INTERSIL HSP50016JC-75

HSP50016
Data Sheet
February 1999
File Number
Digital Down Converter
Features
The Digital Down Converter (DDC) is a single chip
synthesizer, quadrature mixer and lowpass filter. Its input
data is a sampled data stream of up to 16 bits in width and
up to a 75 MSPS data rate. The DDC performs down
conversion, narrowband low pass filtering and decimation to
produce a baseband signal.
• 75 MSPS Input Data Rate
3288.6
• 16-Bit Data Input; Offset Binary or 2’s Complement
Format
• Spurious Free Dynamic Range Through Modulator
>102dB
• Frequency Selectivity: <0.006Hz
The internal synthesizer can produce a variety of signal
formats. They are: CW, frequency hopped, linear FM up
chirp, and linear FM down chirp. The complex result of the
modulation process is lowpass filtered and decimated with
identical real filters in the in-phase (I) and quadrature (Q)
processing chains.
• Identical Lowpass Filters for I and Q
• Passband Ripple: <0.04dB
• Stopband Attenuation: >104dB
• Filter -3dB to -102dB Shape Factor: <1.5
• Decimation Factors from 32 to 131,072
Lowpass filtering is accomplished via a High Decimation
Filter (HDF) followed by a fixed Finite Impulse Response
(FIR) filter. The combined response of the two stage filter
results in a -3dB to -102dB shape factor of better than 1.5.
The stopband attenuation is greater than 106dB. The
composite passband ripple is less than 0.04dB. The
synthesizer and mixer can be bypassed so that the chip
operates as a single narrow band low pass filter.
The chip receives forty bit serial commands as a control
input. This interface is compatible with the serial I/O port
available on most microprocessors.
The output data can be configured in fixed point or single
precision floating point. The fixed point formats are 16,
24, 32, or 38-bit, two’s complement, signed magnitude, or
offset binary.
The circuit provides an IEEE 1149.1 Test Access Port.
• IEEE 1149.1 Test Access Port
• HSP50016-EV Evaluation Board Available
Applications
• Cellular Base Stations
• Smart Antennas
• Channelized Receivers
• Spectrum Analysis
• Related Products: HI5703, HI5746, HI5766 A/Ds
Ordering Information
PART
NUMBER
TEMP. RANGE
(oC)
PKG.
NO.
PACKAGE
HSP50016JC-52
0 to 70
44 Ld PLCC
N44.65
HSP50016JC-75
0 to 70
44 Ld PLCC
N44.65
HSP50016GC-52
0 to 70
48 Ld CPGA
G48.A
Block Diagram
16
HIGH DECIMATION
FILTER
DATA
LOW PASS FIR
FILTER
I
I
HIGH DECIMATION
FILTER
CONTROL
COS
SIN
COMPLEX
SINUSOID
GENERATOR
TEST ACCESS
PORT/CTRL
TEST ACCESS
PORT
3-198
LOW PASS FIR
FILTER
Q
OUTPUT
FORMATTER
CLK
OUTPUT
Q
IQSTRB
CLK
CLK
R
CLK
4R
OR
CLK
2R
IQCLK
CLKSER
CAUTION: These devices are sensitive to electrostatic discharge; follow proper IC Handling Procedures.
http://www.intersil.com or 407-727-9207 | Copyright © Intersil Corporation 1999
HSP50016
Pinouts
48 PIN CPGA
BOTTOM VIEW
48 PIN CPGA
TOP VIEW
1
TDI
TDO
TRST
TCK
TMS
GND
VCC
GND
VCC
CLK
DATA0
DATA1
DATA8
DATA2
DATA9 DATA11
2
3
6
7
8
CS
CDATA
CSTB
RESET
A
I
GND
B
Q
GND
IQSTB
VCC
VCC
CCLK
C
DATA4
VCC
DATA6
DATA15
GND
DATA13
D
VCC
DATA3
DATA5
DATA14
DATA12
GND
E
VCC
GND
DATA2
DATA9
DATA11
VCC
DATA0
DATA1
IQSTRT IQCLK
H
VCC
G
DATA7
DATA10
F
VCC
VCC
GND
VCC
DATA3
DATA4
VCC
DATA6
DATA15
GND
DATA13
Q
GND
IQSTB
VCC
VCC
CCLK
I
IQSTRT
IQCLK
CS
CDATA
CSTB
E
DATA5
DATA14
GND
DATA12
D
C
F
CLK
G
GND
VCC
H
TDI
TDO
DATA8
DATA7
DATA10
GND
VCC
VCC
GND
TRST
TCK
TMS
RESET
B
A
3
6
7
8
VCC
CCLK
CSTB
CDATA
CS
IQSTB
IQSTRT
IQCLK
I
Q
GND
44 LEAD PLCC
TOP VIEW
6 5 4 3 2 1 44 43 42 41 40
VCC
DATA6
DATA5
DATA4
DATA3
VCC
GND
CLK
DATA2
DATA1
DATA0
7
8
9
10
11
12
13
14
15
16
17
39
38
37
36
35
34
33
32
31
30
29
3-199
VCC
GND
RESET
TCK
TMS
VCC
TDO
TRST
TDI
18 19 20 21 22 23 24 25 26 27 28
VCC
2
GND
1
GND
DATA15
DATA14
DATA13
DATA12
GND
DATA11
DATA10
DATA9
DATA8
DATA7
HSP50016
Pin Description
NAME
TYPE
DESCRIPTION
VCC
-
+5V Power.
GND
-
Ground.
DATA0-15
I
Input Data Bus. Selectable between two's complement and offset binary. DATA0 is the LSB.
CLK
I
Clock for input data bus. fS is the frequency of CLK, which is also the input sample rate.
RESET
I
RESET initializes the internal state of the DDC. During RESET, all internal processing stops. RESET
facilitates the synchronization of multiple chips for Auto Three-State operation. If the Force bits in Control
Word 7 are inactive and the IEEE Test Access Port is in an Idle state, RESET causes the IQCLK, IQSTB,
I and Q outputs to go to a high impedance state.
All Control Registers are updated from their respective Control Buffer Registers on the third rising edge
of CLK after the deassertion of RESET. If RESET is deasserted tRS nanoseconds prior to the rising edge
of CLK, the internal reset will deassert synchronously. If tRS is violated, then the circuit contains a synchronizer which will cause reset to be deasserted internally one or more clocks later.
An initial reset is required to guarantee proper operation of the DDC. Active low.
I
O
The I output has three modes: I data; I data followed by Q data; real data.
Q
O
The Q output has two modes: Q data and the carry out of the Phase Adder.
IQCLK
O
IQ Clock: Bit or word clock for the I and Q outputs.
IQSTB
O
IQ Strobe: Beginning or end of word indicator for I and Q.
IQSTRT
I
IQ Start: Initiates output data sequence. Active low.
CDATA
I
Control Data: Port for control data input.
CCLK
I
Control Data Clock: Control data input bit clock.
CSTB
I
Control Data Strobe: Beginning of word indicator for control data.
CS
I
Chip Select: Enables control data loading of DDC. Active low.
TCK
I
Test Clock: Bit Clock for IEEE 1149.1 Data. This signal should be either tied low or pulled high when the
TAP is not used.
TMS
I
Test Port Mode Select: This signal should be either left unconnected or pulled high when the TAP is not
used.
TDI
I
Test Data Input for IEEE Test Port: This signal should be either left unconnected or pulled high when the
TAP is not used.
TDO
O
Test Data Output for IEEE Test Port: This output will be in the high impedance state when the TAP is
not used.
TRST
I
Test Port Reset. Active Low. This signal should be tied low when the TAP is not used.
3-200
DDC Functional Block Diagram
HDF
SECTION
MIXER
SECTION
SCALING
MULTIPLIER
18
17
3-201
†INPUT
FORMAT
17
DATA
RAM
MULTIPLIER/
ACCUMULATOR
+
HDF
SHIFTER
FIR SECTION
18
17
†HDF
SHIFT
CLK
COEFFICIENT
ROM
†HDF
INPUT
REGISTER
COS
17
SIN
SHIFTER
HDF
17
DATA
RAM
SCALE
FACTOR
MULTIPLIER
GAIN
18
LOCAL OSCILLATOR
CLKSER = IQCLK
MULTIPLIER/
ACCUMULATOR
†NUMBER OF OUTPUT BITS
†OUTPUT SENSE
†I FOLLOWED BY Q
†TIME SLOT NUMBER
†IQCLK POLARITY
†IQCLK DUTY CYCLE
†IQCLK DURATION
†IQCLK THREE-STATE CTL
†IQSTRB POLARITY
†IQSTRB LOCATION
FORMATTER
SHIFT
REGISTER
Q
(PARALLEL
TO SERIAL
CONVERTER
AND BUFFER)
IQSTB
†IQSTRB THREE-STATE CTRL
IQSTRT
†I POLARITY AND THREE-STATE CTRL
†Q POLARITY AND THREE-STATE CTRL
†IQ CLK RATE
0
IQCLK
†TEST ENABLE AND CONTROL SIGNALS
CS
1
CSTB
1
TCLK
DECODER
CDATA
2
†CONTROL
2
PARAMETERS
TMS
IEEE 1149.1
TEST ACCESS PORT
CCLK
TDO
TDI
7
7
CLK
RESET
TRST
D
Q
D
Q
CONTROL CONTROL
BUFFERS REGISTERS
† Indicates parameters from control registers.
FIGURE 1. FUNCTIONAL BLOCK DIAGRAM
HSP50016
PHASE
GENERATOR
CLK
4R
22
†WAIT FOR RAM FULL
†OUTPUT FORMAT
†TIME SLOT LENGTH
†SCALING
†MIN PHASE INCR
†DELTA PHASE INCR
†PHASE OFFSET
†MAX PHASE INCR
†MODE
18
17
SIN/COS
GENERATOR
PHASE WORD
18
+
16
I
CLK
R
DECIMATION
COUNTER PRELOAD (DCP)
DATA0-15
SHIFT
REGISTER
HSP50016
Phase Generator Block Diagram
PHASE ACCUMULATOR
1
PHASE REGISTER
0
>
18
R
E
G
MUX
†PHASE OFFSET
>
R
E
G
33
PHASE WORD
(TO THE
SIN/COS
GENERATOR)
18
33
+
+
PHASE ADDER
CARRY OUT
-
†MAXIMUM PHASE
INCREMENT
MUX SELECT 1
32
PHASE INCREMENT ACCUMULATOR
PHASE
INCREMENT
REGISTER
2
>
R
E
G
>
R
E
G
>
0
INCREMENT
1
MUX
32
†MINIMUM PHASE
R
E
G
MUX SELECT 0
INCREMENT
32
ADDER /
SUBTRACTOR
CARRY OUT
>
R
E
G
≤
INITIALIZE
PHASE INCR.
TO MIN
INITIALIZE
PHASE INCR.
TO MAX
>
ADD/SUBTRACT
†DELTA PHASE
MUX SELECT 0
CONTROL
MUX SELECT 1
†MODE
CONTROL
† Indicates parameters set in Control Registers.
FIGURE 2. PHASE GENERATOR BLOCK DIAGRAM
Functional Description
The primary function of the DDC is to extract a narrow
frequency band of interest from a wideband input, convert that
band to baseband and output it in either a quadrature or real
form. This narrow band extraction is accomplished by down
converting and centering the band of interest at DC. The
conversion is done by multiplying the input data with a
quadrature sinusoid. A quadrature lowpass filter is applied to
3-202
the multiplier outputs. Identical real lowpass filters are
provided in the in-phase (I) and quadrature phase (Q)
processing branches. Each filtering chain consists of a
cascaded HDF and FIR filter, which extracts the band of
interest. During filtering, the signal is decimated by a rate
which is proportional to the output bandwidth. The bandwidth
of the resulting signal is the double sided passband width of
the lowpass filters. An Output Formatter manipulates the filter
output to provide the data in a variety of serial data formats.
HSP50016
Local Oscillator
Signal data clocked into the DATA0-15 input of the DDC is
multiplied by a quadrature sinusoid in the Mixer Section (see
Figure 1). The data input to the DDC is a 16-bit real data
stream which is sampled on the rising edges of CLK. It can
be in two's complement or offset binary format.
The input data is passed to a mixer, which is composed of
two real multipliers. One of these multiplies the input data
samples by the in-phase (cosine) component of the
quadrature sinusoid, and the other multiplies the input data
samples by the quadrature (sine) component. The in-phase
and quadrature data paths are designated I and Q
respectively. The sine and cosine are generated in the local
oscillator as shown in Figure 1.
The local oscillator is programmed to produce a quadrature
sinusoid with programmable frequency and phase. The
frequency can be constant (Continuous Wave - CW), linearly
increasing (up chirp), linearly decreasing (down chirp), or
linear up/down chirp. The initial phase of the waveform is set
by the phase offset.
The phase, frequency and chirp limits of the quadrature
sinusoid are controlled by the Phase Generator (Figure 2).
The output of the Phase Generator is an 18-bit phase word
that represents the current phase angle of the complex
sinusoid. The Phase Generator automatically increments the
phase angle by a preprogrammed amount on every rising
edge of CLK. Stepping the output phase from 0 through full
scale (218 - 1) steps the phase angle of the quadrature
sinusoid from 0 to (-2+2-17)π radians. NOTE: The phase is
stepped in a clockwise (decreasing) direction to support
down conversion. The frequency of the complex sinusoid is
determined by the number of clocks needed for the phase to
step though its full range of 2π radians. The required phase
increment for a given local oscillator frequency is calculated by:
minimum phase increment is the phase step taken on every
clock. When the SIN/COS Generator is producing a chirped
sinusoid, the minimum phase increment is the smallest
phase step taken. Maximum phase increment is only used
during Chirped Modes; it is the largest allowable phase
increment. During Chirp Modes, the delta phase increment
is the difference between successive phase increments.
The four phase parameters are stored in their respective
registers in the Phase Generator. The Phase Register stores
the current phase angle. On the first clock following the
deassertion of RESET, the 18 MSBs of the Phase Register
are loaded from the Phase Offset Register. On every rising
edge of CLK thereafter, the output of the Phase Increment
Register is subtracted from the 32 LSBs of the current
phase. The 33-bit difference is stored back in the Phase
Register on the next CLK. The 18 most significant bits of the
Phase Register form the phase word, which is the input to
the SIN/COS Generator.
Figure 3 gives a graphic representation of the phase
parameters for the CW case. To understand their
interrelationships, the phase should be visualized as the
angle of a rotating vector. When the local oscillator in the
DDC is programmed to generate a CW waveform, the
multiplexers are configured so that the Minimum Phase
Increment is stored in the Phase Increment Register; this
value is subtracted from the output of the Phase Register on
every CLK and the difference becomes the new Phase
Register value. The Delta Phase Increment and Maximum
Phase Increment are ignored when generating a CW.
STARTING PHASE
+90o
θINCR
33
Phase Increment = INT [ ( f C ⁄ f S ) 2 ]H
– 33
f C = ( Phase Incr ) f S 2
; 0 < f C < f S /2
θOFFSET
(0)
(EQ. 1)
θINCR
(1)
(2)
where:
±180o
fC is the desired local oscillator frequency
(5)
Phase Increment is the Control Word Value (in Hex)
The phase offset is the initial setting of the phase word going
to the SIN/COS Generator. Subsequent phases of the
sinusoid are calculated relative to this offset. The minimum
phase increment has two mode dependent functions: when
the SIN/COS Generator is forming a CW waveform, the
3-203
0o
(4)
fS is the input sampling frequency
There are five parameters which control the Phase Generator:
Phase offset, minimum phase increment, maximum phase
increment, delta phase increment and Mode Control. These
values are programmed via Control Words 2, 3, and 4. Mode
Control is used to select the function of the other parameters.
θINCR
(3)
θINCR
θINCR
-90o
FIGURE 3. PHASE WORD PARAMETERS FOR CW CASE
In Up Chirp Mode the local oscillator generates a signal
with a linearly increasing frequency (Figure 4A). The Phase
Increment Register is initially loaded with the minimum
Phase Increment value; on every clock, the contents of the
Phase Increment Register is subtracted from the current
output of the Phase Register. Simultaneously, the Delta
HSP50016
Phase Increment Register is added to the 24 LSBs of the
output of the Phase Increment Register. On the next CLK,
that sum is stored back in the Phase Increment Register,
the new phase is stored in the Phase Register and the
process is repeated. The phase increment is allowed to
grow until the next phase increment would equal or exceed
the maximum phase increment value. When this happens,
the Phase Increment Register is reset to the minimum
phase increment and the cycle starts over again.
NOTE: The phase increment is never equal to the
maximum phase increment, since the Phase Increment
Register is reloaded if the next phase increment value
would be greater than the maximum phase increment.
From the time the Phase Generator starts at the minimum
phase increment until it reaches the maximum phase
increment, the phase word on clock n is given by:
Phase Word = Phase Offset - [Minimum Phase Increment (EQ. 2)
+ n (Delta Phase Increment) ]
An example of the outputs of the Phase Increment Register,
Phase Register, and the I output of the SIN/COS Generator
are shown in Figure 4B.
In Down Chirp Mode the local oscillator generates a signal
with a linearly decreasing frequency (Figure 5A). The
maximum phase increment is loaded into the Phase
Increment Register and the phase offset value goes into
the Phase Register. The delta phase increment is
subtracted from the 24 LSBs of the phase increment to
form a new phase increment at each clock. The phase
increment is allowed to diminish until it reaches the
minimum phase increment value, then it is reset to the
maximum phase increment value and the cycle is repeated.
Note that the value of the phase increment can be equal to,
but never less than the minimum phase increment, since
the Phase Increment Register is reloaded if the next phase
increment value would be less than the minimum phase
increment. This feature protects the DDC from exceeding
the Nyquist frequency. In this case, from the time the Phase
Generator starts at the maximum phase increment until it
reaches the minimum phase increment, the phase word on
clock n is given by:
Phase Word = Phase Offset -[Minimum Phase Increment (EQ. 3)
– n (Delta Phase Increment)]
See Figure 5B for a graphical representation of this process.
PHASE INCREMENT
STARTING PHASE
MAXIMUM
+90o
TIME
MINIMUM
θINCR
θINCR + 2θ∆
θOFFSET
(0)
(8)
(1)
PHASE WORD
θINCR + θ∆
(2)
±180o
0o
(7)
θINCR + θ∆
(3)
θINCR + 2θ∆
PHASE
OFFSET
TIME
(6)
(5)
θINCR
(4)
θINCR + 3θ∆
COSINE OUTPUT OF SIN/COS GENERATOR
θINCR + 4θ∆
IF
-90o
TIME
θINCR + 5θ∆ > θMAX INCR
THEN
START NEW RAMP
FIGURE 4A. PHASE WORD DURING UP CHIRP
3-204
FIGURE 4B. UP CHIRP
HSP50016
PHASE INCREMENT
STARTING PHASE
θMAX INCR
-4θ∆
θMAX INCR
-3θ∆
±180o
TIME
MINIMUM
+90o
PHASE WORD
θOFFSET
(0)
(5)
MAXIMUM
PHASE
OFFSET
(4)
TIME
0o
(3)
θMAX INCR
(1)
(2)
COSINE OUTPUT OF SIN/COS GENERATOR
θMAX INCR
-2θ∆
θMAX INCR
- θ∆
TIME
-90o
FIGURE 5A. PHASE WORD DURING DOWN CHIRP
FIGURE 5B. DOWN CHIRP
PHASE INCREMENT
MAXIMUM
TIME
MINIMUM
PHASE WORD
PHASE
OFFSET
TIME
COSINE OUTPUT OF SIN/COS GENERATOR
TIME
FIGURE 6. UP/DOWN CHIRP
In Up/down Chirp Mode, the phase accumulator is set to the
phase offset value and the minimum phase increment is
loaded into the Phase Increment Register. The delta phase
increment is added to the 24 LSBs of the Phase Increment
Register to form a new phase increment at each clock. The
phase increment is allowed to grow until it nears the
maximum phase increment value (as defined in the up chirp
description). The delta phase increment value is then
subtracted from the least significant bits of the Phase
Increment Register to form a new phase increment at each
clock. The phase increment is allowed to diminish until it
3-205
reaches the minimum phase increment value (as defined in
the down chirp description). The Phase Increment Register
is then reloaded with the minimum phase increment, and the
up/down cycle begins again. See Figure 6 for a graphical
representation of this process.
The minimum and maximum phase increments have
allowable values from 0 to 232-1. This corresponds to the
phase increment:
0 < Phase Increment < π ( 1 – 2
– 32
) radians
(EQ. 4)
HSP50016
The Delta Phase Increment parameter can take on values
from 0 to 224 - 1 which corresponds to the Delta Phase
Increment:
0 < DeltaPhase Increment < π ( 2
–8
–2
– 32
) radians
(EQ. 5)
The output of the phase accumulator forms the input to the
SIN/COS Generator which in turn produces a quadrature
vector which rotates clockwise: the outputs are cos(ωn) and
-sin(ωn). The outputs of the SIN/COS Generator are two's
complement values which are scaled to prevent overflow in
subsequent operations in the DDC under normal operation.
The scale factor has a negligible effect on the end to end
DDC gain.
The frequency resolution of the DDC = (frequency of CLK)/
(Number of Phase Register bits). At the maximum clock rate,
this results in a frequency selectivity of 75MHz/233 = 0.009Hz.
The 18-bit phase word yields a phase noise figure of greater
than 102dB.
the input data to the HDF for the maximum dynamic range
while avoiding overflow errors. The shift factor is
programmed into the Shift field of Control Word 4. The value
in this field is calculated by the equation:
Shift = 75 – Ceiling ( 5 log 2 ( R ) )
(EQ. 7)
where R is the HDF decimation factor and Ceiling(X)
denotes the ceiling function of X; i.e., the result is X if X is an
integer, otherwise the result is the next higher integer.
During RESET, the HDF is initialized and will not output any
information until it is filled with new data.
NOTE: The output rate of the HDF is CLK divided by the
HDF decimation factor (CLK/R). The HDF decimation
counter preload (DCP) is programmed in Control Word 5,
bits 21-35 and has the value: DCP = R -I, where R is the
HDF decimation factor.
0
Mixer
-20
-40
GAIN (dB)
The Mixer performs quadrature modulation by multiplying
the output of the SIN/COS Generator by the input data. The
outputs of the I and Q multipliers are symmetrically rounded
to 17 bits to preserve the 102dB spurious free dynamic
range (SFDR). The result of the quadrature modulation
process is passed to the High Decimating Filter (HDF)
Section.
-80
High Decimation Filter
-100
The High Decimation Filter (HDF) Section is comprised of
two real HDF filters, one processing the I data branch and
one processing the Q data branch. Each branch has the
lowpass response shown in Figure 7. The normalized HDF
frequency impulse response is given by the equation:
Sin ( πF S ) 5 I 5
---H ( f ) = -------------------------------R
Sin ( πF S /R )
-60
(EQ. 6)
-120
fS
8R
fS
4R
3fS
fS
5fS
3fS
7fS
fS
8R
2R
8R
4R
8R
R
FREQUENCY (Hz)
FIGURE 7A. FREQUENCY RESPONSE OF HIGH DECIMATION
FILTER FROM DC TO FIRST NULL
(FOR R = 16)
Gain (dB) = 20log [H(f)]
where FS is the input sampling rate; R is the decimation
(rate change) factor.
0
Figure 7A shows this equation plotted from DC to the first
null, while Figure 7B shows the equation plotted from DC
response to fS.
The data path through the HDF was designed to ensure a
true 16-bit noise floor (approximately 98dB) at the output of
the DDC. The structure of the HDF filter used in the DDC is a
five stage decimation filter. The width of each successive
stage decreases such that the LSBs are lost due to
truncation [1]. As a result, the data must be processed in the
MSBs of the filter so that the noise due to truncation is below
the required noise floor. Thus, the input data of the HDF
must be shifted so that its output data fills the HDF output
word. The shift is a function of the desired HDF decimation
rate R and the number of HDF filter stages (which is fixed at
5). The shift is performed by the Data Shifter, which positions
3-206
GAIN (dB)
NOTE: The HDF is a true FIR filter; i.e., the phase is linear.
-20
-40
-60
-80
-100
-120
fS 2fS
R
R
4fS
R
6fS
fS
10fS
R
2
R
FREQUENCY (Hz)
12fS
14fS
R
R
FIGURE 7B. DDC HC FREQUENCY RESPONSE
(FOR R = 16)
Gain (dB) = 20log[H(f)]
fS
HSP50016
Scaling Multipliers
The output of each HDF is passed to a Scaling Multiplier.
The Scaling Multipliers are used to compensate for the HDF
gain, which is between 1 (inclusive) and 0.5 (non-inclusive),
or (0.5, 1.0). The gain through the HDF is dependent on the
decimation factor: when the decimation is an even power of
two, the HDF gain is equal to 1; otherwise, the gain must be
compensated for in the Scaling Multiplier. The HDF gain is
given by the equation:
5
HDF Gain = R /2 CEILING ( 5 log 2 ( R ) )
(EQ. 8)
where R is the HDF decimation factor. The compensating
Scale Factor, which is input to both Scaling Multipliers, is
given by the equation:
5
Scale Factor = 2 CEILING ( 5 log 2 ( R ) ) ⁄ R
(EQ. 9)
where R is the HDF decimation factor.
MAGNITUDE (dB)
a0(-20). a1(2-1) a2(2-2)... a17(2-17)
Scale factor:
a0(20). a1(2-1) a2(2-2)... a15(2-15)
Output:
a0(-20). a1(2-1) a2(2-2)... a16(2-16)
The Scaling Multiplier output is passed to the FIR Filter,
which performs aliasing attenuation, passband roll off
compensation and transition band shaping. The FIR Filter
Section is functionally two identical 121 tap lowpass FIR
filters, one each for the I and Q channel. The two filters are
each implemented as sum of products, each with a single
multiplier, with the coefficients stored in ROM. The filters'
passbands are precompensated to be the inverse of the
response of the HDF. The frequency responses of the HDF,
FIR, and Composite HDF/FIR filters are shown in Figure 8.
The composite passband of the HDF and FIR filter
frequency response is shown in Figure 9. The FIR
coefficients are scaled so that the maximum gain of the
composite filter is less than or equal to 0dB. The composite
passband ripple is less than 0.04dB.
HDF
0
-50
-100
-150
(R = 16)
fS
fS
3fS
4
2
4
0
fS
0
-50
-100
-150
fS
HDF
-20
FIR
4
MAGNITUDE (dB)
Input from HDF:
FIR Filter
fS
3fS
2
4
fS
MAGNITUDE (dB)
MAGNITUDE (dB)
NOTE: The Scale Factor falls in the interval [1, 2). The
output of the scaling multiplier is symmetrically rounded to
17 bits.
The binary formats of the inputs and outputs of the scaling
multiplier are as follows:
-40
-60
-80
COMPOSITE
COMPOSITE
HDF/FIR
-100
0
FIR
-50
-100
-120
-150
fS
fS
3fS
4
2
SAMPLE TIMES
4
FIGURE 8A. DDC HDF, FIR, AND COMPOSITE FILTER
RESPONSE (FOR R = 16)
3-207
fS
2
fS
4 6
fS
128R 64R
8 10 12 14 16 18 20 22 24 26 28 30 32
fS
fS
3fS
fS
32R
16R
32R
SAMPLE TIMES
FIGURE 8B. DDC FILTER RESPONSES (FOR R = 16)
8R
HSP50016
0.5
0.4
MAGNITUDE (dB)
0.3
0.2
0.1
FIR
HDF
COMPOSITE
HDF/FIR
0
-0.1
-0.2
-0.3
-0.4
-0.5
fS
64R
fS
32R
FREQUENCY (Hz)
3fS
64R
fS
16R
FIGURE 9. FIR COMPENSATION FOR HDF ROLL OFF (FOR R = 16)
The coefficients of the filter are quantized to 22 bits to
preserve greater than 106dB of stopband attenuation. The
sum of products of each filter output calculation is a 38-bit
number with 37 fractional bits.
When a quadrature output is selected, the outputs of the
FIR filters are decimated by a factor of four. When real
output is selected, only the I output is active. The output is
decimated by two in this case. When Filter Only Mode is
selected, only the I filter path is active and its output is
decimated by four.
The composite filter bandwidths are a function of the HDF
decimation rate and the FIR Filter shape. The double sided
bandwidths are specified by Equations 10 and 11.
16 < R < 16384
(EQ. 10)
– 102dB BW DS = 0.2002F S ⁄ R 16 < R < 16384
(EQ. 11)
– 3dB BW DS = 0.1375F S ⁄ R
TABLE 1. FIR OUTPUT RATE AND DECIMATION
OUTPUT MODE
FIR OUTPUT RATE
FIR DECIMATION
Real
CLK/2R
2
Complex
CLK/4R
4
Filter Only
CLK/4R
4
R - HDF Decimation Factor
Output Formatter
The circuit has two serial data outputs, I and Q. The timing
of the output bits is referenced to IQCLK and IQSTB. There
are several modes of operation for the data and control line
interface, all of which were designed to be compatible with
common microprocessors. These interface modes are
selected by loading the appropriate control words (see
Tables 3 through 10, with Table 9 containing most interface
parameters).
where FS = CLK; R = HDF Decimation Factor.
Quadrature data output can occur in one of two ways:
simultaneously or sequentially. The simultaneous method
clocks out the I and Q data on their respective serial output
pins. The I followed by Q method clocks I and Q out
sequentially on the I output pin: the entire I word is serially
clocked out first, then the entire Q word. In real data Output
Mode, the Formatter converts the quadrature data to real
and clocks it out serially on the I output pin. In all modes, the
I and Q outputs return to the zero state after the last bit is
transmitted.
NOTE: The output data rate of the FIR is the HDF output
rate divided by either 2 or 4, depending on mode. Recall
the HDF output rate is CLK/R. (See Table 1.)
When the “I followed by Q” signal (CW6, bit 35) is low,
I data will appear on the I output and Q data will appear on
the Q output.
where FS = CLK; R = HDF Decimation Factor.
The single sided bandwidths are specified in Equations 12
and 13.
– 3dB BW SS = 0.06875F S ⁄ R
(EQ. 12)
– 102dB BW SS = 0.100097F S ⁄ R
(EQ. 13)
When the “I followed by Q” signal (CW6, bit 35) is asserted,
the Q output is inactive and I data, followed by Q data
appear on the I output. When in this state, and both the “Test
Enable” signal (CW1, Bit 3), and “Q Strobe on Rollover”
signal (CW7, Bit 10) signal are asserted, the Phase
3-208
HSP50016
Generator Carry Out will appear on the output. Control Word
5 contains fields to set the number of output bits transmitted
to the arithmetic representation and interface control of the
serial output data. Control Word 4, Bits 31-32, allow
selection of baseband centered quadrature on baseband
offset quadrature complex outputs. Control Word 4, Bit 0,
allows selection of spectral inversion. In addition, the output
drivers for I, Q, IQCLK and IQSTB can be individually
enabled or placed in a high impedance state using Control
Word 6, Bits 20-28. These options are explained below.
and trailing zero bits occur before bit 0 and after bit N,
respectively.
CLK
IQCLK Rate = -------------------------------------------------- – 1
IQCLK Frequency
I
The output has a saturation option to prevent possible
overflow due to a step input at power up. When Overflow
Protection is enabled, the output is forced to be either the
most positive or most negative number. Saturation is
available in all four fixed point output options, and is set via
Control Word 7, Bit 0.
Data can also be output in single precision floating point
format (see Table 2). For all output data formats, the
internal calculations are performed in exact two’s
complement integer arithmetic and the resulting data is
converted in the Output Formatter.
TABLE 2. FLOATING POINT FORMAT
SIGN
EXPONENT
-20
27 to 20
MANTISSA
Implied 1
0.2-1 to 2-23
The I and Q pins can be programmed for either
simultaneous or I followed by Q output. In simultaneous
mode, the I and Q data appear on the I and Q pins,
respectively. Each data sample is preceded by a leading
zero bit, followed by the output data, followed by a trailing
zero bit. In I followed by Q Mode, the output data appears
on the I pin, and consists of a leading zero bit, then the I
data, a trailing zero, a leading zero, the Q data, and finally
a trailing zero bit. In Figures 10 through 12, the leading
3-209
N-1
1
0
STOP
BIT
When the “Output Spectrum” signal (CW4, bits 31-32) is set
to “01”, then the real output data appears on the I output and
the Q output in the I/Q separate mode. When in I Mode
followed by Q Mode, the Q slot is also real data since the
real mode outputs at twice the rate of the complex mode
(CW4, bits 31-32 = 00).
When set for fixed point output, the output data can be in
two's complement, offset binary or signed magnitude form.
Data is converted to offset binary by complementing the most
significant bit of a two's complement number. The length of
the output data word can be 16, 24, 32 or 38-bits. The first
three options are symmetrically rounded to the LSB of the
output data; the fourth option represents the full 38-bit
width of the accumulator and so represents exact
arithmetic.
N
(EQ. 14)
START
BIT
N
Q
N-1
STOP
BIT
1
0
START
BIT
N = 16, 24, 32 OR 38
A. SIMULTANEOUS OUTPUT MODE
QN
I
STOP
BIT
QN-1
Q1
Q0
QN
QN-1
START STOP
BIT
BIT
Q1
Q0
START
BIT
Q
B. I FOLLOWED BY Q OUTPUT MODE
FIGURE 10. DATA OUTPUT MODES
IQCLK is used to delineate the bit or word timing of the I and
Q outputs. There are several options on the configuration of
IQCLK, which are controlled with Control Word 6 (see
Table 8). The frequency of IQCLK is programmed to be a
fraction of the CLK frequency, from (CLK rate)/2 to (CLK
rate)/8192 (see Equation 14). If IQCLK Rate = 0, then
IQCLK remains in its inactive state and the output bits
change on the rising edges of CLK.
IQCLK can be programmed to be active continuously, or only
during I or Q data output via the IQCLK duration bit. Using the
IQCLK Duty Cycle bit, IQCLK is selectable as either 50% duty
cycle or to be high for one period of CLK. In addition, the
Formatter can be set so that the data bits are clocked on
either the positive or negative edges of IQCLK with the IQCLK
Polarity bit. Figure 11 shows the various modes of operation
with IQCLK Polarity programmed for active high operation.
Control Word 6 also configures IQSTB, as shown in
Figure 12. When programmed for Active Prior to Data Word,
IQSTB is high for one period of IQCLK and terminates
simultaneously with the beginning of the first data bit;
otherwise it goes active with the beginning of the first bit and
inactive with the end of the last bit. IQSTB can be
programmed to be either active high or low.
HSP50016
CCLK
CLK
IQCLK
CDATA
BIT 0
I OR Q
BIT 1
BIT N-1
BIT39
BIT 36 BIT35
BIT 0
BIT N
A. IQCLK DUTY CYCLE: ACTIVE TIME = CLK PERIOD
(IQCLK POLARITY = 0)
CSTB
CS
CLK
FIGURE 14. CONTROL WORD TIMING DIAGRAM
IQCLK
BIT 0
I OR Q
BIT 1
BIT N-1
BIT N
B. IQCLK DUTY CYCLE: 50%
IQCLK
I OR Q
BIT 0
BIT 1
BIT N-1
BIT N
C. IQCLK DURATION: CONTINUOUS
IQCLK
I OR Q
BIT 0
BIT 1
BIT N-1
BIT N
D. IQCLK DURATION: ACTIVE DURING I OR Q ONLY
FIGURE 11. TIMING FOR CLK, IQCLK, IQSTB, I AND Q
IQCLK
BIT 0
I OR Q
BIT 1
BIT N-1
BIT N
I/QSTB
A. IQSTB ACTIVE PRIOR TO DATA WORD
Data can be read out of the DDC on request through the use
of the IQSTRT pin. After passing through the Output
Formatter, the I and Q data are stored in output buffers,
which are updated at the end of the FIR Filter processing
cycle. The IQSTRT and IQSTB lines form a two line
handshake as shown in Figure 13. IQSTRT initiates the
request. If the buffer has data in it, the DDC will begin an
output data sequence on the next edge of IQCLK. The DDC
will then put out one bit per IQCLK until the output cycle is
complete. In I followed by Q Mode, one IQSTRT will initiate
an I output word followed by a Q output word. In real data
Output Mode, one IQSTRT will initiate two samples of real
data on the I pin.
To avoid the generation of multiple read cycles, IQSTRT
must go inactive within 10 cycles of IQCLK after the initiation
of IQSTB. The DDC will not update the output buffer again
until the current output cycle has completed. When IQSTRT
is used in this handshake mode, it must consist of pulses
that satisfy the set up and hold requirements listed in the AC
Timing Specifications and the pulses must occur at a rate of
at least CLK/(HDF Decimation Factor x 4 -1). This mode of
operation requires the Time Slot Number in Control Word 6
to be 0.
NOTE: When handshake mode is not used, IQSTRT should
be at a logic low.
IQCLK
BIT 0
I OR Q
BIT 1
BIT N-1
BIT N
IQSTB
B. IQSTB ACTIVE DURING DATA WORD
FIGURE 12. IQSTB TIMING
IQCLK
I or Q
LEADING
0
BIT 0
BIT N
TRAILING
0
Auto Three-State Mode for IQCLK, IQSTB, I and Q allows
multiple chips to operate using common data and output
control lines. Each chip is assigned a Time Slot Number on
the bus to use for outputting its data. All outputs
programmed for Auto Three-State Mode are active during
their time slot and are in a high impedance state at all other
times. A time slot starts one CLK period prior to the
beginning of the first bit of I or Q and ends (Time Slot
Length) CLK periods afterwards. Assignment of a time slot is
with reference to the deassertion of RESET. The minimum
possible Time Slot Length for a given application is:
Length MIN = [ ( Numberof Output Bits + 2 ) × Mode ] + 1 ; or
(EQ. 15)
IQSTRT
IQSTB
where Mode = 2 if the DDC is in either Real Output or I
followed By Q Mode; else Mode = 1.
FIGURE 13. REQUESTED DATA OUTPUT TIMING
3-210
HSP50016
Note that Equation 15 is useful in all modes for calculating
the number of IQCLKs necessary to complete one output
data cycle. For a given decimation rate and output word
length, the maximum value in the IQCLK Rate field is:
(R) × 4
IQCLKRate MAX = Floor ------------------------------- – 1;
Length MIN
(EQ. 16)
where Floor(X) represents the integer part of X, R is the
HDF decimation factor, 4 is the FIR decimation factor.
Example Clock Calculations
Clarification of the use of Equations 14-16, the calculation of
the HDF and FIR clocks and the calculation of the IQCLK is
best done by example:
The sample clock, CLK, is 10MHz . . . . . . . . CLK = 10MHz
The HDF Decimation Factor, R, is 100 (which makes the
decimation counter preload = 99). . . . . . . . . . . . . . . R = 100
The Output Mode is I followed by Q . . . . . . . . . . . . Mode = 2
Complex output . . . . . . . . . . . . . . . . . . . FIR Decimation = 4
The desired number of output bits is 32.
1. We begin by identifying the HDF Input Rate:
HDF Input Rate = CLK = 10MHz . . . . . . . CLK = 10MHz
2. Next we calculate the HDF Output Rate:
HDF Output Rate = CLK/R = 10MHz/(100) = 100kHz
. . . . . . . . . . . . . . . . . . . . . . .HDF Output Rate = 100kHz
3. Next we calculate the FIR output Rate:
FIR Output Rate = CLK/4R = 25kHz.
. . . . . . . . . . . . . . . . . . . . . . . . FIR Output Rate = 25kHz
4. Next we calculate the minimum time slot length:
Equation 15:
LengthMIN = [(Number of Output Bits + 2) x Mode] +1
where the number of output bits = 32 and the Mode is 2
because of the I followed by Q output selection.
LengthMIN = [(32 + 2) x 2] +1 = 69 IQCLKs
. . . . . . . . . . . . . . . . . . . . . . . . . LengthMIN = 69 IQCLKs
5. Next we calculate the IQCLK frequency:
IQCLK frequency = [( FS )(LengthMIN)/(R)(4)] - 1
IQCLK frequency = [(10MHz)(69)/(100)(4)] - 1 = 1.725MHz
The IQCLK frequency can be no slower than 1.725MHz if
all of the bits are to be output of the DDC in a time slot.
. . . . . . . . . . . . . Slowest Serial Output Rate = 1.725MHz
6. The Programmed value for the maximum IQCLK Rate,
from Equation 16, is:
IQCLKRATEMAX = Floor[(R) x 4/ LengthMIN] -1
IQCLKRATEMAX = Floor[(100 x 4)/69] - 1 = 4
The IQCLKRATE can be not greater than 4 if all of the bits
are to be output of the DDC in a time slot.
. . . . . . . . . . . Control Word Value for IQCLK Ratemax =
[00004]H; 0 0000 0000 0100LSB
7. Let’s sanity check with Equation 14.
IQCLK Rate = [(CLK/IQCLKfreq)-1] = [10E6/1.725E6] -1 = 4.
This checks!
3-211
Control Word Input
The DDC has eight 40-bit control words which are loaded
through the four pin control interface. The format and timing
of this interface is compatible with the serial interface timing
of most common DSP microprocessors (see Figure 14). The
words are shifted MSB first, where bit 39 of the control word
is the MSB. Bits 39 through 37 are the control word address,
i.e., the target control buffer. CS must go low before bit 35 is
clocked in. All 40 bits of the control word must be loaded.
The formats of the control words are shown in Tables 3
through 10.
The control words are double buffered: each control word is
initially loaded into one of eight control buffers for
subsequent down loading into the corresponding Control
Register. The internal circuitry of the DDC uses the Control
Registers to regulate its operation. Control buffers can be
downloaded in one of two ways. Loading a Buffer Register
with bit 36 = 1 causes all Control Registers to be updated
from their respective control buffers when the current word
is finished loading. If bit 36 = 0, then only that control buffer
is updated and the operation of the DDC is not affected. All
Control Registers are updated from their respective buffers
on the third rising edge of CLK following the deassertion of
RESET. NOTE: Control Word 0 is unique in that it is
only used to update the seven Control Registers, and it
is recognized by the DDC regardless of the state of CS.
In systems with multiple DDCs, this allows the user to
update the configuration of all chips simultaneously without
using RESET.
To ensure that the control information is properly loaded, the
frequency of CLK must be greater than the frequency of
CCLK. In addition, RESET must remain inactive during the
loading of a control word.
HSP50016
TABLE 3. DESTINATION ADDRESS = 0
BIT
POSITION
FUNCTION
DESCRIPTION
39-37
Address
000 = Control Word 0
36
Update
0 = Update Only This Control Register
1 = Update All Control Registers
35-32
Reserved
All Zeroes
TABLE 4. PHASE GENERATOR/TEST ENABLE/OUTPUT REGISTER
DESTINATION ADDRESS = 1
BIT
POSITION
FUNCTION
DESCRIPTION
39-37
Address
001 = Control Word 1
36
Update
0 = Update Only This Control Register
1 = Update All Control Registers
35-4
Minimum Phase
Increment
Bits 35-4 = 231...20. Range: 0 < Minimum Phase Increment < π (1-2-32) radians.
In the CW mode this is the phase increment of the NCO which is added to the NCO intitial phase offset
state. The desired Sin/Cos generator (local oscillator) frequency is set by the equation:
fc = (phase increment)fs2-33;
where fc is the desired local oscillator frequency, fs is the input sampling frequency, and phase increment
is the control word value in hexidecimal.
To calculate the value to be programmed into this field, use this equation:
phase increment = INT[fc / fs)233]hex
Some examples of phase increments and local oscillator frequencies:
00000000h: fc = zero frequency
00000001h: fc = fs /233 - lowest frequency (75MHz x 2-33 = 8.73mHz)
10000000h: fc = fs /32
20000000h: fc = fs /16
40000000h: fc = fs /8
80000000h: fc = fs /4
ffffffffh: fc = (0.49999)fs - highest frequency (75MHz x 2-33 = 37.49MHz)
In the CHIRP modes, this is the smallest allowable phase increment.
In the Filter Only mode, this parameter should be set to 0.
3
Test Enable
0 = Test Features Disabled
1 = Test Features Enabled
2-0
Phase Generator Mode 000 = Filter Only
001 = Normal Mode (CW)
010 = Reserved
011 = Up Chirp
100 = Reserved
101 = Down Chirp
110 = Reserved
111 = Up/Down Chirp
Note that the lsb sets the gain through the DDC as follows:
0 = Gain is1
1 = Gain is 2
3-212
HSP50016
TABLE 5. PHASE GENERATOR REGISTER
DESTINATION ADDRESS = 2
BIT
POSITION
FUNCTION
DESCRIPTION
39-37
Address
010 = Control Word 2
36
Update
0 = Update Only This Control Register
1 = Update All Control Registers
35-32
Reserved
All Zeroes
31-0
Maximum Phase
Increment
Bits 31-0 = 231... 20. Range: is 0 < Maximum Phase Increment < π(1-2-32) radians.
This parameter is only used in the CHIRP modes, and this is the largest allowable phase increment. Set
to 0 in the Filter Only and CW modes.
TABLE 6. PHASE GENERATOR/OUTPUT TIME SLOT REGISTER
DESTINATION ADDRESS = 3
BIT
POSITION
FUNCTION
DESCRIPTION
39-37
Address
011 = Control Word 3
36
Update
0 = Update Only This Control Register
1 = Update All Control Registers
35-32
Reserved
All Zeroes
31-18
Time Slot Length
Time Slot Length in IQCLK Periods; Bits 31-18 = 213... 20. Range is (19,25, 33, 37, 39, 49, 65 and 77)
The equation for calculating the value for this field is:
TSL = [[(Number of Output Bits + 2)Mode ] + 1]Hex;
where mode is 2 if the DDC is in either the real or I followed by Q mode. Mode is 1 for all other DDC
operational modes.
Allowable Minimum Time Slot Lengths:
(18)1 + 1 = 19 (13 hexidecimal)
(18)2 + 1 = 37 (25 hexidecimal); Real Output or I followed by Q
(24)1 + 1 = 25 (19 hexidecimal)
(24)2 + 1 = 49 (31 hexidecimal); Real Output or I followed by Q
(32)1 + 1 = 33 (21 hexidecimal)
(32)2 + 1 = 65 (41 hexidecimal); Real Output or I followed by Q
(38)1 + 1 = 39 (27 hexidecimal)
(38)2 + 1 = 77 (4d hexidecimal); Real Output or I followed by Q
17-0
Phase Offset
3-213
Starting Phase Angle of Phase Accumulator; Range = 0 to 2π. Bits 17-0 = 232... 215.
Some example phase offset hexidecimal values :
0000 - 0
1000 - π/2
2000 - π
3000 - 3π/2
3fff - 2π
HSP50016
TABLE 7. PHASE GENERATION/HDF.OUTPUT REGISTER
DESTINATION ADDRESS = 4
BIT
POSITION
FUNCTION
DESCRIPTION
39-37
Address
100 = Control Word 4
36
Update
0 = Update Only This Control Register
1 = Update All Control Registers
35-33
Reserved
All Zeroes
32-31
Output
Spectrum
00 = No Up Conversion,Complex Output
01 = Up Convert by f”/4, Real Output
10 = Up Convert by f”/2,Complex Output
11 = Reserved Mode
30-7
Delta Phase Increment
24-Bit Delta Phase Increment. Bits 30-7 = 223... 20.
Range: 0 < Delta Phase Increment < π (2-8-2-32)
6-1
HDF Data Shift
(Shift Factor)
16-Bit HDF Gain Compensation Number - the shift portion.
HDF Input Data Shift (Towards LSB). Bits 6-1 = 25...20.
Range: 0 ≤ Shift Factor ≤ 55 decimal; Range: [0 ≤ Shift Factor ≤ 37]hex
Calculate the value for this field using this equation:
HDF Data Shift = [75 - Ceiing(5 log2(R))]hex
Note: log2(x) = (3.32)log(x)
0
Spectral Reverse
0 = Normal Output
1 = Spectrally Reversed Output
TABLE 8. HDF/OUTPUT REGISTER
DESTINATION ADDRESS = 5
BIT
POSITION
FUNCTION
DESCRIPTION
39-37
Address
101 = Control Word 5
36
Update
0 = Update Only This Control Register
1 = Update All Control Registers
35-21
HDF Decimation
Counter Preload
(HDF DCP)
HDF Decimation Counter Preload. Range: 15 < HDF Decimation Counter Preload < 32,767
Calculate the value for this feild using this equation:
HDF DCP = R - 1; where R is the HDF decimation (rate change) factor.
Common HDF decimation (rate change) factors and associated hexidecimal HDF DCP values:
000f: R = 16
00ff: R = 128
01ff: R = 512
03ff: R = 1,024
07ff: R = 2,048
0fff: R = 4,096
1fff: R = 8,192
3fff: R = 16,384
7fff: R = 32,768
3-214
HSP50016
TABLE 8. HDF/OUTPUT REGISTER (Continued)
DESTINATION ADDRESS = 5
BIT
POSITION
20-5
FUNCTION
DESCRIPTION
Scaling Multiplier Gain
(Scale Factor)
16-Bit HDF Gain Compensation Number - the multiplier portion.
Range: 1 ≤ Scale Factor < 2,
Field Format = 20.2-1...2-15.
Calculate the value for this field using this equation:
Scale Factor = 2CEILING(5log2(R))/(R)5; where R is the HDF decimation (rate change) factor and CEILING(x) is equal to x for integer values, otherwise is equal to the next higher integer.
Common HDF decimation factors (R), decimation counter preload (DCP) and Scale Factors (SF) values:
R
DCP(dec)
DCP(hex)
SF(dec)
SF(hex)
16
15
000f
1.000
8000
128
127
007f
1.000
8000
512
511
01ff
1.00
8000
1,024
1,023
03ff
1.000
8000
2,048
2,047
07ff
1.000
8000
4,096
4,095
0fff
1.000
8000
8,192
8,191
1fff
1.00
8000
16,384
16,382
3fff
1.00
8000
32,768
32,768
7fff
1.000
8000
Note that the Scale Factor is 1 (8000hex) for power of 2 decimation factors.
The compensation for the HDF gain is performed with a shifter and a multiplier. Thus to program the HDF
Gain compensation, there is an associated Shift Factor and the Scale Factor. As the Decimation Factor
increases, the multiplier moves away from the value 1 and approaches the value 2. When the calculated
value for the multiplier equals or exceeds 2, the shifter is incremented and the multiplier returns to 1 and
increases towards 2 again as the Decimation Factor increases.
As an example, the table below details the values of Scale Factor for values of R from 16 to 32:
4-3
Output
Format
R
DCP(dec)
DCP(hex)
SF(dec)
16
15
000f
1.000000000
8000
17
16
0010
1.477013647
BD0E
18
17
0011
1.109857915
8E0F
19
18
0012
1.693916116
D8D2
20
19
0013
1.310720000
A7C5
21
20
0014
1.026983417
8374
22
21
0015
1.627707993
D058
23
22
0016
1.303318981
A6D3
24
23
0017
1.053497942
86D9
25
24
0018
1.717986918
DBE6
26
25
0019
1.412060017
B4BE
27
26
001a
1.169233029
95A9
28
27
001b
1.949663831
F98E
29
28
001c
1.635911864
D165
30
29
001d
1.380840823
B0BF
31
30
001e
1.172037271
9605
32
31
001f
1.000000000
8000
00 = Two’s Complement
01 = Offset Binary
10 = Sign Magnitude
11 = Single Precision Floating Point Format
3-215
SF(hex)
HSP50016
TABLE 8. HDF/OUTPUT REGISTER (Continued)
DESTINATION ADDRESS = 5
BIT
POSITION
FUNCTION
DESCRIPTION
2-1
Number Of Output Bits
00 = 16 Bits
01 = 24 Bits
10 = 32 Bits
11 = 38 Bits
0
Output Sense
0 = LSB First
1 = MSB First
TABLE 9. INPUT AND OUTPUT FORMAT REGISTER
DESTINATION ADDRESS = 6
BIT
POSITION
FUNCTION
DESCRIPTION
39-37
Address
110 = Control Word 6
36
Update
0 = Update Only This Control Register
1 = Update All Control Registers
35
I followed by Q
0 = I and Q Output Separately
1 = I and Q Data Output on I Pin
34-29
Time Slot Number
Bits 34-29 = 25...20. Range: 0 < Time Slot Number < 63.
This implies that 64 different channels may be mutliplexed, assigning one time slot per channel.
28
IQCLK
Polarity
0 = Output Data Stable On Rising Edge Of IQCLK; IQCLK High between I or Q Bit Periods when IQCLK
Duration = 0.
1 = Output Data Stable on Falling Edge of IQCLK; IQCLK Low between I or Q Bit Periods when IQCLK
Duration = 0.
27
IQCLK Duty Cycle
0 = IQCLK Active Time = CLK Period.
1 = 50% Duty Cycle
26
IQCLK
Duration
0 = Active During I or Q Output Periods Only
1 = Active Continuously
25-24
IQCLK Three-State
Control
00 = Three-State IQCLK
01 = Enable IQCLK
1x = Auto-Three-State Enable IQCLK (during time slot)
23
IQSTB
Polarity
0 = Active High
1 = Active Low
22
IQSTB
Location
0 = IQSTB Prior to the Beginning of the Data Word.
1 = IQSTB During the Data Word.
21-20
IQSTB Three-State
Control
00 = Three-State IQSTB
01 = Enable IQSTB
1x = Auto Three-State Enable IQSTB (during time slot)
19
I Polarity
0 = True Data
1 = Inverted Data
18-17
I Three-State Control
00 = Three-State I
01 = Enable I
1x = Auto Three-State Enable I (during time slot)
16
Q Polarity
0 = True Data
1 = Inverted Data
3-216
HSP50016
TABLE 9. INPUT AND OUTPUT FORMAT REGISTER (Continued)
DESTINATION ADDRESS = 6
BIT
POSITION
FUNCTION
DESCRIPTION
15-14
Q Three-State Control
00 = Three-State Q
01 = Enable Q
1x = Auto Three-State Enable Q (during time slot)
13
Input Format
0 = Offset Binary
1 = Two’s Complement
12-0
IQCLK Rate Counter
Preload
I/QCLK Rate Counter Preload, Bits 12-0 = 212 ... 20.
Range: 2 ≤ IQCLK Rate Counter Preload ≤ 1701.
To calculate the value in this field use this equation:
IQCLK Rate Counter Preload = [FLOOR[(HDF Decimation Factor x 4)/TSL] - 1]hex; where FLOOR(x)
represents the integer part of x, and TSL is the decimal value of Control Word 3, bit 31-18.
TABLE 10. PHASE OFFSET REGISTER
DESTINATION ADDRESS = 7
BIT
POSITION
FUNCTION
DESCRIPTION
39-37
Address
111 = Control Word 7
36
Update
0 = Update Only This Control Register
1 = Update All Control Registers
35-14
Reserved
All Zeroes
13
Data
0 = Normal Data Input
1 = Force Input Data to 8000 Hex.
12-11
FIR
Accumulator Control
00 = Normal Accumulation - The accumulator is reset on every FIR cycle.
01 = No Accumulation -The accumulator is disabled.
10 = Continuous Accumulation -The accumulator is not reset on every FIR cycle. This test mode was created to allow the user to perform the equivilent of a check sum test. A very long term test could be
run and an accumulated output would yeild a specific numeric value. If the answer differed, the part
is not functioning properly.
11 = Reserved
10
Q Strobe on Roll Over
0 = Q carries Normal Data
1 = Q Strobes When Phase Generator Rolls Over
9
Force Outputs
0 = Normal Output Response
1 = Force Outputs
8
IQCLK Forced Data
If Bit 9 = 1, Force IQCLK = Bit 8; Else Normal
7
IQSTB Forced Data
If Bit 9 = 1, Force IQSTB = Bit 7; Else Normal.
6
I Forced Data
If Bit 9 = 1, Force I = Bit 6; Else Normal.
5
Q Forced Data
If Bit 9 = 1, Force Q = Bit 5; Else Normal.
4
Sin/Cos Generator
Bypass
0 = Sin Cos Generator Normal,
1 = Bypass Sin Cos Generator; Sin = Cos = 0.fffff (approximately 1)
3
Scaling Multiplier
Bypass
0 = Scaling Multiplier Normal,
1 = Scale Factor = 1.
2
Reserved
Must be Zero for Proper Operation while Test Features are Enabled.
1
Wait For RAM Full
If Bit = 0, DDC will Output Data Normally after a Reset, which will Include Unpredictable Data in Data
RAMs. If Bit = 1, No Chip Output will Occur until Sufficient Data RAM Locations are Written.
0
Disable Overflow Protection
0 = Normal Operation
1 = Disable Overflow Protection
3-217
HSP50016
HSP50016 supports two types of testing. Control Word 7 can
be used to verify the operation of the circuit through the
divide and conquer method. Setting the Enable Test Bit
(Control Word 1, Bit 3) equal to a 1 enables the test features
controlled by Control Word 7. (This bit is in Control Word 1
so that Word 7 does not have to be loaded if the test features
are not being used.) The functions allowed by Control Word 7
are shown in Table 10.
NOTE: Asserting bits 9 and 13 of Control Word 7 will put
all outputs to a static mode. This may remove strobe
enables or clocks used to read the data signals. This Test
Mode was intended for interface evaluation at the board
level.
The DDC also has a Test Access Port (TAP). This port is
fully conformant to IEEE Std. 1149.1 - 1990 - IEEE Standard
Test Access Port and Boundary-Scan [2]. The TAP supports
the following instructions: BYPASS, SAMPLE/PRELOAD,
INTEST, EXTEST, RUNBIST and IDCODE. In addition, there
are seven instructions called RDCNTLWD1-7, which read
the contents of the control words over the TAP. The address
bits and bit 36 are only used to determine the destination of
data during loading; they are not stored, so they are not read
out with the RDCNTLWD1-7 instruction.
Summary
To use the DDC in a down conversion application three items
must be considered and designed to compliance. Solutions
must satisfy all three items.
1. The Nyquist Sampling Rate for the bandwidth of interest
FS ≥ 2BW, where BW is the bandwidth of interest.
2. The composite FIR/HDF double sided bandwidth, BW-3dB
= 0.1375FS/R.
3. The desired serial output clock rate (total decimation, plus
parallel to serial conversion rate increase).
f S Length MIN
IQCLK Frequency = ------------------------------------ – 1
(R FIR ) (R)
NOTE:
RFIR = 2 for real mode, 4 for all other modes.
Where x(n) is the real input data sequence, ω = 2πf, and ωc is
the frequency of the signal generated by the SIN/COS
Generator.
For demonstration purposes let x(n) = cos(ωkn). The
multiplication then becomes:
u(n) = cos(ωkn)[cos(ωcn) - jsin(ωcn)]
= 1/2[cos((ωk - ωc)n) + cos((ωk + ωc)n)
- j(sin((ωk + ωc)n) - sin((ωk - ωc)n))]
(EQ. 18)
The signal u(n) is passed through a low pass filter; assuming
that the filter passes the low frequency terms with no
degradation and attenuates the high frequency terms
completely, the filtering operation produces the output:
v(n) = 1/2(cos((ωk - ωc)n) + jsin((ωk - ωc)n))
= 1/2ej(ωk - ωc)n
(EQ. 19)
When the magnitude of the input signal x(n) is one, the
magnitude of v(n) is 1/2. Both the I and Q channels are
multiplied by a factor of two to yield:
w(n) = cos((ωk - ωc)n) + jsin((ωk - ωc)n)
= ej(ωk - ωc)n.
(EQ. 20)
Figure 16 shows an HSP50016 in a single channel down
conversion circuit. Notice that the input data is only 12 bits,
so it is justified to the MSB of the DDC’s input data. If a
smaller sample width is used, it is recommended that the
MSB of the data is input into DATA15. The unused bits are
connected to ground. This alignment makes it easier to
locate the position of the MSB in the output data. Note that
the input is configured for offset binary arithmetic and the
output is set up for I followed by Q, which enables the use of
only one serial connection to the output processor. The
serial data clock of the processor and the Control Clock of
the DDC are driven by a TTL compatible oscillator. (IQCLK
cannot be used for this purpose since its frequency is
indeterminate until the DDC has been configured). Note that
many processors provide a bit clock which eliminates the
need for the external oscillator.
Applications
Down Conversion
The primary spectral operation in the DDC is down
conversion of an input signal to base band, see Figure 15.
This process is done in two steps: multiplication of the input
waveform by an internally generated quadrature sinusoid,
i.e., modulation and lowpass filtering to attenuate the
unwanted spectral components. The unwanted spectral
components have two sources, the input signal and an
artifact of the modulation process.
The modulation process can be written as:
u(n) = x(n)e-jωc = x(n)[cos(ωc) - jsin(ωc)]
3-218
(EQ. 17)
- ωC
A. INPUT SIGNAL SPECTRUM
B. DOWN CONVERSION AND FILTERING
FIGURE 15. DOWN CONVERSION
ωC
HSP50016
MICROPROCESSOR
12-BIT
ADC
VIN D11
.
.
CS
.
OE D0
A0
CONV
HSP50016
DATA15
.
I
.
.
IQCLK
DATA4
IQSTB
DATA0-3
.
.
.
CLK
CDATA
CCLK
CSTB
DR
CLKR
FSR
DX
CLKX
FSX
CS
TABLE 13. SAMPLE FORMAT FOR CONTROL WORD 3 PHASE GENERATOR/OUTPUT TIME SLOT
BIT
POSITION
FUNCTION
DESCRIPTION
39-37
Address
011 = Control Word 3.
36
Update
1 = Control Register Update.
35-32
Reserved
All Zeroes.
31-18
Time Slot Length
All Zeroes.
17-0
Phase Offset
All Zeroes.
0111 0000 0000 0000 0000 0000 0000 0000 0000 0000
5MHz
OSCILLATOR
TABLE 14. SAMPLE FORMAT FOR CONTROL WORD 4 PHASE GENERATOR/HDF/OUTPUT
FIGURE 16. CIRCUIT FOR SINGLE CHANNEL OPERATION
An example of the control word contents for this mode of
operation is given in Tables 11 through 17. In this setup, the
DDC has been configured for a constant down conversion
frequency, decimation by 64 and Test Features disabled. Bit
fields of three bits or less are in binary notation; longer fields
are in hexadecimal. Control Words Zero and Seven are not
used.
TABLE 11. SAMPLE FORMAT FOR CONTROL WORD 1 PHASE GENERATOR/TEST ENABLE
BIT
POSITION
39-37
FUNCTION
Address
BIT
POSITION
39-37
FUNCTION
Address
100 = Control Word 4.
36
Update
1 = Control Register Update.
35-33
Reserved
All Zeroes.
32
Up Convert
0 = Do Not Up convert.
31
Real Mode
0 = Complex Mode.
30-7
Delta Phase
Increment
All Zeroes.
6-1
Shift
37 = Decimal 55, the Shift
Corresponding to HDF
Decimation by 16.
0
Spectral Reverse
0 = No Spectral Reversal
DESCRIPTION
001 = Control Word 1.
DESCRIPTION
1001 0000 0000 0000 0000 0000 0000 0000 0110 1110
36
Update
1 = Control Register Update.
35-4
Minimum Phase
Increment
Minimum Phase Increment
Computed according to
[fC/fS] 233.
TABLE 15. SAMPLE FORMAT FOR CONTROL WORD 5
HDF/OUTPUT
3
Test Enable
0 = Test Features Disabled.
BIT
POSITION
2-0
Phase Generator
Mode
001 = Normal Mode.
39-37
Address
101 = Control Word 5.
36
Update
1 = Control Register Update.
35-21
HDF Decimation
Counter Preload
F = Decimation by 16 in HDF.
TABLE 12. SAMPLE FORMAT FOR CONTROL WORD 2 PHASE GENERATOR
20-5
Scaling Multiplier
Gain
8000 = Scaling Multiplier
Gain of 1.
BIT
POSITION
4-3
Output Format
00 = Two’s Complement.
2-1
Number of Output
Bits
00 = 16-Bits.
0
Output Sense
1 = MSB First.
0011 XXXX XXXX XXXX XXXX XXXX XXXX XXXX XXXX 0001
FUNCTION
DESCRIPTION
39-37
Address
010 = Control Word 2.
36
Update
1 = Control Register Update.
35-32
Reserved
All Zeroes.
31-0
Maximum Phase
Increment
All Zeroes.
0101 0000 0000 0000 0000 0000 0000 0000 0000 0000
3-219
FUNCTION
DESCRIPTION
1011 0000 0000 0001 1111 0000 0000 0000 0000 0000
HSP50016
TABLE 16. SAMPLE FORMAT FOR CONTROL WORD 6 OUTPUT
sideband from aliasing onto the other sideband when the real
part of the output signal is taken.
BIT
POSITION
The spectrum of a quadrature signal which has been over
sampled by 2 is shown in Figure 17A. This represents the
output of the filters. As described in the previous paragraph,
the oversampling is a necessary feature of this process,
since the final signal will occupy twice the bandwidth of the
filter output. To prevent aliasing upon taking the real part of
the signal, it is necessary to perform an up conversion by f”/4,
where f” is the decimated sample frequency. (Note that fs is
defined as the input sampling frequency, f’ is the input
sampling frequency divided by the HDF decimation rate R,
and f” is f’ divided by the FIR decimation rate. f’’ is the FIR
output sampling rate). The up conversion function is:
FUNCTION
DESCRIPTION
39-37
Address
110 = Control Word 6.
36
Update
1 = Control Register Update.
35
I followed by Q
1 = I and Q Data Output on I Pin.
34-29
Time Slot
Time Slot Number = 0.
28
IQCLK Polarity
0 = Data Stable on Rising Edge
of IQCLK.
27
IQCLK Duty Cycle 1 = IQCLK Duty Cycle is 50%.
26
IQCLK Duration
1 = Active Continuously.
25-24
IQCLK ThreeState Control
01 = Enable IQCLK.
23
IQSTB Polarity
0 = IQSTB Active High.
22
IQSTB Location
0 = IQSTB Active Prior to the
Beginning of the Data Word.
21-20
IQSTB
Three-State
01 = Enable IQSTB.
19
I Polarity
0 = I Output Active High.
18-17
I Three-State
Control
01 = Enable I.
16
Q Polarity
0 = Q Output Active High.
15-14
Q Three-State
Control
00 = Disable Q.
13
Input Format
1 = Two’s complement.
12-0
IQCLK Rate
All Zeroes = CLK Used to Clock
Output Bits.
1101 1000 0000 1101 0001 0010 001X XXXX XXXX XXXX
TABLE 17. SUMMARY OF CONTROL WORDS FOR THE EXAMPLE
CONTROL WORD
HEX VALUE
0
0 0 0 0 0 0 0 0 0 0
1
3 X X X X X X X X 1
2
5 0 0 0 0 0 0 0 0 0
3
7 0 0 0 0 0 0 0 0 0
4
9 0 0 0 0 0 0 0 6 E
5
B 0 0 1 F 0 0 0 0 0
6
D 8 0 D 1 2 X X X X
7
0 0 0 0 0 0 0 0 0 0
Quadrature To Real Conversion
After the input data has been processed by the DDC, the output
can be converted into a real signal if desired. In that case, the
baseband centered quadrature signal is upcobaseband. The
real part of the upconverted signal is taken as the output. To
satisfy the Nyquist criteria, the sample rate of the resulting
signal must be at least twice the minimum sample rate of the I
and Q components of the quadrature signal. This prevents one
3-220
ej2πnf’’/4f’’ = ejπn/2
(EQ. 21)
For n = 0, 1, 2, 3, 4,... the output values of the local oscillator
in rectangular representation are: 1 + 0j, 0 + j, -1 + 0j, 0 - j,
1 + 0j,.... Since the real half of the complex multiplication of
the local oscillator values by the filtered signal values (the
desired output is the real part of the product) require only
trivial operations, this up conversion is done in the Formatter.
Figure 17B shows the signal spectrum after up conversion.
Figure 17C shows the spectrum of the real output signal.
Continuing with the single tone example from the previous
section, the quadrature signal output from the FIR filters is:
w(n) = cos((ωk - ωc)n) + jsin((ωk - ωc)n)
(EQ. 22)
= ej(ωk - ωc)n
Multiplying w(n) by the up convert function and summing the
result is equivalent to the output sequence:
y(n) = 1 x cos((ωk - ωc)n),
y(n+1) = j x jsin((ωk - ωc)(n+1)),
y(n+2) = -1 x cos((ωk - ωc)(n+2)),
y(n+3) = -j x jsin((ωk - ωc)(n+3)),
y(n+4) = 1 x cos((ωk - ωc)(n+4)),...
(EQ. 23)
y = cos((ωk - ωc)n), -sin((ωk - ωc)(n+1)),
-cos((ωk - ωc)(n+2)), sin((ωk - ωc)(n+3)),
cos((ωk - ωc)(n+4)),...
Or:
y = RE(w(n)), - IM(w(n+1)), - RE(w(n+2)), IM(w(n+3)),
RE(w(n+4)),...
Since |ejπn/2| = 1 and |w(n)| = 1, no further magnitude
corrections are required.
The setup for this application is similar to that of the down
conversion circuit given above, except the Output Formatter
is set for Real Mode (Bit 31 in Control Word 4). This bit
configures the part for up conversion by f”/4 and summing of
the real and imaginary parts of the filter output.
HSP50016
fS = f1’R
f1’ = 2f”
-f”/2
-f”/4
-f’
4
-fS
4R
-f’
8
-fS
8R
0
f”/4
f”/2
-f”/2
-f”/4
f’
8
fS
8R
f’
4
fS
4R
-f’
4
-fS
4R
-f’
8
-fS
8R
A. OUTPUT OF FIR FILTERS: SIGNAL OVERSAMPLED BY 2
0
f”/4
f”/2
f’
8
fS
8R
f’
4
fS
4R
B. FIRST OPERATION IN FORMATTER:
UP CONVERT BY SAMPLE FREQUENCY / 4
-f”/2
-f”/4
-f’
4
-fS
4R
-f’
8
-fS
8R
0
f”/4
f”/2
f’
8
fS
8R
f’
4
fS
4R
C. SECOND OPERATION IN FORMATTER: I OUTPUT = I + Q
FIGURE 17. QUADRATURE TO REAL CONVERSION OF AN OUTPUT SIGNAL
Up Conversion by f”/2
This operation allows the user to exchange the positions of
the upper and lower halves of a down converted signal
while leaving each half unchanged. Quadrature up
conversion by f”/2 is performed by multiplying the output
signal by ej2πnf’’/2 = cos(2pnf”/2) + j sin(2pnf”/2). When
sampled at a rate of f”, cos(2pnf”/2) takes on the values 1,
-1, 1, -1,... and sin(2pnf”/2) always = 0. Thus, the up
convert LO sequence is:
ejπn = 1 + j0, -1 + j0, 1 + j0, …
= ej(ωk - ωc)n,
The setup is identical to the down conversion configuration,
except that the Up Convert Bit is set in Control Word 4.
(EQ. 24)
This sequence is multiplied by the output of the I and Q
branches of the filter:
w(n) = cos((ωk - ωc)n) + jsin((ωk - ωc)n)
The Formatter contains the circuitry to shift the quadrature
output spectrum up by one half of the output sample
frequency f”. This operation is independent of the function of
the Phase Generator and Mixer. The spectra of the outputs
of the Filter and Formatter are shown in Figure 18.
-f”/2
-f”/4
0
f”/4 f”/2
A. OUTPUT OF FIR FILTERS
(EQ. 25)
yielding an output sequence:
y = (RE(w(n)), IM(w(n))),
-RE(w(n+1)), - IM(w(n+1))),
RE(w(n+2)), IM(w(n+2))), …
3-221
(EQ. 26)
-f”/2
-f”/4
0 f”/4
f”/2
B. FILTER OUTPUT UP CONVERTED BY
OUTPUT SAMPLE FREQUENCY / 2
FIGURE 18. UP CONVERSION BY F” / 2
HSP50016
Quadrature Spectral Reversal
Spectral reversal is often used to negate a spectral reversal
which has occurred due to a previous operation in the
processing chain. Examples of this are spectral reversal in
an analog down conversion or in a constructive aliasing
operation. The DDC gives the user the ability to convert the
signal to baseband in either forward or reverse fashion.
Quadrature spectral reversal is achieved by translating the
lower sideband of the input to baseband rather than the
upper sideband. This is implemented in the DDC by mixing
the input signal with ej2πfcn - that is, up converting the input
rather than down converting it. The resulting signal is:
u(n)= x(n)ej2πfcn = x(n)[cos(ωcn) + jsin(ωcn)]
(EQ. 27)
quadrature to real conversion. The output spectrum is
shown in Figure 20.
-f”/2 -f”/4 0 f”/4 f”/2
A. OUTPUT OF FILTERS: SIGNAL OVERSAMPLED BY 2
-f”2 -f”/4 0 f”/4 f”/2
B. FIRST OPERATION IN FORMATTER:
UP CONVERT BY SAMPLE FREQUENCY / 4
Assuming x(n) = cos(ωkn),
u(n) = cos(ωkn)[cos(ωcn) + jsin(ωcn)]
= [cos((ωk - ωc)n) + cos((ωk + ωc)n)
+ j(sin((ωk + ωc)n) - sin((ωk - ωc)n))]
(EQ. 28)
-f”/2 -f”/4 0 f”/4 f”/2
After quadrature filtering and correcting for the gain of 1/2,
we have:
w(n) = cos((ωk - ωc)n) - jsin((ωk - ωc)n)
The setup for this application is similar to that of down
conversion, except in Control Word 4, where the Spectral
Reverse and Real Output bits are set to one.
= ej(ωc- ωk)n
The appropriate spectral plots are shown in Figure 19. In up
conversion, the sine output of the SIN/COS Generator is
negated so that the vector output of the Local Oscillator
rotates counter clockwise. This is implemented by setting the
Spectral Reverse bit in Control Word 4 to a one. Otherwise,
the setup for this mode is the same as the one for down
conversion.
A. INPUT SIGNAL SPECTRUM
FIGURE 20. QUADRATURE TO REAL CONVERSION OF AN
OUTPUT SIGNAL
(EQ. 29)
= cos(-(ωk - ωc)n) + jsin(-(ωk - ωc)n)
= cos((ωc - ωk)n) + jsin((ωc - ωk)n)
- ωC
C. SECOND OPERATION IN FORMATTER: I OUTPUT = I + Q
ωC
High Decimation Filter Only
The DDC can be operated as a single high decimation
filter. This is done by setting the Phase Generator to Filter
Only and the Minimum Phase Increment and Phase Offset
to 0. This multiplies the incoming data stream by a constant
hexadecimal 3FFFF in the I channel and 0 in the Q
channel. The HDF Section of the circuit requires a
minimum decimation rate of 16 to allow sufficient time for
the FIR to compute its response. This mode of operation
implements a filter which has a decimation rate from 64 to
131,072. The frequency response is shown in Figures 7, 8
and 9. Only the I output has valid data in this mode; the Q
output should be set to high impedance state to reduce
circuit noise.
Multichannel Operation
B. UP CONVERSION AND FILTERING
FIGURE 19. UP CONVERSION OF FILTER OUTPUT SIGNAL
Real Spectral Reversal
Real spectral reversal is simply quadrature spectral
reversal with quadrature to real conversion in the
Formatter. The up converted and filtered signal w(n) is
upconverted again by f”/4 in the Formatter. Each sideband
of the result is spectrally reversed from the sidebands that
would have been produced by down conversion with
3-222
Several DDCs can be placed in parallel with each one
operating on a different frequency band. To minimize
wiring, their outputs can be configured so that they are
connected over a common serial bus. Each DDC is
assigned a time slot number (Control Word 6) and a time
slot length (Control Word 3). Each DDC in turn controls the
bus for long enough to output its data, then relinquishes the
bus. The time slot assignment and length are programmed
at configuration time. Up to 64 chips can be multiplexed in
this manner.
HSP50016
Figure 21 shows a Block Diagram of this configuration. The
DDCs are configured by the microprocessor by first writing a
logical 0 to its Chip Select line. The control words are written
to that part in any order. When the part has been configured,
CS is written high again, and the next part is configured in
the same manner. Collisions are prevented by programming
each DDC with a unique Time Slot number, which holds its
output from 0 to 63 output word times before transmission.
Each part also has a Time Slot Length, whose minimum
value is given in Equation 8. Note that a value greater than
the minimum can be used to give the processor time to
operate on the data
DATA
FROM A/D
MICROPROCESSOR
HSP50016
I
DATA0-15
IQCLK
IQSTB
IQSTRT
CLK
CDATA
CCLK
CSTB
CHIP SELECT
DECODER
A0-15
R/W
STRB
HSP50016
I
DATA0-15
IQCLK
IQSTB
SYSTEM
CLOCK
TIME SLOT 0
TIME SLOT 1
TIME SLOT N-1
DDC 0
DDC 1
DDC N-1
FIGURE 22. TIMING FOR MULTIPLE CHANNEL OPERATION
(AUTO THREE-STATE)
CS
CLK
When operating a set of HSP50016s in the Multiple
Channel Operation Mode, the two control signals that
ensure proper time slot operation are: CS and RESET. The
CS allows unique Control Word loading of each DDC. The
RESET synchronizes all of the DDC’s to the start of the first
time slot.
DR
CLKR
FSR
IQSTRT
DX
CLKX
FSX
CDATA
CCLK
CSTB
OSCILLATOR
CS
FIGURE 21. CIRCUIT FOR MULTIPLE CHANNEL OPERATION
(AUTO THREE-STATE)
The corresponding Configuration Register setup is similar to
that of single channel down conversion, except for the Auto
Three-State fields. In this example, the first DDC in the chain
is set to drive IQCLK; the others have this output set for high
impedance. (It makes no difference which DDC is chosen to
be the one to drive IQCLK, but it must be active
continuously). The unused outputs are put in their high
impedance condition on the other DDCs to minimize power
consumption. Note also that this example shows all DDCs in
I followed by Q Mode so that only one data line to the
microprocessor is necessary. Figure 22 gives the timing of
the output data.
3-223
NOTE: In this mode, it was anticipated that parallel
DDC’s would be programmed identically, except for the
NCO (L.O.) frequency and Time Slot Number. This implies
identical HDF Decimation factors, Time Slot Length’s,
Number of Output Bits, Output Mode are identical in all
DDCs. This means that the output rate of all DDC HDFs,
FIRs and Parallel to Serial Converters are identical.
The DDC keeps an internal count of the number of IQCLK
periods that have transpired since the rising edge of RESET.
The internal counter of each DDC is set to enable the serial
output at the time slot number assigned to that DDC. The
count is based on the time slot length, number of output bits,
HDF decimation, and Output Mode programmed to that part.
NOTE: In the Multiple Channel Operation Mode, all the
time slot lengths should be set to the same value to avoid
output signal contention. The time slot length should be
equal to the largest “minimum time slot length” as calculated by Equation 15, for every DDC in the multichannel
arrangement. Note that Equation 8 is in IQCLK periods, not
CLK periods. Equations 7 and 9 will be helpful in making
the translation to CLK periods.
This mode does not require that all of the time slots be used
(assigned to a DDC). If only two parts were used and the
maximum time signal isolation was desired, one could
assign the first signal time slot 31 and the second signal time
slot 63. This would ensure that the maximum separation in
time occurred. It is the designers responsibility to ensure that
the output rate, including the decimation is consistent with
the time allotted to output each signal.
HSP50016
Let’s return to the clock calculation example found in the Output
Formatter, and add the requirement of 4 time slots. The
calculations in the Example Clock Calculation Section remain
true, but step 8 must be added:
DATA
FROM
A/D
8. Now lets consider the multichannel timing. Each channel
must output the data at no less than 1.725MHz to get all
the I/Q data out in the allocated time for the assigned time
slot. Time margin is created when the output is clocked out
at a higher rate. Because each channel is outputting data
in only one of four channels, the effective output rate for
each channel is 1.725MHz /4 = 431.25kHz, even though
the part is outputting data 1.7MHz in every time slot.
. . . . . . . . . . Effective Channel Output Rate = 431.25kHz
Alternatively, the processor can request data from each of
the DDCs asynchronously. In this setup, Requested Output
Mode is used. The Data Concentrator polls each channel
individually and is responsible for ensuring that each
channel is polled before the output data is lost. The Data
Concentrator is a custom circuit designed by the user. A
Block Diagram of such a system is shown in Figure 23. The
interface between the controller and the DDCs has been
omitted for the sake of clarity.
HSP50016
DATA0-15
I
IQCLK
IQSTB
IQSTRT
CLK
DATA
CONCENTRATOR
MICROPROCESSOR
A0-15
D0
R/W
STRB
HSP50016
I
IQCLK
IQSTB
IQSTRT
DATA0-15
CLK
SYSTEM
CLOCK
FIGURE 23. CIRCUIT FOR MULTIPLE CHANNEL OPERATION
(REQUESTED OUTPUT)
References
[1] Hogenauer, Eugene V., An Economical Class of Digital
Filters for Decimation and Interpolation, IEEE
Transactions on Acoustics, Speech and Signal
Processing, April 1981.
[2] IEEE Standard Test Access Port and Boundary-Scan
Architecture, IEEE Std 1149.1 - 1990.
3-224
HSP50016
Absolute Maximum Ratings
Thermal Information
Supply Voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . +7.0V
Input, Output or I/O Voltage . . . . . . . . . . . .GND -0.5V to VCC +0.5V
ESD Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Class 1
Thermal Resistance (Typical, Note 1)
θJA (oC/W) θJC (oC/W)
CPGA Package . . . . . . . . . . . . . . . . . .
45
7
PLCC Package. . . . . . . . . . . . . . . . . . .
35
N /A
Maximum Junction Temperature
CPGA Package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .150oC
PLCC Package. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .150oC
Maximum Storage Temperature Range . . . . . . . . . . -65oC to 150oC
Maximum Lead Temperature (Soldering 10s) . . . . . . . . . . . . .300oC
Operating Conditions
Operating Voltage Range. . . . . . . . . . . . . . . . . . . . +4.75V to +5.25V
Operating Temperature Range . . . . . . . . . . . . . . . . . . . 0oC to 70oC
Die Characteristics
Gate Count . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65,000 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:
1. θJA is measured with the component mounted on an evaluation PC board in free air.
DC Electrical Specifications
VCC = 5.0V ±5%, TA = 0o to 70oC
PARAMETER
SYMBOL
Power Supply Current
ICCOP
Standby Power Supply Current
ICCSB
Input Leakage Current
II
TEST CONDITIONS
MIN
MAX
UNITS
VCC = Max, CLK Frequency 52.6MHz
Notes 2, 3
-
394
mA
VCC = Max, CLK Frequency 76.9MHz
Notes 2, 3
-
577
mA
-
500
µA
VCC = Max, Input = 0V or VCC
TMS, TDI, TRST
-500
10
µA
VCC = Max, Input = 0V or VCC
All other inputs
-10
10
µA
10
µA
VCC = Max, Outputs Not Loaded
Output Leakage Current
IO
VCC = Max, Input = 0V or VCC
-10
Logical One Input Voltage
VIH
VCC = Max
2.0
-
V
Logical Zero Input Voltage
VIL
VCC = Min
-
0.8
V
Logical One Input Voltage: CLK, TRST
VIHC
VCC = Max
3.0
-
V
Logical One Output Voltage
VOH
IOH = -5mA, VCC = Min
2.6
-
V
Logical Zero Output Voltage
VOL
IOL = 5mA, VCC = Min
-
0.4
V
Input Capacitance
CIN
CLK Frequency 1MHz
All measurements referenced to GND.
TA = 25oC, Note 4
-
10
pF
-
10
pF
Output Capacitance
COUT
NOTES:
2. Power supply current is proportional to frequency. Typical rating is 7.5mA/MHz. Note that operation at maximum clock frequency will exceed
maximum junction temperature of device. Use of a heat sink and/or air flow is required under these conditions: Recommended heat sink is
EG&G Wakefield D10650-40.
3. Output load per test circuit and CL = 40pF.
4. Not tested, but characterized at initial design and at major process/design changes.
AC Electrical Specifications
VCC = 5.0V ±5%, TA = 0o to 70oC, (Note 5)
-52(52.6)MHz
PARAMETER
MIN
MAX
MIN
MAX
UNITS
CLK Period
tCP
19
-
13
-
ns
CLK High
tCH
7
-
5
-
ns
CLK Low
tCL
7
-
5
-
ns
Setup Time DATA0-15 to CLK
tDS
10
-
7
-
ns
Hold Time DATA0-15 from CLK
tDH
1
-
1
-
ns
RESET Pulse Width
tRL
tCP+11
-
tCP+8
-
ns
3-225
SYMBOL
NOTES
-75(76.9)MHz
HSP50016
AC Electrical Specifications
VCC = 5.0V ±5%, TA = 0o to 70oC, (Note 5) (Continued)
-52(52.6)MHz
PARAMETER
SYMBOL
MIN
MAX
MIN
MAX
UNITS
tRS
Note 6
10
-
7
-
ns
RESET, IQSTRT Hold Time to CLK
tRH
Note 6
1
-
1
-
ns
CLK to I, Q, IQSTB, IQCLK Delay
tDO
-
15
-
12
ns
CCLK Period
tCCP
100
-
100
-
ns
CCLK High
tCCH
40
-
40
-
ns
CCLK Low
tCCL
40
-
40
-
ns
CDATA, CSTB, CS Setup to CCLK
tCDS
30
-
30
-
ns
CDATA, CSTB, CS Hold from CCLK
tCDH
CCLK Low Setup to CLK
tCLS
Notes 6, 7
CCLK High Hold from CLK
tCHH
TCK Period
tTCP
TCK High
TCK Low
RESET, IQSTRT Setup Time from CLK
NOTES
-75(76.9)MHz
30
-
30
-
ns
30
-
30
-
ns
Notes 6, 7, 10
30
-
30
-
ns
Note 8
100
-
100
-
ns
tTH
40
-
40
-
ns
tTL
40
-
40
-
ns
TRST Pulse Width
tTRL
100
-
100
-
ns
TCK to TDO, Data Delay
tTDO
-
30
-
30
ns
Setup Time On All Inputs to TCK
tATS
Note 9
30
-
30
-
ns
Hold Time On All Inputs from TCK
tATH
Note 9
30
-
30
-
ns
TCK Setup Time to CLK
tTCS
Note 8
30
-
30
-
ns
TCK Hold Time from CLK
tTCH
Note 8
30
-
30
-
ns
Output Enable Time from CLK
tOE
Note 10
-
18
-
12
ns
Output Disable Time from CLK
tOD
Note 10
-
18
-
12
ns
Output Enable Time from TCK
tTOE
Note 10
-
32
-
32
ns
Output Disable Time from TCK
tTOD
Note 10
-
32
-
32
ns
tRF
Note 10
-
5
-
5
ns
Output Rise, Fall Time
NOTES:
5. AC tests performed with CL= 40pF, IOL = 5mA, and IOH = -5mA. Input reference level for CLK, TRST is 2.0V, all other inputs 1.5V. Test VIH =
3.0V, VIHC = 4.0V, VIL = 0V; VOH = VOL = 2.5V.
6. These are asynchronous inputs; setup and hold times must only be maintained in order to predict which clock cycle they take effect internally.
7. Timing must only be maintained when Update bit is active in control word data being loaded.
8. Special Timing relationship between TCK and CLK is required for Test Instructions RUNBIST, EXTEST and INTEST.
9. All inputs except TRST, and only when TCK is driving internal clock.
10. Controlled via design or process parameters and not directly tested. Characterized upon initial design and after major process and/or design
changes.
AC Test Load Circuit
DUT
S1
CL (NOTE)
±
SWITCH S1 OPEN FOR ICCSB AND ICCOP
IOH
2.5V
EQUIVALENT CIRCUIT
NOTE:
Test head capacitance.
3-226
IOL
HSP50016
Waveforms
tCP
tCH
tCL
CLK
tDS
tDH
DATA0-15, IQSTRT
tRH
tRL
tRS
RESET
tDO
I, Q, IQSTB, IQCLK
tOE
2.7V
I, Q, IQSTB, IQCLK
2.3V
tOD
I, Q, IQSTB, IQCLK
tCHH
tCLS
CCLK
tTCS
tTCH
TCK
FIGURE 24. TIMING RELATIVE TO CLK
tCCP
tCCH
tCCL
CCLK
tCDS
CDATA, CSTB, CS
FIGURE 25. TIMING RELATIVE TO CCLK
3-227
tCDH
HSP50016
Waveforms
(Continued)
tTCP
tTH
tTL
TCK
tTRL
TRST
tTDO
TDO, I, Q,
IQSTB, IQCLK
tATS
tATH
ALL INPUTS
tTOE
2.7V
TDO, I, Q, IQSTB, IQCLK
2.3V
tTOD
TDO, I, Q, IQSTB, IQCLK
FIGURE 26. TIMING RELATIVE TO TCK
tRF
tRF
2.0V
0.8V
FIGURE 27. OUTPUT RISE AND FALL TIMES
All Intersil semiconductor products are manufactured, assembled and tested under ISO9000 quality systems certification.
Intersil semiconductor 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 http://www.intersil.com
Sales Office Headquarters
NORTH AMERICA
Intersil Corporation
P. O. Box 883, Mail Stop 53-204
Melbourne, FL 32902
TEL: (407) 724-7000
FAX: (407) 724-7240
3-228
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 (Taiwan) Ltd.
7F-6, No. 101 Fu Hsing North Road
Taipei, Taiwan
Republic of China
TEL: (886) 2 2716 9310
FAX: (886) 2 2715 3029