Single Sideband Modulation Conventional double sideband(DSB) modulation can be considered wasteful of power and bandwidth because they contain a carrier signal and two identical sidebands. Conversely, single sideband(SSB) modulation, as the name implies, uses only one sideband to provide the final signal. In other words, SSB provides a considerably more efficient form of communication when compared to DSB modulation. It is far more efficient in terms of the radio spectrum used, and also the power used to transmit the signal.In view of its advantages SSB modulation has been widely used for many years, providing effective communications [29]. In terms of mathematics, let’s illustrate DSB as below : A is carrier frequency, B is baseband(data) frequency. Therefore, DSB modulation, as the name implies, provides two sidebands : (A+B) and (A-B). 1
Transcript
Single Sideband Modulation
Conventional double sideband(DSB) modulation can be considered wasteful of
power and bandwidth because they contain a carrier signal and two identical
sidebands. Conversely, single sideband(SSB) modulation, as the name implies,
uses only one sideband to provide the final signal. In other words, SSB provides a
considerably more efficient form of communication when compared to DSB
modulation. It is far more efficient in terms of the radio spectrum used, and also
the power used to transmit the signal.In view of its advantages SSB modulation
has been widely used for many years, providing effective communications [29].
In terms of mathematics, let’s illustrate DSB as below :
A is carrier frequency, B is baseband(data) frequency. Therefore, DSB modulation,
as the name implies, provides two sidebands : (A+B) and (A-B).
1
As for SSB :
or
(A-B) is lower sideband(LSB), and (A+B) is upper
And as illustrated below, there is exactly
cosine.
B) is lower sideband(LSB), and (A+B) is upper sideband(USB).
elow, there is exactly 90º phase offset between sine and
phase offset between sine and
2
In terms of Unit Circle, the definition of Sine and Cosine are as below :
As illustrated above, Cosine is in-phase, so we call it “I” signal. And Sine is
quadrature-phase, so we call it “Q”.
3
Thus, if we want to generate a (A-B) signal by means of SSB modulation, the
block diagram is as below :
4
I/Q Imbalance
As mentioned above, for SSB modulation, there should be only one desired
sideband in the spectrum in theory. Nevertheless, in reality, there will be at least
three tones in the spectrum. As shown below, one of the three tones is undesired
sideband, so-called image [16].
The undesired sideband, so-called image, resulting from I/Q imbalance [2,4,27].
Especially, in the case of broadband operation, compared to the narrowband case,
the I/Q imbalance among the differential I/Q input channels becomes more
serious and thus brings about the image product, which aggravates the system
performance [18].
5
Direct up-conversion (DUC) transmitter has the inherent advantage of
conceptual simplicity and high integration level [5]. Thus, it has become popular
in recent years [39], especially for handset device, e.g. cellphone.
As illustrated below [18], the DUC, just as its name implies, baseband converts to
RF directly.
6
To assure high signal quality, the ideal IQ modulator would have perfectly
symmetrical in-phase and quadrature arms [16]. That is to say, in theory, the I
and Q channels should have identical gains, and should be exactly 90º out of
phase.
For DUC transmitter, due to the high frequency of the LO, it is not possible to
implement the IQ modulator digitally. Nevertheless, while developers strive for a
symmetrical IQ modulator circuit, manufacturing process variations cause slight
differences between the in-phase and quadrature paths on the same die[16].
Besides, an analog IQ modulator exhibits gain and phase imbalances between the
two branches [29,40].
In other words, in a practical DUC quadrature modulator, the I and Q channels
may have different gains and the LO signals may not be exactly 90◦ out of phase
[21]. The symptom that I and Q channels have different gains is I/Q gain
imbalance, or I/Q amplitude imbalance. And the symptom that I and Q channels
are not exactly 90◦ out of phase is I/Q phase imbalance.
7
Both I/Q gain and phase imbalance are known collectively as I/Q imbalance. In
terms of constellation, as illustrated below :
Of course, due to constellation distortion, I/Q imbalance results in EVM
degradation and degrades modulation accuracy. In addition, as mentioned above,
I/Q imbalance leads to undesired sideband(i.e. image) [2,27]. The amplitude
difference between signal and image is defined as sideband suppression.
8
The following figure shows a plot that can be used to relate sideband
suppression to I/Q gain imbalance and quadrature imbalance. It is notable in this
example that improving the quadrature phase imbalance has no effect on the
sideband suppression unless the gain imbalance is also improved [41].
In other words, generally speaking, I/Q gain imbalance has more effect on
sideband suppression than I/Q phase imbalance.
9
Although the I/Q imbalance is inevitable, we are able to diminish it as much as
possible. We can make use of the fact that the sideband suppression can be
optimized by adjusting phase and amplitude offsets between I and Q channel
[16,18].
As shown in the figure above, in the first pass, the gain delta between I and Q
is adjusted. The sweep yields a null of around −57 dBc for a gain difference of
approximately −0.1 dB. Next, adjust the skew between I and Q. This drives the
null down further to −60 dBc for a phase adjust of −0.05°[41]. In this case, the
first-pass gain adjust yields a deep trough that is only slightly improved during
the phase sweep. The phenomenon proves that I/Q gain imbalance has more
effect on sideband suppression than I/Q phase imbalance again, as mentioned
above. Thus, gain and phase need to be adjusted consecutively in several
steps until the undesired sideband leakage is minimized[16].
10
In terms of frequency domain, the sideband suppression does improve with
adjustment [18].
Besides, in the RF scenario, to alleviate the performance degradation caused by
the image product, attention should be paid to the PCB layout process where the
differential I/Q channels should be identical in their physical layout [18, 40].
According to [36], both Tx I/Q and Rx I/Q signals adopt differential form to avoid
being interfered by outside interference, and then degrading the modulation and
demodulation accuracy.
11
There will be four I/Q signals : I+、I-、Q+、Q-. And the phase relationship is as
shown in the figure below :
Ideally, the four traces on the IQ signal path from the DAC output to the
modulator input should be symmetrical between the I channel and Q channel and
between the positive side and negative side within a channel. In reality, due to
variations of PCB design rules and manufacturing limitations, trace lengths are
not perfectly matched. The mismatches cause the signal in one channel to be
skewed from the other, and, therefore, result in IQ phase errors. Typically there
are two types of trace length mismatches as shown in the figure below [40] :
12
Trace mismatches between I and Q channels degrade IQ phase imbalance.
Mismatches between the positive and negative side in a channel distort a
differential signal by skewing the two sides away from 180° out of phase. This
causes both gain and phase imbalance. Typically, the traces in a differential pair
are laid out very close to each other. Its potential mismatch is relatively small.
However, when the differential pair is long, every time it makes a turn on the PCB,
the external trace adds a little bit more in the total length than the internal one. It
can accumulate to a certain level where the mismatch starts to have an impact on
sideband suppression[40]. Thus, we have to make use of some methods to
alleviate the mismatch caused by turns on the PCB[36].
As mentioned above, the I/Q imbalance is inevitable. In terms of PCB, what we
can do is to try our best to make the four I/Q signals (i.e. I+、I-、Q+、Q-) have
identical lengths. Of course, if possible, make the PCB trace lengths of the four
I/Q signals as short as possible to reduce the potential mismatch. Otherwise, the
sideband suppression will aggravate.
13
Besides, we should consider the effect of temperature as well [17,24]. On the
whole, lower the temperature, more the sideband suppression.
According to[19], higher-order modulation schemes such as 64-QAM are much
more susceptible to IQ gain imbalance. One easy way to visualise this effect
is to observe a constellation plot of varying orders of modulation.
Thus, it is important to minimise gain or phase imbalance when designing an
RFIC that supports complex modulation schemes[19].
14
As illustrated in the figure above, several constellation plots with increasing
orders of modulation and constant gain imbalance. Consequently, in LTE
application, the EVM specifications vary with modulation schemes due to the fact
that 64QAM is the worst case in terms of modulation[42].
15
Carrier leakage
As mentioned above, for SSB modulation, in reality, there will be at least three
tones in the spectrum. One is signal(i.e. desired sideband), another is undesired
sideband(i.e. image), and the other is carrier leakage(i.e. LO leakage). As shown
in the figure below :
Carrier leakage is also known as carrier feedthrough and I/Q origin offset, mainly
results from two factors :
� LO leakage
� DC Offset of I/Q channels.
16
To get low conversion loss from a passive mixer, typically a high LO power is
needed. Due to the finite mixer port to port isolation, and strong LO power, the
LO signal can leak through the RF port, which may result in significant LO
leakage [2,6].
Besides, excessive DC offsets in I/Q channels cause high levels of carrier leakage
as well [3,8,18].
17
In DUC transmitter, with LO, b
DC offsets in I/Q channels,
well.
In terms of constellation, as illustrated below
Of course, due to constellation distortion,
EVM degradation and degrades modulation accuracy.
measurement result screen of CMW500, there is I/Q offset value as well.
As shown in the figure below[28] :
with LO, baseband converts to RF directly. Thus,
DC offsets in I/Q channels, with LO, DC offsets converts to LO leakage directly as
In terms of constellation, as illustrated below [7] :
Of course, due to constellation distortion, DC offsets in I/Q channels
EVM degradation and degrades modulation accuracy. Thus, in the LTE EVM
measurement result screen of CMW500, there is I/Q offset value as well.
below[28] :
aseband converts to RF directly. Thus, if there are
DC offsets converts to LO leakage directly as
DC offsets in I/Q channels result in
Thus, in the LTE EVM
measurement result screen of CMW500, there is I/Q offset value as well.
18
In EDGE application, DC offsets in I/Q channels degrade origin offset suppression
as well [9,40]. Besides, in CDMA application, it affects rho measurement result as
well [3].
19
As mentioned above, in the case of broadband operation, compared to the
narrowband case, the I/Q imbalance becomes more serious. Similarly, the carrier
leakage becomes more serious in the case of broadband operation [25].
As shown in the figure above, in the case of broadband operation and DUC
transmitter, the carrier leakage and signal overlap[22, 25].
20
According to[5], without countermeasures, the carrier leakage stays constant
while the signal is reduced. Therefore, as shown in the figures above, in low
power mode, the carrier leakage is even larger than signal, which degrades SNR.
As illustrated below, EVM varies inversely with SNR :
That is to say, with carrier leakage, EVM begins to exceed the set limit when too
much gain reduction is exercised, less the output power, higher the EVM [5].
21
Thus, in LTE application, the LO leakage specifications vary with output power
[28].
In WCDMA application, the step E and step F of Inner Loop Power Control (ILPC)
need 73 dB dynamic range(-50 dBm ~ 23 dBm).
22
Nevertheless, as shown in the figure below, with carrier leakage, it is impossible
for the output power to be lower than -30 dBm. That is to say, the carrier leakage
may reduce the dynamic range and make ILPC fail.
23
Of course, ideally, without DC offsets in I/Q channels, there should be completely
no carrier leakage, as shown in the formula below :
Or as shown in the figure below[4] :
24
Nevertheless, carrier leakage is inevitable, and can’t be rejected by means of DC
block :
As illustrated above, the DC block is a high pass filter. Before mixer, both the DC
offset and baseband data are rejected. Whereas after mixer, both the DC offset
and baseband data pass the DC block [11, 27]. Thus, DC block is not the solution
to carrier leakage.
25
In order to solve the carrier leakage issue, some transceivers integrate
calibration circuit [11,17].
For example, according to [26,27,43], the BCM4356 of Broadcom integrates LO
feedthrough (LOFT) calibration circuit.
26
The amplitude difference between signal and carrier leakage is defined as carrier
suppression.
In terms of frequency domain, afrter calibration, the carrier suppression
improves indeed [10].
27
In addition, several methods to suppress the carrier leakage have been reported
recently. Balancing techniques are frequently used. As illustrated in the figures
below[12] :
In terms of constellation[12] :
28
In a typical RF transmitter implementation, individual components, including the
digital-to-analog-converter are subject to slight errors in gain and DC offset. Thus,
when considering a DAC or direct quadrature modulator, it is important to apply
gain or DC offset adjustments to the baseband I or Q signals. Take RTR6285A of
Qualcomm for example, after iterative adjustment of the DC offset of the I/Q
differential input, the carrier suppression improves indeed, as shown below
[18,22] :
29
As mentioned above, I/Q imbalance and carrier leakage are inevitable. Thus, the
chip vendors need to measure carrier feed-through and sideband suppression on
the bench and specify them in the datasheet[16]. For example, the RTR6285A of
Qualcomm provides carrier suppression and image suppression measurement
[44].
As mentioned above, for I/Q imbalance, we should consider the effect of
temperature. Similarly, for carrier leakage, we should also consider the effect of
temperature[24].
30
In LTE application, both image suppression and carrier suppression are included
in in-band emission requirements [21,28,45].
31
As mentioned above, in terms of PCB, what we can do is to try our best to make
the four I/Q signals (i.e. I+、I-、Q+、Q-) have identical lengths. Fortunately, this
method improves not only sideband suppression, but also carrier suppression.
According to[36], if the IQ modulator has perfectly symmetrical in-phase and
quadrature arms[16], the DC offset will cancel.
This is the reason why we should try our best to make the four I/Q signals have
identical PCB layout trace lengths. Of course, similarly, if possible, make the PCB
trace lengths of the four I/Q signals as short as possible.
32
Besides, don’t short those unused I/Q pins to ground. For example, the
MDM9X35 of Qualcomm :
As shown above, TX_DAC1_IP, TX_DAC1_IM, TX_DAC1_QP, TX_DAC1_QM, these
four pins are unused. Because these I/Q pins are all related within chip. If you
short the four unused pins to ground, the DC offset on the ground will flows into
these unused pins, then leakages to the used I/Q pins and generates carrier
leakage.
33
According to[5], undesired sideband and carrier leakage are the inherent
shortcomings of DUC transmitter. As illustrated in the detailed figure below [22] :
34
Reference
[1] Optical modulation with a single sideband and carrier suppressed
[2] EVM estimation by analyzing transmitter imperfections mathematically and
graphically
[3] Understanding CDMA Measurements for Base Stations and Their Components
[4] Quadrature Mixer LO Leakage Suppression Through Quadrature DC Bias
[5] Carrier Leakage Suppression in Direct-Conversion WCDMA Transmitters
[6] Cancellation Techniques for LO Leakage and DC Offset in Direct
Conversion Systems
[7] IQ Offset (GSM/EDGE/EDGE Evolution)
[8] Understanding GSM/EDGE Transmitter and Receiver Measurements for Base
Transceiver Stations and their Components, KEYSIGHT
[9] A 65nm CMOS Low-Noise Direct-Conversion Transmitter with
Carrier Leakage Calibration for Low-Band EDGE Application
[10] A Carrier Leakage Auto-Calibration Circuit with a Direct DC-Offset
Comparison Technique for a WiMAX Transmitter
[11] Local Quadrature Signal and Carrier Leakage Calibration Techniques
for a Mobile-WiMAX Transceiver
[12] Direct Carrier Six-Port Modulator Using a Technique to Suppress Carrier
Leakage
[13] Investigation of LO-leakage cancellation and DC-offset influence on
flicker-noise in X-band Mixers
[14] An Overview of Transmitter Calibration Techniques
[15] WTR4905 Wafer-level RF Transceiver, Qualcomm
35
[16] Characterization of IQ Modulators Counts On Flexible Signal Generator
Stimulus
[17] A carrier leakage calibration and compensation technique for wideband
wireless transceiver
[18] Design of a Broadband MIMO RF Transmitter for Next-generation
