS, and -60dBFS undithered, and -80.65dBFS dithered, respectively. (These signals were obtained using the Hi-Fi News & Record Review Test Disc 2.) Each jitter spectrum consists of a large fundamental harmonic of the audio signal as well as higher harmonics and noise superimposed upon the frequency response of the PLL loop filter. Note the changes in the relative magnitudes of the harmonies, as well as the absolute jitter level, as the magnitude of the audio signal changes. These measured results can be directly compared to simulations of the recovered clock-jitter spectra for the same audio |
|
|
|
|
|
Fig19 Measured jitter spectra at interface receiver for 1kHz audio signals at a) 0dBFS, b) -20dBFS, c) -40dBFS, d) -60dBFS, and e) -80.65dBFS. |
to previous page Stereophile, March 1996 to next page |