Z87200
Zilog Spread-Spectrum Transceiver
4-47
4
Differential Demodulation
As noted in the preceding text, computation of the “Dot”
and “Cross” products is fundamental to operation of the
DPSK Demodulator and Frequency Discriminator. Let I
k
and Q
k
represent the I and Q channel inputs, respectively,
for the k
th
symbol after downconversion and despreading.
The Dot and Cross products can then be defined as:
Dot(k) = I
k
I
k-1
+ Q
k
Q
k-1
; and,
Cross(k) = Q
k
I
k-1
- I
k
Q
k-1
In the complex domain, these products can be seen to
have been defined to form the complex conjugate product
between two input samples, one symbol apart. Let the k
th
input sample, sin(k), be defined as:
s
in
(k) = I(k) + j Q(k),
where I(k) and Q(k) are the 8-bit peak power PN Matched
Filter I and Q channel outputs directed to the DPSK De-
modulator. In polar form, s
in
(k) may be conveniently de-
fined as:
s
in
(k) = A(k)e
jØ(k)
with
A(k)
Ø(k) = arctan
Simple substitution then shows that the complex conjugate
product between consecutive symbols (with an arbitrary
phase shift introduced to the previous symbol value) may
be expressed as:
s
out
(k)= s
in
(k) [s
in
(k–1) . ω
fixed
]
*
= Dot(k) + j Cross(k)
where
ω
fixed
= arbitrary fixed phase rotation;
Dot(k)= Re[s
out
(k)]; and,
Cross(k)= Im[s
out
(k)]
*
The fixed phase rotation ω
fixed
has been introduced to later
simplify the decision criteria. The ability to express real and
imaginary parts of the complex conjugate product between
consecutive symbols with the Dot and Cross products is
the key to their use in DPSK demodulation.
DBPSK Demodulation
In DPSK, the phase difference between successive sam-
ples is due to the data modulation phase differences,
Ømod, plus any induced phase rotation between sym-
bols, Ørot, resulting from, for example, a frequency offset
between the received signal’s I.F. and that provided by the
Downconverter. For DBPSK, the data modulation differ-
ences Ømod can take only the values of 0° or 180°. Ex-
pressing the complex phase difference [Ø(k)-Ø(k-1)] in
terms of these components, the decision can be seen to be
based on:
Sout
(k)
=A(k) A(k-1) e
jØ(k)*
e
-jØ(k-1)
= A(k)
*
A(k1)
*
e
j[Ø
mod
(k)+Ø
rot
(k)]
For DBPSK, only the real part of sout(k), Dot(k), is needed
to determine the modulated phase transition:
Dot(k)= A(k)
*
A(k-1)
*
cos(Ø
mod
(k)+Ørot(k))
= ±A(k)
*
A(k-1)
*
cos(Ø
rot
(k))
where the sign is determined by the transmitted data since
cos[Ø
mod
(k)] = ±1
*
As a result,
Dot(k) ±A
2
(k)
if the amplitude of the signal is constant for consecutive
symbols and if the phase rotation Ø
rot
(k) between sym-
bols is small. The Z87200 DPSK Demodulator can thus
use the sign of the Dot product in order to make DBPSK
symbol decisions without the introduction of any fixed
phase rotation.
PS010202-0601
Z87200
Spread-Spectrum Transceiver Zilog
4-48
THEORY OF OPERATION (Continued)
DQPSK Demodulation
For DQPSK modulation, the possible phase shifts be-
tween successive symbols due to the modulation are 0°,
90°, 180°, and 270°. Here, introduction of a phase shift
(ω
fixed
) of ±45° to the previous symbol in the calculation of
the Dot and Cross products is desired in order shift the
possible phase differences to 45°, 135°, 225°, or 315° so
that the DQPSK decision boundaries coincide with the
signs of the Dot and Cross products. In the Z87200 DPSK
demodulator, phase rotation is accomplished in the signal
rotation block by the following transformation of the I and
Q channel values:
I
rot
(k)=[ I(k) - Q(k)]/2 for 45° rotation
I
rot
(k)=[ I(k) + Q(k)]/2 for –45° rotation
Q
rot
(k)=[ I(k) + Q(k)]/2 for 45° rotation
Q
rot
(k)=-[ I(k) + Q(k)]/2 for –45° rotation
The divide-by-2 is part of the signal rotation function. This
transformation is equivalent to multiplying by (1 ± j)/2 or
(1/2)e
jØ(fixed)
where Ø
fixed
is ±45°. In this case, s
out
(k) be-
comes:
s
out
(k)=A(k).A(k-1)*e
jØ(k)*
e
–jØ(k-1)*
[ω
fixed
]*
=A(k).A(k-1)*e
j[Ø
mod
(k)+Ø
rot
(k)]*
(1/2)ej
Ø(fixed)
so that
Dot(k)(1/2)A(k)*A(k-1)*cos(Ø
mod
(k) - Ø
fixed
)
Cross(k)(1/2)A(k)*A(k-1)*sin(Ø
mod
(k) - Ø
fixed
)
where the phase rotation Ø
rot
(k) due to the frequency off-
set between symbols has been assumed negligible.
A summary of the Dot(k) and Cross(k) products for the
possible values of Ø
mod
(k) and Ø
fixed
is shown below, il-
lustrating how the sign of the Dot and Cross products allow
the symbol decision to be made:
π/4 QPSK Demodulation
The Z87200 DPSK Demodulator decision logic is de-
signed so that correct DQPSK decisions are made with a
signal rotation of Ø
fixed
= –45°. For π/4 QPSK modulation,
however, the modulator itself inserts 45° between consec-
utive symbols, and the possible phase shifts between suc-
cessive symbols due to modulation are 45°, 135°, 225°,
and 315°. As a result, the DPSK Demodulator should be
configured for π/4 QPSK with Ø
fixed
=0°.
DQPSK Phasing and I/Q Channel Reversal
The Z87200 uses Differential BPSK and QPSK modulation
and demodulation, meaning that the data is modulated on
the carrier as phase changes. At the demodulator, the data
is recovered by monitoring the phase change over a sym-
bol period.
The Z87200 provides configuration control to specifically
address DPSK phasing and I/Q channel reversal: the Sig-
nal Rotation control register, bits 0 and 1 of address 33
H
,
and the Reverse I and Q control register, bit 0 of address
36
H
. The first register causes an insertion of ±45° in phase
between consecutive symbols at the receiver, while the
second register switches the I and Q channels presented
to the DPSK demodulator. As discussed in the Z87200 ap-
pendix, the introduction of a phase shift between consec-
utive symbols changes the mapping of the input data with
respect to the decision boundaries defined by the "Cross"
and "Dot" product axes.
Assuming that the transmitted DQPSK modulation phas-
ing is differentially encoded as defined in Table 3, the
phase shift between consecutive symbols should always
be set to –45°; that is, bits 1 and 0 of address 33
H
should
be set to 11. Similarly, when the transmission path from
modulator to demodulator does not introduce a frequency
(or phase direction) reversal, the "reverse I and Q" control
function should be disabled; that is, bit 0 of address 36H
should be set to 0. Note that, in the case of DBPSK, the
phase increments are either 0 or 180° and frequency re-
versal has no impact.
If frequency reversal does take place, however, correct
DQPSK demodulation can be achieved by enabling I and
Q reversal; that is, the entry into bit 0 of address 36
H
should be set to 1. Frequency reversal may occur in the up
or down conversion process, depending on which mixing
product is selected for further processing. No reversal oc-
curs when the following conditions exist: when the mixing
()
Q(k)
I(k)
.
I
2
(K)+Q
2
(k)
Ø
fixed
= -45° Ø
fixed
= +45°
Ø
mod
(k)
Dot(k)
Cross
(k)
Ø
mod
(k) Dot(k)
Cross
(k)
0°
+A
2
+A
2
0°
+A
2
–A
2
90°
–A
2
+A
2
90°
+A
2
+A
2
180°
–A
2
–A
2
180°
–A
2
+A
2
270°
+A
2
–A
2
270°
–A
2
–A
2
PS010202-0601
Z87200
Zilog Spread-Spectrum Transceiver
4-49
4
at the transmitter is performed by processing the sum fre-
quency of the local oscillator and the modulator; when the
mixing at the receiver is performed by subtracting the local
oscillator from the incoming signal; and when the in-phase
and quadrature inputs into the I and Q analog-to-digital
converters are correctly connected such that the in-phase
component leads the quadrature component by 90°. Un-
der these conditions, bit 0 of address 36
H
should be set to
0; otherwise, the I and Q channels may need to be re-
versed at the DPSK demodulator (by setting bit 0 of ad-
dress 36
H
to 1) in order to achieve proper demodulation.
Frequency Error Signal Generation
The frequency discriminator function or error signal is gen-
erated based on the Dot and Cross products. The objec-
tive is an error signal that is proportional to the sine of the
phase difference between the present and prior symbol af-
ter correcting for the estimated phase increments due to
data modulation. In the Z87200 Frequency Discriminator,
the frequency error is calculated through a decision-direct-
ed cross-product algorithm and is then used with the Loop
Filter to correct the NCO frequency. Assuming an input
sin(k), where:
s
in
(k) = I(k) + j Q(k),
the algorithm calculates the frequency discriminator func-
tion for DBPSK, s
AFC/BPSK
(k), as:
S
AFC/BPSK
(k)=SIGN[Dot(k)]*Cross(k)
=SIGN[Dot(k)]*A(k)*A(k-1)*sin(Ø(k)-Ø(k-1))
=SIGN[Dot(k)]*A(k)*A(k-1)*sin(Ømod(k) + Ørot(k))
SIGN[Dot(k)]*A2(k)*cos[Ømod(k)]*sin[Ørot(k)]
A2(k)*sin[Ørot(k)]*
The final result assumes that the amplitude of the signal is
constant over consecutive symbols and shows that the
discriminator function is directly related to the change in
phase between successive symbols. Since the interval be-
tween successive symbols is fixed, the discriminator func-
tion can be interpreted as a frequency error signal.
For DQPSK signals, the Z87200 computes the discrimina-
tor function S
AFC/QPSK
(k) as:
S
AFC/QPSK(
k)=SIGN[Dot(k)] Cross(k) - SIGN[Cross(k)]
Dot(k),
where the above expression can be reduced to the same
as for DBPSK,
S
AFC/QPSK
(k)A2(k)*sin(Ørot(k)).
BPSK/QPSK Modulation
The Z87200 incorporates a Direct Digital Synthesizer
(DDS) to implement its on-chip BPSK/QPSK modulator. In
the Z87200 design, the NCO and thus the sampling clock
for the modulator is driven by fRXIFCLK; for this reason,
both TXIFCLK and RXIFCLK must be common if the on-
chip BPSK/QPSK modulator is to be used. The
BPSK/QPSK modulator can then be used to generate the
transmit output signal at a programmable IF frequency,
thereby eliminating the need for an external modulator.
Because it is a sampled data system like the Downconvert-
er of the Z87200, however, care must be taken to ensure
that the results of aliasing do not adversely affect the out-
put transmit signal.
PS010202-0601

Z8720045FSG

Mfr. #:
Manufacturer:
ZiLOG
Description:
RF Transceiver SS MODEM
Lifecycle:
New from this manufacturer.
Delivery:
DHL FedEx Ups TNT EMS
Payment:
T/T Paypal Visa MoneyGram Western Union