ADN2812 Data Sheet
Rev. E | Page 14 of 28
THEORY OF OPERATION
The ADN2812 is a delay- and phase-locked loop circuit for
clock recovery and data retiming from an NRZ encoded data
stream. The phase of the input data signal is tracked by two
separate feedback loops that share a common control voltage.
A high speed delay-locked loop path uses a voltage controlled
phase shifter to track the high frequency components of input
jitter. A separate phase control loop, comprised of the VCO,
tracks the low frequency components of input jitter. The initial
frequency of the VCO is set by yet a third loop, which compares
the VCO frequency with the input data frequency and sets the
coarse tuning voltage. The jitter tracking phase-locked loop
controls the VCO by the fine-tuning control.
The delay- and phase-loops together track the phase of the
input data signal. For example, when the clock lags input data,
the phase detector drives the VCO to a higher frequency and
also increases the delay through the phase shifter; both these
actions serve to reduce the phase error between the clock and
data. The faster clock picks up phase, while the delayed data
loses phase. Because the loop filter is an integrator, the static
phase error is driven to zero.
Another view of the circuit is that the phase shifter implements
the zero required for frequency compensation of a second-order,
phase-locked loop. This zero is placed in the feedback path and,
thus, does not appear in the closed-loop transfer function. Jitter
peaking in a conventional second-order phase-locked loop is
caused by the presence of this zero in the closed-loop transfer
function. Because this circuit has no zero in the closed-loop
transfer, jitter peaking is minimized.
The delay- and phase-loops together simultaneously provide
wide-band jitter accommodation and narrow-band jitter filter-
ing. The linearized block diagram in Figure 17 shows that the
jitter transfer function, Z(s)/X(s), is a second-order low-pass
providing excellent filtering. Note that the jitter transfer has no
zero, unlike an ordinary second-order phase-locked loop. This
means that the main PLL loop has virtually zero jitter peaking
(see Figure 18), making this circuit ideal for signal regenerator
applications, where jitter peaking in a cascade of regenerators
can contribute to hazardous jitter accumulation.
The error transfer, e(s)/X(s), has the same high-pass form as an
ordinary phase-locked loop. This transfer function is free to be
optimized to give excellent wide-band jitter accommodation
because the jitter transfer function, Z(s)/X(s), provides the
narrow-band jitter filtering.
04228-017
X(s)
Z(s)
RECOVERED
CLOCK
e(s)
INPUT
DATA
d/sc
psh
o/s
1/n
d = PHASE DETECTOR GAIN
o = VCO GAIN
c = LOOP INTEGRATOR
psh = PHASE SHIFTER GAIN
n = DIVIDE RATIO
=
1
cn
do
s
2
+
n psh
o
s + 1
Z(s)
X(s)
JITTER TRANSFER FUNCTION
=
s
2
s
2
d psh
c
s+ +
do
cn
e(s)
X(s)
TRACKING ERROR TRANSFER FUNCTION
Figure 17. PLL/DLL Architecture
ADN2812
Z(s)
X(s)
04228-018
FREQUENCY (kHz)
JITTER PEAKING
IN ORDINARY PLL
JITTER GAIN (dB)
o
n psh
d psh
c
Figure 18. Jitter Response vs. Conventional PLL
The delay- and phase-loops contribute to overall jitter accom-
modation. At low frequencies of input jitter on the data signal,
the integrator in the loop filter provides high gain to track large
jitter amplitudes with small phase error. In this case, the VCO
is frequency modulated and jitter is tracked as in an ordinary
phase-locked loop. The amount of low frequency jitter that
can be tracked is a function of the VCO tuning range. A wider
tuning range gives larger accommodation of low frequency
jitter. The internal loop control voltage remains small for small
phase errors, so the phase shifter remains close to the center of
its range and thus contributes little to the low frequency jitter
accommodation.
