Upload
jatayu2011
View
216
Download
1
Embed Size (px)
Citation preview
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD
Radio System Design II
Transceiver Architectures
Prof. Bhaskar Banerjee
EERF 6330- RF IC Design
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD
Outline
• Receiver Architecture
• Transmitter Architectures
• Reading:
✓Chapter 4: Fundamental of Microelectronics, B. Razavi
2
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 3
Frequency Planning
• depends directly on – receiver topology– number of down-conversions– mode of operation
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 4
Frequency Planning
• Blockers– Understanding the wireless applications that co-exist in the frequency spectrum
surrounding the band of interest is needed– Operating close to the frequency bands of other applications places great
demands on front-end filter selectivity
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 5
Frequency Planning
• Blockers– Solution by frequency planning
• IF frequency selection–to avoid blockers that can interfere with the IF chain–Must check the availability of IF surface-acoustic wave
(SAW) filters at the chosen frequency
– Another solution• Filtering
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 6
Frequency Planning
• Spurs and De-Sensing– Spurs
• unwanted spurious frequencies that are generated by interaction between various components of our own transceiver
– De-sensing• Spurs with a higher power level lands either directly in the
band or adjacent to the band and saturates the transceiver
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 7
• Transmitter leakage– a major concern for any RF subsystems, especially in duplex systems
• de-sense the receiver by saturating the receiver front-end or causing oscillations
• makes it difficult to integrate both transmit and receive functions of a duplex system on a single chip
– Solution• stringent filtering/isolation to maintain confinement to the transmit band
Frequency Planning
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 8
Heterodyne Receivers
• Also called Super-Heterodyne• have been in use for a long time and are quite popular from
the early days of radio communication systems • two (or more) stages of down conversion
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 9
LO Leakage
• LO leakage– LO signals and their harmonics are major sources of spurious interference– Leakage paths
• IC substrate, package, or board
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 10
LO Leakage
• LO leakage– Solution by frequency planning
• Decide LO frequencies where all LO frequencies and their harmonics, and frequencies resulting from higher-order mixing of these signals do not fall in the RF or IF bands
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 11
Problem of Image Frequency
• Image frequency– One of the most problematic issues with designing traditional
heterodyne receivers – It is located on the opposite side of the LO frequency and folds
on top of the IF band as the signal in down converted in a mixer– It needs to be addressed by either filtering or image-reject mixing
topologies
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 12
Why is “Image” Frequency Critical?
• Down conversion thru a mixer generates f1-f2 component.
• ‘Image frequency(f3)’ has equal spectral distance from the LO(f2) to the RFin(f1) on the opposite side.
• Mixer generates an unwanted signal placed exactly at the same frequency as the down converted IF signal, where f1-f2 = f2-f3.
• Unwanted signal located on the ‘Image frequency’ CANNOT be filtered out once down converted to the IF.
Pass Mixer
Pass Mixer
Pass Mixer
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 13
Band-pass filter used for ‘image-rejection’.
Image-reject Filter + Mixer
• Regardless of the existence of an Image Frequency, Image Filter is always placed BEFORE a Mixer to protect the RFin signal from possibly existing spurs at ‘image’ frequency.
• Image filter prevents the unwanted spurs at image frequency from being down converted to the IF frequency.
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 14
What IF frequency to choose?
• With higher IF frequency, subsequent channel-select BPF requires “higher Q” filter due to higher frequency, resulting in lower selectivity for given Q-factor of the channel-select BPF.
• On the contrary, lower IF frequency provides better channel-selectivity on the subsequent channel-select BPF block.
• Hence, the choice of IF frequency has a trade-off between channel-selectivity and image-rejection.
• Dual IF stage can be used to achieve both benefits at higher costs.
With lower IF, BPF cannot effectively filter out the image frequency from the
RF signal.
With higher IF, BPF can effectively filter out the Image resulting in clean IF
signal.
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 15
Typical IF frequency choice
• In reality, IF frequency choice is limited by commercially available SAW filters.
• Specific application tend to use specific IF frequency value:
Typical IF frequency usedTypical IF frequency used
455kHz General equipments
10.7MHz General purpose receivers
21.4MHz Hi-performance receivers
45MHz TV’s, Cellular phones
70MHz Satellite TV’s, Military
160MHz Satellite equipments
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 16
Problem of Half IF
• Half IF frequency– It is located directly between the LO and the RF – It can double in the LNA or RF amplifiers in the front-end and get
down-converted into the IF band by mixing with the second harmonic of the LO signal
Spurious signal
12 · (!RF � !LO)
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 17
What LO frequency to choose?
• For a down mixer, two local oscillator frequency can be chosen. A Low Side LO (LSLO) and a High Side LO (HSLO). Both will down mix the RF signal to the desired IF frequency.
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 18
Heterodyne Receivers
• Selectivity: Lower Q Required
• Sensitivity: Reverse Isolation on IF stage
• Stability: Lower Gain per stage required
• Repeatability: Reuse IF stage and below, multiple carrier support.– e.g. Dual-IF Topology
• Not fully integrable
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 19
Image Reject Receivers
• To maintain an acceptable receiver signal quality, most modern wireless standards require 60 to 90 dB of image rejection
• The traditional method of image rejection is use of band stop filters at the image frequency
• Due to the stringent requirements, image rejection is typically performed through a combination of filtering and image rejection mixing techniques
• The two most popular image rejection techniques are – Hartley architecture – Weaver architecture– These techniques yield about 30 to 35 dB of image rejection
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 20
Image Reject Receivers• Hartley Architecture
– RF signal is down-converted by quadrature LO signals – Problem
• sensitivity to phase and amplitude imbalance between the quadrature mixers or the two IF paths
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 21
Hartley Architecture
• Drawbacks• Sensitivity to Gain and Phase Matching• Linearity requirement on the Adder circuit
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 22
Image Reject Receivers
• Weaver Architecture– Utilizes an additional pair of mixers to perform the phase shifting prior to IF
combination – Problem
• a second image → zero IF for the second mixing• sensitivity to phase and amplitude imbalance
sinω1t
cosω1t
sinω2t
cosω2t
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 23
• Also called Homodyne Receivers and Zero-IF Receivers• The desired signal is directly down converted to baseband
Direct Conversion Receivers (DCR)
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 24
Direct Conversion Receivers
• Main Features– Integration possible
• No Image reject filter, LNA can drive the Mixer directly• IF SAW filter - replaced by LPF and Baseband Amps
– Channel selection harder
– Flicker Noise is a big issue
– have been proposed in the early radio days, but the system performance has been usually poor
– With the advances in the digital communication area, a lower performance can be accepted with the advantage of highest possible on chip integration
– DCR is a demodulator with much higher dynamic range and higher frequency of operation than a typical IF demodulator
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 25
Direct Conversion Receivers
• The order of the baseband building blocks arrangement(Low Pass Filter, Variable Gain Amplifier, Analog Digital
Converter)– Depending on the chosen configuration, the linearity and noise
performance of the front-end and baseband blocks can be defined → need to find optimum order
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 26
Direct Conversion Receivers• The order of the baseband building blocks arrangement
– By placing the low-pass filter (LPF) in the front we can filter much of the extraneous signals that result from the mixing and relax the linearity requirements of the variable-gain amplifier (VGA) and the analog-to-digital converter (A/D)
– However, the loss through the LPF reduces the gain before the VGA and A/D and consequently increases the overall noise figure (NF) of the receiver
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 27
Direct Conversion Receivers
• The order of the baseband building blocks arrangement – As we move the LPF further down the baseband chain the NF
improves while the linearity requirements of the VGA and A/D increase
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 28
Direct Conversion Receivers• Frequency schemes of DCR and heterodyne receivers
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 29
Issues in Direct Conversion Receivers
• DC Offset– Extraneous DC voltages in the demodulated spectrum of a DCR not only
corrupt the output, but also propagate through the baseband circuitry and saturate the subsequent stages
– DC offsets are mostly generated through • self-mixing the LO signal → time-varying DC offset• mismatch in the mixers → constant DC offset
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 30
Issues in Direct Conversion Receivers
• Phase and Amplitude Imbalance – Most modern wireless modulation scheme requires I and Q signal
separation and demodulation to fully recover the information– implementing accurate phase-shifters at higher frequency becomes a
challenging task
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 31
Issues in Direct Conversion Receivers
• LO Leakage and Radiation – Since a typical DCR requires a LO signal frequency identical to the RF
input carrier frequency, the LO signal is considered an in-band interference – This LO can couple into the antenna and not only radiate out in the receive
band of other users but also penetrate and saturate the RF front-end – Solution:
• High reverse isolation in the front-end and good shielding of the receiver
• sub-harmonic mixing by moving the LO signal out of the RF frequency band of interest
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 32
Issues in Direct Conversion Receivers
• Solution for DC Offset– AC coupling of the mixer output.
• This will not only remove the unwanted DC offsets, but at the same time will corrupt the down-converted signal by attenuating the components near DC → not acceptable for demodulating most modulation schemes that exhibit a DC peak in their signal spectrum
– Therefore, DC offset cancellation techniques using DSP are necessary to accommodate the use of direct conversion topology in today’s wireless applications
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 33
Issues in Direct Conversion Receivers
• Intermodulations– susceptible to both odd- and even-order intermodulations– even-order intermodulations
• by the non-linearities of mixer and LNA• Solution:
– Balanced topologies – even-harmonic mixing
Fundamental
Frequency890, 900
Harmonic
1780 (2*890), 1800 (2*900),2670 (3*890), 2700 (3*900)
2nd IM1790 (890 + 900),
10 (900 - 890)
3rd IM
2680 (2*890 + 900),2690 (890 + 2*900),880 (2*890 - 900), 910 (2*900 - 890 )
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 34
Other Receiver Architectures
• Digital-IF Receivers– Heterodyne - but 2nd IF is digitized– Digital processing alleviates I/Q mismatch problem– A/D Converter performance needs to be very good
• Low thermal, quantization noise• Low non-linearity, high dynamic range
• Sub-sampling Receivers– Sampling at 2Δf - where Δf is the bandwidth of the narrowband
RF signal• Aliasing of Noise
• Digital Radio– Digitize at the RF Front End
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 35
Transmitter Architectures
• Much relaxed requirements:– Noise– Interference Rejection– Band Selectivity
• Direct Conversion Transmitters• Two-step Transmitters
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 36
Direct Conversion Transmitter
• Transmit Carrier Freq = LO Frequency– Modulation and upconversion occur in same circuit.
sinωct
cosωct
PA MatchingNetwork
Duplexer
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 37
Direct Conversion Transmitter
• Major Drawback: PA output disturbs transmit VCO
ωLO
Leakage: Injection pulling
sinωct
cosωct
PABPF
Bhaskar Banerjee, EERF 6330, Sp‘2013, UTD 38
Transmitter Architecture
• Direct Conversion– Frequency Pulling
• Can be alleviated by “offsetting” LO freq (Use f1 + f2)• Use Two-step architecture
• Two-step Architecture– Quadrature mixing at low frequency
• Lesser mismatch - lesser cross-talk• Channel filter to improve adjacent channel rejection
– BPF needed to reject large unwanted side-band• High center frequency - ‘off-chip’ filtering (expensive)
– Higher Power Consumption owing to more components