6-33. Mixer IF. IF Amp LO. Transmitter

Transcription

1 6-33 Power Amplifier (PA) Design Antenna Mixer IF BPF Filter PA IF Amp LO Transmitter A PA is used in the final stage of wireless transmitters to increase the radiated power level. Typical PA output powers may be 0.3 to W for a handheld cellular or PCS phone, or 0-00 W for base station transmitters. Important considerations for PAs are Output power, Gain Efficiency (for battery using time) Nonlinear intermodulation effects,.. S parameters of linear devices are well defined not depend on the input power level or output load impedance For high input powers (near the P db ), transistors do not behave linearly. the impedances seen at the input & output port will depend on the input power level greatly complicates the design of PAs. 04 H.-R. Chuang, EE NCKU

2 6-34 Characteristics of PAs and Amplifiers Classes The PA is usually the primary consumer of DC power in most handheld wireless devices Efficiency is a very important consideration One measure of efficiency is the ratio of RF output power to DC input power One drawback of this definition is that it does not account for the RF power delivered at the input to the amplifier Since most PAs have relatively low gains, the efficiency of (*) tends to overate the actual efficiency. A better measure that includes the effect of input power is the power added efficiency (PAE) Antenna P DC P t P out P in P rad PA Transmitter P diss Efficiency: P P out DC Power added efficiency(pae): P out Power stage 3-stage PA Driver stage Gain stage P in PAE ( or PAE) Pout Pin PDC Pout ( Pout / G) P ( ) out PDC G PDC ( ) G PAs are often designed to provide the best efficiency, even if this means that the resulting gain is less than the maximum possible. Nonlinearities can lead to the generation of spurious frequencies & intermodulation distortion. This is a serious issue in wireless transmitters, especially in a multicarrier system, where spurious signals may appear in adjacent channels! Linearity is also critical for nonconstant envelope modulations, such as amplitude shift keying (ASK) & higher quadrature amplitude modulation methods. 04 H.-R. Chuang, EE NCKU

3 Amplifier Classes of Operation 6-35 Class-A PAs are inherently linear circuits, where the transistor is biased to conduct over the entire range of the input signal cycle. theoretical maximum efficiency of 50%. Most small-signal and low-noise amplifiers operate as class A circuits. Class-B PAs are biased to conduct only during one-half of the input signal cycle. Usually two complementary transistors are operated in a class B push-pull amplifier to provide amplification over the entire cycle. theoretical efficiency of a class B amplifier is 78%. Class C PAs are operated with the transistor near cutoff for more than half of the input signal cycle generally use a resonant circuit in the output stage to recover the fundamental. can achieve efficiencies near 00%, but only with constant envelope modulations. Higher classes, such as class D, E, P, and S, use the transistor as a switch to pump a highly resonant tank circuit, and achieve very high efficiencies. The majority of wireless transmitters operating at UHF frequencies or above rely on class A, AB, or B PAs because of the need for low distortion product 04 H.-R. Chuang, EE NCKU

4 6-36 Large-Signal Characterization of Transistors (for Power Amplifier: PA) A transistor behaves linearly for signal powers well below P -db the small-signal S-parameters should not depend on either the input power level or the output termination impedance. For power levels comparable to or greater than P -db the nonlinearity of the transistor becomes apparent the measured S parameters will depend on input power level & the output termination impedance. Large-signal S parameters are not uniquely defined and do not satisfy linearity, and cannot be used in place of small-signal parameters. A more useful way to characterize transistors under large-signal operating conditions is to measure the gain & output power as a function of source and load impedances. One way of doing this is to determine the large-signal source & load reflection coefficients, SP & LP, that maximize power gain for a particular output power (often chosen as P -db ) & versus frequency. Nonlinear equivalent circuit models can also be developed and used to predict the large-signal performance of FETs and BJTs The dominant nonlinear parameters for a microwave FET are Cg,, g n,, Cgd, Rds Equivalent circuit models can be very useful when combined with computer-aided design software. An important consideration in modeling large-signal transistors is that parameters dependents on device temperature, which increases with output power. 04 H.-R. Chuang, EE NCKU

5 6-37 Load-Pull Method ( 負載調整法 ) Plot contours of constant power output on a Smith chart as a function of the load reflection coefficient, LP, with the transistor conjugately matched at input. => Load-pull contours Load-pull contours can be obtained using an automated measurement set-up with computer-controlled electromechanical stub tuners (load-pull system) Vector Network Analyzer Vector Signal Generator Directional RF Relay Couplers RF Relay DUT RF Relay 50 Ohm Load Tuning stubs Control Computer Tuning stubs Power meter Power meter Spectrum Analyzer 04 H.-R. Chuang, EE NCKU

6 HBTone_LoadPul_G Com psub_palib X 6-38 A 900-MHz three-stage PA Measurement Results of 5V PA Frequency Range 90~98 MHz VDD 5V VGG -4V Gain 33 db Gain Flatness +/ db VSWR (Input) <.5 Power Output.4W (3.5 dbm) db Compression Point 3.5 dbm 3rd Order Intercept Point dbm PAE 45% ADS Software Load-Pull Method C_model C_model chip_cap7 chip_cap33 Cap= pf Cap=0 nf fo=.483 GHz fo=55.06 MHz Sdb= Sdb=-49.9 db db MSub MSUB MSub H= mm Er=4.7 Mur= Cond=4.e7 Hu=.0e+033 mm T= um TanD=0.08 Rough=0 mm V_DC SRC Vdc=Vlow P_Tone PORT Num= Z=Z_s P=dbmtow(4) Freq=RFfreq I_Probe Is_low C C C=4 pf Set these values: Var VAR Eqn STIMULUS RFfreq=5775 MHz Vhigh=3.6 Vlow=3.6 db_gain_comp=.0 I_Probe Is_high MLIN One Tone Load Pull Simulation; X db Gain TL4 Compression output power and PAE f ound at MLIN Subst="MSub" each f undamental load impedance TL W= mm Subst="MSub" L=0 mm W= mm L=0 mm L_model chip_ind4 L_model Lp=3.3 nh chip_ind5 C_model C_model fo= GHz Lp=3.3 nh chip_cap35 chip_cap34 Sdb=-9.85 fo= db GHz V_DC Cap= pf Cap=0 nf Specify desired Load Tuner coverage: MLIN Sdb=-9.85 db SRC fo=.483 GHz fo=55.06 MHz s_rho is the radius of the circle of reflection coefficients TL MLIN Vdc=Vhigh Sdb= Sdb=-49.9 db db Subst="MSub" TL3 generated. However, the radius of the circle will be W= Subst="MSub" mm reduced if it would otherwise go outside the Smith Chart. L=3 mm W= mm s_center is the center of the circle of generated reflection L L=3 mm coefficients L pts is total number of reflection coefficients generated L=3 nh Z0 is the system reference impedance R= L L Va r VAR Eq n L=3 nh SweepEquations R R= s_rho =0.75 R s_center =0.0 +j*0.0 R=0.6 kohm C pts=00 C Z0=50 C=4 pf I_Probe Iload R R Term R=0.75 QP8X5X3_R_VIA kohm Term QP8X5X3_R_VIA Num= Z=ZLoadTuner BVIA BVIA PARAMETER SWEEP ParamSweep Sweep PARAMETER SWEEP ParamSweep Sweep GAIN COMPRESSION XDB HB Freq[]=RFfreq Order[]=5 GC_XdB=dB_Gain_Comp Set Load and Source impedances at harmonic frequencies Va r VAR Eq n VAR Z_l_ =Z0 + j*0 Z_l_3 = Z0 + j*0 Z_l_4 = Z0 + j*0 Z_l_5 = Z0 + j*0 Z_s_fund = 0 + j*0 Z_s_ = Z0 + j*0 Z_s_3 = Z0 + j*0 Z_s_4 = Z0 + j*0 Z_s_5 =Z0 + j*0 Pdel_contours_p m Maximum Power Delivered, dbm 4.85 indep(pdel_contours_p) (0.000 to 9.000) m indep(m)=0 Pdel_contours_p=0.630 / 4.94 level= , number= impedance = Z0 * (0.5 + j H.-R. Chuang, EE NCKU

7 6-39 MRF858S transistor at 900 MHz f (MHz) S S S S Input & output impedance (for large signal ) to have P-dB = 3.6W, G= db ( V CE =4 V. I C = 0.5 A) Z Z in out. 9.0 j3.5 j4.50 in out o o 04 H.-R. Chuang, EE NCKU

8 H.-R. Chuang, EE NCKU

9 6-4.4 GHz Hybrid PA 04 H.-R. Chuang, EE NCKU

10 6-4 Broadband Transistor Amplifier Design 寬頻放大器 Balanced amplifier V V V A A G A V A Z 0 V B Z 0 o 90 hybrid V B B G B o 90 hybrid V A balanced amplifier using two 90 o hybrid couplers Symmetrical Hybrid Coupler : Branch-line or Lange Coupler ( Ch 4) S 0 j 0 j 0 0 j j 0 => S j equally split & 90 phase shift S 3 S4 0 no power flowing to port 4 04 H.-R. Chuang, EE NCKU

11 6-43 V V V A A G A V A Z 0 V B Z 0 o 90 hybrid V B B G B Balanced amplifier o 90 hybrid V Reflections are absorbed in the coupler terminations, improving input/output matching as well as the stability of the individual amplifiers. The balanced amplifier has a gain equal to that of a single amplifier and provides a graceful degradation of a -6 db loss in gain if a single amplifier section fails. Bandwidth can be an octave or more, primarily limited by BW of the coupler. V V A V j VB V (from 90 o -coupler property) j VA V B V ( ) A B ( AV A ) j ( ) BVB A[ V V A B if S ( ) A B V ] A B V S 0 => V => no refrection (totally match) within the bandwidth of the coupler j j [ ] B V j j j j V V A V B GAV A GBV B G A V GB V ( ) ( ) [ ] [ ] j V G A GB j V GA G B j if G G G => S GA GB A B S jg V Summary If the amplifiers are identical: GA GB G & A B => S 0 & total gain = G It can also be shown that the noise figure of the balanced amplifier is F ( F A FB ) /, where F A & F B are the NF of the individual amplifiers. 04 H.-R. Chuang, EE NCKU

12 to.6 GHz broadband balanced power amplifier (with Lange couplers) EMcom Lab EE NCKU, 995 Frequency: Gain: Gain flatness: VSWR(in): VSWR(out): P db : OIP3: Harmonics:.3 ~.6 GHz 7.5 db +.0 db.44: (max).8: (max) 7 dbm@.3 GHz 39.5 dbm (min) <-30 out < 9 dbm 04 H.-R. Chuang, EE NCKU