REV. 0–22–
AD9860/AD9862
RECEIVE SECTION COMPONENTS
The receive block is configurable to process input signals of dif-
ferent formats and has special features such as an input buffer,
gain stage, and decimation filters. The AD9860/AD9862 receive
path block diagram is shown in Figure 6. The block diagram can
be broken into the following stages: Input Buffer (Block A),
RxPGA (Block B), dual, 10-/12-bit, 64 MSPS ADC (Block C),
Decimation filter (Block D), Digital Hilbert Block (Block E),
and a Data Output Multiplexer. The function of each stage is
explained in the following paragraphs.
Input Buffer Stage
The input buffer stage buffers the input signal on-chip for both
receive paths. The buffer stage has two main benefits, providing
a constant input impedance and reducing any kick-back noise
that might be generated on-chip, affecting the analog input signal.
The Rx path sampling mode can be split into two categories,
depending on the frequency of the input signal. When sampling
input signals up to Nyquist of the ADC, the sampling is referred to
as Nyquist sampling. When sampling at rates above ADC Nyquist
rate, the sampling is referred to as IF sampling or undersampling.
For Nyquist sampling, the input buffer provides a constant 200 W
impedance over the entire input signal range. The constant input
impedance accommodates matching networks to ensure proper
transfer of signal to the input of the device. The input buffer is
self-biased to ~ 2 V, and therefore the input signal should be
ac-coupled to the Rx differential input or have a common-mode
voltage of about 2 V. If an external buffer is present, the internal
input buffer can be bypassed and powered down to reduce power
consumption. The input buffer accepts up to a 2 V p-p input
signal for maximum SNR performance. Optimal THD perfor-
mance occurs with 1 V p-p input signal.
For IF sampling, the input buffer can be used with input signals
up to about 100 MHz, the 3 dB bandwidth of the buffer. When
undersampling the input signal, the output spectrum will contain
an aliased version of the original, higher frequency signal. As was
the case with Nyquist sampling, the input signal should be
ac-coupled to the Rx differential input or have a common-mode
voltage of ~ 2 V. For input signals over 100 MHz to about 250 MHz,
the input buffer needs to be bypassed and an external input
buffer is required. In the case that the input buffer is bypassed,
the input circuit is a switched capacitor network. The switching
input impedance during the sample phase is about 1/(2()FC),
where F is the input frequency and C is the input capacitance
(about 4 pF). During hold mode, the input impedance is > 1 MW.
RxPGA
The RxPGA stage has a Programmable Gain Amplifier that can be
used to amplify the input signal to utilize the entire input range
of the ADC. The RxPGA stage provides a 0 dB to 20 dB gain
range in steps of about 1 dB. The Rx channel independent gain
control is accomplished through two 5-bit SPI programmable
RxPGA A/B registers. The gain curve is linear in dB with a minimum
gain setting (0 dB, nominally) of hex00 and a maximum gain
setting (20 dB, nominally) of hex14.
The RxPGA stage can provide up to a 2 V p-p signal to the
ADC input.
Analog-to-Digital (A/D) Converter
The analog-to-digital converter (ADC) stage consists of two
high performance 10-/12-bit, 64 MSPS analog-to-digital (A/D)
converters. The dual A/D converter paths are fully independent,
except for a shared internal bandgap reference source, V
REF
. Each
of the A/D converters paths consists of a front-end sample and
hold amplifier followed by a pipelined, switched capacitor, A/D
converter. The pipelined A/D converter is divided into three sections,
consisting of a 4-bit first stage followed by eight 1.5-bit stages and
a final 3-bit flash. Each stage provides sufficient overlap to correct
for flash errors in the preceding stages. The quantized outputs
from each stage are combined into a final 12-bit result through
a digital correction logic block. The pipelined architecture permits
the first stage to operate on a new input sample while the remain-
ing stages operate on preceding samples. Sampling occurs on the
rising clock edge.
Each stage of the pipeline, excluding the last, consists of a low
resolution flash A/D connected to a switched capacitor DAC and
interstage residue amplifier (MDAC). The residue amplifier magni-
fies the difference between the reconstructed DAC output and the
flash input for the next stage in the pipeline. One bit of redundancy
is used in each one of the stages to facilitate digital correction of
flash errors. The last stage simply consists of a flash A/D.
A stable and accurate 1.0 V bandgap voltage reference is built into
the AD9860/AD9862 and is used to set a 2 V p-p differential input
range. The internally generated reference should be decoupled
at the V
REF
pin using a 10 mF and a 0.1 mF capacitor in parallel
to ground. Separate top and bottom references, V
RT
and V
RB
,
for each converter are generated from V
REF
and should also be
decoupled. Recommended decoupling for the top and bottom
references consists of using 10 mF and 0.1 mF capacitors in parallel
between the differential reference pins, and a 0.1 mF capacitor
ADC
LOW-PASS
DECIMATION
FILTER
PGA
1
VIN+A
VIN–A
ADC
PGA
1
VIN+B
VIN–B
HILBERT
FILTER
BLOCK A BLOCK B BLOCK C BLOCK D BLOCK E
RxA DATA
[0:11]
RxB DATA
[0:11]
Figure 6. Receive Section Block Diagram
REV. 0
AD9860/AD9862
–23–
from each to ground. The internal references can also be disabled
(powered down) and driven externally to provide a different input
voltage range or low drift reference. If an external V
REF
reference
is used, it should not exceed 1.0 V.
A Shared Reference mode allows the user to connect the differen-
tial references from both ADCs together externally for superior
gain matching performance. If the ADCs are to function inde-
pendently, then the reference can be left separate and will provide
superior isolation between the dual channels. Shared Reference
mode can be enabled through the Shared Ref register.
A power-down option allows the user to power down both ADCs
(sleep mode) or either ADC individually to reduce power
consumption.
Decimation Stage
For signals with maximum frequencies less than or equal to 3/16 the
ADC sampling rate, f
ADC
, the decimate by 2 filter (or half-band
filter) can be used to provide on-chip suppression of out-of- band
images and noise. When data is present in frequencies greater
than 1/4 f
ADC
, the decimate by 2 filter can be disabled by switching
the filter out of the circuit. The decimation filter allows the ADC
to oversample the input while decreasing the output data rate by
half. The two main benefits are a simplification of the input anti-
aliasing filter and a slower data interface rate with the external
digital ASIC. The decimation filter is an 11 tap filter and suppresses
out of band noise by 38 dB.
Hilbert Block
The Hilbert filter is available to provide a Hilbert Transform of the
data from the ADC in Channel B. The Digital Hilbert Transform,
in combination with an external complex downconverter, enables
a receive image rejection architecture (similar to Hartley image
rejection architecture). The Hilbert filter pass-band (< 0.1 dB
ripple) is between 25% to 75% of the Nyquist rate of its input data
rate. The maximum data rate of the Rx Hilbert filter is 32 MSPS.
At ADC rates higher than this, the decimation filters should be
enabled. The Hilbert filter transfer function plots are shown in
Figure 7.
NORMALIZED – f
S
–40
–0.5
MAGNITUDE – dB
–80
–120
–0.4 –0.3 –0.2 –0.1 0 0.1 0.2 0.3 0.4 0.5
0
Figure 7. Rx Hilbert Filter, Keeping Positive Frequencies
Response
Data Output Multiplexer Stage
The Rx data output format can be configured for either twos
complement or offset binary. This is controlled by the Rx Twos
Complement register.
The output data from the dual ADCs can be multiplexed onto
a single 10-/12-bit output bus. The multiplexing is synchronized
using the RxSYNC output pin that indicates which channel data
is on the output bus.
RECEIVE APPLICATIONS SECTION
The AD9860/AD9862 receive path (Rx) includes two high speed,
high performance, 10-/12-bit ADCs. Figure 6 shows a detailed
block diagram of the Rx data path and can be referred to through-
out the explanation of the various modes of operation. The various
Rx modes of operation are broken into three parts determined by
the type of input signal:
1. Single Channel ADC Signal
2. Dual Channel Real ADC Signal (diversity or dual channel)
3. Dual Channel Complex ADC Signal (I and Q or Single
Sideband).
Each one of these parts is further divided into two cases, sampling
input signals up to Nyquist of the ADC (Nyquist sampling) and
sampling at rates above ADC Nyquist rate (IF sampling or
undersampling).
The AD9860/AD9862 uses oversampling and decimation filters to
ease requirements on external filtering components. The decima-
tion filters (for both receive paths) can be used or bypassed so as
to accommodate different signal bandwidths and provide different
output data rates to allow easy integration with several different
data processing schemes.
Nonbaseband data can be used in an effort to avoid the dc offsets
in the receive signal path that can cause errors. By receiving
nonbaseband data, the requirements of external filtering may be
greatly reduced.
In each of the different receive modes, the input buffer, Program-
mable Gain Amplifier (RxPGA), and output multiplexer remain
within the receive path.
Single Channel ADC Signal
In this mode, a single input signal to be digitized is connected to
the differential input pins, VIN+A and VINA. The 10-/12-bit
output Rx data is latched using either CLKOUT1 or CLKOUT2
edges as defined in the Clock Overview section. The Rx path
available options include bypassing the input buffer, Rx PGA
control and using the Decimation Filter. By default, both Rx paths
are enabled and the unused one should be powered down using
the appropriate bit in the Rx Power-Down register, d1.
The input buffer description above explains the conditions under
which the buffer should be bypassed.
If the input signal, or the undersampled alias signal for the
IF sampling case, falls below 40% of the ADC Nyquist rate, the
decimation filter can be enabled to suppress out-of-band noise and
spurious signals by 40 dB or more. With the decimation filter
enabled, the SNR of the Rx path improves by about 2.3 dB.
Dual Channel Real ADC Signal
The Dual Channel Real ADC Signal mode is used to receive
diversity signals or dual independent channel signals that will be
processed independent of each other. In this mode, the two input
signals to be digitized are connected to the differential input pins
of the AD9860/AD9862, VIN+A, VINA, VIN+B, and VINB.
The two 10-/12-bit Rx outputs can be either interleaved onto a
single 10-/12-bit bus or output in parallel on two 10-/12-bit buses.
REV. 0–24–
AD9860/AD9862
The output will be latched using some configuration of CLKOUT1
or CLKOUT2 edges as defined in the Clock Overview section of
the data sheet. The Rx path available options include bypassing
the input buffer, RxPGA control and using the decimation filter.
The input buffer description above explains the conditions under
which the buffer should be bypassed.
If the input signal, or the undersampled alias signal for the
IF sampling case, falls below 40% of the ADC Nyquist rate, the
decimation filter can be enabled to suppress out-of-band noise
and spurious signals by 40 dB or more. With the decimation
filter enabled the SNR of the Rx path improves by about 2.3 dB.
Dual Channel Complex ADC Signal
The Dual Channel Complex ADC Signal mode is used to receive
baseband I and Q signals or a single sideband signal at some IF.
In this mode, a complex input signal is generated from an external
quadrature demodulator. The in-phase channel (I channel) is
connected to VIN+A and VINA, and the Quadrature Data
(Q channel) is connected to the VIN+B and VINB differential
pins. The Rx path available options include bypassing the input
buffer, RxPGA control, the decimation filter, and using the digital
Hilbert filter. Shared Reference mode is also discussed below.
The RxPGA provides 0 dB to 20 dB gain control for both chan-
nels. The input buffer description above explains the conditions
under which the buffer should be bypassed.
If the input signal, or the undersampled alias signal for the IF sam-
pling case, falls below 40% of the ADC Nyquist rate, the decimation
filter can be enabled to suppress out-of-band noise and spurious
signals by 40 dB or more. With the decimation filter enabled,
the SNR of the Rx path improves by about 2.3 dB.
A digital Hilbert filter can be enabled to provide a receive image
rejection architecture on-chip. The digital Hilbert filter combines
the I data and a phase shifted version of the Q data to produce a
single combined Rx signal. The filter can provide 50 dB image
suppression in the pass band (less than 0.1 dB ripple). The pass
band of the filter is from 25% to 75% of Nyquist rate of the data
entering the Hilbert filter. Note, the Hilbert filters maximum
input data rate is 32 MSPS, at ADC rates above 32 MSPS. The
decimation filter is required to reduce the data rate. With the
decimation filter also enabled, the pass band of the Hilbert filter
will be 12.5% to 37.5% of the ADC Nyquist rate (still 25% to 75%
of the Nyquist rate of the data entering the Hilbert filter).
An optional Shared Reference mode allows the user to connect the
differential references from the dual ADC together externally for
superior gain matching performance. To enable the Shared Ref-
erence mode, the Shared Ref register (d4, b1) should be set high.
TIMING GENERATION BLOCK
The AD9860/AD9862 Timing Generation block uses a single
external clock reference to derive all internal clocks to operate
the transmit and receive channels. The input clock reference
can consist of either an external single ended clock applied to
the OSC1 pin, with the OSC2 pin left floating or an external
crystal connected between the clock input pins (OSC1 and OSC2).
By default, the AD9860/AD9862 can accept either an external
reference clock or a crystal to generate the input clock. The
internal oscillator, if not used, should be disabled by setting the
Input Control Clock register. The OSC1 input impedance is a
relatively high resistive impedance (typically, about 500 kW).
An internal Delay Lock Loop (DLL) based clock multiplier pro-
vides a low noise, 2 or 4 multiplication of the input clock over
an output frequency range of 32 MHz to 128 MHz. The DLL
Fast register should be used to optimize the DLL performance.
For DLL output frequencies between 32 MHz and 64 MHz, this
bit should be set low. For output frequencies between 64 MHz
to 128 MHz, the Fast bit should be set high (for a 64 MHz out-
put frequency, the register can be set either high or low). The DLL
can be bypassed by setting a 1 multiplication factor in the DLL
Multiplier register. The DLL can be powered down when it is
bypassed for power savings by setting the DLL PwrDwn register.
For applications where an external crystal is desired, the AD9860/
AD9862 internal oscillator circuit and the DLL clock multiplier
enable a low frequency, lower cost quartz crystal to be used to
generate the input reference clock. The quartz crystal would be
connected between the OSC1 and OSC2 pins with parallel
resonant load capacitors as specified by the crystal manufacturer.
An internal Duty Cycle Stabilizer (DCS) can be enabled on the
AD9860 by setting the Clk Duty register. This provides a stable
50% duty cycle to the ADC for high speed clock rates between
40 MSPS to 64 MSPS when proper duty cycle is more critical.
System Clock Distribution Circuitry
There are many variables involved in the timing distribution.
External variables include CLKIN, CLKOUT1, CLKOUT2,
Rx Data Rate, Tx Data Rate. Internal variables include ADC
conversion rate, DAC update rate, interpolation rate, decimation
rate, Rx data multiplexing and Tx data demultiplexing. Many of
these parameters are interrelated and based on CLKIN. Optimal
power versus performance and ease of integration options can
be chosen to suit a particular application.
ADC
DATA
MUX
AND
LATCH
DATA
LATCH
AND
DEMUX
NO DECIMATION, 2
DECIMATE:
REG D6 B0
MUX OUT:
REG D5 B0
Rx RETIME:
REG D5 B3
2 DATA PATHS: REG D19 B4
Q/I ORDER: REG D18 B5
Tx RETIME: REG D18 B6
NO INTERP
2, 4
INTERPOLATION:
REG D19 B0, 1
DIV INV
NO INVERSION,
INVERT
INV1: REG D25 B1
1, 1/2
CLKSEL
1, 1/2
ADC DIV2:
REG D24 B5
DLL MULTIPLIER:
REG D24 B3, 4
DIV INVDLL DIV
1, 1/2, 1/4
NO INVERSION,
INVERT
INV2:
REG D25 B5
1, 2, 4
CLKOUT2
DIV FACTOR:
REG 25 B6, 7
DAC
CLOCK PATH
DATA PATH
CLKIN
Rx DATA
[0:23]
CLKOUT1
CLKOUT2
Tx DATA
[0:13]
Figure 8. Normal Operation Timing Block Diagram
One of two possible timing operation modes can be selected. The
typical timing mode is called Normal Operation mode; a block
diagram is shown in Figure 8. The other mode is called Alterna-
tive Operation mode, and a block diagram is shown in Figure 12.

AD9860BSTZ

Mfr. #:
Manufacturer:
Analog Devices Inc.
Description:
Analog Front End - AFE Mixed Signal Front-End Processor
Lifecycle:
New from this manufacturer.
Delivery:
DHL FedEx Ups TNT EMS
Payment:
T/T Paypal Visa MoneyGram Western Union