Documente Academic
Documente Profesional
Documente Cultură
RFICs, ECE-5022
DATE RECEIVED:
Section A: 16-QAM Modulator and Receiver
[Question 1.1] A random bit stream is used to model the data of interest. The bits are split into two channels,
using every other bit, to form parallel bit streams. Each channel is then converted to a two bit waveform,
represented by an amplitude modulated signal with four levels. A quadrature modulator is implemented by
using two mixers and two LO signals (at the carrier frequency) with a 90 phase difference. The two data
waveforms are then up-converted and combined in quadrature. The quadrature amplitude modulated (QAM)
signal is then passed to the transmission channel. A QAM demodulator will use the reverse process to recover
the data.
[Question 1.2] As shown in Figure 2, the necessary ratio of the bit energy to the noise spectral density (Eb/No)
to achieve a bit error rate of 10-5 is determined to be 13.7 dB. Using the expression
13.7
= = 10 log (10 10 (4)),
where R is the symbol bit rate and B is the bandwidth (or 1/T for digital systems) the corresponding SNR
necessary to maintain the desired bit error rate is calculated to be 19.7 dB.
[Question 1.4] Since each symbol requires four bits, the symbol period is TS = 4*TB = 4 s. In Figure 4 we see
the symbol rate of 4 s as expected. As shown in Figure 5, the main lobe bandwidth noticeably decreases from
0.25 MHz to 0.17 MHz from the input to the output of the filter. The unfiltered bandwidth is one sided in the
baseband and is 1/4 of the data rate as expected. This is expected since the bandwidth efficiency of 16QAM is
four. The side lobes are also significantly suppressed by the RRC filter.
Higher roll-off factor of the RRC filter will reduce the side lobe levels. The bandwidth of the spectrum after the
filter can be calculated by using BW = RS(1+R), where RS is the symbol rate (1 Mbps), R is the root raised
cosine filter roll-off factor ( = 0.35):
It is apparent from the equation above that increasing the RRC filter roll-off increases the channel bandwidth in
the time domain, however as shown in Figure 5, the bandwidth in the frequency domain is decreased.
Figure 4 Even and odd bit waveforms for 16-QAM, 1Mbps data rate
3
Figure 5 RRC filter input (blue) and output (red) spectra
[Question 1.5] The roll-off of the RRC pulse shaping filter was chosen as 0.35 to provide a reasonable tradeoff
between side-lobe reduction and bandwidth for the filtered pulses. Figure 5 shows the reduction in the side-lobe
levels of the data signal.
Since the LO frequency is 850 MHz, the mixers output should appear very close to 850 MHz. Consequently
the BPF center frequency is set to 850 MHz and the bandwidth is set to cover the bandwidth of the data signal
after the RRC filtering. In order to maintain a Q-factor between 20 and 30, the bandwidth of the BPF was
chosen to be 40 MHz. The 40 MHz bandwidth was centered around 850 MHz.
The loss of the BPF is chosen to provide an output signal strength of 0dBm. Two filter reactances are used to
control the filter roll-off. More reactances in the filter mean a sharper roll-off in the stop band.
[Question 1.6] The spectrum of the transmitted signal (output of the PGA) is shown in Figure 6. The loss of the
BPF and gain of the transmitter PGA was adjusted such that the transmitted signal power is approximately 0
dBm.
4
[Question 1.7] The constructed receiver demodulates the signal using commercial off-the-shelf (COTS)
components. The SKY67101-396LF LNA was chosen for its low noise figure of 0.57 dB and operating
frequency from 0.4 1.2 GHz. This LNA operating frequency must be in the range of the image reject BPF
used in the receiver. An ideal image reject BPF with 2 dB of loss and a 2 MHz bandwidth was used after the
LNA component to remove the image, which has an offset of twice the intermediate frequency (IF) from the
desired channel signal. The ADL5801 RF mixer was used to down-convert the signal to baseband. The local
oscillator (LO) was modelled using and ideal tone source. Finally, a HMC960LP4E power amplifier was used
to amplify the IF signal for detection. See attached Appendix for component datasheets. From the received
signal spectrum it is apparent that the commercial components functioned well, yielding an SNR of 47 dB
which meets the EVM and BER requirement.
[Question 1.8] Figure 8 shows the 16-QAM receiver constellation with 16 distinct constellation points. Since
the constellation points are well defined, we can infer the BER is low and the SNR is high. The transmitted
signal constellation maintains the 16 constellation points, however the magnitude is significantly reduced, as
expected from the channel loss of 90 dB. Nonetheless, the constellation points are still relatively well defined,
indicating the chosen receiver components are functioning well. The received signal SNR of 47 dB meets the
SNR requirement calculated in Question 1.2, and consequently the BER requirement of less than 10-5 is met. It
is noted that increasing the isolation on the mixer improves the receive signal BER. Specifically, better isolation
of the LO from the output has most impact on BER.
Figure 8 Modulated signal constellation before transmit channel (left) and after PGA (right)
5
[Question 1.9] See the attached pages for the full system schematic for the Part A simulations.
[Question 2.1] To examine the effect of IQ amplitude and phase mismatch of the transmitted signal, an IQ
imbalance sub circuit was implemented in the modulator. As shown in Figure 9, the 0.5 dB IQ amplitude
mismatch slightly compressed the constellation diagram. Even with the slight compression, the SNR, and
consequently the EVM and BER is still acceptable, as pictured in the received signal spectrum of Fig. 9. To
exaggerate the effect of IQ amplitude mismatch, a 5 dB mismatch was introduced to the transmission circuit and
the constellation diagram is plotted in Figure 10. It is apparent that the amplitude mismatch compresses the Q-
axis of the constellation proportional to the level of mismatch.
Figure 9 Constellation diagram w/ 0.5 dB amplitude mismatch (left) and resulting SNR
6
[Question 2.2] Next a 1 IQ phase mismatch was introduced to the transmission system to examine the resulting
BER and EVM. As shown in Figure 11, the effect of the 1 IQ imbalance is minimal and the BER, EVM, and
SNR specifications are still met.
To exaggerate the effect of IQ phase mismatch, a 20 mismatch was introduced to the transmission circuit and
the constellation diagram is plotted in Figure 12. It is apparent that the phase mismatch introduces a rotational
distortion to the constellation that is proportional to the level of mismatch.
[Question 3.1] After introducing phase noise to the LO in the receiver, the received signal constellation is
significantly distorted as shown in Figure 16. Consequently, the BER of the received data significantly
increased. The addition of phase noise in the LO signal causes symbol rotation and results in shifting of
adjacent constellation points. This results in the observed increase in the BER that does not meet the 10-5
specification.
7
Figure 13 Configuring phase noise in the receiver LO source (left) resulting signal constellation
[Question 4.1] The system is shown to tolerate up to a 1.4 dB IQ amplitude mismatch while still meeting the
BER/SNR specification, as shown in Figure 14.
BER
ID=BER5
VARNAME="AMPLITUDE"
VALUES=stepped(0,6,0.5)
OUTFL=""
BER
Figure 14 Varying the IQ amplitude mismatch and measuring BER (< 1.4 dB meets spec)
[Question 4.2] The system is shown to tolerate up to a 2.75 IQ phase mismatch while still meeting the
BER/SNR specification, as shown in Figure 15.
[Question 4.3] Next the IQ amplitude and phase mismatch were set to zero and an 850 MHz jamming signal
was introduced at the receiver input. The receiver system is shown to tolerate up to a -59.42 dBm jamming
signal while still meeting the BER/SNR specification, as shown in Figure 16.
8
BER
ID=BER5
VARNAME="PHASE"
VALUES=stepped(0,20,2)
OUTFL=""
BER
Figure 15 Varying the IQ phase mismatch and measuring BER (< 2.75 meets spec)
BER
ID=BER5
VARNAME="JAM_POWER"
VALUES=stepped(-80,-30,5)
OUTFL=""
BER
Figure 16 Varying the 850 MHz jamming signal power and measuring BER (< -59.42 dBm meets spec)
9
ECE-5022 Final Project
16 QAM Modulator and Receiver System Design
The Ohio State University, AU 2014
By: Stephen Watt
Date Submitted: December 11, 2014
- IQ Imbalance
- AGWN
- Path Loss
- Jamming Signal
From 16-QAM IQ
Modulator
From Rx PGA
From IQ Modulator
DATA SHEET
Skyworks Solutions, Inc. Phone [781] 376-3000 Fax [781] 376-3100 sales@skyworksinc.com www.skyworksinc.com
201266I Skyworks Proprietary Information Products and Product Information are Subject to Change Without Notice July 10, 2014 1
High IP3,
10 MHz to 6 GHz, Active Mixer
Data Sheet ADL5801
FEATURES FUNCTIONAL BLOCK DIAGRAM
VPLO GND NC IFON IFOP GND
Broadband upconverter/downconverter 24 23 22 21 20 19
Power conversion gain of 1.8 dB
Broadband RF, LO, and IF ports
GND 1 18 VPRF
SSB noise figure (NF) of 9.75 dB ADL5801
Input IP3: 28.5 dBm GND 2 17 GND
08079-001
7 8 9 10 11 12
VPLO GND ENBL VSET DETO GND
Radio link downconverters
Broadband block conversion Figure 1.
Instrumentation
GENERAL DESCRIPTION
The ADL5801 uses a high linearity, doubly balanced, active The balanced active mixer arrangement provides superb LO-to-
mixer core with integrated LO buffer amplifier to provide high RF and LO-to-IF leakage, typically better than 40 dBm. The IF
dynamic range frequency conversion from 10 MHz to 6 GHz. outputs are designed to provide a typical voltage conversion
The mixer benefits from a proprietary linearization architecture gain of 7.8 dB when loaded into a 200 load. The broad
that provides enhanced input IP3 performance when subject to frequency range of the open-collector IF outputs allows the
high input levels. A bias adjust feature allows the input linearity, ADL5801 to be applied as an upconverter for various transmit
SSB noise figure, and dc current to be optimized using a single applications.
control pin. An optional input power detector is provided for The ADL5801 is fabricated using a SiGe high performance IC
adaptive bias control. The high input linearity allows the device process. The device is available in a compact 4 mm 4 mm,
to be used in demanding cellular applications where in-band 24-lead LFCSP package and operates over a 40C to +85C
blocking signals may otherwise result in degradation in dynamic temperature range. An evaluation board is also available.
performance. The adaptive bias feature allows the part to provide
high input IP3 performance when presented with large blocking
signals. When blockers are removed, the ADL5801 can auto-
matically bias down to provide low noise figure and low power
consumption.
For price, delivery and to place orders: Hittite Microwave Corporation, 2 Elizabeth Drive, Chelmsford, MA 01824
978-250-3343 tel 978-250-3373 fax Order On-line at www.hittite.com
1
Application Support: apps@hittite.com