Introduction to CMOS RF Integrated Circuits Design

Size: px

Start display at page:

Download "Introduction to CMOS RF Integrated Circuits Design"

Jody Floyd
6 years ago
Views:

1 II. RFIC System Overview Fall 0, Prof. JianJun Zhou II-

2 Outline Introduction RF Transceiver rchitectures RF System Considerations Sensitivity and Selectivity Noise Figure Dynamic Range -db CP and IP Fall 0, Prof. JianJun Zhou II-

3 RFIC is nalog Circuit KCL KVL Ohm s Law Current Source Voltage Source R L C Wire Transistors Fall 0, Prof. JianJun Zhou II-

4 Unit-length Inductors LC L > L > L > L > L 4 > L 5 > L o f < f <f < f < f 4 < f 5 < f o Fall 0, Prof. JianJun Zhou II-4

5 Capacitors h w C wh d d d Unit length C > C > C > C > Co f < f < f < f < fo Fall 0, Prof. JianJun Zhou II-5

6 Challenges in RF IC Design RF IC Designs = Device Models + Simulators + Experience RF IC Designer = nalogue circuit designer (Simulation) + Component maker (layouts) +System designer Fall 0, Prof. JianJun Zhou II-6

7 RFIC Development Flow Target System Spec RF System Design System partitioning & sub-block sepcs Device Model High-cost design: Manpower ED tool Measurement systems Test chips (~5 rounds) RF Library Circuit Design 8~4 Months PCB Design for Evaluation Layout Design and DRC / LVS, parasitic extraction (Wafer Process) (ssembly Process) Design Rule On-wafer measurement Evaluation OK NG Mass Production Transfer Fall 0, Prof. JianJun Zhou II-7

8 General Design Considerations Portable Universal Multi-functions Low Costs Miniaturization Low-power From the customers point of view Fall 0, Prof. JianJun Zhou II-8

9 Broad-knowledge Required Device Physics EMC/EMI Material Packaging nalog Circuit Microwave Theory Transceiver rchitectures IC Design nalog/rf RF IC Design IC Designer Wireless Standard Semiconductor Processes Comm. Theory CD Tools Distributed Parameters Fall 0, Prof. JianJun Zhou II-9

10 Teamwork Required Global Standard Customer Wireless System Spec / Evaluation Packaging RF System/Circuit Design RF Circuit Layout/Parasitic extraction Digital Design Device Model Design Rule / CD / Library Foundry Fall 0, Prof. JianJun Zhou II-0

11 Trade-Offs Wide-band Low-Noise Low-Power Consumption High-linearity Wide-dynamic range High-frequency Low-Voltage High-Gain Fall 0, Prof. JianJun Zhou II-

12 Outline Introduction RF Transceiver rchitectures RF System Considerations Sensitivity and Selectivity Noise Figure Dynamic Range -db CP and IP Fall 0, Prof. JianJun Zhou II-

13 Multi-Mode Mode Wireless Future receivers need Co-exist with transmitters of different standards. (i.e. simultaneous operation SoC/SiP) Fall 0, Prof. JianJun Zhou II-

14 (Super-) Heterodyne Receiver Out-of-band rejection BPF LO BPF Image problem (st down-conversion) BPF st channel selection (nd down-conversion) BPF4 nd channel selection *Razavi Fall 0, Prof. JianJun Zhou II-4

15 Out-of of-band Interference Rejection (Band( Selection) Out-of-band In-band Out-of-band Fall 0, Prof. JianJun Zhou II-5

16 Image Problem in Heterodyne System Fall 0, Prof. JianJun Zhou II-6

17 Image Rejection by Filter Fall 0, Prof. JianJun Zhou II-7

18 Channel Selection Filter Fall 0, Prof. JianJun Zhou II-8

19 Image Rejection and Channel Selection High Q Filter High Q Filter Fall 0, Prof. JianJun Zhou II-9

20 In-band Interference () Fall 0, Prof. JianJun Zhou II-0

21 In-band Interference () Near-Far Problem Comes from Wandering into djacent Cells Fall 0, Prof. JianJun Zhou II-

22 In-band Interference () How does Non-linear mplifier (IMD) affect the signal? Fall 0, Prof. JianJun Zhou II-

23 Receiver : Heterodyne dvantages (High Selectivity) Relaxation of linearity requirements due to the use of IF SW BPF (High Sensitivity) Less DC-offset impairment, Easier I/Q match at lower frequencies Disadvantages Bulky off-chip RF/IF SW BPFs good frequency plan is essential Image problem Half-IF spurious response at lower IF frequencies Need at least two LO sources Integration level is low due to filter Fall 0, Prof. JianJun Zhou II-

24 Receiver : Direct-Conversion (Zero-IF) dvantages No Image or half-if issues High level integration and lower cost (No IF filters) Disadvantages DC offset problems are extremely challenging (IM/IP) LO leakage re-radiation (LO pulling) /f noise (CMOS) can substantially corrupt the D/C signal Even-order distortion of great concert More difficult I/Q match at RF frequencies Fall 0, Prof. JianJun Zhou II-4

25 DC offset Fall 0, Prof. JianJun Zhou II-5

Receiver : Low-IF dvantages Integration of channel filters is possible Less susceptible to /f noise and DC offsets (C coupling) Low-frequency IM product can be easily blocked.

26 Receiver : Low-IF dvantages Integration of channel filters is possible Less susceptible to /f noise and DC offsets (C coupling) Low-frequency IM product can be easily blocked. Disadvantages Image is still a problem, which entails precise I/Q match Complex signal processing is essential to obtain necessary selectivity Fall 0, Prof. JianJun Zhou II-6

27 Receiver: Image Rejection Low-IF To achieve 0dB Image rejection IQ amplitude imbalance is less than 0.5dB IQ phase imbalance is less than.5 degree Fall 0, Prof. JianJun Zhou II-7

28 Receiver: Image Rejection Low-IF (Digital) dvantages Digital signal process avoids the problem of I/Q mismatch\ Less susceptible to process variations Disadvantages DC performance is a great concern Fall 0, Prof. JianJun Zhou II-8

29 Zero-IF Receiver Channel Selection IP Fall 0, Prof. JianJun Zhou II-9

30 Transmitter : Heterodyne Fall 0, Prof. JianJun Zhou II-0

31 Transmitter: Direct Conversion Fall 0, Prof. JianJun Zhou II-

32 No interference Fall 0, Prof. JianJun Zhou II-

33 With Interference (LO Pulling) Fall 0, Prof. JianJun Zhou II-

34 Indirect VCO Frequency (Sub-harmonic LO) - Or using sub-harmonic modulator or mixer Fall 0, Prof. JianJun Zhou II-4 *PMC006

35 Duplexer Freq. Response Fall 0, Prof. JianJun Zhou II-5

36 P Leakage to Rx MHz Rx Tx TDD TRx MHz, FDD TRx Fall 0, Prof. JianJun Zhou II-6

37 Desensitization Through Compression Fall 0, Prof. JianJun Zhou II-7

38 TRx rchitecture Selection Fall 0, Prof. JianJun Zhou II-8

39 Outline Introduction RF Transceiver rchitectures RF System Considerations Sensitivity and Selectivity Noise Figure Dynamic Range -db CP and IP Fall 0, Prof. JianJun Zhou II-9

40 System Considerations Sensitivity Selectivity Spurious Rejection Dynamic Range IM Rejection Frequency EMI stability Linearity Fall 0, Prof. JianJun Zhou II-40

41 Sensitivity RF Receiver sensitivity: quantifies the ability to respond to a weak signal. Defined as the minimum detectable signal power level, satisfying the requirement of the specified signal-to-noise ratio (SNR) for an analog receiver and bit-error-rate (BER) for a digital receiver. Fall 0, Prof. JianJun Zhou II-4

42 dbm dbm 0 log (mw)=0log(w) 0dB Boltzmann constant k = JK - Room temp=00k 00W/Hz kt.80 = 7.8 dbm / Hz Fall 0, Prof. JianJun Zhou II-4

43 Equations Noise Floor P (dbm) ktb(dbm) F (db) Sensitvity P (dbm) ktb(dbm) F (db) SNR (db) in,min nf receiver 74 dbm / Hz 0logB F SNR receiver receiver min min Desired Signal Receiver dded Noise Receiver Thermal Noise Fall 0, Prof. JianJun Zhou II-4

44 Selectivity The ability to reject unwanted signals on adjacent channels (channel selectivity) and/or the outside of the wanted band (band selectivity). 70 to 90 db rejections are normally required Desired RF band Wanted Channel Desired RF Band f Band selectivity Channel selectivity Fall 0, Prof. JianJun Zhou II-44

45 Spurs and Intermodulation Spurious Response Rejection The ability to reject undesired channels to reduce the interference. Rejection of 70dB to 00 db is usually required for wireless communications; Intermodulation (IM) Rejection The receiver has the tendency to generate its own on-channel interference from one or more RF signals due to the nonlinearity of the receiver. These interference signals are called IM products. Greater than 70 db rejection is desirable Fall 0, Prof. JianJun Zhou II-45

46 Others Frequency Stability Stable frequency operation is important in order to capture the desired frequency channel. PLL/synthesizers are commonly employed to obtain an accurately controlled LO frequency. EMI: Electromagnetic Interference From one part to another part within an RF front-end receiver or from interconnects as well as the silicon substrates Fall 0, Prof. JianJun Zhou II-46

47 Noise Figure Signal-to-noise ratio (SNR): ratio of the signal power to the total noise power SNR wanted signal power unwanted noise power Noise figure is a figure of merit quantitatively specifying how noisy a system/component is. The noise factor F is defined for the two-port network: F SNR SNR in put out put NF=0log(F) (db) S S i o / / N N i o Fall 0, Prof. JianJun Zhou II-47

48 Noise Figure cont. S i N i F G N n S o N o S Two-port system with power gain G, added noise power N n and the noise factor F F o GS S / i N N N i i o GSi / No GNi o Nn GN GN i i N N o n (W) FGN i Fall 0, Prof. JianJun Zhou II-48

49 Cascaded Noise Figure S i F G N n F G N n F G N n F m G m N nm S o N i N o m systems in cascade NF contribution Friis equation: F F F F m total F G GG GG GGm Fall 0, Prof. JianJun Zhou II-49

50 Dynamic Range P in P out G (db) For an RF system, operation is normally in a region where the output power is linearly proportional to the input power, while the coefficient is the desired power gain. This region is called as the dynamic range (DR). DR is the rang between the maximum power level that the system is still in linear region to the minimum detectable signal (MDS) power level The range could be specified in terms of input power or output power. Higher DR is desirable Fall 0, Prof. JianJun Zhou II-50

51 Nonlinear Effects It is desired that no matter how high the input signal power is, the output power will be the linearly amplified input signal. Nonlinearities often exist in practical systems and lead to interesting phenomena, those phenomena limit the linear operating range of a system. For simplicity, the output input relationship can be approximately modelled as (Taylor Series expansion): y( t) a a x( t) a x ( t) a x ( t) o y(t) is the output and x(t) is the input signal. a o is the DC component, a the gain, a and a (less than zero) the coefficients of the second and third-order nonlinear terms. Fall 0, Prof. JianJun Zhou II-5

52 -db Compression Point P in,-db = P out,-db G + db Linear relationship DR = P in,db MDS (db) db P out (db) DR - db - Noise Floor MDS P in (db) P in,db Gain-compression of a realistic RF system Fall 0, Prof. JianJun Zhou II-5

53 Fall 0, Prof. JianJun Zhou II-5 -db Compression Point: Equations db Compression Point: Equations ) cos cos cos ( ) cos cos cos ( ) cos cos cos ( ) ( t t t a t t t a t t t a a t y o o o o o o o Fundamental components: t a a t a a t a a cos )]} ( 4 [ { cos )]} ( 4 [ { cos )]} ( 4 [ {

54 -db CP If 0 is the desired signal then the gain will be, a decreasing gain because of a <0. If the unwanted signal strengths and are so strong, the gain of the wanted signal drops to or lower when: a 0 a Now the wanted signal is blocked by the unwanted strong signal, because the wanted signal cannot be amplified by the RF section. Many RF sections in wireless applications must be able to withstand blocking signals 60 to 70 db stronger than the wanted signal Fall 0, Prof. JianJun Zhou II-54

55 db CP vs a The -db compression point can be obtained from three-tone for the wanted channel as (assuming input tones are at the same power): 5 0 log( a db ) 0 log( a db a db ) (db) 4 Or a a a 0.09 or 0.09 db db a Thus, from the measured linear gain a and the input level at the - db compression point, one can calculate the nonlinear coefficient a Fall 0, Prof. JianJun Zhou II-55

56 Intermodulation Intermodulation or intermodulation distortion (IMD), or intermod for short, is the result of two or more signals of different frequencies being mixed together, forming additional signals at frequencies that are not, in general, at harmonic frequencies (integer multiples) of either. Intermodulation should not be confused with general harmonic distortion. Intermodulation specifically creates non-harmonic tones ("offkey" notes, in the audio case) due to unwanted mixing of closely spaced frequencies. Fall 0, Prof. JianJun Zhou II-56

57 IMD in a -Tone Case Desired Channel IM RF Section djacent Channels Fall 0, Prof. JianJun Zhou II-57

58 Fall 0, Prof. JianJun Zhou II-58 There are IM effects between any two channels others } ) cos( ) cos( { 4 } ) ]cos( [ ) ]cos( {[ ) cos( )]} ( 4 ) ( [ { ) cos( )]} ( 4 ) ( [ { cos )]} ( 4 [ { ) ( t t a t t a t a a a t a a a t a a a t y Intermodulation Equations for Intermodulation Equations for -Tone Case Tone Case

59 Intermodulation Examples One channel Output Signal Magnitude Three-tone Two-tone Input Signal Magnitude in The output signal vs. the input signal amplitude for the three-tone and the two-tone tests, respectively Fall 0, Prof. JianJun Zhou II-59

60 Intercept Point (IP) The intercept point is obtained graphically by plotting the output power versus the input power both on logarithmic scales (e.g., db). Two curves are drawn; one for the linearly amplified signal at an input tone frequency, one for a nonlinear product. On a logarithmic scale, the function xn translates into a straight line with slope of n. Therefore, the linearly amplified signal will exhibit a slope of. third-order nonlinear product will increase by db in power when the input power is raised by db. The intercept point is a purely mathematical concept, and does not correspond to a practically occurring physical power level. In many cases, it lies beyond the damage threshold of the device. Fall 0, Prof. JianJun Zhou II-60

61 IP Plots 00 IM of three-tone 0log(Output) 50 0log(a ) IP IP 0 IM of two-tone -50 0log() Fall 0, Prof. JianJun Zhou II-6

62 Nonlinear Effects of Cascaded RF Systems P P db db X(t) y (t) G G P IP P db P and IP IP, G P IP, GG P IP, P db P db, G P db, GG P db, GGG P db,4 P IP N N dbm P, dbm dbm 0log[ IP N Gi PIP, N i i P IP, i j N i G j ] Fall 0, Prof. JianJun Zhou II-6

63 Trade-off between NF & IP Higher gains lower NF lower IP P db P IP G P db, P P IP, db, G P IP, GG P db, GG P IP, GG G P db,4 P IP N N dbm P, dbm dbm 0log[ IP N Gi PIP, N F i F i N i PIP, i G F F m total F G GG GG GGm j j ] Fall 0, Prof. JianJun Zhou II-6

64 Spurious-Free Dynamic Range IM of three-tone IP IP 0log(Output) log(a ) Noise-floor : slope P SFDR in, max IM of two-tone P 0log() IIP : slope F characterizes a receiver with more than one signal applied to the input Fall 0, Prof. JianJun Zhou II-64

65 Spurious-Free Dynamic Range P IIP P in P out P IM,out IIP IM,in The maximum input level for which the IM products become Equal to the noise floor: PIIP F Pin, max, F 74dBm NF 0log B PIIP F ( PIIP F) SFDR ( F SNRmin) SNR E.g. if a receiver with NF=9dB, P IIP =-5dBm, and B=00kHz SNR min =db, then, SFDR5dB. The SFDR represents the maximum relative level of interferers that a receiver can tolerate while producing an acceptable signal quality from a small input level. P in P P min Fall 0, Prof. JianJun Zhou II-65

66 Total System IIP Transfer all input intercept points to system input, subtracting gains and adding losses decibel for decibel Convert intercept points to powers (dbm to mw). We have IP, IP,. IPN for N elements ssuming all input intercepts points are independent and uncorrelated, add powers in parallel : Convert IIP from power (mw) to dbm. IIP ( IP IP IP IP N ) (mw) Fall 0, Prof. JianJun Zhou II-66

TSEK02: Radio Electronics Lecture 8: RX Nonlinearity Issues, Demodulation. Ted Johansson, EKS, ISY

TSEK02: Radio Electronics Lecture 8: RX Nonlinearity Issues, Demodulation Ted Johansson, EKS, ISY RX Nonlinearity Issues: 2.2, 2.4 Demodulation: not in the book 2 RX nonlinearities System Nonlinearity