receiver l losses w when using quadr drature ban bandpas
play

Receiver L Losses w when using Quadr drature Ban Bandpas ass - PowerPoint PPT Presentation

Receiver L Losses w when using Quadr drature Ban Bandpas ass Samplin ling Andrew Dempster, Ediz Cetin How Bandpass Sampling Works exploits aliasing brings all images with it must attenuate out of band images received


  1. Receiver L Losses w when using Quadr drature Ban Bandpas ass Samplin ling Andrew Dempster, Ediz Cetin

  2. How Bandpass Sampling Works • exploits aliasing • brings all images with it – must attenuate out of band “ images ” received signal sampling frequency IGNSS 2016, UNSW, Sydney 2

  3. Bandpass Sampling: GPS L1 • For GPS L1 signal, with carrier 1.6 GHz, BW 2MHz: – Lowpass sampling would require sampling rates of 3.2Gs/s – Bandpass sampling requires a minimum rate of twice the BW, i.e. 4Ms/s IGNSS 2016, UNSW, Sydney 3

  4. Bandpass Filter Design • For Bandpass sampling: N p in-band noise S = N 0 out-of-band noise SNR + − s N ( n 1 ) N n subsampling ratio p 0 • Max subsampling ratio for GNSS is for L1 GPS:   f = = = max n 1576/2 788   max   B • Requires N p /N 0 of 29dB for 3dB SNR loss IGNSS 2016, UNSW, Sydney 4

  5. Bandpass Sampling • Sampling rate must be at least twice the signal bandwidth (Nyquist) • It must also, because a virtual downconversion results, ensure each downconverted band does not: – overlap dc, – overlap the Nyquist rate, or – overlap any other downconverted signal band IGNSS 2016, UNSW, Sydney 5

  6. 2B is not to be • Need more than twice the bandwidth sometimes: Vaughan, R.G.; Scott, N.L.; White, D.R.; “ The theory of bandpass sampling”, IEEE Trans Signal Processing, Volume 39, Issue 9, Sept. 1991 Page(s):1973 - 1984 IGNSS 2016, UNSW, Sydney 6

  7. Sampling at 2B f s B IGNSS 2016, UNSW, Sydney 7

  8. Sampling at the Minimum Rate IGNSS 2016, UNSW, Sydney 8

  9. Exceeding the Minimum is not enough either Vaughan, R.G.; Scott, N.L.; White, D.R.; “ The theory of bandpass sampling”, IEEE Trans Signal Processing, Volume 39, Issue 9, Sept. 1991 Page(s):1973 - 1984 Slide 9 IGNSS 2016, UNSW, Sydney 9

  10. Exceeding the Minimum is not enough either 2 • fu=2.5B, 1.5 fs=3.5B 1 0.5 0 -0.5 -1 -1 0 1 2 3 4 5 6 multiples of bandwidth B IGNSS 2016, UNSW, Sydney 10

  11. “Normal” Quadrature Sampling • RF signal split into I/ Q by 90° phase shift in downconversion ~ ~ i o (n) X ~ ADC s ( t ) 0° ω LO T s 90° ~ ~ X q o (n) ~ ADC IGNSS 2016, UNSW, Sydney 11

  12. Quadrature Bandpass Sampling • I/ Q from delay in sampling: ∆ t = 1/4f c i 1 (n) ADC s ( t ) ~ ~ 0 ~ T s ∆ t q 1 (n) ADC IGNSS 2016, UNSW, Sydney 12

  13. Quadrature Bandpass Sampling • Sampling rate must be at least twice the signal bandwidth (Nyquist) • Overlaps are allowed: – overlap dc, – overlap the Nyquist rate – overlap any other downconverted signal band • BPS: no band overlaps • QBPS: 2 bands can overlap but not 3 IGNSS 2016, UNSW, Sydney 13

  14. Dempster, Andrew G. , “Quadrature Bandpass Sampling Rules for Single- and Multiband Communications and Satellite Navigation Receivers”, IEEE Transactions on Aerospace and Electronic Systems, vol 47, no 4, Oct 2011, pp 2308 – 2316 IGNSS 2016, UNSW, Sydney 14

  15. BPS vs QBPS rates “ channel ” rate “ total ” rate Dempster, Andrew G. , “Quadrature Bandpass Sampling Rules for Single- and Multiband Communications and Satellite Navigation Receivers”, IEEE Transactions on Aerospace and Electronic Systems, vol 47, no 4, Oct 2011, pp 2308 – 2316 IGNSS 2016, UNSW, Sydney 15

  16. Distortion due to QBPS vs Sampled Quadrature Downconversion • Signal • Sampled I/Q distortion • QBPS A.G. Dempster , E. Cetin, “QBPS in RF front-ends”, Electronics Letters, vol 52 no 23, 3 Nov 2016, pp1965 - 1967 IGNSS 2016, UNSW, Sydney 16

  17. Simple Remedy i o (n), q o (n) Baseband or IF Processing I/Q s ( t ) i 1 (n) i’ 1 (n) ∆ t QBPS q 1 (n) Naïve reconstruction Simple Remedy IGNSS 2016, UNSW, Sydney 17

  18. Image Rejection Ratio • For – x(t) = e j ω t input – y(t) = α e j ω t + β e -j ω t output • IRR = 20 log 10 ( β/α ) IGNSS 2016, UNSW, Sydney 18

  19. IRR for corrected, uncorrected • Naïve single frequency • Naïve band • Simple remedy A.G. Dempster , E. Cetin, “QBPS in RF front-ends”, Electronics Letters, vol 52 no 23, 3 Nov 2016, pp1965 - 1967 IGNSS 2016, UNSW, Sydney 19

  20. IRR: Naive IGNSS 2016, UNSW, Sydney 20

  21. IRR: Band • GPS L1 IGNSS 2016, UNSW, Sydney 21

  22. IRR: Band • Galileo E5 IGNSS 2016, UNSW, Sydney 22

  23. IRR: remedy • GPS L1 IGNSS 2016, UNSW, Sydney 23

  24. Achievable IRR IGNSS 2016, UNSW, Sydney 24

  25. Conclusions • QBPS can readily be used without remedy and achieve good IRR • Remedy improves at some frequencies IGNSS 2016, UNSW, Sydney 25

Recommend


More recommend