Power Amplifier Seminar, John Brown, Month, 200x Saguaro Cactus Sabino Canyon Tucson, Arizona

Transcription

1 Power Amplifier Seminar, John Brown, Month, 200x Saguaro Cactus Sabino Canyon Tucson, Arizona Max Specs - Height: 16 meters or 50 feet Age: 200 years Weight: 8 tons Specifications Deserts are places of extremes but house many plant and animal life forms, including unique landscapes 1

2 Excite Low Resistance Transducers, e.g. Load Cells Power Amplifier Applications AMP Signal Conditioning Data Conversion Signal Chain Processing ucontroller or Data Conversion Signal Conditioning Driving Sensor AMP ADC DAC AMP Transducer Motor Solenoid Actuator Relay Positioner TEC Heater Lamp 2

3 Power Op Amps Crossover Distortion Principles & Circuit Applications John Brown, 9/12/006 3

4 Essential Diodes / Transistors Power Output Stage, Class C Complementary Symmetry V V- V Tina Simulation V1 15 V2 15 Rseries 10 Vbases Q1!NPN Emitters Output Vout pwr amp ±2V T Iout Sine Wave Gen m m A p-p V- Q2!PNP Watch out for on-resistance of PNP Iout A RL 2 Vbases 3.97V p-p m Vout pwr amp 2.00V p-p u 50.00u 75.00u u Time (s) 4

5 Power Op Amps Crossover Distortion Amplitude Fundamental Harmonics 1. Harmonic Distortion time/freq domain Appearance - collective frequencies (Fourier Series) Consequence - harsh sounding audio signals or errors in analyzing spectral purity Frequency Nice looking and pure Dead Zone 2. Dead Band time domain Appearance fast change in signal level, oscillations, hunting Consequence servo systems going in the wrong direction 5

6 Power Op Amps Crossover Distortion 2. Dead Band time domain Usually expressed as % of Span. Unintentional dead band problem in electronics can cause control loss or make control loop go in wrong direction. Intentional dead band to avoid mechanical valve wear out problem. PID loop calculated output must leave this region before the actual output will change. Power Amp Dead Zone Can Cause HUNTING, which is movement back and forth around the set point 6

7 Dead Band or Blind Spot in Human Eye No Optical Sensors in Retina at Optic Nerve = Blind Spot 7

8 Dead Band or Blind Spot in Human Eye Directions to find your optical blind spot. Hold head still. Close Left Eye, focus on letters until black dot disappears Close Right Eye, focus on letters until black dot disappears By moving head back and forth you can HUNT for spot. 8

9 Power Op Amps - Crossover Distortion V V- V1 15 V2 15 Tina Simulation Power Amplifier: Just External Transistors, No Diodes Class C, Lots of Crossover Distortion JMB 5/6/2006 V Class C no feedback loop Rseries 10 Vbases Q1!NPN Emitters Output Vout pwr amp ±2V Sine Wave Gen Examples of Output Complementary Power Transistors MOSFET - IRF530 N-ch, 100V, 14A; IR9530 P-Ch, 12A, 100V V- Bipolar Darlingtons - MJE802 NPN; MJE702 PNP 80V, 4A, hfe=750 Bipolar Darlingtons-2N ; 2N , 60V,15A,hfeT>1000 Bipolar FZT851 NPN (6A,200V); FZT951 PNP (5A,100V), hfe=100 ZTEX Q2!PNP Watch out for on-resistance of PNP Iout A RL 2 9

10 T Iout m m 2.00 Power Amplifier: Just External Transistors, No Diodes Class C, Lots of Crossover Distortion JMB 5/6/2006 Power Op Amps - Crossover Distortion Tina Simulation Iout=0.5Ap=1Vp/2ohms Class C op amp feedback loop Vbases Vbases=2Vp Vout pwr amp f=10khz, Vin=1Vp Vout =1Vp Not Symmetrical u 50.00u 75.00u u Time (s) Rload = 2Ω Watch out for onresistance of PNP Notice: Doesn't swing to -1V 10

11 T Iout 80.00m m 2.00 Power Amplifier: Just External Transistors, No Diodes Class C, Lots of Crossover Distortion JMB 5/6/2006 Power Op Amps - Crossover Distortion Tina Simulation Iout=0.5Ap=1Vp/2ohms Class C op amp feedback loop Vbases Vbases=2Vp Vout pwr amp f=10khz, Vin=2Vp Vout =1Vp Symmettrical Time (s) Rload = 20Ω Watch out for on-resistance of PNP Watch out for onresistance of PNP Does Swing Symmetrically with logher 20 ohm load u 50.00u 75.00u u 11

12 Power Op Amps Little ro (static & dynamic) Essential Principles Output Resistance (See Tim Green s Presentation) 12

13 Power Amplifier: Just External Transistors, No Diodes Class C, Lots of Crossover Distortion JMB 5/6/2006 Power Op Amps - Crossover Distortion Tina Simulation Class C no feedback loop NPN Remitter RL=2Ω Watch out for onresistance of PNP Re=0.305Ω in Simulator 13

14 Power Amplifier: Just External Transistors, No Diodes Class C, Lots of Crossover Distortion JMB 5/6/2006 Power Op Amps - Crossover Distortion Tina Simulation Class C no feedback loop PNP Remitter RL=2Ω Watch out for onresistance of PNP Re=3.18Ω in Simulator Lot s or Ro 14

15 Power Op Amps - Crossover Distortion NPN-PNP Complementary Symmetry Vin NPN from Emitter PNP from Emitter Vout Class AB no feedback loop R, C, or L load 15

16 Power Amplifier: Just External Transistors, 2 Diodes & Remitters Class AB, Less Crossover Distortion JMB 5/6/2006 Power Op Amps - Crossover Distortion Tina Simulation V V- V1 15 V2 15 R2 100 V Class AB no feedback loop Rseries 0 Vbases Q1!NPN ±1V Sine Wave Gen This circuit biases the transistors just above cut-off and reduces crossover distortion, because both output devices are turned on. This results in higher Iq. D1 1N1183 D2 1N1183 R1 100 V- Re npn 500mVout pwr amp Re pnp 500m Q2!PNP Iout A Watch out for on-resistance of PNP RL 2 16

17 T Power Op Amps - Crossover Distortion m Tina Simulation Power Amplifier: Just External Transistors, 2 Diodes & Remitters Class AB, Less Crossover Distortion JMB 5/6/2006 Iout Iout=0.5Ap=1Vp/2ohms m 1.00 Vbases Vbases=1Vp Vout pwr amp f=10khz, Vin=1Vp Vout =1Vp Time (s) What out for on-resistance of PNP Watch out for onresistance of PNP Notice: Doesn't swing to -1V u 50.00u 75.00u u 17

18 Power Amplifier: Just External Transistors, 2 Diodes & Remitters & Vbe Mul Class AB, Less Crossover Distortion JMB 5/6/2006 V V- Power Op Amps - Crossover Distortion Tina Simulation V1 15 V2 15 R2 100 V Class AB no feedback loop Rseries 0 Vbases Q1!NPN Vbe Multiplier ±1V Sine Wave Gen R1 be 10k Vb-e multiplier R2 be 10k Vce = 1 (R1 / R2 x Vbe) = ((R1 R2) / R2) x Vbe = 2 x Vbe R1 100 Q3 vbe mul!npn V- Re npn 500m Vout pwr amp Re pnp 500m Q2!PNP What out for onresistance of PNP Iout A RL 2 This circuit biases the transistors just above cut-off and reduces crossover distortion, because both output devices are turned on. This results in higher Iq. 18

19 T Iout Power Op Amps - Crossover Distortion 2.00 Tina Simulation Power Amplifier: Just External Transistors, 2 Diodes & Remitters & Vbe Mul Class AB, Less Crossover Distortion JMB 5/6/2006 Iout=1.75Ap=3.5Vp/2ohms Vbases Vbases=1Vp Vbe Multiplier Vout pwr amp f=10khz, Vin=1Vp Vout =3.5Vp What out for on-resistance of PNP Watch out for onresistance of PNP u 50.00u 75.00u u Time (s) 19

20 Power Op Amps - Crossover Distortion V V- Tina Simulation Power Amplifier: Op Amp with External Transistors, No Loop V1 15 V2 15 Class C, Lots of Crossover Distortion JMB 5/6/2006 V Ri 1k Rf 9k Loop Does Not Attenuate Crossover - V- Vout opa Q1!NPN Vout pwr amp ±0.2V Sine Wave Gen V What other (more dc accurate) op amp is useful here? U1 OPA134 Q2!PNP Watch out for on-resistance of PNP V- Iout A RL 2 20

21 T Iout m m 2.00 Power Amplifier: Op Amp with External Transistors, No Loop Class C, Lots of Crossover Distortion JMB 5/6/2006 Power Op Amps - Crossover Distortion Tina Simulation Iout = 500mA RLoad = 2 ohms Iout=0.5Ap=1Vp/2ohms Heavy Load Class C op amp no feedback loop Iout = -250mA Vout opa Vbases=2Vp Vout pwr amp Vout = 1Vp f=10khz, Vin=2Vp Vout =1Vp Rload = 2Ω Loop Does Not Attenuate Crossover u 50.00u 75.00u u Time (s) Due to on-resistance of PNP Notice: Doesn't swing to -1V Vout = -0.5Vp Watch out for onresistance of PNP 21

22 T Iout 80.00m m 2.00 Power Amplifier: Op Amp with External Transistors, No Loop Class C, Lots of Crossover Distortion JMB 5/6/2006 Power Op Amps - Crossover Distortion Tina Simulation Iout = 50mA RLoad = 20 ohms Iout=0.05Ap=1Vp/20ohms Lighter Load Class C op amp no feedback loop Iout = -50mA Vout opa Vbases=2Vp Vout pwr amp Vout = 1Vp f=10khz, Vin=2Vp Vout =1Vp Rload = 20Ω Loop Does Not Attenuate Crossover u 50.00u 75.00u u Time (s) Symmetrical with RLoad = 20 ohms Vout = -1Vp 22

23 Power Op Amps - Crossover Distortion Tina Simulation V V- V1 15 V2 15 Power Amplifier: Op Amp with External Transistors Class C, Lots of Crossover Distortion JMB 5/6/2006 V Ri 1k Rf 9k Loop Gain Attenuates Crossover ±0.1V Sine Wave Gen - V- U1 OPA134 V What other (more dc accurate) op amp is useful here? Vout opa Q1!NPN Q2!PNP Watch out for on-resistance of PNP V- Iout Vout pwr amp A RL 2 ±1V 23

24 T Iout m m 2.00 Power Amplifier: Op Amp with External Transistors Class C, Lots of Crossover Distortion JMB 5/6/2006 Power Op Amps - Crossover Distortion Tina Simulation Iout=0.5Ap=1Vp/2ohms Class C op amp feedback loop Vout opa Vbases=2Vp Vout pwr amp f=10khz, Vin=2Vp Vout =1Vp Loop Attenuates Crossover u 50.00u 75.00u u Time (s) 24

25 Power Op Amps - Crossover Distortion Tina Simulation V V- V1 15 V2 15 Power Amplifier: Op Amp with External Transistors, 1 Diode Class AB, Less Crossover Distortion JMB 5/6/2006 V Ri 1k Rf 9k Loop Gain Attenuates Crossover ±0.1V Sine Wave Gen V- - V U1 OPA134 Vout opa D1 1N1183 R1 1k Q1!NPN Q2!PNP Watch out for on-resistance of PNP V- Iout Vout pwr amp A RL 2 ±1V What other (more dc accurate) op amp is useful here? Bias makes only a little difference in crossover, due to loop gain 25

26 T Iout m m 2.00 Power Amplifier: Op Amp with External Transistors, 1 Diode Class AB, Less Crossover Distortion JMB 5/6/2006 Power Op Amps - Crossover Distortion Tina Simulation Iout=0.5Ap=1Vp/2ohms Class AB op amp feedback loop Vout opa Vbases=2Vp Vout pwr amp f=10khz, Vin=2Vp Vout =1Vp Loop Attenuates Crossover Bias makes only a little difference in crossover u 50.00u 75.00u u Time (s) 26

27 Power Amplifier: Just External Transistors, 2 Diodes & Remitters Class AB, Less Crossover Distortion JMB 5/6/2006 Power Op Amps - Crossover Distortion Tina Simulation V V- V1 15 V2 15 R2 100 V Class AB no feedback loop Rseries 0 Vbases Q1!NPN ±1V Sine Wave Gen Re npn 500mVout pwr amp Re pnp 500m RL 2 This circuit biases the transistors just above cut-off and reduces crossover distortion, because both output devices are always turned on. D1 1N1183 D2 1N1183 R1 100 V- Q2!PNP Iout A Watch out for on-resistance of PNP 27

28 T Power Op Amps - Crossover Distortion m Tina Simulation Power Amplifier: Just External Transistors, 2 Diodes & Remitters Class AB, Less Crossover Distortion JMB 5/6/2006 Iout Iout=0.5Ap=1Vp/2ohms m 1.00 Vbases Vbases=1Vp Vout pwr amp f=10khz, Vin=1Vp Vout =1Vp Time (s) What out for on-resistance of PNP Watch out for onresistance of PNP Notice: Doesn't swing to -1V u 50.00u 75.00u u 28

29 Power Op Amps All NPN Power Output Stage (not NPN-PNP complementary symmetry) 29

30 Power Op Amps Crossover Distortion Notice Darlington Connection All NPN Power Output Stage (not NPN-PNP complementary symmetry) NPN from Emitter Vout NPN from Collector 30

31 Power Op Amps Crossover Distortion All NPN Power Output Stage (not NPN-PNP complementary symmetry) Actually Darlington Connection Local Compensation Loop, Independent of Overall Op Amp Compensation OPA547 NPN from Emitter Vout NPN from Collector 31

32 OPA547 Power Op Amp Measurement OPA547 RLoad = 32 ohms V2 30 Vin = 2.25Vp, Iout = 70mA ±2.25V Vin - Rlim 15.8k Ilim E/S V1 30 U1 OPA547 R 1M A RL 32 Vout Iout ±2.25V Rload Only 32

33 OPA547 Power Op Amp Measurement RLoad = ohms, Satisfactory Within OPA547 Bandwidth Limitation f = 20kHz Green Waveform Notice output current is in phase with output voltage. The cross-over occurs at Iout = 0 amps, which is at Vout = 0V. The slew rate of 20kHz at 2.25V peak is 0.3V/us. This does not exceed the overall 6V/us in the OPA547 data sheet, and negative slew condition causes very little crossover. The OPA548 behaves similarly. OPA547 Vout Iout 33

34 OPA547 Power Op Amp Measurement RLoad = ohms, Satisfactory Within OPA547 Bandwidth Limitation f = 20kHz Green Waveform Notice output current is in phase with output voltage. The cross-over occurs at Iout = 0 amps, which is at Vout = 0V. The slew rate of 20kHz at 2.25V peak is 0.3V/us. This does not exceed the overall 6V/us in the OPA547 data sheet, and negative slew condition causes very little crossover. The OPA548 behaves similarly. OPA547 Vout Iout 34

35 OPA547 Power Op Amp Measurement RLoad = ohms, Satisfactory Within OPA547 Bandwidth Limitation f = 100kHz Green Waveform Notice output current is in phase with output voltage. The cross-over occurs at Iout = 0 amps, which is at Vout = 0V. The slew rate of 100kHz at 2.25V peak is 1.4V/us. Although this does not exceed the overall 6V/us in the OPA547 data sheet, negative slew condition causes crossover. The OPA548 behaves similarly. OPA547 Vout Iout 35

36 OPA547 Power Op Amp Measurement RLoad = ohms, Satisfactory Within OPA547 Bandwidth Limitation f = 100kHz Green Waveform Notice output current is in phase with output voltage. The cross-over occurs at Iout = 0 amps, which is at Vout = 0V. The slew rate of 100kHz at 2.25V peak is 1.4V/us. Although this does not exceed the overall 6V/us in the OPA547 data sheet, negative slew condition causes crossover. The OPA548 behaves similarly. OPA547 Iout Vout 36

37 OPA547 Power Op Amp Measurement OPA547 RLoad = 3 ohms and 100nF V2 30 ±2.25V Vin - Rlim 15.8k Ilim E/S V1 30 U1 OPA547 CL 100n R 1M A RL 3 Vin = 2.25Vp, Iout = 140mA at 100kHz Vout Iout ±2.25V Rload - Cload Xc = 1/(2πfC) = 16 Ω -90 Iout = 2.25V / 16.3 Ω = 140mA -73 R = 3 Ω 0 Z = ( ) = 16.3Ω

38 OPA547 Power Op Amp Measurement CL = 100nF, Rseries = 3ohms, Heavy CLoad per OPA547 Data Sheet With heavy capacitance load, beyond that specified in the data sheet, the OPA547 output does show some instability. This occurs during negative slewing in the OPA ohms in series with 100nF (16 ohms at -90 degrees phase shift) results in a vectored load of about 16 ohms at 100kHz. Although the steady state vectored current of 140mA is within the OPA547 drive capability, the OPA547 will need an application circuit (shown in the following slides) to remain stable. Iout OPA547 Vout Green Waveform i(t) = CdV/dt initially, then op amp recovery. 38

39 Power Op Amps Crossover Distortion (COD) is dependent on: 1. Highest Signal Frequency 2. Largest Signal Amplitude 3. Highest Slew Rate 4. Circuit Techniques 39

40 Power Op Amps Crossover Distortion (COD) can be improved by: 1. Reducing input signal frequency, given a constant amplitude 2. Reducing input signal amplitude, given a constant frequency 3. Employing circuit techniques - adding noise gain or using pull-down resistor 4. Choosing a higher bandwidth amplifier for signal frequency and amplitude 40

41 Power Op Amps Crossover Distortion All NPN Power Output Stages 1. Can You Reduce It? Somewhat. Mitigation Fix #1 2. Can It Be Totally Fixed? No, Without More Iq! But Yes, External Pull-Down Resistor. Application Fix #2 41

42 OPA547 Power Op Amp Measurement Mitigation Fix #1 CL = 100nF, Rseries = 3ohms, R-C Summing Junction to Vin CL = 100nF, Rseries = 3ohms, Heavy CLoad per OPA547 Data Sheet Noise gain = 3V/V at higher frequencies. Gnoise = 3 = 1 (20k/10k) V2 30 V1 30 Rsj junction & Csj junction shift loop phase to reduce oscillation effect. Rsj 10k Csj 100n Vin ±2.25V RC Comp Increrases Noise Gain - Rf 20k Rlim 15.8k Ilim E/S U1 OPA547 CL 100n R 1M A RL 3 Vout Iout ±2.25V Reduces Ugliness but Increases Noise Gain, and Reduces BW Rload - Cload 42

43 OPA547 Power Op Amp Measurement Mitigation Fix #1 CL = 100nF, Rseries = 3ohms, R-C Summing Junction to Vin CL = 100nF, Rseries = 3ohms, Heavy CLoad per OPA547 Data Sheet Green Waveform Compensation on Summing Junction, Cleans Up Output Waveform Somewhat. Iout OPA547 Vout Reduces Ugliness but Increases Noise Gain, and Reduces BW 43

44 OPA547 Power Op Amp Measurement Application Fix #2 CL = 100nF, Rseries = 3ohms, 91 ohms connected to -15V CL = 100nF, Rseries = 3ohms, Heavy CLoad per OPA547 Data Sheet Ipull down = about 190mA, which keeps the upper output power transistor always turned on. Vin ±2.25V Pull-Down Resistor Eliminates Crossover but adds Iquiescent - V2 30 Rlim 15.8k Ilim E/S V1 30 U1 OPA547 V2 15 Rpd 91 CL 100n R 1M A RL 3 Vout ±2.25V Iload = ±140mA Pull Down Resistor Makes All Clean, but adds Iquies Rload - Cload Iout Ipull down = (15V2.25V) / 91Ω = 190mA Ipull down - = (15V-2.25V) / 91Ω = -140mA 44

45 OPA547 Power Op Amp Measurement Application Fix #2 CL = 100nF, Rseries = 3ohms, 91 ohms connected to -15V CL = 100nF, Rseries = 3ohms, Heavy CLoad per OPA547 Data Sheet Green Waveform Negative Pull Down Keeps Upper Output Power Transistor Turned On, Which Provides a Clean Output Waveform. OPA547 Iout Vout Pull Down Resistor Makes All Clean, but adds Iquies 45

46 Power Op Amps Crossover Distortion Time and Frequency Domain For Pull Down Resistor Application fix 46

47 OPA547 Power Op Amp Measurement Application Fix #1 CL = 100nF, Rseries = 3ohms, 91 ohms connected to -15V CL = 100nF, Rseries = 3ohms, Heavy CLoad per OPA547 Data Sheet Input Voltage Signal Spectrum, 100kHz Distortion in this case comes from Input Signal Generator Input Pull Down Resistor Makes All Clean, but adds Iquies 47

48 OPA547 Power Op Amp Measurement Application Fix #1 CL = 100nF, Rseries = 3ohms, 91 ohms connected to -15V CL = 100nF, Rseries = 3ohms, Heavy CLoad per OPA547 Data Sheet Pull Down Resistor Makes All Clean, but add Iquies Output VOLTAGE Signal Spectrum, 100kHz Output No Rpull-down Output With Rpull-down Different than Input Current Signal Distortion in this case mostly comes from Input Signal Generator 48

49 OPA547 Power Op Amp Measurement Application Fix #1 CL = 100nF, Rseries = 3ohms, 91 ohms connected to -15V CL = 100nF, Rseries = 3ohms, Heavy CLoad per OPA547 Data Sheet Pull Down Resistor Makes All Clean, but add Iquies Output CURRENT Signal Spectrum, 100kHz Output No Rpull-down Output With Rpull-down Different than Input Voltage Signal Distortion in this case mostly comes from Input Signal Generator 49

50 OPA547 Power Op Amp Measurement Application Fix #1 CL = 100nF, Rseries = 3ohms, 91 ohms connected to -15V CL = 100nF, Rseries = 3ohms, Heavy CLoad per OPA547 Data Sheet Pull Down Resistor Makes All Clean, but add Iquies Output CURRENT Signal Spectrum, 100kHz Output No Rpull-down Output With Rpull-down Different than Input Voltage Signal Distortion in this case mostly comes from Input Signal Generator 50

51 Power Op Amps The Real Fix For Reducing Crossover Distortion No Matter How You Look At It, The Circuit Requires More Quiescent Current, Especially in the Output Stage. 51

52 Power Op Amps Swing to the Output Rail It s Based on Linearity in Both Cases 1. Maintaining Aol 2. Keeping THD Below Some Level 52

53 Power Op Amps Swing to Output Rail with good linearity is dependent on Aol Test Circuit for Measuring Aol Calculate 20 Log b10 (Vx / Vin) 53

54 Power Op Amps, OPA547 Swing to Output Rail With Good Signal Purity is At the Zero Crossing SR = 2π * Vp * BWfp Example #1 SR = 2π * 2.25V * 100kHz = 1.4V/us dependent on Slew Rate Example #2 SR = 2π * 27Vp max * 35kHz = 6V/µs min 54

55 Power Op Amps What affects Crossover Distortion (COD)? 1. Output Swing is a Function of Frequency 2. Distortion is a Function of Frequency 3. Distortion or Fidelity is a Function of Open Loop Gain, Aol 4. Overshoot or Ringing is a Function of Load Capacitance 55

56 Power Op Amps, OPA547 Output Swing is a Function of Frequency 27Vp 35kHz 56

57 Power Op Amps, OPA547 Distortion is a Function of Frequency At the Zero Crossing, SR = 2π * Vp * BWfp 0.01% 3kHz 57

58 Power Op Amps, OPA547 Distortion or Fidelity is a Function of Aol 40dB 100V/V 10kHz 58

59 Power Op Amps, OPA547 Overshoot or Ringing is a function of Load Capacitance 15% 0.001µF = 1nF 0.012µF = 12nF 59

60 Power Op Amps Happy Crossover Distortion Thank You For Your Participation John Brown 60