(LVPECL/LVDS only at 155.52 MHz)
Optional 19.44 MHz on-chip oscillator to be used with
external crystal
Loss-of-lock indicator
Loopback mode for high speed test data
Output squelch and bypass features
Single-supply operation: 3.3 V
Low power: 540 mW typical
7 mm × 7 mm, 48-lead LFCSP
The ADN2807 provides the receiver functions of quantization,
signal level detect, and clock and data recovery at rates of OC-3,
OC-12, and 15/14 FEC. All SONET jitter requirements are met,
including jitter transfer, jitter generation, and jitter tolerance. All
specifications are quoted for –40°C to +85°C ambient
temperature, unless otherwise noted.
The device is intended for WDM system applications and can
be used with either an external reference clock or an on-chip
oscillator with external crystal. Both native rates and 15/14 rate
digital wrappers are supported by the ADN2807, without any
change of reference clock.
This device, together with a PIN diode and a TIA preamplifier,
can implement a highly integrated, low cost, low power, fiber
optic receiver.
The receiver front end signal detect circuit indicates when the
input signal level has fallen below a user adjustable threshold.
The signal detect circuit has hysteresis to prevent chatter at the
output.
The ADN2807 is available in a compact 7 mm × 7 mm 48-lead
chip-scale package (LFCSP).
CF1CF2
LOL
ADN2807
2
PIN
QUANTIZER
NIN
VREF
Rev. A
Information furnished by Analog Devices is believed to be accurate and reliable.
However, no responsibility is assumed by Analog Devices for its use, nor for any
infringements of patents or other rights of third parties that may result from its use.
Specifications subject to change without notice. No license is granted by implication
or otherwise under any patent or patent rights of Analog Devices. Trademarks and
registered trademarks are the property of their respective owners.
Single-Ended Output Swing VSE (See Figure 7) 400 488 540 mV
Differential Output Swing V
Output High Voltage V
(See Figure 7) 850 975 1100 mV
DIFF
OH
Output Low Voltage VOL, referred to VCC –0.60
Rise Time 20% to 80%
Fall Time 80% to 20%
Setup Time TS (See Figure 3)
OC-12 750
OC-3 3145
Hold Time TH (See Figure 3)
OC-12 750
OC-3 3150 ps
REFCLK DC INPUT CHARACTERISTICS
Input Voltage Range @ REFCLKP or REFCLKN 0 VCC V
Peak-to-Peak Differential Input 100 mV
Common-Mode Level DC-coupled, single-ended VCC/2 V
TEST DATA DC INPUT CHARACTERISTICS4
CML inputs
(TDINP/N)
Peak-to-Peak Differential Input Voltage 0.8 V
LVTTL DC INPUT CHARACTERISTICS
Input High Voltage V
Input Low Voltage V
IH
IL
Input Current VIN = 0.4 V or VIN = 2.4 V –5 +5 µA
Input Current (SEL0 and SEL1 Only)
5
VIN = 0.4 V or VIN = 2.4 V –5 +50 µA
LVTTL DC OUTPUT CHARACTERISTICS
Output High Voltage VOH, IOH = –2.0 mA 2.4 V
Output Low Voltage VOL, IOL = +2.0 mA 0.4 V
1
PIN and NIN should be driven differentially, ac-coupled for optimum sensitivity.
2
PWD measurement made on quantizer outputs in BYPASS mode.
3
Jitter tolerance measurements are equipment limited.
4
TDINP/N are CML inputs. If the drivers to the TDINP/N inputs are anything other than CML, they must be ac-coupled.
5
SEL0 and SEL1 have internal pull-down resistors, causing higher IIH.
140 200 kHz
48 85 kHz
0.004
0.002
dB
dB
0.003 UI rms
0.02 0.04 UI p-p
0.002 UI rms
0.02 0.04 UI p-p
100
UI p-p
UI p-p
UI p-p
UI p-p
VCC
–0.30 V
150 ps
150 ps
UI p-p
UI p-p
UI p-p
V
ps
ps
ps
2.0 V
0.8 V
Rev. A | Page 4 of 20
ADN2807
www.BDTIC.com/ADI
ABSOLUTE MAXIMUM RATINGS
Table 2.
Parameter Rating
Supply Voltage (VCC) 5.5 V
Minimum Input Voltage (All Inputs) VEE – 0.4 V
Maximum Input Voltage (All Inputs) VCC + 0.4 V
Maximum Junction Temperature 165°C
Storage Temperature –65°C to +150°C
Lead Temperature (Soldering 10 sec) 300°C
Stresses above those listed under Absolute Maximum Ratings
may cause permanent damage to the device. This is a stress
rating only; functional operation of the device at these or any
other conditions above those indicated in the operational
section of this specification is not implied. Exposure to absolute
maximum rating conditions for extended periods may affect
device reliability.
ESD CAUTION
ESD (electrostatic discharge) sensitive device. Electrostatic charges as high as 4000 V readily accumulate on
the human body and test equipment and can discharge without detection. Although this product features
proprietary ESD protection circuitry, permanent damage may occur on devices subjected to high energy
electrostatic discharges. Therefore, proper ESD precautions are recommended to avoid performance
degradation or loss of functionality.
THERMAL CHARACTERISTICS
Thermal Resistance
48-Lead LFCSP, 4-layer board with exposed paddle soldered
to VCC
1 THRADJ AI LOS Threshold Setting Resistor.
2, 26, 28, Pad VCC P Analog Supply.
3, 9, 16, 19,
VEE P Ground.
22, 27, 29, 33,
34, 42, 43, 46
4 VREF AO Internal VREF Voltage. Decouple to GND with a 0.1 µF capacitor.
5 PIN AI Differential Data Input.
6 NIN AI Differential Data Input.
7 SLICEP AI Differential Slice Level Adjust Input.
8 SLICEN AI Differential Slice Level Adjust Input.
10 LOL DO Loss-of-Lock Indicator. LVTTL active high.
11 XO1 AO Crystal Oscillator.
12 XO2 AO Crystal Oscillator.
13 REFCLKN DI Differential REFCLK Input. LVTTL, LVCMOS, LVPECL, LVDS (LVPECL, LVDS only at 155.52 MHz).
14 REFCLKP DI Differential REFCLK Input. LVTTL, LVCMOS, LVPECL, LVDS (LVPECL, LVDS only at 155.52 MHz).
15 REFSEL DI Reference Source Select. 0 = on-chip oscillator with external crystal. 1 = external clock source, LVTTL.
17 TDINP AI Differential Test Data Input. CML.
18 TDINN AI Differential Test Data Input. CML.
20, 47 VCC P Digital Supply.
21 CF1 AO Frequency Loop Capacitor.
23 REFSEL1 DI Reference Frequency Select (See Table 6) LVTTL.
24 REFSEL0 DI Reference Frequency Select (See Table 6) LVTTL.
25 CF2 AO Frequency Loop Capacitor.
30 SEL1 DI Data Rate Select (See Table 5) LVTTL.
31 NC No Connect.
32 SEL0 DI Data Rate Select (See Table 5) LVTTL.
35, 36 VCC P Output Driver Supply.
37 DATAOUTN DO Differential Retimed Data Output. CML.
38 DATAOUTP DO Differential Retimed Data Output. CML.
39 SQUELCH DI Disable Clock and Data Outputs. Active high. LVTTL.
40 CLKOUTN DO Differential Recovered Clock Output. CML.
41 CLKOUTP DO Differential Recovered Clock Output. CML.
44 BYPASS DI Bypass CDR Mode. Active high. LVTTL.
45 SDOUT DO Loss-of-Signal Detect Output. Active high. LVTTL.
48 LOOPEN DI Enable Test Data Inputs. Active high. LVTTL.
1
Type: P = Power, AI = Analog Input, AO = Analog Output, DI = Digital Input, DO = Digital Output
Rev. A | Page 6 of 20
ADN2807
www.BDTIC.com/ADI
CLKOUTP
T
S
T
H
DATAOUTP/N
10
9
8
7
6
5
4
FREQUENCY
3
2
1
0
01123456789
Figure 5. LOS Hysteresis OC-3, −40°C, 3.6 V,
23
2
HYSTERESIS (dB)
–1 PRBS Input Pattern, RTH = 90 kΩ
OUTP
OUTN
OUTP–OUTN
V
CML
V
0V
SE
Figure 3. Output Timing
18
16
14
12
10
mV
8
6
4
2
0
0100
THRADJ RESISTOR VS. LOSTRIP POINT
102030405060708090
RESISTANCE (kΩ)
Figure 4. LOS Comparator Trip Point Programming
18
16
14
12
10
8
FREQUENCY
6
4
2
0
0
03877-0-005
01123456789 0
Figure 6. LOS Hysteresis OC-12, −40°C, 3.6 V,
V
DIFF
Figure 7. Single-Ended vs. Differential Output Specifications
03877-0-003
03877-0-004
HYSTERESIS (dB)
23
2
–1 PRBS Input Pattern, RTH = 90 kΩ
V
SE
03877-0-007
03877-0-006
Rev. A | Page 7 of 20
ADN2807
S
www.BDTIC.com/ADI
DEFINITION OF TERMS
MAXIMUM, MINIMUM, AND TYPICAL
SPECIFICATIONS
Specifications for every parameter are derived from statistical
analyses of data taken on multiple devices from multiple wafer
lots. Typical specifications are the mean of the distribution of
the data for that parameter. If a parameter has a maximum (or a
minimum) value, that value is calculated by adding to (or
subtracting from) the mean six times the standard deviation of
the distribution. This procedure is intended to tolerate production variations. If the mean shifts by 1.5 standard deviations,
the remaining 4.5 standard deviations still provide a failure rate
of only 3.4 parts per million. For all tested parameters, the test
limits are guardbanded to account for tester variation, and
therefore guarantee that no device is shipped outside of data
sheet specifications.
SINGLE-ENDED VS. DIFFERENTIAL
AC coupling is typically used to drive the inputs to the
quantizer. The inputs are internally dc-biased to a commonmode potential of ~0.6 V. Driving the ADN2807 single-ended
and observing the quantizer input with an oscilloscope probe at
the point indicated in Figure 9 shows a binary signal with
average value equal to the common-mode potential and
instantaneous values both above and below the average value. It
is convenient to measure the peak-to-peak amplitude of this
signal and call the minimum required value the quantizer
sensitivity. Referring to Figure 8, since both positive and
negative offsets need to be accommodated, the sensitivity is
twice the overdrive.
10mV p-p
VREF
INPUT SENSITIVITY AND INPUT OVERDRIVE
Sensitivity and overdrive specifications for the quantizer involve
offset voltage, gain, and noise. The relationship between the
logic output of the quantizer and the analog voltage input is
shown in Figure 8. For sufficiently large positive input voltage,
the output is always Logic 1; similarly for negative inputs, the
output is always Logic 0. However, the transitions between
output Logic Levels 1 and 0 are not at precisely defined input
voltage levels, but occur over a range of input voltages. Within
this zone of confusion, the output may be either 1 or 0, or it may
even fail to attain a valid logic state. The width of this zone is
determined by the input voltage noise of the quantizer. The
center of the zone of confusion is the quantizer input offset
voltage. Input overdrive is the magnitude of signal required to
guarantee a correct logic level with a 1 × 10
OUTPUT
1
0
OFFSET
OVERDRIVE
SENSITIVITY
(2× OVERDRIVE)
Figure 8. Input Sensitivity and Input Overdrive
NOISE
–10
confidence level.
INPUT (V p-p)
03877-0-008
SCOPE
PROBE
Figure 9. Single-Ended Sensitivity Measurement
COPE
PROBE
Figure 10. Differential Sensitivity Measurement
PIN
VREF
PIN
NIN
VREF
ADN2807
5mV p-p
ADN2807
50Ω 50Ω
50Ω 50Ω
+
QUANTIZER
+
QUANTIZER
VREF
03877-0-007
03877-0-010
Driving the ADN2807 differentially (Figure 10), sensitivity
seems to improve by observing the quantizer input with an
oscilloscope probe. This is an illusion caused by the use of a
single-ended probe. A 5 mV p-p signal appears to drive the
ADN2807 quantizer. However, the single-ended probe measures
only half the signal. The true quantizer input signal is twice this
value since the other quantizer input is complementary to the
signal being observed.
Rev. A | Page 8 of 20
ADN2807
www.BDTIC.com/ADI
LOS RESPONSE TIME
The LOS response time is the delay between the removal of the
input signal and indication of loss of signal (LOS) at SDOUT.
The ADN2807’s response time is 300 ns typ when the inputs are
dc-coupled. In practice, the time constant of ac coupling at the
quantizer input determines the LOS response time.
JITTER SPECIFICATIONS
The ADN2807 CDR is designed to achieve the best bit-errorrate (BER) performance, and has exceeded the jitter transfer,
generation, and tolerance specifications proposed for
SONET/SDH equipment defined in the Telcordia Technologies
specification. Jitter is the dynamic displacement of digital signal
edges from their long-term average positions measured in UI
(unit intervals), where 1 UI = 1 bit period. Jitter on the input
data can cause dynamic phase errors on the recovered clock
sampling edge. Jitter on the recovered clock causes jitter on the
retimed data. The following sections briefly summarize the
specifications of the jitter generation, transfer, and tolerance in
accordance with the Telcordia document (GR-253-CORE, Issue
3, September 2000) for the optical interface at the equipment
level, and the ADN2807 performance with respect to those
specifications.
Jitter Generation
The jitter generation specification limits the amount of jitter
that can be generated by the device with no jitter and wander
applied at the input.
Jitter Transfer
The jitter transfer function is the ratio of the jitter on the output
signal to the jitter applied on the input signal versus the
frequency. This parameter measures the limited amount of jitter
on an input signal that can be transferred to the output signal
(Figure 11).
0.1
SLOPE = –20dB/DECADE
JITTER GAIN (dB)
ACCEPTABLE
RANGE
f
C
JITTER FREQUENCY (kHz)
Figure 11. Jitter Transfer Curve
Jitter Tolerance
The jitter tolerance is defined as the peak-to-peak amplitude of
the sinusoidal jitter applied on the input signal that causes a
1 dB power penalty. This is a stress test intended to ensure that
no additional penalty is incurred under the operating
conditions (Figure 12).
15
SLOPE = –20dB/DECADE
1.5
0.15
INPUT JITTER AMPLITUDE (UI)
f
0
Figure 12. SONET Jitter Tolerance Mask
f
1
JITTER FREQUENCY (Hz)
f2f
3
f
4
03877-0-012
03877-0-011
Table 4. Jitter Transfer and Tolerance: SONET Specifications vs. ADN2807
Jitter tolerance measurements are limited by test equipment capabilities.
Rev. A | Page 9 of 20
SONET Spec
(UI p-p)
ADN2807
(UI p-p)
Implementation
2
Margin
ADN2807
T
A
A
www.BDTIC.com/ADI
THEORY OF OPERATION
The ADN2807 is a delay-locked 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 the
input jitter. A separate phase control loop, comprised of the
VCO, tracks the low frequency components of the input jitter.
The initial frequency of the VCO is set by a third loop, which
compares the VCO frequency with the reference frequency and
sets the coarse tuning voltage. The jitter tracking phase-locked
loop controls the VCO by the fine tuning control. The delayand phase-locked 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 of 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. Since the loop filter is an integrator, the static phase
error will be driven to zero.
Another view of the circuit is that the phase shifter implements
the zero required for the frequency compensation of a secondorder phase-locked loop. This zero is placed in the feedback
path and, therefore, does not appear in the closed-loop transfer
function. Jitter peaking in a conventional second-order phaselocked loop is caused by the presence of this zero in the closedloop transfer function. Since this circuit has no zero in the
closed-loop transfer, jitter peaking is minimized.
The delay- and phase-locked loops together simultaneously
provide wideband jitter accommodation and narrow-band jitter
filtering. The linearized block diagram in Figure 13 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 the main PLL loop has low jitter peaking (Figure 14),
which makes this circuit ideal for signal regenerator
applications where jitter peaking in a cascade of regenerators
can contribute to hazardous jitter accumulation.
psh
INPUT
DAT A
RECOVERED
d = PHASE DETEC
o = VCO GAIN
c = LOOP INTEGR
psh = PHASE SHIFTER GAIN
n = DIVIDE R
OR GAIN
TOR
TIO
Figure 13. PLL/DLL Architecture
CLOCK
e(s)X(s)
d/sc
Z(s)
JITTERTRANSFER FUNCTION
Z(s)
X(s)
TRACKING ERRORTRANSFER FUNCTION
e(s)
X(s)
o/s
1/n
=
cn
2
s
+ s +1
do
=
d psh
s2 + s+
1
2
s
c
n psh
o
do
cn
03877-0-013
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 wideband jitter accommodation
since the jitter transfer function, Z(s)/X(s), provides the narrowband jitter filtering. See Table 4 for error transfer bandwidths
and jitter transfer bandwidths at the various data rates. The
delay-locked and phase-locked loops contribute to overall jitter
accommodation. 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.
Rev. A | Page 10 of 20
ADN2807
www.BDTIC.com/ADI
At medium jitter frequencies, the gain and tuning range of the
VCO are not large enough to track the input jitter. In this case,
the VCO control voltage becomes large and saturates, and the
VCO frequency dwells at one or the other extreme of its tuning
range. The size of the VCO tuning range, therefore, has only a
small affect on the jitter accommodation. The delay-locked loop
control voltage is now larger; therefore, the phase shifter takes
on the burden of tracking the input jitter. The phase shifter
range, in UI, can be seen as a broad plateau on the jitter
tolerance curve. The phase shifter has a minimum range of 2 UI
at all data rates.
The gain of the loop integrator is small for high jitter
frequencies, so larger phase differences are needed to make the
loop control voltage big enough to tune the range of the phase
shifter. Large phase errors at high jitter frequencies cannot be
tolerated. In this region, the gain of the integrator determines
the jitter accommodation. Since the gain of the loop integrator
declines linearly with frequency, jitter accommodation is lower
with higher jitter frequency. At the highest frequencies, the loop
gain is very small, and little tuning of the phase shifter can be
expected. In this case, jitter accommodation is determined by
the eye opening of the input data, the static phase error, and the
residual loop jitter generation. The jitter accommodation is
roughly 0.5 UI in this region. The corner frequency between the
declining slope and the flat region is the closed loop bandwidth
of the delay-locked loop, which is roughly 5 MHz for OC-12
data rates and 600 kHz for OC-3 data rates.
JITTER PEAKING
IN ORDINARY PLL
JITTER
GAIN
(dB)
d psh
o
n psh
Figure 14. Jitter Response vs. Conventional PLL
c
ADN2807
Z(s)
X(s)
f
(kHz)
03877-0-014
Rev. A | Page 11 of 20
ADN2807
www.BDTIC.com/ADI
FUNCTIONAL DESCRIPTION
MULTIRATE CLOCK AND DATA RECOVERY
The ADN2807 recovers clock and data from serial bit streams at
OC-3, OC-12 data rates as well as the 15/14 FEC rates. The
output of the 2.5 GHz VCO is divided down in order to support
the lower data rates. The data rate is selected by the SEL[2..0]
inputs (Table 5).
The limiting amplifier has differential inputs (PIN/NIN) that
are internally terminated with 50 Ω to an on-chip voltage
reference (VREF = 0.6 V typically). These inputs are normally
ac-coupled, although dc-coupling is possible as long as the input
common-mode voltage remains above 0.4 V (Figure 24 to
Figure 26 in the Applications Information section). Input offset
is factory trimmed to achieve better than 4 mV typical
sensitivity with minimal drift. The limiting amplifier can be
driven differentially or single-ended.
SLICE ADJUST
The quantizer slicing level can be offset by ±100 mV to mitigate
the effect of ASE (amplified spontaneous emission) noise by
applying a differential voltage input of ±0.8 V to SLICEP/N
inputs. If no adjustment of the slice level is needed, SLICEP/N
should be tied to VCC.
LOSS-OF-SIGNAL (LOS) DETECTOR
The receiver front end level signal detect circuit indicates when
the input signal level has fallen below a user adjustable
threshold. The threshold is set with a single external resistor
from THRADJ (Pin 1) to GND. The LOS comparator trip point
versus the resistor value is illustrated in Figure 4 (this is only
valid for SLICEP = SLICEN = VCC). If the input level to the
ADN2807 drops below the programmed LOS threshold,
SDOUT (Pin 45) will indicate the loss-of-signal condition with
a Logic 1. The LOS response time is ~300 ns by design but will
be dominated by the RC time constant in ac-coupled
applications. If the LOS detector is used, the quantizer slice
adjust pins must both be tied to VCC. This is to avoid
interaction with the LOS threshold level.
Note that it is not expected to use both LOS and slice adjust at
the same time. Systems with optical amplifiers need the slice
adjust to evade ASE. However, a loss-of-signal in an optical link
that uses optical amplifiers causes the optical amplifier output
to be full-scale noise. Under this condition, the LOS would not
detect the failure. In this case, the loss-of-lock signal indicates
the failure because the CDR circuitry is unable to lock onto a
signal that is full-scale noise.
REFERENCE CLOCK
There are three options for providing the reference frequency to
the ADN2807: differential clock, single-ended clock, or crystal
oscillator. See Figure 15 to Figure 17 for example configurations.
ADN2807
REFCLKP
BUFFER
REFCLKN
VCC
VCC
VCC
Figure 15. Differential REFCLK Configuration
VCC
CLK
OSC
REFCLKP
OUT
REFCLKN
NC
VCC
VCC
VCC
Figure 16. Single-Ended REFCLK Configuration
XO1
XO2
XO1
XO2
100kΩ 100kΩ
CRYSTAL
OSCILLATOR
REFSEL
ADN2807
100kΩ 100kΩ
CRYSTAL
OSCILLATOR
REFSEL
VCC/2
BUFFER
VCC/2
03877-0-015
03877-0-016
Rev. A | Page 12 of 20
ADN2807
www.BDTIC.com/ADI
Table 7. Required Crystal Specifications
Parameter Value
Mode Series Resonant
Frequency/Overall Stability 19.44 MHz ± 100 ppm
Frequency Accuracy ±100 ppm
Temperature Stability ±100 ppm
Aging ±100 ppm
ESR 50 Ω max
REFSEL must be tied to VCC when the REFCLKN/P inputs are
active or to VEE when the oscillator is used. No connection
between the XO pin and REFCLK input is necessary (Figure 15
to Figure 17). Note that the crystal should operate in series
resonant mode, which renders it insensitive to external
parasitics. No trimming capacitors are required.
LOCK DETECTOR OPERATION
The lock detector monitors the frequency difference between
the VCO and the reference clock and deasserts the loss-of-lock
signal when the VCO is within 500 ppm of center frequency.
This enables the phase loop, which then maintains phase lock,
unless the frequency error exceeds 0.1%. Should this occur, the
loss-of-lock signal is reasserted and control returns to the
frequency loop, which will reacquire and maintain a stable clock
signal at the output. The frequency loop requires a single
external capacitor between CF1 and CF2. The capacitor
specification is given in Table 8.
Table 8. Recommended CF Capacitor Specification
Parameter Value
Temperature Range –40°C to +85°C
Capacitance >3.0 µF
Leakage <80 nA
Rating >6.3 V
LOL
XO1
XO2
ADN2807
100kΩ 100kΩ
CRYSTAL
OSCILLATOR
REFSEL
BUFFER
VCC/2
03877-0-017
VCC
REFCLKP
NC
REFCLKN
19.44MHz
Figure 17. Crystal Oscillator Configuration
The ADN2807 can accept any of the following reference clock
frequencies: 19.44 MHz, 38.88 MHz, and 77.76 MHz at LVTTL/
LVCMOS/LVPECL/LVDS levels, or 155.52 MHz at LVPECL/
LVDS levels via the REFCLKN/P inputs, independent of data
rate. The input buffer accepts any differential signal with a
peak-to-peak differential amplitude of greater than 100 mV
(e.g., LVPECL or LVDS) or a standard single-ended low voltage
TTL input, providing maximum system flexibility. The
appropriate division ratio can be selected using the REFSEL0/1
pins according to Table 6. Phase noise and duty cycle of the
reference clock are not critical, and 100 ppm accuracy is
sufficient.
An on-chip oscillator to be used with an external crystal is also
provided as an alternative to using the REFCLKN/P inputs.
Details of the recommended crystal are given in Table 7.
Table 6. Reference Frequency Selection
Applied Reference
REFSEL REFSEL[1..0]
Frequency (MHz)
1 00 19.44
1 01 38.88
1 10 77.76
1 11 155.52
0 XX
REFCLKP/N Inactive. Use 19.44 MHz
XTAL on Pins XO1, XO2 (Pull REFCLKP
to VCC)
1
100050005001000f
Figure 18. Transfer Function of LOL
Rev. A | Page 13 of 20
ERROR
VCO
(ppm)
03877-0-018
ADN2807
www.BDTIC.com/ADI
ADN2807
PIN
NIN
VREF
+
QUANTIZER
50Ω 50Ω
50Ω 50Ω
VCC
TDINP/NLOOPEN BYPASS
Figure 19. Test Modes
SQUELCH MODE
When the squelch input is driven to a TTL high state, both the
clock and data outputs are set to the zero state to suppress
downstream processing. If desired, this pin can be directly
driven by the LOS (loss-of-signal) detector output (SDOUT). If
the squelch function is not required, the pin should be tied to
VEE.
TEST MODES—BYPASS AND LOOP-BACK
When the bypass input is driven to a TTL high state, the
quantizer output is connected directly to the buffers driving the
data out pins, thus bypassing the clock recovery circuit
(Figure 19). This feature can help the system to deal with
nonstandard bit rates.
0
CDR
FROM
QUANTIZER
1
OUTPUT
RETIMED
DATACLK
10
DATAOUTP/NCLKOUTP/N SQUELCH
03877-0-019
The loopback mode can be invoked by driving the LOOPEN
pin to a TTL high state, which facilitates system diagnostic
testing. This will connect the test inputs (TDINP/N) to the
clock and data recovery circuit (per Figure 19). The test inputs
have internal 50 Ω terminations and can be left floating when
not in use. TDINP/N are CML inputs and can be dc-coupled
only when being driven by CML outputs. The TDINP/N inputs
must be ac-coupled if driven by anything other than CML
outputs. Bypass and loop-back modes are mutually exclusive;
only one of these modes can be used at any given time. The
ADN2807 is put into an indeterminate state if both BYPASS
and LOOPEN pins are set to Logic 1 at the same time.
Rev. A | Page 14 of 20
ADN2807
www.BDTIC.com/ADI
APPLICATION INFORMATION
PCB DESIGN GUIDELINES
Proper RF PCB design techniques must be used for optimal
performance.
Power Supply Connections and Ground Planes
Use of one low impedance ground plane to both analog and
digital grounds is recommended. The VEE pins should be
soldered directly to the ground plane to reduce series
inductance. If the ground plane is an internal plane and
connections to the ground plane are made through vias,
multiple vias may be used in parallel to reduce the series
inductance, especially on Pins 33 and 34, which are the ground
returns for the output buffers.
Use of a 10 µF electrolytic capacitor between VCC and GND is
r
ecommended at the location where the 3.3 V supply enters the
PCB. Use of 0.1 µF and 1 nF ceramic chip capacitors should be
placed between IC power supply VCC and GND as close as
possible to the ADN2807’s VCC pins. Again, if connections to
the supply and ground are made through vias, the use of
multiple vias in parallel will help to reduce series inductance,
especially on Pins 35 and 36, which supply power to the high
speed CLKOUTP/N and DATAOUTP/N output buffers. Refer
to the schematic in Figure 20 for recommended connections.
Transmission Lines
Use of 50 Ω transmission lines are required for all high
frequency input and output signals to minimize reflections,
including PIN, NIN, CLKOUTP, CLKOUTN, DATAOUTP, and
DATAOUTN (also REFCLKP/N for a 155.52 MHz REFCLK). It
is also recommended that the PIN/NIN input traces are
matched in length and that the CLKOUTP/N and
DATAOUTP/N traces are matched in length. All high speed
CML outputs, CLKOUTP/N and DATAOUTP/N, also require
100 Ω back termination chip resistors connected between the
output pin and VCC. These resistors should be placed as close
as possible to the output pins. These 100 Ω resistors are in
parallel with on-chip 100 Ω termination resistors to create a
50 Ω back termination (Figure 21).
The high speed inputs, PIN and NIN, are internally terminated
wi
th 50 Ω to an internal reference voltage (Figure 22). A 0.1 µF
capacitor is recommended between VREF (Pin 4) and GND to
provide an ac ground for the inputs.
As with any high speed mixed-signal design, care should be
t
aken to keep all high speed digital traces away from sensitive
analog nodes.
Soldering Guidelines for Chip Scale Package
The leads on the 48-lead LFCSP are rectangular. The printed
circuit board pad for these should be 0.1 mm longer than the
package lead length and 0.05 mm wider than the package lead
width. The land should be centered on the pad. This ensures
that solder joint size is maximized. The bottom of the LFCSP
has a central exposed pad. The pad on the printed circuit board
should be at least as large as this exposed pad. The user must
connect the exposed pad to analog VCC. If vias are used, they
should be incorporated into the pad at 1.2 mm pitch grid. The
via diameter should be between 0.3 mm and 0.33 mm, and the
via barrel should be plated with 1 oz. copper to plug the via.
Rev. A | Page 15 of 20
ADN2807
www.BDTIC.com/ADI
VCC
TRANSMISSION
LINES
VCC
36
VCC
35
VEE
34
VEE
33
SEL0
32
NC
31
SEL1
30
VEE
29
VCC
28
VEE
27
VCC
26
CF2
25
µC
50Ω
CLKOUTP
CLKOUTN
DATAOUTP
DATAOUTN
1nF
µC
NC
µC
1nF
4.7µF
(SEETABLE8 FOR SPECS)
VCC
0.1µF
0.1µF
VCC
VCC
VCC
TIA
0.1µF
50Ω
50Ω
10µF
C
IN
VCC
µC
19.44MHz
R
1nF
0.1µF
TH
THRADJ
VCC
VEE
VREF
PIN
NIN
SLICEP
SLICEN
VEE
LOL
XO1
XO2
4 × 100Ω
µC
1nF0.1µF
LOOPEN
VCC
VEE
SDOUT
BYPASS
VEE
VEE
CLKOUTP
CLKOUTN
48 47 46 45 44 43 42 41 40 39 38 37
1
2
3
4
5
6
7
8
9
10
11
ADN2807
12
13 14 15 16 17 18 19
REFCLKN
NC
REFSEL
REFCLKP
VCC
EXPOSED PAD
TIED OFFTO
VCC PLANE
WITH VIAS
1nF
VEE
TDINP
TDINN
NC
NC
0.1µ F
20 21 22 23 24
CF1
VEE
VCC
DATAOUTP
SQUELCH
VEE
REFSEL1
µC
DATAOUTN
REFSEL0
100Ω
VCC
100Ω
VCC
100Ω 100Ω
0.1µ F
0.1µ F
50Ω
50Ω
ADN2807
Figure 21. AC-Coupled Output Configuration
1nF
Figure 20. Typical Application Circuit
V
TERM
50Ω
50Ω
V
TERM
03877-0-021
VCC
0.1µF
VCC
C
0.1µ F
IN
C
IN
TIA
50Ω
50Ω
Figure 22. AC-Coupled Input Configuration
ADN2807
PIN
NIN
50Ω50Ω
VREF
03877-0-020
03877-0-022
Rev. A | Page 16 of 20
ADN2807
www.BDTIC.com/ADI
CHOOSING AC COUPLING CAPACITORS
The ac coupling capacitors at the input (PIN, NIN) and output
(DATAOUTP, DATAOUTN) of the ADN2807 must be chosen
so that the device works properly at both OC-3 and OC-12 data
rates. When choosing the capacitors, the time constant formed
with the two 50 Ω resistors in the signal path must be
considered. When a large number of consecutive identical digits
(CIDs) are applied, the capacitor voltage can drop due to
baseline wander (Figure 23), causing pattern dependent jitter
(PDJ). For the ADN2807 to work robustly at both OC-3 and
OC-12, a minimum capacitor of 0.1 µF to PIN/NIN and 0.1 µF
on DATAOUTP/DATAOUTN should be used. This is based on
the assumption that 1000 CIDs must be tolerated, and that the
PDJ should be limited to 0.01 UI p-p.
DC-COUPLED APPLICATION
The inputs to the ADN2807 can also be dc-coupled. This may
be necessary in burst mode applications where there are long
periods of CIDs, and where baseline wander cannot be
tolerated. If the inputs to the ADN2807 are dc-coupled, care
must be taken not to violate the input range and common-mode
level requirements of the ADN2807 (Figure 24 to Figure 26). If
dc coupling is required and the output levels of the TIA do not
adhere to the levels shown in Figure 25 and Figure 26, there
must be level shifting and/or an attenuator between the TIA
outputs and the ADN2807 inputs.
LOL TOGGLING DURING LOSS OF INPUT DATA
If the input data stream is lost due to a break in the optical link
(or for any reason), the clock output from the ADN2807 stays
within 1000 ppm of the VCO center frequency as long as there
is a valid reference clock. The LOL pin will toggle at a rate of
several kHz. This is because the LOL pin will toggle between a
Logic 1 and Logic 0 while the frequency loop and phase loop
swap control of the VCO. The chain of events is as follows:
• The ADN2807 is locked to the input data stream; LOL = 0.
• The input data stream is lost due to a break in the link. The
VCO frequency drifts until the frequency error is greater
than 1000 ppm. LOL is asserted to a Logic 1 as control of
the VCO is passed back to the frequency loop.
•The frequency loop pulls the VCO to within 500 ppm of its
center frequency. Control of the VCO is passed back to the
phase loop and LOL is deasserted to Logic 0.
•The phase loop tries to acquire, but there is no input data
present so the VCO frequency drifts.
•The VCO frequency drifts until the frequency error is
greater than 1000 ppm. LOL is asserted to a Logic 1 as
control of the VCO is passed back to the frequency loop.
This process is repeated until a valid input data stream is
re-established.
C
IN
V2V1
PIN
TIA
1
V1
V1b
V2
V2b
V
DIFF
V
= V2–V2b
DIFF
VTH = ADN2807 QUANTIZERTHRESHOLD
NOTES
1. DURING DATA PATTERNS WITH HIGH TRANSITION DENSITY, DIFFERENTIAL DC VOLTAGE AT V1 AND V2 IS 0.
2. WHENTHE OUTPUT OFTHE TIA GOESTO CID, V1 AND V1b ARE DRIVEN TO DIFFERENT DC LEVELS. V2 AND V2b
DISCHARGETOTHE V
3. WHENTHE BURST OF DATA STARTS AGAIN,THE DIFFERENTIAL DC OFFSET ACROSSTHE AC COUPLING CAPACITORS IS APPLIEDTOTHE INPUT LEVELS,
CAUSING A DC SHIFT INTHE DIFFERENTIAL INPUT. THIS SHIFT IS LARGE ENOUGH SUCHTHAT ONE OFTHE STATES, EITHER HIGH OR LOW DEPENDING ON
THE LEVELS OF V1 AND V1bWHENTHE TIA WENT TO CID, IS CANCELLED OUT. THE QUANTIZERWILL NOT RECOGNIZETHIS AS A VALID STATE.
4. THE DC OFFSET SLOWLY DISCHARGES UNTILTHE DIFFERENTIAL INPUT VOLTAGE EXCEEDSTHE SENSITIVITY OF THE ADN2807. THE QUANTIZERWILL BE
ABLETO RECOGNIZE BOTH HIGH AND LOW STATES ATTHIS POINT.
LEVEL,WHICH EFFECTIVELY INTRODUCES A DIFFERENTIAL DC OFFSETACROSSTHE AC COUPLING CAPACITORS.