Keysight Technologies Inc. · NYSE:KEYS

Keysight Technologies — Agilent Technologies — Technical Presentation (April 2001)

53 slides from Keysight Technologies's technical presentation, Apr 17, 2001

  1. Slide 1: Presentation on Fundamentals of Mixer Design

    Agilent EEsof EDA Presentation on Fundamentals of Mixer Design This document is owned by Agilent Technologies, but is no longer kept current and may contain obsolete or inaccurate references. We regret any inconvenience this may cause. For the latest information on Agilent’s line of EEsof electronic

  2. Slide 2: Fundamentals of Mixer Design: Abstract

    •4/17/01 H DesignSeminar Agilent EEsof Customer Education and Applications Fundamentals of Mixer Design Abstract The mixer is used in nearly all RF/Microwave systems for frequency translation. This seminar explains how mixers provide this function and why they often generate so many spurious outputs

  3. Slide 3: About the Author

    •4/17/01 H About the Author Steve Long • University of California, Santa Barbara • Professor, Electrical and Computer Engineering • Consultant to Industry Page 2 BIOGRAPHICAL SKETCH Stephen Long received his BS degree in Engineering Physics from UC Berkeley and MS and PhD in Electrical Engineering f

  4. Slide 4: Choosing the right mixer for the task: Learning objectives

    •4/17/01 H Basic engineering problem: Choosing the right mixer for the task... Learning Objectives: • Understand operating principles of the mixer • What makes a good mixer? • Choices: Nonlinear/switching mode; single/double balance; active/passive • Specify performance: Gain, NF, P1dB, TOI, SFDR, i

  5. Slide 5: Why study mixers?

    •4/17/01 H Why study mixers? • Receivers • up or down conversion • demodulation of SC SSB or SC DSB • input must support large dynamic range • AGC • Transmitters • up conversion • modulation: amplitude and phase • input has optimum signal level for high performance Page 4 Mixers have a wide variety

  6. Slide 6: What is a mixer?

    •4/17/01 H What is a mixer? • Frequency translation device A AB • Ideal mixer: • Doesn’t “mix”; it multiplies B ( A sin ω1t )( B sin ω 2t ) = AB [cos(ω1 − ω 2 )t − cos(ω1 + ω 2 )t ] 2 Downconvert Upconvert ω1 and ω2 suppressed No harmonics ω1 − ω2 ω1 + ω2 Page 5 A mixer doesn’t really “mix” or sum s

  7. Slide 7: Images

    •4/17/01 H Images • Two inputs (RF & Image) will mix to the same output (IF) frequency. • The image frequency must be removed by filtering • fIF and fLO must be carefully selected • Image rejection ratio: dB(PIF desired/PIF image) DIFFERENCE fIF fIF OUTPUTS { { INPUTS SUM OUTPUTS fIF fIM fLO fRF fLO

  8. Slide 8: Image rejection preselector

    •4/17/01 H Image rejection preselector BPF RF IF LO DIFFERENCE fIF fIF OUTPUTS { { INPUTS SUM OUTPUTS fIF fIM fLO fRF fLO+fIM fLO+fRF Page 7 A bandpass preselection filter is often used ahead of the mixer to suppress the image signal. The IF and LO frequencies must be carefully selected to avoid ima

  9. Slide 9: Mixer operating mechanisms

    •4/17/01 H Mixer operating mechanisms • Nonlinear transfer function • use device nonlinearities creatively! • useful at mm-wave frequencies • Switching or sampling • a time-varying process • preferred; fewer spurs Page 8 High performance RF mixers use nonlinear characteristics to generate the multip

  10. Slide 10: Nonlinear mixer operation

    •4/17/01 H Nonlinear mixer operation Any diode or transistor will exhibit nonlinearity in its transfer characteristic at sufficiently high signal levels. RS + VO(t) − RL } VLO(t) vin(t) second-order product term VRF(t) VO (t ) = a0 + a1vin (t ) + a 2 vin2 (t ) + a3vin3 (t ) + Λ Page 9 We see that ou

  11. Slide 11: Unbalanced diode mixer output

    •4/17/01 H Unbalanced diode mixer output FRF = 110 MHz |VRF| = 0.1V VDC = 0.6V FLO = 100 MHz |VLO| = 0.2V Page 10 We can see that there are a lot of spurious outputs generated. Ideally, we would like to see outputs only at 10 MHz and 210 MHz. So, we prefer the switching type mixer when the RF and LO

  12. Slide 12: Switching or sampling mixers

    •4/17/01 H Switching or sampling mixers • Let VIN(t) = VR cos (ωRFt) • Multiply by the LO switching function T(t) RS VIN(t) + VO(t) − RL VRF(t) 1 2 sin( 3ω LO t ) sin( 5ω LO t )  T (t ) = + sin(ω LO t ) + + +Λ  2 π 3 5  T(t) no even order harmonics Page 11 This simple switch is operated by the

  13. Slide 13: Mixer output

    •4/17/01 H Mixer output VR 2V  cos(ω RF t ) sin(3ω LO t )  Vo (t ) = cos(ω RF t ) + R cos(ω RF t ) sin(ω LO t ) + +Λ  π  { { 2 3  { RF feedthrough 2nd-order product 4th-order spurs VO(ω) ω RF−LO LO RF RF+LO 3LO−RF 3LO 3LO+RF Page 12 The product of VRF(t)T(t) produces the desired output frequen

  14. Slide 14: Isolation between ports

    •4/17/01 H Isolation between ports • The mixer is not perfectly unilateral - leakage between: • LO to IF • LO to RF • RF to IF • Determine the magnitude of these leakage components at the IF and RF ports using harmonic balance. • Use the mix function to select frequencies. Page 13 Isolation can be q

  15. Slide 15: Conversion gain or loss

    •4/17/01 H Conversion gain or loss • Generally expressed as a voltage gain or as a transducer power gain RS + + VRF VLO(t) RL VIN VIF VIF − − AV = VIN  v IF 2    =  2 RL 2  Output power at FIF ConvGain = RF available input power  v RF     8 RS  Page 14 Conversion gain is usually defined

  16. Slide 16: Ideal Single-balanced mixer

    •4/17/01 H Ideal Single-balanced mixer +V •Linear I -> V RL conversion IF output •Switch polarity LO Input of RF current •Linear V -> I +gm conversion RF input Page 15 The RF feedthrough can be eliminated by using a differential IF output and a polarity reversing LO switch. The polarity reversing LO

  17. Slide 17: LO Switching Function

    •4/17/01 H LO Switching Function T(t) 1 2 1  T1(t ) = + sin(ω LOt ) + sin(3ω LO t ) + .... 2 π 3  1 2 1  T2 (t ) = − + sin(ω LOt ) + sin(3ω LOt ) + .... 2 π 3  SUM OF SWITCHING FUNCTIONS T1(t) T2(t) T(t) = T1(t) + T2(t) Page 16 When added together, the DC terms (1/2 & -1/2) cancel. The D

  18. Slide 18: Output spectrum: SB mixer

    •4/17/01 H Output spectrum: SB mixer RF input VIF = V1 − V2 LO DC offset LO feedthrough 3FLO - FRF FRF = 200 MHz FIF FIF FLO = 1.0 GHz 3FLO 5FLO FIF @ 800 & 1200 MHz Page 17 As you can see, the output spectrum of the single-balanced switching mixer is much less cluttered than the nonlinear mixer spe

  19. Slide 19: Ideal Double Balanced Mixer

    •4/17/01 H Ideal Double Balanced Mixer +V •Linear I -> V RL RL conversion IF output •Switch polarity LO Input of RF current •Linear V -> I -gm +gm conversion RF input Page 18 An ideal double balanced mixer consists of a switch driven by the local oscillator that reverses the polarity of the RF input

  20. Slide 20: Output Spectrum: DB mixer

    •4/17/01 H Output Spectrum: DB mixer Differential output voltage NO LO or RF feedthrough Polarity reversing switch function FRF = 200 MHz is easily seen here FLO = 1.0 GHz FIF @ 800 & 1200 MHz Page 19 The differential output voltage and frequency spectrum are simulated using a transient analysis in

  21. Slide 21: Mixer Performance Specifications

    •4/17/01 H Mixer Performance Specifications • Image rejection • Conversion gain: voltage or power • Port-to-port isolation: dBc • Large signal performance: • gain compression: P1dB • intermodulation distortion spec: third-order intercept (TOI) • Small signal performance: noise figure • Operating ran

  22. Slide 22: Large-signal mixer performance

    •4/17/01 H Large-signal mixer performance • Signal statistics • 4 sources • Pavs = 0 dBm • in phase @ t=0 ∆f = 10 MHz • Peak signal voltage, not average power dictates when distortion becomes excessive. Page 21 In receiver applications, a mixer is often exposed to several signals within its preselec

  23. Slide 23: Gain Compression

    •4/17/01 H Gain Compression • Conversion gain degrades at large input signal levels due to nonlinearity in the signal path. • Assume a simple nonlinear transfer function: VRF (t ) = vin − a3vin3  3a V 2  1 VRF (t ) = VR 1 − 3 R  sin(ω RF t ) + a3VR3 sin(3ω RF t )  4  4 { { Gain compression T

  24. Slide 24: Use Harmonic Balance simulation

    •4/17/01 H Use Harmonic Balance simulation •Highest order of IM products •Fundamental Frequencies: put highest power source first •Number of harmonics of sources [1] & [2] •Sweep P_RF in 1 dB steps from −15 dBm •Can be used to pass variables to data display Page 23 Harmonic balance is the method of

  25. Slide 25: Gain Compression: P1dB

    •4/17/01 H Gain Compression: P1dB P1dB Page 24 The RF mixer behavioral model in ADS has been used to illustrate the gain compression phenomenon. The input RF power (P_RF) was swept from -15 dBm to +5 dBm. On the left, we see the simulated IF output power vs. the ideal output power. Ideal output powe

  26. Slide 26: Intermodulation distortion

    •4/17/01 H Intermodulation distortion • IMD consists of the higher order signal products that are generated when two RF signals are present at the mixer input. The IMD will be down and up converted by the LO as will the desired RF signal. • IMD generation is a good indicator of large signal performa

  27. Slide 27: Intermodulation Distortion

    •4/17/01 H Intermodulation Distortion • Let’s consider the 3rd order nonlinearity: a3vin3 • two inputs: vin = V1 sin(ω1t) + V2 sin(ω2t) Vout 3 = a3[V13 sin 3 (ω1t ) + V23 sin 3 (ω 2 t ) + 3V12V2 sin 2 (ω1t ) sin(ω 2 t ) + 3V1V22 sin(ω1t ) sin 2 (ω 2 t )] 3V12V2 a3 {sin(ω 2t ) − 12 [sin( 2ω1 − ω 2 )t

  28. Slide 28: IMD simulation

    •4/17/01 H IMD simulation • Use a two-tone generator at the mixer input. • The two input frequencies are separated by Fspacing and each have an input power of P_RF dBm. •Large MaxOrder and LO Order are needed for accurate HB TOI predictions with highly nonlinear elements •Oversampling (Param menu) s

  29. Slide 29: IF output spectrum

    •4/17/01 H IF output spectrum Third-order intermodulation products at 2f1 - f2 - fLO and 2f2 - f1 - fLO will be present in the IF output. Page 28 Note that the third-order (m2) and fifth-order products are quite close in frequency to the desired signal (m1). This means that they are often impossible

  30. Slide 30: Third-order intercept definition

    •4/17/01 H Third-order intercept definition OIP3 slope = 1 IIP3 slope = 3 Page 29 A widely-used figure of merit for IMD is the third-order intercept (TOI) point. This is a fictitious signal level at which the fundamental and third-order product terms would intersect. In reality, the intercept power

  31. Slide 31: Noise figure & SFDR

    •4/17/01 H Noise figure & SFDR • We have been concentrating on the large signal limitations of the mixer. Noise determines the other end of the mixer dynamic range.  ( S / N )in  • Spurious-free dynamic range: NF = 10 log10    (S / N )out  SFDR = 2 [IIP3 − (10 log kT∆f + NF )] 3 Pout (dBm) tal

  32. Slide 32: Determining Noise Figure

    •4/17/01 H Determining Noise Figure • Use harmonic balance simulator for mixer NF. • takes into account any nonlinearities and harmonics that could mix noise down into the IF band. • If P_RF << P_ LO, either a 1-tone generator or a passive termination can be used at the input with equal accuracy. •

  33. Slide 33: SSB or DSB Noise Figure?

    •4/17/01 H SSB or DSB Noise Figure? • There are two definitions used for noise figure with mixers - often a source of confusion. • SSB NF assumes signal input from only one sideband, but noise inputs from both sidebands. • Relevant for heterodyne architectures signal image noise SSB NF S N Page 32 M

  34. Slide 34: SSB or DSB Noise Figure?

    •4/17/01 H SSB or DSB Noise Figure? • DSB NF includes both signal and noise inputs from both sidebands. Appropriate for direct conversion architectures. signal image noise DSB NF S N Page 33 A DSB NF is easier to measure; wideband excess noise is introduced at both the signal and image frequencies.

  35. Slide 35: Passive or Active Mixers?

    •4/17/01 H Passive or Active Mixers? • Passive nonlinear devices or switches • conversion loss, not gain • high tolerance to IMD • external baluns or transformers needed • Active mixers • can provide conversion gain • active baluns - better for IC implementation • more difficulty in achieving good I

  36. Slide 36: Mixer circuit examples

    •4/17/01 H Mixer circuit examples • OK, now let’s look at some examples • 2 mixer circuits will be reviewed: • diode DB quad: • familiar and widely used • wide bandwidth, limited by baluns • Low IMD FET mixer: • not as well known, but good performance Page 35 35

  37. Slide 37: Diode DB quad mixer

    •4/17/01 H Diode DB quad mixer + LO − LO VLO(t) VRF(t) VIF(t) VRF(t) VIF LO Page 36 The diode double-balanced quad mixer is a very popular design and available in a wide variety of frequency bandwidths and distortion specs. The MiniCircuits™ catalog [7] is full of these. The diodes act as a polarity

  38. Slide 38: Design procedure

    •4/17/01 H Design procedure • Determine correct RF and LO impedances to match through the diode ring to the 50 ohm IF load. Transformer ratios can be swept using HB or XDB. • Sweep LO power to maximize conversion gain and gain compression • Or, buy one from MiniCircuits™! Page 37 The RF and LO imped

  39. Slide 39: Diode mixer performance

    •4/17/01 H Diode mixer performance 1-tone 2-tone P1dB = + 5 dBm P1dB = 0 dBm Page 38 A harmonic balance simulation can be used to estimate the mixer performance. With the DB diode mixer, the currents in the diodes are half wave, so a high number of LO harmonics and maximum order and some oversamplin

  40. Slide 40: Need good match at each port

    •4/17/01 H Need good match at each port Image, HD, IMD RF VRF(t) IF VLO(t) IF term P1dB TOI 50Ω 0 dBm 15 dBm BPF −2 −2 Page 39 Since these passive diode switch mixers are bilateral, that is, the IF and RF ports can be reversed, the performance of the mixer is very sensitive to the termination impeda

  41. Slide 41: Mixer termination methods

    •4/17/01 H Mixer termination methods • Passive diplexer series resonant at F IF RF 50 IF LO 50 out of band reflections LO, RF harmonics, feedthrough, IMD parallel resonant at FIF Page 40 This passive diplexer provides a low-loss forward path through the series resonant branch. At FIF, the parallel r

  42. Slide 42: Mixer termination methods

    •4/17/01 H Mixer termination methods • Active wideband termination RF LO Zin ≈ 1/gm gm ≈ IC/VT Page 41 This common base stage provides a wideband resistive impedance provided the maximum frequency at the input is well below the fT of the transistor. The bias current can be set to provide a 50 ohm in

  43. Slide 43: Low IMD FET mixer

    •4/17/01 H Low IMD FET mixer • LO modulates channel resistance • Channel resistance is very linear at small VDS • LPF and HPF act as diplexer HPF LPF RF IF LO VT Page 42 Another popular mixer utilizes only one FET in its simplest form. [10] The LO signal switches the FET on and off. The gate is bias

  44. Slide 44: ID - VDS of MOSFET

    •4/17/01 H ID - VDS of MOSFET • Highly linear at low VDS • Device width and LO voltage can be optimized for performance Page 43 The mixer RF to IF path will be quite linear if the total drain voltage (VDS) remains small. As can be seen from the DC simulation, the MOSFET exhibits quite linear channel

  45. Slide 45: Design process

    •4/17/01 H Design process • Select nominal FET channel width and optimize the source and load impedances for conversion gain • Design input (HPF) and output (LPF) networks which present these impedances to the FET drain • Optimize FET width and matching impedances for highest P1dB Page 44 The design

  46. Slide 46: FET mixer performance

    •4/17/01 H FET mixer performance Conv Gain (1 tone input) Conv Gain (2 tone input) P1dB = + 12 dBm P1dB = + 7 dBm Page 45 The mixer performance was simulated. Surprisingly, we see that the P1dB is higher than that simulated for the diode mixer. The OIP3 is also reasonably high. [See ADS design files

  47. Slide 47: FET mixer isolation

    •4/17/01 H FET mixer isolation LO to RF isolation is poor. Good LO to IF and RF to IF HPF doesn’t attenuate LO due to LPF on IF port Solutions: double-balancing, LO trap or higher-order HPF Page 46 So far, so good. But, now we see the problem with this design. There is a very large LO feedthrough to

  48. Slide 48: Conclusion

    •4/17/01 H Conclusion Learning objectives: • Understand operating principles of the mixer • What makes a good mixer? • Choices: nonlinear/switching mode; single/double balance; active/passive • Specify performance: Gain, NF, P1dB, TOI, SFDR, isolation • Mixer examples - numerous other possibilities

  49. Slide 49: Exercises

    •4/17/01 H Exercises 1. Use ADS to evaluate the performance of a single-balanced or double-balanced FET mixer based on the mixer example in this seminar. You will need to add transformers to do this. Evaluate and compare the conversion gain, isolation, P1dB, and TOI with the single FET version. Furt

  50. Slide 50: Exercises

    •4/17/01 H Exercises 2. Refer to the notes page... VGG VLO(t) VRF(t) VIF(t) Page 49 The mixer shown above is the FET equivalent of the diode ring mixer. This was first described by Ed Oxner [10] and can provide impressive large signal performance with a sufficiently strong LO voltage. The substrate

  51. Slide 51: References

    •4/17/01 H References • [1] Gray, P. R. and Meyer, R. G., Design of Analog Integrated Circuits, 3rd Ed., Chap. 10, Wiley, 1993. • [2] Gilbert, B., “Design Considerations for BJT Active Mixers”, Analog Devices, 1995. • [3] Lee, T. H., The Design of CMOS Radio-Frequency Integrated Circuits, Chap. 11,

  52. Slide 52: End of Design Seminar...

    •4/17/01 H End of Design Seminar... Page 51 51

  53. Slide 53: Agilent contact information

    For more information about Agilent Email Updates www.agilent.com Agilent EEsof EDA, visit: www.agilent.com/find/emailupdates For more information on Agilent Technologies’ products, applications or services, please Get the latest information on the www.agilent.com/find/eesof products and applications y

Agilent Technologies — Technical Presentation (April 2001) on 6ix