INTERSIL HSP50110

HSP50110
TM
Data Sheet
March 2001
File Number
3651.6
Digital Quadrature Tuner
Features
The Digital Quadrature Tuner (DQT) provides many of the
functions required for digital demodulation. These functions
include carrier LO generation and mixing, baseband
sampling, programmable bandwidth filtering, baseband AGC,
and IF AGC error detection. Serial control inputs are provided
which can be used to interface with external symbol and
carrier tracking loops. These elements make the DQT ideal
for demodulator applications with multiple operational modes
or data rates. The DQT may be used with HSP50210 Digital
Costas Loop to function as a demodulator for BPSK, QPSK,
8-PSK OQPSK, FSK, FM, and AM signals.
• Input Sample Rates to 52MSPS
The DQT processes a real or complex input digitized at rates
up to 52 MSPS. The channel of interest is shifted to DC by a
complex multiplication with the internal LO. The quadrature
LO is generated by a numerically controlled oscillator (NCO)
with a tuning resolution of 0.012Hz at a 52MHz sample rate.
The output of the complex multiplier is gain corrected and fed
into identical low pass FIR filters. Each filter is comprised of a
decimating low pass filter followed by an optional
compensation filter. The decimating low pass filter is a 3
stage Cascaded-Integrator-Comb (CIC) filter. The CIC filter
can be configured as an integrate and dump filter or a third
order CIC filter with a (sin(X)/X)3 response. Compensation
filters are provided to flatten the (sin(X)/X)N response of the
CIC. If none of the filtering options are desired, they may be
bypassed. The filter bandwidth is set by the decimation rate of
the CIC filter. The decimation rate may be fixed or adjusted
dynamically by a symbol tracking loop to synchronize the
output samples to symbol boundaries. The decimation rate
may range from 1-4096. An internal AGC loop is provided to
maintain the output magnitude at a desired level. Also, an
input level detector can be used to supply error signal for an
external IF AGC loop closed around the A/D.
• Fixed Decimation from 1-4096, or Adjusted by NCO
Synchronization with Baseband Waveforms
• Internal AGC Loop for Output Level Stability
• Parallel or Serial Output Data Formats
• 10-Bit Real or Complex Inputs
• Bidirectional 8-Bit Microprocessor Interface
• Frequency Selectivity <0.013Hz
• Low Pass Filter Configurable as Three Stage CascadedIntegrator-Comb (CIC), Integrate and Dump, or Bypass
• Input Level Detection for External IF AGC Loop
• Designed to Operate with HSP50210 Digital Costas Loop
• 84 Lead PLCC
Applications
• Satellite Receivers and Modems
• Complex Upconversion/Modulation
• Tuner for Digital Demodulators
• Digital PLLs
• Related Products: HSP50210 Digital Costas Loop;
A/D Products HI5703, HI5746, HI5766
• HSP50110/210EVAL Digital Demod Evaluation Board
Ordering Information
PART NUMBER
The DQT output is provided in either serial or parallel formats
to support interfacing with a variety DSP processors or digital
filter components. This device is configurable over a general
purpose 8-bit parallel bidirectional microprocessor control bus.
TEMP.
RANGE (oC)
PACKAGE
PKG. NO.
HSP50110JC-52
0 to 70
84 Ld PLCC
N84.1.15
HSP50110JI-52
-40 to 85
84 Ld PLCC
N84.1.15
Block Diagram
COMPLEX
MULTIPLIER
LOOP
FILTER
GCA
10
10
LOW PASS FIR
FILTER
I DATA
90o
REAL OR COMPLEX
INPUT DATA
0o
CARRIER
TRACKING CONTROL
NCO
10
LOW PASS FIR
FILTER
10
GCA
IF AGC
CONTROL
LEVEL
DETECT
LEVEL
DETECT
CONTROL/STATUS
BUS
8
1
Q DATA
DUMP
PROGRAMMABLE
CONTROL
INTERFACE
RE-SAMPLING
NCO
SAMPLE STROBE
SAMPLE RATE
CONTROL
CAUTION: These devices are sensitive to electrostatic discharge; follow proper IC Handling Procedures.
1-888-INTERSIL or 321-724-7143 | Intersil and Design is a trademark of Intersil Americas Inc.
Copyright © Intersil Americas Inc. 2001, All Rights Reserved
HSP50110
Pinout
IIN6
IIN7
IIN8
IIN9
HI/LO
SSTRB
SPH4
VCC
SPH3
SPH2
SPH1
SPH0
LOTP
OEI
IOUT9
IOUT8
IOUT7
GND
IOUT6
IOUT5
IOUT4
HSP50110 (PLCC)
TOP VIEW
11 10 9 8 7 6 5 4 3 2 1 84 83 82 81 80 79 78 77 76 75
IIN5
IIN4
IIN3
IIN2
GND
IIN1
IIN0
ENI
QIN9
QIN8
QIN7
QIN6
QIN5
QIN4
VCC
QIN3
QIN2
QIN1
QIN0
PH1
PH0
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
74
73
72
71
70
69
68
67
66
65
64
63
62
61
60
59
58
57
56
55
54
IOUT3
IOUT2
IOUT1
IOUT0
DATARDY
VCC
CLK
GND
QOUT9
QOUT8
QOUT7
QOUT6
QOUT5
GND
QOUT4
QOUT3
QOUT2
QOUT1
QOUT0
OEQ
VCC
CFLD
WR
RD
A2
GND
A1
A0
C7
C6
C5
C4
C3
C2
VCC
C1
C0
COF
COFSYNC
SOF
SOFSYNC
GND
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53
Pin Descriptions
NAME
TYPE
DESCRIPTION
VCC
-
+5V Power Supply.
GND
-
Ground.
IIN9-0
I
In-Phase Input. Data input for in-phase (real) samples. Format may be either two’s complement or offset binary format
(see I/O Formatting/Control Register in Table 9). IIN9 is the MSB.
QIN9-0
I
Quadrature Input. Data input for quadrature (imaginary) samples. Format may be either two’s complement or offset binary format (see I/O Formatting/Control Register in Table 9). QIN9 is the MSB.
ENI
I
Input Enable. When ENI is active ‘low’, data on IIN9-0 and QIN9-0 is clocked into the processing pipeline by the rising
edge of CLK. This input also controls the internal data processing as described in the Input Controller Section of the
data sheet. ENI is active ‘low’.
PH1-0
I
Carrier Phase Offset. The phase of the internally generated carrier frequency may be shifted by 0, 90, 180, or 270 degrees by controlling these pins (see Synthesizer/Mixer Section). The phase mapping for these inputs is given in Table 1.
CFLD
I
Carrier Frequency Load. This input loads the Carrier Frequency Register in the Synthesizer NCO (see
Synthesizer/Mixer Section). When this input is sampled ‘high’ by clock, the contents of the Microprocessor Interface
Holding Registers are transferred to the carrier frequency register in the Synthesizer NCO (see Microprocessor Interface Section). NOTE: This pin must be ‘low’ when loading other configuration data via the Microprocessor Interface. Active high Input.
COF
I
Carrier Offset Frequency Input. This serial input is used to load the Carrier Offset Frequency into the Synthesizer NCO
(see Serial Interface Section). The new offset frequency is shifted in MSB first by CLK starting with the clock cycle after
the assertion of COFSYNC.
COFSYNC
I
Carrier Offset Frequency Sync. This signal is asserted one CLK cycle before the MSB of the offset frequency data word
(see Serial Interface Section).
2
HSP50110
Pin Descriptions
(Continued)
NAME
TYPE
DESCRIPTION
SOF
I
Sampler Offset Frequency. This serial input is used to load the Sampler Offset Frequency into the Re-Sampler NCO
(see Serial Interface Section). The new offset frequency is shifted in MSB first by CLK starting with the clock cycle after
assertion of SOFSYNC.
SOFSYNC
I
Sampler Offset Frequency Sync. This signal is asserted one CLK cycle before the MSB of Sampler Offset Frequency
data word (see Serial Interface Section).
A2-0
I
Address Bus. These inputs specify a target register within the Microprocessor Interface (see Table 5). A2 is the MSB.
This input is setup and held to the rising edge of WR.
C7-0
I/0
Control Bus. This is the bidirectional data bus for reads and writes to the Microprocessor Interface (see Microprocessor
Interface Section). C7 is the MSB.
WR
I
Write. This is the write strobe for the Microprocessor Interface (see Microprocessor Interface Section).
RD
I
Read. This is the read enable for the Microprocessor Interface (see Microprocessor Interface Section).
IOUT9-0
O
In-Phase Output. The data on these pins is output synchronous to CLK. New data on IOUT9-0 is indicated by the assertion of the DATARDY pin. Data may be output parallel or serial mode (see Output Formatter Section). In the parallel
mode, IOUT9 is the MSB. When the serial mode is used, IOUT0 is data, and IOUT9 is the serial clock. Other pins not
used in serial mode may be set high or low via the control interface.
QOUT9-0
O
Quadrature Output. The data on these pins is output synchronous to CLK. New data on the QOUT(9-0) pins is indicated
by the DATARDY pin. Data may be output parallel or serial mode. In the parallel mode, IOUT9 is the MSB. When the
serial mode is used, QOUT0 is data.
DATARDY
O
Data Ready. This output is asserted on the first clock cycle that new data is available on the IOUT and QOUT data
busses (see Output Formatter Section). This pin may be active ‘high’ or ‘low’ depending on the configuration of the I/O
Formatting/Control Register (see Table 9). In serial mode, DATARDY is asserted one IQ clock before for first bit of serial
data.
OEI
I
In-Phase Output Enable. This pin is the three-state control for IOUT9-0. When OEI is ‘high’, the IOUT bus is held in the
high impedance state.
OEQ
I
Quadrature Output Enable. This pin is the three-state control for QOUT9-0. When OEQ is ‘high’, the QOUT bus is held
in the high impedance state.
LOTP
0
Local Oscillator Test Point. This output is the MSB of the Synthesizer NCO phase accumulator (see Synthesizer/Mixer
Section). This is provided as a test point for monitoring the frequency of the Synthesizer NCO.
SSTRB
0
Sample Strobe. This is the bit rate strobe for the bit rate NCO. SSTRB has two modes of operation: continuous update
and sampled. In continuous update mode, this is the carry output of the Re-Sampler NCO. In sampled mode, SSTRB
is active synchronous to the DATARDY signal for parallel output mode. The sampled mode is provided to signal the
nearest output sample aligned with or following the symbol boundary. This signal can be used with SPH(4-0) below to
control a resampling filter to time shift its impulse response to align with the symbol boundaries.
SPH4-0
0
Sample Phase. These are five of the most significant 8 bits of the Re-Sampler NCO phase accumulator. Which five bits
of the eight is selected via the Chip Configuration Register (see Table 11). These pins update continuously when the
SSTRB output is in the continuous update mode. When the SSTRB pin is in the sampled mode, SPH4-0 update only
when the SSTRB pin is asserted. In the sampled mode, these pins indicate how far the bit phase has advanced past
the symbol boundary when the output sample updates. SPH4 is the MSB.
HI/LO
0
HI/LO. The output of the Input Level Detector is provided on this pin (see Input Level Detector Section). The sense of
the HI/LO pin is set via the Chip Configuration Register (see Table 11). This signal can be externally averaged and used
to control the gain of an amplifier to close an AGC loop around the A/D converter. This type of AGC sets the level based
on the median value on the input.
CLK
I
Clock. All I/O’s with the exception of the output enables and the microprocessor interface are synchronous to clock.
3
HSP50110
UPPER LIMIT †
LOWER LIMIT †
LOOP GAIN †
AGC THRESHOLD †
HI/LO OUTPUT SENSE †
LEVEL
DETECT
HI/LO
THRESHOLD FOR
EXTERNAL AGC †
AGC
CLK
QIN0-9
10
INPUT
CONTROLLER
10
12
11
12
11
ENI
COS
DECIMATING
FILTER
SIN
10
INPUT FORMAT †
SYNTHESIZER
NCO
SHIFT REG
8
CENTER
FREQUENCY †
PHASE
OFFSET †
LOTP
COF EN †
10
F
O
R
M
A
T
COMPENSATION
FILTER
IOUT0-9
DATARDY
QOUT0-9
OEQ
10
32
PH0-1
CFLD
COF
10
COMPLEX
MULTIPLIER
INPUT MODE †
COFSYNC
OEI
LOW PASS FILTERING
SYNTHESIZER/MIXER
IIN0-9
LEVEL
DETECT
LOOP
FILTER
DIVIDER
CLK
RE-SAMPLER
NCO
32
WORD WIDTH †
SOF
SOFSYNC
SHIFT REG
A0-2
SSTRB
SPH0-4
5
SAMPLER CENTER
FREQUENCY †
SOF EN †
WORD WIDTH †
RE-SAMPLER
WR
RD
MICROPROCESSOR INTERFACE
† Indicates data downloaded via microprocessor interface
C0-7
FIGURE 1. FUNCTIONAL BLOCK DIAGRAM OF HSP50110
Functional Description
The Digital Quadrature Tuner (DQT) provides many of the
functions needed for digital demodulation including: carrier
LO generation, mixing, low-pass filtering, baseband
sampling, baseband AGC, and IF AGC error detection. A
block diagram of the DQT is provided in Figure 1. The DQT
processes a real or complex input at rates up to 52 MSPS.
The digitized IF is input to the Synthesizer/Mixer where it is
multiplied by a quadrature LO of user programmable
frequency. This operation tunes the channel of interest to DC
where it is extracted by the Low Pass FIR Filtering section.
The filter bandwidth is set through a user programmable
decimation factor. The decimation factor is set by the ReSampler which controls the baseband sampling rate. The
baseband sample rate can be adjusted by an external
symbol tracking loop via a serial interface. Similarly, a serial
interface is provided which allows the frequency of the
Synthesizer/Mixer’s NCO to be controlled by an external
carrier tracking loop. The serial interfaces were designed to
mate with the output of loop filters on the HSP50210 Digital
Costas Loop.
The DQT provides an input level detector and an internal
AGC to help maintain the input and output signal
magnitudes at user specified levels. The input level detector
4
compares the input signal magnitude to a programmable
level and generates an error signal. The error signal can be
externally averaged to set the gain of an amplifier in front of
the A/D which closes the AGC loop. The output signal level
is maintained by an internal AGC loop closed around the
Low Pass Filtering. The AGC loop gain and gain limits are
programmable.
Input Controller
The input controller sets the input sample rate of the
processing elements. The controller has two operational
modes which include a Gated Input Mode for processing
sample rates slower than CLK, and an Interpolated Input
Mode for increasing the effective time resolution of the
samples. The mode is selected by setting bit 1 of the I/O
Formatting Control Register in Table 9.
In Gated Input Mode, the Input Enable (ENI) controls the
data flow into the input pipeline and the processing of the
internal elements. When this input is sampled “low” by CLK,
the data on IIN0-9 and QIN0-9 is clocked into the processing
pipeline; when ENI is sampled “high”, the data inputs are
disabled. The Input Enable is pipelined to the internal
processing elements so that they are enabled once for each
time ENI is sampled low. This mode minimizes the
HSP50110
processing pipeline latency, and the latency of the part’s
serial interfaces while conserving power. Note: the
effective input sample rate to the internal processing
elements is equal to the frequency with which ENI is
asserted “low”.
In Interpolated Input Mode, the ENI input is used to insert
zeroes between the input data samples. This process
increases the input sample rate to the processing elements
which improves the time resolution of the processing chain.
When ENI is sampled “high” by CLK, a zero is input into the
processing pipeline. When ENI is sampled “low” the input
data is fed into the pipeline. Note: Due to the nature of the
rate change operation, consideration must be given to
the scaling and interpolation filtering required for a
particular rate change factor.
In either the Gated or Interpolated Input Mode, the
Synthesizer NCO is gated by the ENI input. This only allows
clocking of the NCO when external samples are input to the
processing pipeline. As a result, the NCO frequency must be
set relative to the input sample rate, not the CLK rate (see
Synthesizer/Mixer Section). NOTE: Only fixed
interpolation rates should be used when operating the
part in Interpolated Mode at the Input Controller.
Note: an external AGC loop using the Input Level
Detector may go unstable for a real sine wave input
whose frequency is exactly one quarter of the sample
rate (FS/4). The Level Detector responds to such an
input by producing a square wave output with a 50%
duty cycle for a wide range of thresholds. This square
wave integrates to zero, indicating no error for a range
of input signal amplitudes.
Synthesizer/Mixer
The Synthesizer/Mixer spectrally shifts the input signal of
interest to DC for subsequent baseband filtering. This
function is performed by using a complex multiplier to
multiply the input with the output of a quadrature numerically
controlled oscillator (NCO). The multiplier operation is:
IOUT = IIN x cos (ωc) - QIN x sin (ωc)
(EQ. 3)
QOUT = IIN x sin (ωc) + QIN x cos (ωc)
(EQ. 4)
The complex multiplier output is rounded to 12 bits. For real
inputs this operation is similar to that performed by a
quadrature downconverter. For complex inputs, the
Synthesizer/Mixer functions as a single-sideband or image
reject mixer which shifts the frequency of the complex
samples without generating images.
Input Level Detector
TO COMPLEX MULTIPLIER
The Input Level Detector generates a one-bit error signal for
an external IF AGC filter and amp. The error signal is
generated by comparing the magnitude of the input samples
to a user programmable threshold. The HI/LO pin is then
driven “high” or “low” depending the relationship of its
magnitude to the threshold. The sense of the HI/LO pin is
programmable so that a magnitude exceeding the threshold
can either be represented as a “high” or “low” logic state.
The threshold and the sense of the HI/LO pin are configured
by loading the appropriate control registers via the
Microprocessor Interface (see Tables 7 and 11).
COS
10
SIN
10
REG
REG
† Controlled via
microprocessor interface.
SIN/COS
ROM
R
E
G
PH0-1
LOTP
11
2
+
REG
0
REG
The high/low outputs can be integrated by an external loop
filter to close an AGC loop. Using this method the gain of the
loop forces the median magnitude of the input samples to
the threshold. When the magnitude of half the samples are
above the threshold and half are below, the error signal is
integrated to zero by the loop filter.
The algorithm for determining the magnitude of the complex
input is given by:
Mag(I,Q) = |I| + .375 x |Q| if |I| > |Q|
32
COF
0
REG
COFSYNC
32
CF
PHASE
ACCUMULATOR
REG
SYNC
COF
SHIFT REG
R
E
G
SYNC
CARRIER
FREQUENCY †
LOAD CARRIER
FREQUENCY †
(EQ. 2)
Using this algorithm, the magnitude of complex inputs can
be estimated with an error of <0.55dB or approximately
6.5%. For real inputs, the magnitude detector reduces to a
an absolute value detector with negligible error.
5
LOAD †
MUX
COF
ENABLE †
CFLD
Mag(I,Q) = |Q| + .375 x |I| if |Q| > |I|,
MUX
+
(EQ. 1)
or:
8 R PHASE OFFSET †
E
G
FIGURE 2. SYNTHESIZER NCO
The quadrature outputs of the NCO are generated by driving
a sine/cosine lookup table with the output of a phase
accumulator as shown in Figure 2. Each time the phase
accumulator is clocked, its sum is incremented by the sum of
HSP50110
the contents of the Carrier Frequency (CF) Register and the
Carrier Offset Frequency (COF) Register. As the
accumulator sum transitions from 0 to 232, the SIN/COS
ROM produces quadrature outputs whose phase advances
from 0o to 360o. The sum of the CF and COF Registers
represent a phase increment which determines the
frequency of the quadrature outputs. Large phase
increments take fewer clocks to transition through the sine
wave cycle which results in a higher frequency NCO output.
The NCO frequency is set by loading the CF and COF
Registers. The contents of these registers set the NCO
frequency as given by the following,
FC = FS x (CF + COF)/232,
As an example, a value of 32, (20HEX), loaded into the Phase
Offset Register would produce a phase offset of 45o.
An alternative method for controlling the NCO Phase uses
the PH0-1 inputs to shift the phase of NCO’s output by 0o,
90o, 180o, or 270o. The PH0-1 inputs are mapped to phase
shifts as shown in Table 1. The phase may be updated every
clock supporting the π/2 phase shifts required for modulation
or despreading of CDMA signals.
The output of the complex multiplier is scaled by 2-36. See
“Setting DQT Gains” below.
TABLE 1. PH0-1 INPUT PHASE MAPPING
(EQ. 5)
where fS is the sample rate set by the Input Controller, CF is
the 32-bit two’s complement value loaded into the Carrier
Frequency Register, and COF is the 32-bit two’s
complement value loaded into the Carrier Offset Frequency
Register. This can be rewritten to have the programmed CF
and COF value on the left:
(CF + COF) = INT [ ( F C /F S )2
32
] HEX
(EQ. 5A)
As an example, if the CF Register is loaded with a value of
3000 0000 (Hex), the COF Register is loaded with a value of
1000 0000 (Hex), and the input sample rate is 40 MSPS, an
the NCO would produce quadrature terms with a frequency
of 10MHz. When the sum of CF and COF is a negative
value, the cos/sin vector generated by the NCO rotates
clockwise which downconverts the upper sideband; when
the sum is positive, the cos/sin vector rotates
counterclockwise which upconverts the lower sideband.
Note: the input sample rate FS is determined by the rate
at which ENI is asserted low (see Input Controller
Section). If ENI is tied low, the input sample rate is equal
to the CLK rate.
The Carrier Frequency Register is loaded via the
Microprocessor Interface and the Carrier Offset Frequency is
loaded serially using the COF and COFSYNC inputs. The
procedure for loading these registers is discussed in the
Microprocessor Interface Section and the Serial Input
Section.
The phase of the NCO’s quadrature outputs can be adjusted
by adding an offset value to the output of the phase
accumulator as shown in Figure 2. The offset value can be
loaded into the Phase Offset (PO) Register or input via the
PH0-1 inputs. If the PO Register is used, the phase can be
adjusted from -π to π with a resolution of ~1.4o. The phase
offset is given by the following equation,
PH1-0
PHASE SHIFT
00
0o
01
90o
10
270o
11
180o
AGC
The level of the Mixer output is gain adjusted by an AGC
closed around the Low Pass Filtering. The AGC provides the
coarse gain correction necessary to help maintain the output
of the HSP50110 at a signal level which maintains an
acceptable dynamic range. The AGC consists of a Level
Detector which generates an error signal, a Loop Gain
multiplier which amplifies the error, and a Loop Filter which
integrates the error to produce gain correction (see
Figure 4).
The Level Detector generates an error signal by comparing
the magnitude of the DQT output against a user
programmable threshold (see AGC Control Register in Table
8). In the normal mode of operation, the Level Detector
outputs a -1 for magnitudes above the threshold and +1 for
those below the threshold. The ±1 outputs are then multiplied
by a programmable loop gain to generate the error signal
integrated by the Loop Filter. The Level Detector uses the
magnitude estimation algorithm described in the Input Level
Detector Section. The sense of the Level Detector Output may
be changed via the Chip Configuration Register, bit 0 (see
Table 11).
(EQ. 6)
The Loop Filter consists of a multiplier, an accumulator and a
programmable limiter. The multiplier computes the product of
the output of the Level Detector and the Programmable Loop
Gain. The accumulator integrates this product to produce the
AGC gain, and the limiter keeps the gain between preset
limits (see AGC Control Register, Table 8). The output of the
AGC Loop Filter Accumulator can be read via the
Microprocessor Interface to estimate signal strength (see
Microprocessor Interface Section).
where PO is the 8-bit two’s complement value loaded into the
Phase Offset Register (see Phase Offset Register in Table 12).
The Loop Filter Accumulator uses a pseudo floating point
format to provide up to ~48dB of gain correction. The format
φ = π x (PO/128),
6
HSP50110
GainAGC = (1.0 + M) x 2E
(EQ. 7)
MAPS TO AGC
UPPER AND LOWER LIMITS
most significant bits of the Loop Filter Accumulator.
Examples of how to set the limits for a specific output signal
level are provided in the “Setting DQT Gains” Section below.
NOTE: A fixed AGC gain may be set by programming
the upper and lower limits to the same value.
MANTISSA
0.0 to 0.9375
GAIN (LINEAR)
L L L L L L L L
22 21 20 . 2-1 2-2 2-3 2-4 2-5 2-6 2-7 2-8 2-9 2-10 2-11 2-12 2-13
E E E M M M M X G G G G G G G G
PROGRAMMABLE
LOOP GAIN
EXPONENT
0 TO 7
MAPS TO µP AGC
LOOP FILTER ACCUMULATOR PARAMETER
This Value Can Be Read By The Microprocessor.
See The Microprocessor Interface Section.
256
240
224
208
192
176
160
144
128
112
96
80
64
48
32
16
0
48
42
36
dB
30
24
LINEAR
18
GAIN (dB)
of the accumulator output is shown in Figure 3. The AGC
gain is given by:
12
6
0 16 32 48 64 80 96 112 128 144 160 176 192 208 224 240
0
GAIN CONTROL WORD
(8 MSBs OF LOOP FILTER ACCUMULATOR)
FIGURE 3. BINARY FORMAT FOR LOOP FILTER
ACCUMULATOR
FIGURE 5. GAIN CONTROL TRANSFER FUNCTION
where M is the 4-bit mantissa value ranging from 0.0 to
0.9375, and E is the three bit exponent ranging from 0 to 7.
The result is a piece wise linear transfer function whose
overall response is logarithmic, as shown in Figure 5. The
exponent bits provide a coarse gain setting of 2(EEE). This
corresponds to a gain range from 0dB to 42dB (20 to 27) with
the MSB representing a 24dB gain, the next bit a 12dB gain,
and the final bit a 6dB gain. The four mantissa bits map to an
additional gain of 1.0 to 1.9375 (0 to ~6dB). Together, the
exponent and the mantissa portion of the limit set a gain
range from 0 to ~48dB.
AGC L.D. SENSE †
AGC THRESHOLD
†
AGC
DISABLE †
AGC GAIN
AGC LOOP FILTER
REG
LIMIT
UPPER
GAIN
LIMIT †
†
REG
+
LOWER
GAIN
LIMIT †
LEVEL
DETECTOR
Q DATA
I DATA
PROGRAMMABLE
LOOP GAIN †
Indicates data downloaded via microprocessor interface.
FIGURE 4. AGC BLOCK DIAGRAM
The limiter restricts the AGC gain range by keeping the
accumulator output between the programmed limits. If the
accumulator exceeds the upper or lower limit, then the
accumulator is held to that limit. The limits are programmed
via eight bit words which express the values of the upper and
lower limits as eight bit pseudo floating point numbers as
shown in Figure 3 (see AGC Control Register, Table 8). The
format for the limits is the same as the format of the eight
7
The response time of the AGC is determined by the
Programmable Loop Gain. The Loop Gain is an unsigned
8-bit value whose significance relative to the AGC gain is
shown in Figure 3. The loop gain is added or subtracted from
the accumulator depending on the output of the Level
Detector. The accumulator is updated at the output sample
rate. If the accumulator exceeds the upper or lower limit, the
accumulator is loaded with that limit. The slew rate of the
AGC ranges between ~0.001dB and 0.266dB per output
sample for Loop Gains between 01(HEX) and FF (HEX)
respectively.
The user should exercise care when using maximum loop
gain when the (x/sin(x)) or the (x/sin(x))3 compensation filter is
enabled. At high decimation rates, the delay through the
compensation filter may be large enough to induce
oscillations in the AGC loop. The Basic Architectural
Configurations Section contains the necessary detailed block
diagrams to determine the loop delay for different matched
filter configurations.
Low Pass Filtering
The gain corrected signal feeds a Low Pass Filtering Section
comprised of a Cascaded Integrator Comb (CIC) and
compensation filter. The filtering section extracts the channel
of interest while providing decimation to match the output
sample rate to the channel bandwidth. A variety of filtering
configurations are possible which include integrate and
dump, integrate and dump with x/sin(x) compensation, third
order CIC, and third order CIC with ((x)/sin(x))3
compensation. If none of these filtering options are desired,
the entire filtering section may be bypassed.
HSP50110
I
H ( f ) = ---- sin ( πfR )/sin(πf )
R
(EQ. 8)
where f is normalized frequency relative to the input sample
rate, FS, and R is the decimation rate [1]. The decimation
rate is equivalent to the number of samples in the integration
period. As an example, the frequency response for an
integrate and dump filter with decimation of 64 is shown in
Figure 6. The decimation rate is controlled by the
Re-Sampler and may range in value from 2 to 4096 (see
Re-Sampler Section).
10
COMPOSITE FILTER
MAGNITUDE (dB)
0
COMPENSATION
FILTER
-10
-20
examples of compensation filter performance for the Integrate
and dump and third order CIC filter are shown overlaid on the
frequency responses of the uncompensated filters in Figure 6
and Figure 7. The coefficients for the compensation filters are
given in Table 2.
10
0
COMPENSATION
FILTER
COMPOSITE FILTER
MAGNITUDE (dB)
The Integrate and Dump filter exhibits a frequency response
given by
-10
-20
CIC
FILTER
-30
-40
-50
CIC
FILTER
-60
fS
fS
3fS
2R
R
2R
R
2R
SAMPLE TIMES
-30
NOTE:
-40
2fS
5fS
3fS
7fS
4fS
R
2R
R
Example plotted is for R = 64 with 64 samples/symbol.
FIGURE 7. THIRD ORDER CIC FREQUENCY RESPONSE
-50
TABLE 2. COMPENSATION FILTER COEFFICIENTS
-60
COEFFICIENT INDEX
x/sin(x) [2]
[x/sin(x)]3
0
-1
-1
1
2
4
2
-4
-16
3
10
32
4
-34
-64
5
384
136
(EQ. 9)
6
-34
-352
where f is normalized frequency relative to the input sample
rate, and R is the decimation rate [1]. As with the integrate
and dump filter, the decimation rate is controlled by the ReSampler. The decimation rate may range in value from
2-4096 when using CLK, or 3-4096 when using the ReSampler NCO as a CLK source to the filter. The frequency
response for the third order CIC with a decimation rate of 64
is shown in Figure 7.
7
10
1312
8
-4
-352
9
2
136
10
-1
-64
fS
fS
3fS
2fS
5fS
2R
R
2R
R
2R
SAMPLE TIMES
3fS
7fS
4fS
R
2R
R
NOTE: Example plotted is for R = 64 with 64 samples/symbol.
FIGURE 6. INTEGRATE AND DUMP FILTER (FIRST ORDER
CIC) FREQUENCY RESPONSE
For applications requiring better out of band attenuation, the
Third Order CIC filter may be selected. This filter has a
frequency response given by
H(f) = [sin(πfR)/sin(πf)]3 [1/R]3
Compensation filters may be activated to flatten the frequency
responses of the integrate and dump and third order CIC
filters. The compensation filters operate at the decimated data
rate, and flatten the roll off the decimating filters from DC to
approximately one half of the output sample rate. Together,
the Integrate and Dump filter and x/sin(x) compensation filter
typically yield a lowpass frequency response that is flat to
0.45FS with 0.03dB of ripple, and the third order CIC with
((x)/sin(x))3 compensation typically yields a flat passband to
0.45FS with 0.08dB of ripple. The overall passband ripple
degrades slightly for decimation rates of less than 10. Some
8
11
32
12
-16
13
4
14
-1
The out of band channels and noise attenuated by the
decimating filters are aliased into the output spectrum as a
result of the decimating process. A summation of the alias
terms at each frequency of the output spectrum produce
alias profiles which can be used to determine the usable
output bandwidth. A set of profiles representative of what
would be observed for decimation factors of ~10 or more are
shown in Figures 8 through 11. The Integrate and Dump
filter is typically used as a matched filter for square pulses
HSP50110
and less as a high order decimating filter. This is evident by
the narrow alias free part of the output bandwidth as shown
in Figures 8 and 9. The more rapid roll off of the third order
CIC produces an output spectrum containing a much higher
usable bandwidth versus output sample rate as shown in
Figures 10 and 11. For example, the aliasing noise at FS/4
for the uncompensated third order CIC filter is approximately
~29dB below the full scale input.
Understanding the Alias Profile
For digital filters that utilize decimation techniques to reduce
the rate of the digital processing, care must be taken to
understand the ramifications, in the frequency domain, of
decimation (rate reduction). Of primary concern is the “noise”
level increase due to signals that may be aliased inside the
band of interest. The potential magnitude of these signals
may render significant portions of the previously thought
usable bandwidth, unusable for applications that require
significant (>60dB) attenuation of undesired signals.
Consider a digital filter with sampling frequency fs, whose
frequency response shown in Figure 12A, the top spectrum.
At first glance the usable bandwidth would appear to be the
3dB bandwidth of the main lobe. This filter is to be
decimated to a rate of 1/8 fS. We concern ourselves with
those elements less than fS/2, as shown in Figure 12B. The
decimation process will fold the various lobes of the
frequency response around the new sampling folding
frequency of fS/2R. The first lobe is folded over the dotted line
and a significant portion of the first lobe appears in the
passband of the filter. Any unwanted signals in this part of the
spectrum will appear in the band of interest with the greatest
amplitude. The second lobe is translated down to be centered
on the dashed line. The third lobe is spectrally inverted and
translated to be centered on the dotted line. The fourth lobe is
simply translated to be centered on the dotted line. If there
were more lobes to the filter, the process would continue to
spectrally invert the odd numbered lobes prior to translation to
fS/2R. This process is shown in the “C” portion of Figure 12.
10
10
0
0
-10
MAGNITUDE (dB)
MAGNITUDE (dB)
FILTER RESPONSE
FILTER RESPONSE
-20
ALIAS PROFILE
-30
-40
-50
-60
ALIAS PROFILE
-20
-30
-40
-50
(EXAMPLE PLOTTED IS FOR R = 64
WITH 64 SAMPLES/SYMBOL)
0
-10
fS
fS
3fS
fS
5fS
3fS
7fS
fS
16R
8R
16R
4R
16R
8R
16R
2R
-60
(EXAMPLE PLOTTED IS FOR R = 64
WITH 64 SAMPLES/SYMBOL)
0
fS
fS
3fS
fS
5fS
3fS
7fS
fS
16R
8R
16R
4R
16R
8R
16R
2R
SAMPLE TIMES
SAMPLE TIMES
FIGURE 9. ALIAS PROFILE: INTEGRATE/DUMP FILTER
WITH COMPENSATION
10
10
0
0
-10
MAGNITUDE (dB)
MAGNITUDE (dB)
FIGURE 8. ALIAS PROFILE: INTEGRATE/DUMP FILTER, NO
COMPENSATION
FILTER RESPONSE
-20
-30
ALIAS PROFILE
-40
-50
-60
fS
fS
3fS
fS
5fS
3fS
7fS
fS
16R
8R
16R
4R
16R
8R
16R
2R
SAMPLE TIMES
FIGURE 10. ALIAS PROFILE: 3RD ORDER CIC, NO
COMPENSATION
9
-20
-30
ALIAS PROFILE
-40
-50
(EXAMPLE PLOTTED IS FOR R = 64
WITH 64 SAMPLES/SYMBOL)
0
FILTER RESPONSE
-10
-60
(EXAMPLE PLOTTED IS FOR R = 64
WITH 64 SAMPLES/SYMBOL)
0
fS
fS
3fS
fS
5fS
3fS
7fS
fS
16R
8R
16R
4R
16R
8R
16R
2R
SAMPLE TIMES
FIGURE 11. ALIAS PROFILE: 3RD ORDER CIC WITH
COMPENSATION
HSP50110
constant for decimation factors of over ~50. A summary of
equivalent IF BN’s for different filter configurations and
decimation rates is given in Table 3. These noise bandwidths
are provided so that output SNR can be calculated from input
SNR. In detection applications this bandwidth indicates the
detection bandwidth.
A
B
TABLE 3. DOUBLE SIDED NOISE EQUIVALENT BANDWIDTH
FOR DIFFERENT FILTER CONFIGURATIONS AND
OUTPUT SAMPLE RATES
3RD
ORDER
CIC
3RD
ORDER
CIC W/
[x/sin(x)]3
C
D
fS
2R
fS
2
fS
FIGURE 12.
To create the alias profile, a composite response, the
components of which are shown in the ”D” portion of
Figure 12, is made from the sum of all the alias elements.
The primary use of an alias profile is used to determine what
bandwidth yields the desired suppression of unwanted
signals for a particular application.
Reviewing Figures 9 through 11, note the following
observations:
1. The uncompensated I&D (1st order CIC) filter yields
about 12dB of alias suppression at fS/16R. This usable
bandwidth is considerably narrower than the 3dB filter
bandwidth. The I&D filter is the matched filter for square
wave data that has not been bandlimited.
2. The compensated I&D filter offers a flatter, wider bandwidth than just the I&D alone. This filter compensates for
the frequency roll off due to the A/D converter.
3. The uncompensated 3rd order CIC filter yields over 60dB
of alias suppression at fS/16R. Typical application is
found in tuners, where the DQT is followed by a very narrow band filter.
4. The 3rd order CIC with compensation yields alias suppression comparable to the 3rd order CIC, but with the
flatter, wider passband. This filter is selected for most
SATCOM applications.
Which filter is selected, is dependent on the application. It is
important to utilize these alias responses in calculating the
filter to be used, so that the signal suppression prediction will
accurately reflect the digital filter performance.
Noise Equivalent Bandwidth
The noise equivalent bandwidth (BN) performance of the
channel filter is dependent on the combination of Decimation
Filter and Compensation Filter chosen. For configurations
using the Integrate and Dump filter, BN is constant
regardless of decimation rate. However, for configurations
which use the third order CIC filter, BN converges to a
10
DEC
INTEGRATE/
DUMP
INTEGRATE/
DUMP W/
x/sin(x)
2
1.0000
1.3775
0.6250
1.3937
10
1.0000
1.3775
0.5525
1.0785
18
1.0000
1.3775
0.5508
1.0714
26
1.0000
1.3775
0.5504
1.0698
34
1.0000
1.3775
0.5502
1.0691
42
1.0000
1.3775
0.5501
1.0688
50
1.0000
1.3775
0.5501
1.0687
58
1.0000
1.3775
0.5501
1.0686
66
1.0000
1.3775
0.5501
1.0685
74
1.0000
1.3775
0.5500
1.0684
82
1.0000
1.3775
0.5500
1.0684
90
1.0000
1.3775
0.5500
1.0684
98
1.0000
1.3775
0.5500
1.0684
106
1.0000
1.3775
0.5500
1.0684
114
1.0000
1.3775
0.5500
1.0683
1224096
1.0000
1.3775
0.5500
1.0683
Re-Sampler
The Re-Sampler sets the output sample rate by controlling
the sample rate of the decimation filters (see Low Pass Filter
Section). The output sample rate may be fixed or adjusted
dynamically to synchronize with baseband waveforms. The
reduction in sample rate between the Low Pass Filter input
and output represents the decimation factor.
The Decimating filter output is sampled by the programmable
divider shown in Figure 13. The divider is a counter which is
decremented each time it is clocked. When the divider reaches
its terminal count, the output of the decimating filter is sampled.
The divider may be programmed with a divisor of from 1 to
4096 (see Table 10 Decimating Filter Configuration Register).
One of two internal clock sources are chosen for the divider
based on whether a fixed or adjustable sample rate is
desired. For fixed output sample rates, a clock equal to the
input sample rate is selected (see Decimating Filter
Configuration Table 10). For adjustable output sample rates,
a clock generated by the carry out from the Re-Sampler
NCO is chosen.
HSP50110
TO DECIMATING FILTERS
PROGRAMMABLE
DIVIDER
SAMPLE PHASE
OUT CONTROL †
DATARDY
MODE †
REG
SSTRB
REG
SPH0-4
MUX
SYNC
CLK
32-BIT ADDER
CARRY OUTPUT
5
SHIFTER
8
RE-SAMPLER
NCO
0
32
REG
MUX
+
SOF ENABLE †
LOAD
RESAMPLER
NCO †
MUX
32
32
SOF
REG
0
SCF
REG
SYNC
Center Frequency Register, and SOF is the 32-bit value
loaded into the Sample Offset Frequency Register. The SCF
Register is loaded through the Microprocessor Interface (see
Microprocessor Interface Section), and the SOF Register is
loaded serially via the SOF and SOFSYNC inputs (see Serial
Input Section). The sample rate Fs is a function of the Input
Controller Mode. If the Controller is in Gated Input Mode, Fs is
the frequency with which ENI is asserted. In Interpolated Input
Mode, Fs is the CLK frequency (see Input Controller Section).
The carry out and 5 of the most significant 8 bits of the
NCO’s phase accumulator are output to control a resampling
filter such as the HSP43168. The resampling filter can be
used to provide finer time (symbol phase) resolution than
can be achieved by the sampling clock alone. This may be
needed to improve transmit/receive timing or better, align a
matched filter’s impulse response with the symbol
boundaries of a baseband waveform at high symbol rates.
The carry out of the NCO’s phase accumulator is output on
SSTRB, and a window of 5 of the 8 most significant 8 bits of
the Phase Accumulator are output on SPH0-4.
Output Formatter
SOFSYNC
SOF
†
SYNC
SAMPLER
CENTER
FREQUENCY †
SHIFT REG
LOAD
ON CF
WRITE
Controlled via microprocessor interface.
FIGURE 13. RE-SAMPLER
The calculation of the decimation factor depends on whether
the output sample rate is fixed or adjusted dynamically. For a
fixed sample rate, the decimation factor is equal to the divisor
loaded into the programmable divider. For example, if the
divider is configured with a divisor of 8, the decimation factor
is 8 (i.e., the output data rate is Fs/8). If the decimation factor
is adjusted dynamically, it is a function of both the
programmable divisor and the frequency of carry outs from
the Re-Sampler NCO (FCO) as given by:
Decimation Factor =
(Programmable Divisor) x Fs/FCO
(EQ. 10)
CLK
IOUT9-0/
QOUT9-0
NOTE:
DATARDY may be programmed active high or low.
FIGURE 14. PARALLEL OUTPUT TIMING
NOTE: The CIC filter architecture only supports
decimation factors up to 4096.
The phase accumulator in the Re-Sampler NCO generates
the carry outs used to clock the programmable divider. The
frequency at which carry outs are generated (FCO) is
determined by the values loaded into the Sampler Center
Frequency (SCF) and Sampler Offset Frequency (SOF)
Registers. The relationship between the values loaded into
these registers and the frequency of the carry outs is given by:
(EQ. 11)
where Fs is the input sample rate of the Low Pass Filter
Section, SCF is the 32-bit value loaded into the Sampler
11
In parallel output mode, the in-phase and quadrature
samples are output simultaneously at rates up to the
maximum CLK. The DATARDY output is asserted on the first
CLK cycle that new data is available on IOUT0-9 and
QOUT0-9 as shown in Figure 14. Output enables (OEI,
OEQ) are provided to individually three-state IOUT0-9 and
QOUT0-9 for output multiplexing.
DATARDY
For example, if the programmable divisor is 8 and Fs/FCO =
40, the decimation factor would be 320.
FCO = Fs x (SCF + SOF)/232
The Output Formatter supports either Word Parallel or Bit
Serial output modes. The output can be chosen to have a
two’s complement or offset binary format. The configuration
is selected by loading the I/O Formatting/Control Register
(see Table 9).
When bit serial output is chosen, two serial output modes are
provided, Simultaneous I/Q Mode and I Followed by Q Mode.
In Simultaneous I/Q Mode, the 10-bit I and Q samples are
output simultaneously on IOUT0 and QOUT0 as shown in
Figure 15. In I Followed by Q Mode, both samples are output
on IOUT0 with I samples followed by Q samples as shown in
Figure 16. In this mode, the I and Q samples are packed into
separate 16-bit serial words (10 data bits + 6 zero bits). The
10 data bits are the 10 MSBs of the serial word, and the I
sample is differentiated from the Q sample by a 1 in the LSB
position of the 16-bit data word. A continuous serial output
clock is provided on IOUT9 which is derived by dividing the
HSP50110
Gain Distribution
CLK by a programmable factor of 2, 4, or 8. When the
programmable clock factor is 1, IOUT9 is pulled high, and the
CLK signal should be used as the clock. The beginning of a
serial data word is signaled by the assertion of DATARDY one
serial clock before the first bit of the output word. In I followed
by Q Mode, DATARDY is asserted prior to each 16-bit data
word. For added flexibility, the Formatter may be configured to
output the data words in either MSB or LSB first format.
The gain distribution in the DQT is shown in Figure 17. These
gains consist of a combination of fixed, programmable, and
adaptive gains. The fixed gains are introduced by processing
elements like the Synthesizer/Mixer and CIC Filter. The
programmable and adaptive gains are set to compensate for
the fixed gains as well as variations in input signal strength.
The bit range of the data path between processing elements
is shown in Figure 17. The quadrature inputs to the data path
are 10-bit fractional two’s complement numbers. They are
multiplied by a 10-bit quadrature sinusoid and rounded to
12-bits in the Synthesizer/Mixer. The I and Q legs are then
scaled by a fixed gain of 2-36 to compensate for the worst
case gain of the CIC filter. Next, a gain block with an adaptive
and programmable component is used to set the output signal
level within the desired range of the 10-bit output (see Setting
DQT Gains Section). The adaptive component is produced by
the AGC and has a gain range from 1.0 to 1.9375*27. The
programmable component sets the gain range of the CIC
shifter which may range from 20 to 263. Care must be taken
when setting the AGC gain limits and the CIC Shifter gain
since the sum of these gains could shift the CIC Scaler output
beyond the bit range (-28 to 2-46) of the CIC Filter input. The
CIC Filter introduces a gain factor given by RN where R is the
decimation rate of the filter and N is the CIC order. The CIC
order is either 1 (integrate and dump filter) or 3. Depending on
configuration, the CIC Filter introduces a gain factor from 20 to
236. The output of the CIC Filter is then rounded and limited to
an 11-bit window between bit positions 21 to 2-9. Values
outside this range saturate to these 11 bits. The
Compensation Filter introduces a final gain factor of 1.0, 0.65,
IOUT9
DATARDY
IOUT0/
QOUT0
LSB
MSB
LSB
DATARDY LEADS 1st BIT
NOTE: Assumes data is being output LSB first.
FIGURE 15. SERIAL TIMING (SIMULTANEOUS I/Q MODE)
IOUT9
DATARDY
MSB
IOUT0
LSB
1
0
MSB
I DATA WORD
Q DATA
WORD
I OUTPUT IDENTIFIED
BY 1 IN LSB OF DATA WORD
DATARDY
LEADS 1st BIT
NOTE: Assumes data is being output MSB first.
DATARDY may be programmed active high or low.
FIGURE 16. SERIAL TIMING (I FOLLOWED BY Q MODE)
AGC GAIN
MANTISSA
1.0 - 1.9375
(0.0625 STEPS)
SYNTHESIZER/
MIXER
CIC
SCALER
G = 0.9990
G = 2-36
EXPONENT CIC BARREL
SHIFTER
20-27
20-263
COMPENSATION
FILTER
GAIN
CIC
FILTER
G = 1.0 - 1.9375*270
G = 1.0, 0.65, 0.77
(BYPASS, x/sin(x), (x/sin(x))3
G = 20- 236
(RN)
LIMIT
GdB = 0dB
GdB = -216.74dB
GdB = GAGC +
GSHIFTER
GdB = 0dB
GdB = 20log[fS/fD]N
8
-28
= 20log[R]N -2
20
20
GdB =
G = 0dB
3 dB
0dB BYPASS -2
-3.74dB
-21
-2.27dB
20
20
-20
2-1
2-1
2-1
2-46
2-9
2-9
RND
RND
N = 1, 3
-21
INPUT
-20
20
2-1
2-1
2-9
2-10
G = -6.02dB RND
BINARY POINT
-2-35
2-46
LIMIT
BIT RANGE OF DATA PATH
FIGURE 17. GAIN DISTRIBUTION AND INTERMEDIATE BIT WEIGHTINGS
12
2-1
2-1
2-9
2-9
RND
OUTPUT
HSP50110
or 0.77 depending on whether the bypass, x/sin(x) or
(x/sin(x))3 configuration is chosen. The Compensation Filter
output is then rounded and limited to a 10-bit output range
corresponding to bit positions 20 to 2-9.
Setting DQT Gains
The AGC and CIC Shifter gains are programmed to maintain
the output signal at a desired level. The gain range required
depends on the signal levels expected at the input and the A/D
backoff required to prevent signal + noise from saturating the
A/D. The signal level at the input is based on the input SNR
which itself is derived from the either output SNR or output
ES/N0. Below are two examples which describe setting the
gains using either an output SNR or ES/N0 specification.
In applications based on the transmission of digital data, it is
useful to specify the DQT’s output in terms of ES/N0. The
following example uses this parameter and the others given in
Table 4 to show how the DQT’s gain settings can be derived.
SETTING
(2)
40 MSPS
(8), (9)
32 KSPS
Input Filter Noise Bandwidth (NBW)
(10)
10MHz
Minimum Output ES/N0
(15)
-3dB
(18), (19)
12dB
Output Signal Magnitude (0 to 1)
(21)
0.5
Number of CIC stages
(11)
3
Compensation Filter
(11)
(x/sin(x))3
Noise Eq. Bandwidth of Comp. Filter
(BN*FSOUT)
N/A
34.18kHz
Input Type (Real/Complex)
(4)
Real
Output Sample Rate (FSOUT) (Notes 1, 2)
Signal + Noise Backoff at A/D Input
The output signal is related to the input signal by:
SOUT = SIN x GMIXER x GSCALER x GAGC x
(EQ. 13)
GSHIFTER x GCIC x GCOMP
(EQ. 14)
Using this equation, limits for GAGC and GSHIFTER can be
determined from the minimum and maximum input signal
conditions as given below (all gains specified in dB):
Min Input Level (Maximum Gain Required):
-6.02dB ≥ -42.96 - 6.02 - 216.74 + GAGC + GSHIFTER+
20 x log((40 x 106/32 x 103)3) - 2.27
(EQ. 15)
-6.02dB ≤ -12 - 6.02 - 216.74 + GAGC + GSHIFTER +
MAIN MENU
ITEM
Input Sample Rate
Thus, the minimum input signal will then be -42.96dB below
full scale (-30.96dB -12dB for A/D backoff). Note: in this
example the symbol rate is assumed to be one half of the
output sample rate (i.e., there are 2 samples per symbol).
Max Input Level (Minimum Input Gain Required)
TABLE 4. EXAMPLE SYSTEM PARAMETERS
PARAMETER
NOTE: 10log10(x) is used because these items are power
related.
20 x log((40 x 106/32 x 103)3) - 2.27
(EQ. 16)
NOTE: 20log10(x) is used because these items are
amplitude related.
Solving the above inequalities for GAGC and GSHIFTER, the
gain range can be expressed as,
45.20dB < (GAGC + GSHIFTER) < 76.16dB.
(EQ. 17)
The shifter gain provides a programmable gain which is a
factor of 2. Since GAGC ≥ 1.0, GSHIFTER is set as close to
the minimum gain requirement as possible:
GSHIFTER = 2N,
(EQ. 18)
where
N = floor(log2(10(GMIN/20)))
= floor(log2(10(45.20/20))) = 7
NOTES:
1. Two samples per symbol assumed.
2. Decimation = 40 MSPS/32 KSPS = 1250.
The limits on the AGC gain can then be determined by
substituting the shifter gain into Equation 18 above. The
resulting limits are given by:
First, the maximum and minimum input signal levels must be
determined. The maximum input signal level is achieved in a
noise free environment where the input signal is attenuated by
12dB as a result of the A/D backoff. The minimum input signal
is determined by converting the minimum output ES/N0
specification into an Input SNR. Using the example parameters
in Table 4 the minimum input SNR is given by:
3.05dB < GAGC <34.02dB.
SNRIN = 10log10(ES/N0) + 10log10(Symbol Rate)
-10log10(NBW)
= -3dB + 10log10(0.5x32 x 103) - 10log10(10 x 106)
= -30.96dB
(EQ. 12)
13
(EQ. 19)
In some applications it is more desirable to specify the DQT
output in terms of SNR. This example, covers derivation of
the gain settings based on an output SNR of 15dB. The
other system parameters are given in Table 4.
As in the previous examples the minimum and maximum
input signal levels must be determined. The minimum input
signal strength is determined by from the minimum output
SNR as given by:
SNRIN = SNROUT - 10log(NBW) + 10log(BN x FSOUT)
= 15 - 10log(10 x 106) + 10log(34.18 x 103)
= -9.66dB
(EQ. 20)
HSP50110
Thus, the minimum input signal will be -21.66dB below full
scale (-9.66 -12 for A/D Backoff). As before the maximum
input signal in the absence of noise is -12dB down due to
A/D backoff.
From Equation 14, the gain relationships for maximum and
minimum input can be written as follows:
Min Input Level
-6.02dB ≥ -21.66 -6.02 - 216.74 + GAGC + GSHIFTER +
20*log((40 x 106/32 x 103)3)-2.27
(EQ. 21)
Max Input Level
-6.02dB ≤ -12 - 6.02 - 216.74 + GAGC + GSHIFTER +
20 x log((40 x 106/32 x 103)3) -2.27
(EQ. 22)
Using the upper and lower limits found above, the gain range
can be expressed as,
45.20dB < GAGC + GSHIFTER < 54.86dB.
(EQ. 23)
Using Equation 2 in the previous example, the shifter gain is
determined to be 27, resulting in an AGC gain range of 3.05dB
< GAGC <12.72dB.
(EQ. 24)
14
Basic Architectural Configurations
Detailed architectural diagrams are presented in Figures 18
through 20 for the basic configurations, Integrate/Dump filtering
with optional compensation, 3rd Order CIC filtering with
optional compensation, and Decimating Filter bypass. Only one
of the data paths is shown since the processing on either the
inphase or quadrature legs is identical. These diagrams are
useful for determining the throughput pipeline delay or the loop
delay of the AGC as all the internal registers are shown.
All registers with the exception of those denoted by daggers ( †)
are enabled every CLK rate to minimize pipeline latency. The
registers marked by daggers are enabled at the output sample
rate as required by the filtering operation performed. The Loop
Filter accumulator in the AGC is enabled once per output
sample, and represents a delay of one output sample. The
accumulators in the CIC filter each represent a delay of one
CLK, but they are enabled for processing once per input
sample. In Interpolated Input Mode the accumulators are
enabled every CLK since the sample rate is determined by the
CLK rate (see Input Controller Section). In Gated Input Mode,
the processing delay of the accumulators is one CLK but they
are only enabled once for each sample gated into the
processing pipeline. As a result, the latency through the
accumulators is 3 CLKs rather than 3 input sample periods
when configured as a 3rd order CIC filter.
HI/LO
SIN/COS
CIC SCALER
VECTOR FROM
2-36
CARRIER NCO
R
E
G
R
E
G
CIC SHIFTER
20-2-63
+
R R R
E E E
G G G
X
LEGEND:
R
E
G
ACCUMULATOR
W/ PROGRAMMABLE
LIMITS
EXPONENT (20 - 27)
LEVEL
DETECT
IIN0-9/
QIN0-9
†
MANTISSA (1.0 - 1.9375)
R R
E E
G G
31
15
COMPLEX
MULTIPLIER
LOOP
GAIN
1ST ORDER CIC (I & D FILTER)
A
C
C
ACC = ACCUMULATOR
LMT = LIMIT
R = DOWN SAMPLER
MUX = MULTIPLEXER
REG = REGISTER
€
† INDICATES ELEMENTS RUNNING
AT THE OUTPUT SAMPLE RATE
LEVEL
DETECT
R
-1
REG
SERIALIZE
1
†
†
FROM
PROGRAMMABLE
DIVIDER
REG
COMPENSATION
FILTER
R
E
G
L
M
T
+
M
U
X
11 TAP
FIR FILTER
L
M
T
M
U
X
R R R R R
E E E E E
G G G G G
IOUT0-9/
QOUT0-9
FIGURE 18. DATA FLOW FOR INTEGRATE/DUMP CONFIGURATION
†
HI/LO
EXPONENT (20 - 27)
R R R
E E E
G G G
X
LEVEL
DETECT
CIC SHIFTER
20-2-63
+
55
A
C
C
A
C
C
COMPLEX
MULTIPLIER
ACC = ACCUMULATOR
LMT = LIMIT
R = DOWN SAMPLER
MUX = MULTIPLEXER
REG = REGISTER
† INDICATES ELEMENTS RUNNING
€
AT THE OUTPUT SAMPLE RATE
LOOP
GAIN
(TOP BITS ALIGNED)
R
E
G
R
E
G
ACCUMULATOR
W/ PROGRAMMABLE
LIMITS
A
C
C
3RD ORDER CIC FILTER
R
-1
FROM
BIT RATE
NCO
3
-3
†
†
†
R
E
G
R
E
G
R
E
G
+
+
REG
COMPENSATION
FILTER
REG
1
SERIALIZE
†
+
L
M
T
M
U
X
15 TAP
FIR FILTER
L
M
T
M
U
X
IOUT0-9/
R R R R R QOUT0-9
E E E E E
G G G G G
FIGURE 19. DATA FLOW FOR 3RD ORDER CIC CONFIGURATION
LEGEND:
MANTISSA (1.0 - 1.9375)
HI/LO
R R
E E
G G
EXPONENT (20 - 27)
SIN/COS
VECTOR FROM
CARRIER NCO
LEVEL
DETECT
IIN0-9/
QIN0-9
R
E
G
R
E
G
X
R R R
E E E
G G G
CIC SCALER
2-36
+
CIC SHIFTER
20-2-63
21
ACCUMULATOR
W/ PROGRAMMABLE
LIMITS
R
E
G
LOOP
GAIN
REG
SERIALIZE
REG
COMPENSATION
FILTER
L
M
T
ACC = ACCUMULATOR
LMT = LIMIT
R = DOWN SAMPLER
MUX = MULTIPLEXER
REG = REGISTER
LEVEL
DETECT
11-15 TAP
FIR FILTER
COMPLEX
MULTIPLIER
FIGURE 20. DATA FLOW WITH CIC STAGE BYPASSED
M
U
X
L
M
T
M
U
X
R R R R R
E E E E E
G G G G G
IOUT0-9/
QOUT0-9
HSP50110
SIN/COS
CIC SCALER
VECTOR FROM
2-36
CARRIER NCO
LEVEL
DETECT
IIN0-9/
QIN0-9 R
E
G
LEGEND:
MANTISSA (1.0 - 1.9375)
R R
E E
G G
HSP50110
Serial Input Interfaces
Frequency control data for the NCOs contained in the
Synthesizer/Mixer and the Re-Sampler are loaded through two
separate serial interfaces. The Carrier Offset Frequency
Register controlling the Synthesizer NCO is loaded via the COF
and COFSYNC pins. The Sample Offset Frequency Register
controlling the Re-Sampler NCO is loaded via the SOF and
SOFSYNC pins.
CLK
COFSYNC/
SOFSYNC
COF/
SOF
MSB
LSB
MSB
OTE: Data must be loaded MSB first.
SHIFT COUNTER VALUE
IGURE 21. SERIAL INPUT TIMING FOR COF AND SOF INPUTS
32 †
30
28
26
24 †
22
20
18
16 †
14
12
10
8†
6
4
2
0
ASSERTION OF
COFSYNC, SOFSYNC
DATA TRANSFERRED
TO HOLDING REGISTER
(8)
(24) (32)
32-bit holding register and the LSBs of the holding register will
be zeroed. See Figure 22 for details. Note: serial data must
be loaded MSB first, and COFSYNC or SOFSYNC should
not be asserted for more than one CLK cycle.
Test Mode
The Test Mode is used to program each of the output pins to
“high” or “low” state via the Microprocessor Interface. If this
mode is enabled, the output pins are individually set or
cleared through the control bits of the Test Register in Table
13. When serial output mode is selected, the Test Register
may be used to set the state of the unused output bits.
Microprocessor Interface
The Microprocessor Interface is used for writing data to the
DQT’s Control Registers and reading the contents of the
AGC Loop accumulator (see AGC Section). The
Microprocessor Interface consists of a set of four 8-bit
holding registers and one 8-bit Address Register. These
registers are accessed via a 3-bit address bus (A0-2) and an
8-bit data bus (C0-7). The address map for these registers is
given in Table 5. The registers are loaded by setting up the
address (A0-2) and data (C0-7) to the rising edge of WR.
TABLE 5. ADDRESS MAP FOR MICROPROCESSOR
INTERFACE
A2-0
REGISTER DESCRIPTION
0
Holding Register 0. Transfers to bits 7-0 of the 32-bit Destination Register. Bit 0 is the LSB of the 32-bit register.
1
TD ††
TD ††
TD ††
Holding Register 1. Transfers to bits 15-8 of a 32-bit Destination Register.
2
Holding Register 2. Transfers to bits 23-16 of a 32-bit Destination Register.
Serial word width can be: 8, 16, 24, 32 bits wide.
3
† TD is determined by the COFSYNC, COFSYNC rate. Note that
TD can be 0, and the fastest rate is with 8-bit word width.
Holding Register 3. Transfers to bits 31-24 of a 32-bit
Destination Register. Bit 31 is the MSB of the 32-bit register.
4
This is the Destination Address Register. On the fourth CLK
following a write to this register, the contents of the Holding
Registers are transferred to the Destination Register. The
lower 4 bits written to this register are decoded into the Destination Register address. The destination address map is
given in Tables 6-15.
(16)
2
6
10 14 18 22 26 30 34 38 42 46 50 54
CLK TIMES
TD ††
FIGURE 22. SERIAL DATA LOAD TO HOLDING REGISTERS
SEQUENCE
The procedure for loading data through these two pin
interfaces is identical. Each serial word has a programmable
word width of either 8, 16, 24, or 32 bits (see Chip
Configuration Register in Table 11). On the rising edge CLK,
data on COF or SOF is clocked into an Input Shift Register.
The beginning of a serial word is designated by asserting
either COFSYNC or SOFSYNC “high” one CLK prior to the
first data bit as shown in Figure 21. The assertion of the
SOFSYNC starts a count down from the programmed word
width. On following CLKs, data is shifted into the register until
the specified number of bits have been input. At this point data
shifting is disabled and the contents of the register are
transferred from the Shift Register to the respective 32-bit
Holding Register. The Shift Register is enabled to accept new
data on the following CLK. If the serial input word is defined to
be less than 32 bits, it will be transferred to the MSBs of the
16
The HSP50110 is configured by loading a series of nine 32-bit
Control Registers via the Microprocessor Interface. A Control
Register is loaded by first writing the four 8-bit Holding
Registers and then writing the destination address to the
Address Register as shown in Figure 23. The Control
Register Address Map and bit definitions are given in Tables
6-15. Data is transferred from the Holding Registers to a
Control Register on the fourth clock following a write to the
Address Register. As a result, the Holding Registers should
not be updated any sooner than 4 CLK’s after an Address
Register write (see Figure 23). NOTE: the unused bits in a
Control Register need not be loaded into the Holding
Register.
HSP50110
Configuration Register can be loaded one CLK after CFLD
has been loaded on the rising edge of CLK.
For added flexibility, the CFLD input provides an alternative
mechanism for transferring data from the Microprocessor
Interfaces’s Holding Registers to the Center Frequency
Register. When CFLD is sampled “high” by the rising edge
of clock, the contents of the Holding Registers are
transferred to the Center Frequency Register as shown in
Figure 23. Using this loading mechanism, an update of the
Center Frequency Register can be synchronized with an
external event. Caution should be taken when using the
CFLD since the Holding Register contents will be
transferred to the Center Frequency Register whenever
CFLD is asserted. NOTE: CFLD should not be asserted
any sooner than 2 CLK’s following the last Holding
Register load. As Shown in Figure 24, the next
The Microprocessor Interface can be used to read the upper
8 bits of the AGC Loop Filter Accumulator. The procedure for
reading the Loop Accumulator consists of first sampling the
loop accumulator by writing 9 to the Destination Address
Register and then reading the loop accumulator value on
C0-7 by asserting RD. The sampled value is enabled for
output on C0-7 by forcing RD “low” no sooner than 6 CLK’s
after the writing the Destination Register as shown in
Figure 25. The 8-bit output corresponds to the 3 exponent
bits and 5 fractional bits to the right of the binary point (see
Figure 3). The 3 exponent bits map to C7-5 with C7 being
the most significant. The fractional bits map to C4-0 in
decreasing significance from C4 to C0.
PROCESSOR
SIGNALS
WR
RD
DON’T CARE
0
A0-2
1
2
3
4
0
1
C0-7
CLK
LOAD
CONFIGURATION
DATA
LOAD ADDRESS OF
TARGET CONTROL
REGISTER AND
WAIT 4 CLKs
1
2
3
4
EARLIEST TIME
ANOTHER LOAD
CAN BEGIN
LOAD NEXT
CONFIGURATION
REGISTER
NOTE: These processor signals are representative. The actual shape of the waveforms will be set by the microprocessor used. Verify that the microprocessor waveforms meet the parameters in the Waveforms Section of this data sheet to ensure proper operation. While the microprocessor
waveforms are not required to be synchronous to CLK, they are shown as synchronous waveforms for clarity in the illustration.
FIGURE 23. CONTROL REGISTER LOADING SEQUENCE
1
1
2
CLK
CLK
WR
WR
2
3
4
5
6
RD
CFLD
A0-2
DON’T CARE
0
1
2
0
3
1
2
3
C0-7
LOAD NEXT
LOAD
CONFIGURATION NEXT CLK CONFIGURATION
REGISTER
FOLLOWING
DATA
CFLD
NOTE: These processor signals are meant to be representative.
The actual shape of the waveforms will be set by the microprocessor
used. Verify that the processor waveforms meet the parameters in
the Waveforms Section of this data sheet to ensure proper operation.
The Processor waveforms are not required to be synchronous to
CLK. They are shown that way to clarify the illustration.
FIGURE 24. CENTER FREQUENCY CONTROL REGISTER
LOADING SEQUENCE USING CF LOAD
17
A0-2
4
C0-7
9
THREE-STATE
LOAD ADDRESS
INPUT BUS
OF TARGET
CONTROL REGISTER
AND WAIT 6 CLK’S
ASSERT RD
TO ENABLE DATA
OUTPUT ON C0-7
NOTE: These processor signals are meant to be representative.
The actual shape of the waveforms will be set by the microprocessor
used. Verify that the processor waveforms meet the parameters in
the Waveforms Section of this data sheet to ensure proper operation.
The Processor waveforms are not required to be synchronous to
CLK. They are shown that way to clarify the illustration.
FIGURE 25. AGC READ SEQUENCE
HSP50110
TABLE 6. CENTER FREQUENCY REGISTER
DESTINATION ADDRESS = 0
BIT
POSITIONS
31-0
FUNCTION
Center Frequency
DESCRIPTION
This register controls the center frequency of the Synthesizer/Mixer NCO. This 32-bit two’s complement
value sets the center frequency as described in the Synthesizer/Mixer Section. Center
F C 32
˙
Center Frequency = CF H = -------2
– COF
H
FS
H
Format: [XXXXXXXX]H
Range: (0000000 - FFFFFFF)H.
TABLE 7. SAMPLER CENTER FREQUENCY REGISTER
DESTINATION ADDRESS = 1
BIT
POSITION
31-0
FUNCTION
DESCRIPTION
This register controls the center frequency of the Re-Sampler NCO. This 32-bit value together with the
setting of a programmable divider set the decimation factor of the CIC Filter (see Re-Sampler and Low
Pass Filter Sections).
F NCO 32
SamplerCenter Frequency = SCF H = ---------------- 2
– SCOF H.
FS
Sampler Center
Frequency
H
Format: [XXXXXXXX]H
Range: (0000000 - FFFFFFF)H.
TABLE 7. INPUT THRESHOLD REGISTER
DESTINATION ADDRESS = 2
BIT
POSITION
7-0
FUNCTION
Input Level Detector
Threshold
DESCRIPTION
This register sets the magnitude threshold for the Input Level Detector (see Input Level Detector Section).
This 8-bit value is a fractional unsigned number whose format is given by:
20. 2-1 2-2 2-3 2-4 2-5 2-6 2-7.
The possible threshold values range from 0 to 1.9961 (00 - FF)H. The magnitude range for complex inputs
is 0.0 - 1.4142 while that for real inputs is 0.0 - 1.0. Threshold values of greater than 1.4142 will never be
exceeded.
31-8
Reserved.
TABLE 8. AGC CONTROL REGISTER
DESTINATION ADDRESS = 3
BIT
POSITION
7-0
FUNCTION
AGC Level Detector
Threshold
DESCRIPTION
Magnitude threshold for the AGC Level Detector (see AGC Section). The magnitude threshold is represented as an 8-bit fractional unsigned value with the following format:
20. 2-1 2-2 2-3 2-4 2-5 2-6 2-7.
The possible threshold values range from 0 to 1.9961. However, the usable range of threshold values
span from 0 to 1.4142, since full scale outputs on both I and Q correspond to a magnitude of
2
I +Q
2
=
2 = 1.4142 . Threshold values of greater than 1.4142 will force the AGC gain to the upper limit.
15-8
Loop Filter Upper Limit
(Maximum Gain)
Upper limit for Loop Filter’s accumulator (see AGC Section). The three most significant bits are the exponent and the five least significant bits represent the mantissa (see Figure 3). The three exponent bits map
to bit positions 15-13 (15 is the MSB) and the five mantissa bits map to bit positions 12-8 (12 is the MSB).
(EEE.MMMMM)2.
23-16
Loop Filter Lower Limit
(Minimum Gain)
Lower limit for Loop Filter’s accumulator (see AGC Section). The format is the same as that for the upper
limit described above. The 3 exponent bits map to bit positions 23-21 (23 is the MSB) and the mantissa
bits map to bit positions 20-16 (20 is the MSB). (EEE.MMMMM)2.
31-24
Programmable Loop
Gain
Programmable part of Loop Gain word (see AGC Section). The Loop Gain value increments or decrements the Loop Filter’s Accumulator at bit positions 2-6 through 2-13 as shown in Figure 3. The 8-bit loop
gain is loaded into bit positions 31-24 (31 is the MSB and maps to the 2-6 position in the Accumulator).
(GGGGGGGG)2.
18
HSP50110
TABLE 9. I/O FORMATTING/CONTROL
DESTINATION ADDRESS = 4
BIT
POSITION
FUNCTION
DESCRIPTION
0
Input Format
0 = Two’s complement input format, 1 = Offset binary input format.
Note: if a real input with offset binary weighting is used, the unused quadrature input pins should be tied
to 1000000000.
1
Input Mode
0 = Input Controller operates in Interpolated Input Mode.
1 = Input Controller operates in Gated Input Mode.
(See Input Controller Section).
2
Serial/Parallel Output
Select
1 = Serial Output, 0 = Parallel Output. (See Output Formatter Section).
Test Enable
0 = Test Mode Disabled, 1 = Test Mode Enabled. (See Test Mode Section).
5-4
3
Serial Output Clock
Select
Bits 5-4
Serial Output Clock Rate
00
CLK (Serial Output Clock Pin = High)
01
Clk/2
10
CLK/4
11
CLK/8
(See Output Formatter Section).
6
Serial Output Mode
1 = I Followed by Q Mode, 0 = Simultaneous I and Q Mode. (See Output Formatter Section)
7
Serial Output Word
Orientation
1 = MSB First, 0 = LSB First.
8
Output Data Format
1 = Offset Binary, 0 = Two’s Complement.
9
DATARDY Polarity
1 = Active Low, 0 = Active High.
This applies to both serial and parallel output modes. (See Output Formatter Section).
10
Output Clock Polarity
1 = High to Low clock transition at midsample.
0 = Low to High clock transition at midsample.
31-11
Reserved.
TABLE 10. DECIMATING FILTER CONFIGURATION REGISTER
DESTINATION ADDRESS = 5
BIT
POSITION
FUNCTION
DESCRIPTION
5-0
CIC Shifter Gain
These 6 bits set the fixed gain of the CIC shifter. The gain factor is of the form, 2N, where N is the valued
stored in this location. A gain range from 20 to 263 is provided. Since the CIC shifter sets the signal level
at the input to the CIC FIlter, care must be taken so that the signal is not shifted outside of the input bit
range of the filter. (See Gain Distribution Section).
17-6
Programmable
Divider
These 12 bits specify the divisor for the programmable divider in the Re-Sampler. The actual divisor is
equal to the 12-bit value +1 for a total range of 1 to 4096. For example, a value of 7 would produce a
sampling rate of 1/8 the CLK or 1/8 the carry-out frequency of the Re-Sampler NCO depending on configuration.
(See Re-Sampler).
SOURCE
PROGRAMMABLE DIVIDER RANGE
18
CLK
1-4096
ReSampler
2-4096
Programmable
Divider Clock Source
1 = Divider clocked at sample rate of data input to the Low Pass Filter.
0 = Divider clocked by Re-Sampler NCO.
(See Re-Sampler).
20-19
CIC Filter
Configuration
0 0 3 stage CIC filter.
0 1 1 stage CIC (Integrate and dump) filter.
1 X bypass CIC.
When a 3 stage CIC filter is chosen, a decimation factor >3 must be used if the Re-Sampler NCO is used
to set the output sampling rate. (See Re-Sampler Section and Low Pass Filtering Section).
22-21
Compensation
Filtering
0 0 x/sinx filtering.
0 1 (x/sinx)3 filtering.
1 X bypass compensation filter.
(See Low Pass Filtering Section).
19
HSP50110
TABLE 10. DECIMATING FILTER CONFIGURATION REGISTER (Continued)
DESTINATION ADDRESS = 5
BIT
POSITION
FUNCTION
31-23
DESCRIPTION
Reserved.
TABLE 11. CHIP CONFIGURATION REGISTER
DESTINATION ADDRESS = 6
BIT
POSITION
FUNCTION
DESCRIPTION
0
HI/LO Output Sense
1 = HI/LO output of 1 means input > threshold.
0 = HI/LO output of 1 means input ≤ threshold.
(See Input Level Detector Section).
1
AGC Disable
1 = AGC disabled, gain forced to 1.0 (0dB), 0 = Normal operation.
(See AGC Section).
2
AGC Level Detector
Sense
1 = Error signal is 1 when output > threshold, -1 otherwise.
0 = Error signal is -1 when output > threshold, 1 otherwise.
Set to 0 for normal operation. (See AGC Section).
4-3
Carrier Offset
Frequency Word Width
0 0 = 8 bits
0 1 = 16 bits
1 0 = 24 bits
1 1 = 32 bits
(See Synthesizer/Mixer Section).
6-5
Sample Rate Offset
Frequency Word Width
0 0 = 8 bits
0 1 = 16 bits
1 0 = 24 bits
1 1 = 32 bits
(See Re-Sampler Section).
7
Carrier Offset
Frequency Enable
1 = Enable Offset Frequency, 0 = Zero Offset Frequency.
(See Synthesizer/Mixer Section).
8
Sample Rate Offset
Frequency Enable
1 = Enable Offset Frequency, 0 = Zero Offset Frequency.
(See Re-Sampler Section).
9
Load Synthesizer NCO
1 = Accumulation enabled.
0 = Feedback in accumulator is zeroed.
(See Synthesizer/Mixer Section) Set to 1 for normal operation.
10
Load Re-Sampler NCO 1 = Accumulation enabled.
0 = Feedback in accumulator is zeroed.
(See Re-Sampler Section) Set to 1 for normal operation.
12-11
Sample Phase Output
Select
Selects 5 of the 8 MSBs of the Re-Sampler NCO’s phase accumulator for output on SPH0-4. (See ReSampler Section).
0 0 Bits 28:24.
0 1 Bits 29:25.
1 0 Bits 30:26.
1 1 Bits 31:27.
13
Sample Phase
Output Control
Selects whether the sample phase output pins and SSTRB update continuously or only when the DATARDY is active. (See Re-Sampler Section).
1 = Continuous Update.
0 = Updated by DATARDY.
14
Clear Accumulators
Writing a 1 to the Clear Accumulator bit forces the contents of all accumulators to 0. Accumulators will
remain at 0 until a 0 is written to this bit. The following accumulators are affected by this bit.
• Carrier NCO Accumulator
• Cascode CIC Filter Accumulator
• AGC Loop Filter Accumulator
• Serial Output Shifter Counter
• Serial Output Clock Logic
• ReSampler NCO Carry Output Programmable Divider
31-15
Reserved.
20
HSP50110
TABLE 12. PHASE OFFSET REGISTER
DESTINATION ADDRESS = 7
BIT
POSITION
7-0
FUNCTION
DESCRIPTION
This 8 bit two’s complement value specifies a carrier phase offset of π(n/128) where n is the two’s complement value. This provides a range of phase offsets from -π to π*(127/128). (See Synthesizer/Mixer
Section).
Phase Offset
31-8
Reserved.
TABLE 13. TEST REGISTER
DESTINATION ADDRESS = 8
BIT
POSITION
FUNCTION
DESCRIPTION
4-0
Force SPH4-0
When Test Mode enabled*, SPH4-0 is forced to the values programmed in these bit locations. Bit position
4 maps to SPH4. (See Test Mode Section).
5
Force SSTRB
When Test Mode enabled*, SSTRB is forced to state of this bit.
6
Force HI/LO
When Test Mode enabled*, HI/LO is forced to state of this bit.
Force IOUT9-0
When Test Mode enabled*, IOUT9-0 if forced to the values programmed in these bit locations. Bit position
16 maps to IOUT9.
17
Force DATARDY
When Test Mode enabled*, DATARDY is forced to state of this bit.
18
Force LOTP
When Test Mode enabled*, LOTP is forced to state of this bit.
Force QOUT9-0
When Test Mode enabled*, QOUT9-0 is forced to the values programmed in these bit locations. Bit position 16 maps to QOUT9.
16-7
28-19
31-29
Reserved.
* Test Mode Enable is Destination Address = 4, bit-3.
TABLE 14. AGC SAMPLE STROBE REGISTER
DESTINATION ADDRESS = 9
BIT
POSITION
7-0
FUNCTION
AGC Read
DESCRIPTION
Writing this address samples the accumulator in the AGC’s Loop Filter. The procedure for reading the
sampled value out of the part on C0-7 is discussed in the Microprocessor Interface Section. (See Microprocessor Interface Section).
References
[1] Hogenauer, Eugene, “An Economical Class of Digital
Filters for Decimation and Interpolation”, IEEE
Transactions on Acoustics, Speech and Signal
Processing, Vol. ASSP-29 No. 2, April 1981.
[2] Samueli, Henry “The Design of Multiplierless FIR filters
for Compensating D/A Converter Frequency Response
Distortion”, IEEE Transaction Circuits and Systems,
Vol. 35, No. 8, August 1988.
21
HSP50110
Absolute Maximum Ratings
Thermal Information (Typical)
Supply Voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . +7.0V
Input, Output Voltage . . . . . . . . . . . . . . . . .GND -0.5V to VCC +0.5V
ESD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Class 3
Thermal Resistance (Typical, Note 3)
θJA (oC/W)
PLCC Package. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23
Maximum Junction Temperature . . . . . . . . . . . . . . . . . . . . . . .150oC
Maximum Storage Temperature. . . . . . . . . . . . . . . . -65oC to 150oC
Maximum Lead Temperature (Soldering 10s) . . . . . . . . . . . . .300oC
(PLCC - Lead Tips Only)
Operating Conditions
Voltage Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . +4.75V to +5.25V
Temperature Range
Commercial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0oC to 70oC
Industrial. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . -40oC to 85oC
Die Characteristics
Gate Count . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38,000 Gates
CAUTION: Stresses above those listed in “Absolute Maximum Ratings” may cause permanent damage to the device. This is a stress only rating and operation of the
device at these or any other conditions above those indicated in the operational sections of this specification is not implied.
NOTE:
3. θJA is measured with the component mounted on an evaluation PC board in free air.
DC Electrical Specifications
PARAMETER
VCC = 5.0V ±5%, TA = 0o to 70oC Commercial, TA = -40o to 85oC Industrial
SYMBOL
TEST CONDITIONS
MIN
MAX
UNITS
Power Supply Current
ICCOP
VCC = Max, CLK = 52.6MHz
Notes 4, 5
-
350
mA
Standby Power Supply Current
ICCSB
VCC = Max, Outputs Not Loaded
-
500
µA
Input Leakage Current
II
VCC = Max, Input = 0V or VCC
-10
10
µA
Output Leakage Current
IO
VCC = Max, Input = 0V or VCC
-10
10
V
Clock Input High
VIHC
VCC = Max, CLK
3.0
-
V
Clock Input Low
VILC
VCC = Min, CLK
-
0.8
V
Logical One Input Voltage
VIH
VCC = Max
2.0
-
V
Logical Zero Input Voltage
VIL
VCC = Min
-
0.8
V
Logical One Output Voltage
VOH
IOH = -400µA, VCC = Min
2.6
-
V
Logical Zero Output Voltage
VOL
IOL = 2mA, VCC = Min
-
0.4
V
Input Capacitance
CIN
CLK = 1MHz
All measurements referenced to GND.
TA = 25oC, Note 6
-
10
pF
-
10
pF
Output Capacitance
COUT
NOTES:
4. Power supply current is proportional to frequency. Typical rating is 7mA/MHz.
5. Output load per test circuit and CL = 40pF.
6. Not tested, but characterized at initial design and at major process/design changes.
AC Electrical Specifications
Note 8, VCC = 5.0V ±5%, TA = 0o to 70oC Commercial, TA = -40o to 85oC Industrial
-52 (52.6MHz)
PARAMETER
SYMBOL
NOTES
MIN
MAX
UNITS
CLK Period
TCP
19
-
ns
CLK High
TCH
7
-
ns
CLK Low
TCL
7
-
ns
Setup Time IIN9-0, QIN9-0, ENI, PH1-0, CFLD, COF, SOF, COFSYNC,
and SOFSYNC to CLK
TDS
7
-
ns
Hold Time IIN9-0, QIN9-0, ENI, PH1-0, CFLD, COF, SOF,
COFSYNC, and SOFSYNC from CLK
TDH
1
-
ns
Setup Time A0-2, C0-7 to Rising Edge of WR
TWS
15
-
ns
Hold Time A0-2, C0-7 from Rising Edge of WR
TWH
0
-
ns
22
HSP50110
AC Electrical Specifications
Note 8, VCC = 5.0V ±5%, TA = 0o to 70oC Commercial, TA = -40o to 85oC Industrial (Continued)
-52 (52.6MHz)
PARAMETER
SYMBOL
MIN
MAX
UNITS
TDO
-
8
ns
WR High
TWRH
16
-
ns
WR Low
TWRL
16
-
ns
RD Low
TRDL
16
-
ns
RD LOW to Data Valid
TRDO
-
15
ns
RD HIGH to Output Disable
TROD
-
8
ns
-
8
ns
CLK to IOUT9-0, QOUT9-0, DATARDY, LOTP, SSTRB, SPH4-0, HI/LO
NOTES
Note 8
Output Enable
TOE
WR to CLK
TWC
Note 9
8
-
ns
Output Disable Time
TOD
Note 8
-
8
ns
Output Rise, Fall Time
TRF
Note 8
-
3
ns
NOTES:
7. AC tests performed with CL = 40pF, IOL = 2mA, and IOH = -400µA. Input reference level for CLK is 2.0V, all other inputs 1.5V.
Test VIH = 3.0V, VIHC = 4.0V, VIL = 0V.
8. Controlled via design or process parameters and not directly tested. Characterized upon initial design and after major process and/or changes.
9. Set time to ensure action initiated by WR or SERCLK will be seen by a particular clock.
AC Test Load Circuit
S1
DUT
CL
†
IOH
±
1.5V
EQUIVALENT CIRCUIT
SWITCH S1 OPEN FOR ICCSB AND ICCOP
†
Test head capacitance.
23
IOL
HSP50110
Waveforms
t CP
t CL
t CH
CLK
t DS
t DH
IIN9-0, QIN9-0,
ENI, PH1-0,
CFLD, COF,
SOF, COFSYNC,
SOFSYNC
t WRL
t WRH
IOUT9-0, QOUT9-0,
DATARDY, LOTP,
SSTRB, SPH4-0,
HI/LO
WR
t WS
t WH
t DO
t WC
C0-7, A0-2
WR
FIGURE 26. TIMING RELATIVE TO WR
FIGURE 27. TIMING RELATIVE TO CLK
OEI
OEQ
1.5V
1.5V
t OD
t OE
t RF
t RF
1.7V
OUTI9-0
OUTQ9-0
2.0V
0.8V
FIGURE 28. OUTPUT RISE AND FALL TIMES
1.3V
FIGURE 29. OUTPUT ENABLE/DISABLE
t RDL
RD
C0-7
t RDO
t ROD
FIGURE 30. TIMING RELATIVE TO READ
24
HSP50110
Plastic Leaded Chip Carrier Packages (PLCC)
0.042 (1.07)
0.048 (1.22)
PIN (1) IDENTIFIER
N84.1.15 (JEDEC MS-018AF ISSUE A)
0.042 (1.07)
0.056 (1.42)
0.004 (0.10)
C
0.025 (0.64)
R
0.045 (1.14)
0.050 (1.27) TP
C
L
D2/E2
C
L
E1 E
D2/E2
VIEW “A”
A1
A
D1
D
0.020 (0.51) MAX
3 PLCS
0.020 (0.51)
MIN
84 LEAD PLASTIC LEADED CHIP CARRIER PACKAGE
INCHES
MILLIMETERS
SYMBOL
MIN
MAX
MIN
MAX
NOTES
A
0.165
0.180
4.20
4.57
-
A1
0.090
0.120
2.29
3.04
-
D
1.185
1.195
30.10
30.35
-
D1
1.150
1.158
29.21
29.41
3
D2
0.541
0.569
13.75
14.45
4, 5
E
1.185
1.195
30.10
30.35
-
E1
1.150
1.158
29.21
29.41
3
E2
0.541
0.569
13.75
14.45
4, 5
N
84
84
6
Rev. 2 11/97
SEATING
-C- PLANE
0.026 (0.66)
0.032 (0.81)
0.045 (1.14)
MIN
0.013 (0.33)
0.021 (0.53)
0.025 (0.64)
MIN
VIEW “A” TYP.
NOTES:
1. Controlling dimension: INCH. Converted millimeter dimensions are
not necessarily exact.
2. Dimensions and tolerancing per ANSI Y14.5M-1982.
3. Dimensions D1 and E1 do not include mold protrusions. Allowable
mold protrusion is 0.010 inch (0.25mm) per side. Dimensions D1
and E1 include mold mismatch and are measured at the extreme
material condition at the body parting line.
4. To be measured at seating plane -C- contact point.
5. Centerline to be determined where center leads exit plastic body.
6. “N” is the number of terminal positions.
All Intersil products are manufactured, assembled and tested utilizing ISO9000 quality systems.
Intersil Corporation’s quality certifications can be viewed at website www.intersil.com/design/quality
Intersil products are sold by description only. Intersil Corporation reserves the right to make changes in circuit design and/or specifications at any time without notice.
Accordingly, the reader is cautioned to verify that data sheets are current before placing orders. Information furnished by Intersil is believed to be accurate and reliable. However, no responsibility is assumed by Intersil or its subsidiaries for its use; nor for any infringements of patents or other rights of third parties which may result from its use.
No license is granted by implication or otherwise under any patent or patent rights of Intersil or its subsidiaries.
For information regarding Intersil Corporation and its products, see web site www.intersil.com
Sales Office Headquarters
NORTH AMERICA
Intersil Corporation
2401 Palm Bay Rd., Mail Stop 53-204
Palm Bay, FL 32905
TEL: (321) 724-7000
FAX: (321) 724-7240
25
EUROPE
Intersil SA
Mercure Center
100, Rue de la Fusee
1130 Brussels, Belgium
TEL: (32) 2.724.2111
FAX: (32) 2.724.22.05
ASIA
Intersil Ltd.
8F-2, 96, Sec. 1, Chien-kuo North,
Taipei, Taiwan 104
Republic of China
TEL: 886-2-2515-8508
FAX: 886-2-2515-8369