1. Trang chủ
  2. » Kỹ Thuật - Công Nghệ

Lecture Digital signal processing: Lecture 4 - Zheng-Hua Tan

18 48 0

Đang tải... (xem toàn văn)

Tài liệu hạn chế xem trước, để xem đầy đủ mời bạn chọn Tải xuống

THÔNG TIN TÀI LIỆU

Thông tin cơ bản

Định dạng
Số trang 18
Dung lượng 1,06 MB

Nội dung

Lecture Digital signal processing - Lecture 4 present sampling and reconstruction. The main contents of this chapter include all of the following: Periodic sampling, „frequency domain representation, reconstruction, changing the sampling rate using discretetime processing.

Digital Signal Processing, Fall 2006 Lecture 4: Sampling and reconstruction Zheng-Hua Tan Department of Electronic Systems Aalborg University, Denmark zt@kom.aau.dk Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Course at a glance MM1 Discrete-time signals and systems MM2 Fourier-domain representation Sampling and reconstruction System System structures System analysis MM6 MM5 Filter MM4 z-transform MM3 DFT/FFT Filter structures MM9,MM10 MM7 Filter design MM8 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Part I: Periodic sampling Periodic sampling Frequency domain representation Reconstruction Changing the sampling rate using discretetime processing „ „ „ „ Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Periodic sampling From continuous-time xc (t ) to discrete-time x[n] „ x[n] = xc (nT ), ‰ ‰ −∞ < n < ∞ T Sampling period Sampling frequency f s = / T Ω s = 2π / T Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Two stages In practice? Mathematically „ Impulse train modulator Conversion of the impulse train to a sequence ‰ ‰ ∞ ∑ δ (t − nT ) s(t ) = n = −∞ x s (t ) = xc (t ) s (t ) ∞ = xc (t ) ∑ δ (t − nT ) n = −∞ ∞ = ∑ xc (nT )δ (t − nT ) n = −∞ x[n] = xc (nT ), −∞ < n < ∞ ∞ xc (t ) = ∫− ∞ xc (τ )δ (t −τ )dτ Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Periodic sampling Tow-stage representation „ ‰ ‰ ‰ ‰ Strictly a mathematical representation that is convenient for gaining insight into sampling in both the time and frequency domains Physical implementation is different x s (t ) a continuous-time signal, an impulse train, zero except at nT x[n] a discrete-time sequence, time normalization, no explicit information about sampling rate Many-to-many Ỉ in general not invertible „ Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Part II: Frequency domain represent Periodic sampling Frequency domain representation Reconstruction Changing the sampling rate using discretetime processing „ „ „ „ Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Frequency-domain representation From xc(t) to xs(t) „ s(t ) = ∞ ∑ δ (t − nT ) n = −∞ x s (t ) = xc (t ) s (t ) ∞ = xc (t ) ∑ δ (t − nT ) n = −∞ = The Fourier transform of a periodic impulse train is a periodic impulse train 2π ∞ ↔ S ( jΩ) = ∑ δ (Ω − kΩ s ) T k = −∞ ? ↔ X s ( jΩ ) = X c ( jΩ) * S ( jΩ) 2π ∞ = ∑ X c ( j (Ω − kΩ s )) T k = −∞ ∞ ∑ xc (nT )δ (t − nT ) n = −∞ The Fourier transform of xs(t) consists of periodic repetition of the Fourier transform of xc(t) „ Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Frequency-domain S ( jΩ ) = X s ( jΩ ) = 2π T T ∞ ∑ δ (Ω − kΩ s ) k = −∞ ∞ ∑ X c ( j (Ω − kΩ s )) k = −∞ Ω s − Ω N > Ω N or Ω s > 2Ω N Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Recovery X r ( jΩ ) = H r ( jΩ ) X s ( jΩ ) Ideal lowpass filter with gain T and cutoff frequency Ω c Ω N < Ω c < (Ω s − Ω N ) X r ( jΩ ) = X c ( jΩ ) 10 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Aliasing distortion „ „ 11 Due to the overlap among the copies of X c ( jΩ) , due to Ω s ≤ 2Ω N X c ( jΩ) not recoverable by lowpass filtering Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Aliasing – an example xc (t ) = cos Ω t a.1 X s ( jΩ ) = T ∞ ∑ X c ( j (Ω − kΩ s )) k = −∞ b.1 xr (t ) = cos Ω t a.2 xr (t ) = cos(Ω s − Ω )t b.2 12 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Nyquist sampling theorem Given bandlimited signal X c ( jΩ) = 0, Then for | Ω |≥ Ω N is uniquely determined by its samples xc (t ) x[n] = xc (nT ), If with xc (t ) −∞ < n < ∞ 2π ≥ 2Ω N T Ωs = ΩN is called Nyquist frequency 2Ω N is called Nyquist rate 13 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Fourier transform of x[n] From xs (t ) to x[n] x s (t ) = ∞ ∑ xc (nT )δ (t − nT ) n = −∞ x[n] = xc (nT ), −∞ < n < ∞ From X s ( jΩ) to X (e jω ) By taking continuous-time Fourier transform of xs(t) X s ( jΩ ) = ∞ ∑ xc (nT )e − jΩTn ( n = −∞ ∞ X ( jΩ) = ∫− ∞ x(t )e − jΩt dt ) By taking discrete-time Fourier transform of x[n] X ( e jω ) = ∞ ∑ x(n)e − jωn n = −∞ 14 X s ( jΩ) = X (e jω ) |ω = ΩT = X (e jΩT ) Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Fourier transform of x[n] X s ( jΩ) = X (e jω ) |ω = ΩT = X (e jΩT ) X s ( jΩ ) = T X ( e jω ) = ∞ ∑ X c ( j (Ω − kΩ s )) k = −∞ T ∞ ∑ k = −∞ X c ( j( ω T − 2πk )) = T T ∞ ∑ k = −∞ Xc( j ω − 2πk T ) X ( e jω ) is simply a frequency-scaled version of X with ω = ΩT x s (t ) retains a spacing between samples equal to the sampling period T while x[n] always has unity space 15 s ( jΩ ) Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Sampling and reconstruction of Sin Signal xc (t ) = cos(4000πt ) → Ω = 4000π T = / 6000 → Ω s = 2π / T = 12000π ∴ no aliasing x[n] = xc (nT ) = cos(4000πnT ) = cos((2π / 3)n) = cos(ω0 n) xc (t ) ↔ X c ( jΩ) = πδ (Ω − 4000π ) + πδ (Ω + 4000π ) X s ( jΩ) = T ∞ ∑ X c ( j (Ω − kΩ s )) k = −∞ X (e jω ) = X s ( jΩ) |Ω =ω / T = X s ( jω / T ) with normalized frequency ω = ΩT How about xc (t ) = cos(16000πt ) 16 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Part III: Reconstruction „ „ „ „ Periodic sampling Frequency domain representation Reconstruction Changing the sampling rate using discretetime processing 17 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Requirement for reconstruction „ On the basis of the sampling theorem, samples represent the signal exactly when: ‰ ‰ ‰ 18 Bandlimited signal Enough sampling frequency + knowledge of the sampling period Ỉ recover the signal Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Reconstruction steps (1) Given x[n] and T, the impulse train is x s (t ) = (2) ∞ ∞ n = −∞ n = −∞ ∑ xc (nT )δ (t − nT ) = ∑ x[n]δ (t − nT ) i.e the nth sample is associated with the impulse at t=nT The impulse train is filtered by an ideal lowpass CT filter with impulse response hr (t ) ↔ H r ( jΩ) xr (t ) = ∞ ∑ x(n)hr (t − nT ) n = −∞ X r ( j Ω ) = H r ( jΩ ) X ( e j Ω T ) 19 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Ideal lowpass filter „ Commonly choose cutoff frequncy as Ωc = Ω s / = π / T hr (t ) = 20 sin(πt / T ) πt / T Digital Signal Processing, IV, Zheng-Hua Tan, 2006 10 Ideal lowpass filter interpolation CT signal Modulated impulse train xr (t ) = 21 ∞ ∑ n = −∞ x[n] sin(π (t − nT ) / T ) π (t − nT ) / T Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Ideal discrete-to-continuous-time converter 22 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 11 discrete-to-continuous-time converter “Practical DACs not output a sequence of dirac impulses (that, if ideally low-pass filtered, result in the original signal before sampling) but instead output a sequence of piecewise constant values or rectangular pulses” Ideally sampled signal Piecewise constant signal typical of a practical DAC output From http://en.wikipedia.org/wiki/Digital-to-analog_converter 23 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Applications 24 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 12 Part IV: Changing the sampling rate „ „ „ „ 25 Periodic sampling Frequency domain representation Reconstruction Changing the sampling rate using discretetime processing Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Downsampling x[n] = xc (nT ) x'[n] = xc (nT ' ) By reconstruction & re-sampling though not desirable Using DT processing only: „ Sampling rate reduction by an integer factor – downsampling by “sampling” it xd [n] = x[nM ] = xc (nMT ) 26 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 13 Frequency domain xd [n] = x[nM ] = xc (nMT ) DT Fourier transform „ 2πk )) T T k = −∞ ∞ ω 2πr X d (e jω ) = ∑ X c ( j ( T ' − T ' )) T ' r = −∞ ∞ 2πr ω = ∑ X c ( j ( MT − MT )) MT r = −∞ X ( e jω ) = T ∞ ∑ X c ( j( ω − r = i + kM , − ∞ < k < ∞, ≤ i ≤ M − 1, − ∞ < r < ∞ X d ( e jω ) = 27 M M −1 ∞ ∑[T ∑ X i =0 k = −∞ j ( ω − 2π i ) / M Since X (e → X d ( e jω ) = M )= c ( j( ω MT − 2π k 2π i − ))] T MT ∞ ω 2π i 2π k ∑ X c ( j ( MT − MT − T )) T k = −∞ M −1 ∑ X (e j ( ω − 2π i ) / M ) i =0 Similar to the Eq above! Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Frequency domain - an example This example „ „ „ Sampling results in copies at nΩ s = 2nπ / T Same, downsampling generates M copies of (e jω ) with frequency X scaled by M and shifted Aliasing can be avoided if X (e jω ) is bandlimited Ω s = 4Ω N X (e jω ) = 0, ω N ≤| ω |≤ 2π and 2π / M ≥ 2ω N 28 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 14 Frequency domain - an example Downsampling factor is too large, causing aliasing, resulting in the need for DT ideal lowpass filter with cutoff frequency pi/M To avoid aliasing in downsampling by a factor of M requires that ωN M < π 29 ωN = Ω NT ' Digital Signal Processing, IV, Zheng-Hua Tan, 2006 A general downsampling system „ „ 30 The system is also called decimator The process is called decimation Digital Signal Processing, IV, Zheng-Hua Tan, 2006 15 Increasing sampling rate – upsampling „ „ Downsampling Ỉ analogous to sampling a CT signal Upsampling Ỉ analogous to D/C conversion x[n] = xc (nT ) xi [n] = xc (nT ' ), T '= T L x i [ n ] = x[ n / L ] = x c ( nT / L ), n = , ± L , ± L , Expander ⎧ x[n / L], n = 0, ± L, ± L, xe [ n] = ⎨ otherwise ⎩ 0, xe [ n] = 31 ∞ ∑ x[k ]δ [n − kL] k = −∞ Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Fourier domain „ Fourier transform of the output of expander xe [ n] = ∞ ∑ x[k ]δ [n − kL] k = −∞ X e ( e jω ) = = ∞ ∞ ∑ ( ∑ x[k ]δ [n − kL])e − jωn n = −∞ k = −∞ ∞ ∑ x[k ]e − jωLk = X (e jωL ) k = −∞ Which is s frequency scaled version, w is replaced by wL so ω = ΩT ' 32 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 16 An example DTFT of x[n] = xc (nT ) X e (e jω ) = X (e jωL ) System: interpolator Process: interpolation 33 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Summary „ „ „ „ 34 Periodic sampling Frequency domain representation Reconstruction Changing the sampling rate using discretetime processing Digital Signal Processing, IV, Zheng-Hua Tan, 2006 17 Course at a glance MM1 Discrete-time signals and systems MM2 Fourier-domain representation Sampling and reconstruction System System structures System analysis MM6 MM5 Filter MM4 z-transform MM3 35 DFT/FFT Filter structures MM9,MM10 MM7 Filter design MM8 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 18 ... http://en.wikipedia.org/wiki /Digital- to-analog_converter 23 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 Applications 24 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 12 Part IV: Changing... Signal Processing, IV, Zheng-Hua Tan, 2006 Ideal discrete-to-continuous-time converter 22 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 11 discrete-to-continuous-time converter “Practical... if X (e jω ) is bandlimited Ω s = 4 N X (e jω ) = 0, ω N ≤| ω |≤ 2π and 2π / M ≥ 2ω N 28 Digital Signal Processing, IV, Zheng-Hua Tan, 2006 14 Frequency domain - an example Downsampling factor

Ngày đăng: 11/02/2020, 17:00

TỪ KHÓA LIÊN QUAN