EE247 Lecture 16 •D/A Converters (continued) –DAC reconstruction filter •ADC Converters –Sampling • Sampling switch considerations – Thermal noise due to switch resistance – Clock jitter related non-idealities – Sampling switch bandwidth limitations – Switch conductance non-linearity induced distortion • Sampling switch conductance dependence on input voltage • Clock voltage boosters – Sampling switch charge injection & clock feedthrough EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 1 Summary Last Lecture • D/A converters – Practical aspects of current-switched DACs (continued) – Segmented current-switched DACs – DAC dynamic non-idealities – DAC design considerations – Self calibration techniques • Current copiers • Dynamic element matching EECS 247 Lecture 16: Data Converters-DAC Design © 2010 Page 2 DAC In the Big Picture Analog Input • Learned to build DACs Analog Anti-Aliasing Preprocessing Filter –Convert the incoming digital A/D Sampling signal to analog Conversion +Quantization 000 • DAC output DSP ...001... staircase form 110 D/A "Bits to • Some applications Conversion Staircase" require filtering (smoothing) of DAC Reconstruction output Analog Post processing Filter Reconstruction filter Analog Output EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 3 DAC Reconstruction Filter B f/2 s •Need for and tu 1 requirements depend on pn I C0.5 application A D 0 0 0.5 1 1.5 2 2.5 3 1 x 106 •Tasks: c nis 0.5 – Correct for sincdroop 0 0 0.5 1 1.5 2 2.5 3 – R(setamiro-cvaes “ea laiapsperos”ximation) tuptu 1 x 106 O C0.5 A D 0 0 0.5 1 1.5 2 2.5 3 Normalized Frequency f/fs EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 4 Reconstruction Filter Options Reconstruction Filters Digital SC CT DAC Filter Filter Filter • Reconstruction filter options: –Continuous-time filter only –CT + SC filter • SC filter possible only in combination with oversampling (signal bandwidth B << f /2) s • Digital filter –Band limits the input signal prevent aliasing –Could also provide high-frequency pre-emphasis to compensate in- band sinx/xamplitude droop associated with the inherent DAC S/H function EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 5 DAC Reconstruction Filter Example: Voice-Band CODEC Receive Path Receive Output f= 8kHz s fs= 8kHz fs= 128kHz fs= 128kHz GSR Reconstruction Filter f= 128kHz & sinx/x Compensator s Note: f max = 3.4kHz sig fDAC = 8kHz s sin(p f max x T )/(p f max xT) sig s sig s = -2.75 dB droop due to DAC sinx/x shape Ref: D. Senderowicz et. al, “A Family of Differential NMOS Analog Circuits for PCM Codec Filter Chip,” IEEE Journal of Solid-State Circuits,Vol.-SC-17, No. 6, pp.1014-1023, Dec. 1982. EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 6 Summary D/A Converter • D/A architecture – Unit element –complexity proportional to 2B-excellent DNL – Binary weighted-complexity proportional to B-poor DNL – Segmented-unit element MSB(B )+ binary weighted LSB(B ) 1 2 Complexity proportional ((2B1-1) + B ) -DNL compromise between the two 2 • Static performance – Component matching • Dynamic performance – Time constants, Glitches • DAC improvement techniques – Symmetrical switching rather than sequential switching – Current source self calibration – Dynamic element matching • Depending on the application, reconstruction filter may be needed EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 7 What Next? Analog Input • ADC Converters: Analog Anti-Aliasing Preprocessing Filter A/D Sampling – Need to build Conversion +Quantization circuits that "sample“ 000 DSP ...001... 110 D/A "Bits to – Need to build Conversion Staircase" circuits for amplitude Analog Reconstruction quantization Post processing Filter Analog Output EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 8 Analog-to-Digital Converters •Two categories: – Nyquist rate ADCs f max ~ 0.5xf sig sampling • Maximum achievable signal bandwidth higher compared to oversampled type • Resolution limited to <14bits – Oversampled ADCs f max << 0.5xf sig sampling • Maximum achievable signal bandwidth significantly lower compared to nyquist • Maximum achievable resolution high (18 to 20bits!) EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 9 MOS Sampling Circuits EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 10 Ideal Sampling • In an ideal world, zero 1 resistance sampling v v switches would close for IN OUT S1 the briefest instant to sample a continuous C voltage v onto the IN capacitor C Output Dirac-like pulses with amplitude 1 equal to V at the time IN of sampling T=1/f S • In practice not realizable! EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 11 Ideal Track & Hold Sampling 1 v v 1 IN OUT S1 C T=1/f S • V tracks input for ½ clock cycle when switch is closed out • Ideally acquires exactvalue of V at the instant the switch opens in • "Track and Hold" (T/H) (often called Sample & Hold!) EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 12 Ideal T/H Sampling Continuous time Time kca dlo T/H signal rT H (Sampled-Data Signal) Clock Discrete-Time Signal EECS 247 Lecture 15: Data Converters-DAC Design & Intro. to ADCs © 2010 Page 13 Practical Sampling Issues 1 v v IN OUT M1 C • Switch induced noise due to M1 finite channel resistance • Clock jitter (edge variation of ) 1 • Finite R limited bandwidth finite acquisition time sw • R = f(V )distortion sw in • Switch charge injection & clock feedthrough EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 14 Sampling Circuit kT/C Noise 1 4kTRDf v IN v v v IN OUT OUT M1 R S1 C C • Switch resistance & sampling capacitor form a low-pass filter • Noise associated with the switch resistance results in Total noise variance= kT/C@ the output (see noise analysis in Lecture 1) • In high resolution ADCs with such sampling circuit right at the input, kT/C noise at times dominates overall minimum signal handling capability (power dissipation considerations). EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 15 Sampling Network kT/C Noise For ADCs sampling capacitor size is usually chosen based on having thermal noise smaller or equal or at times slightly larger compared to quantization noise: Assumption: Nyquist rate ADC D2 For a Nyquist rate ADC: Total quantization noise power 12 Choose C such that thermal noise level is less (or equal) than Q noise k T D2 B C 12 2 2B1 C12k T B V FS 22B C12k T B V 2 FS EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 16 Sampling Network kT/C Noise 2B 2 C12k T B 2 V FS Required C as a Function of ADC Resolution min B C (V = 1V) C (V = 0.5V) min FS min FS 8 0.003 pF 0.012 pF 12 0.8 pF 2.4 pF 14 13 pF 52 pF 16 206 pF 824 pF 20 52,800 pF 211,200 pF The large area required for C limit highest achievable resolution for Nyquist rate ADCs Oversampling results in reduction of required value for C (will be covered in oversampled converter lectures) EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 17 Practical Sampling Issues 1 v v IN OUT M1 C • Switch induced noise due to M1 finite channel resistance • Clock jitter (edge variation of ) 1 • Finite R limited bandwidth finite acquisition time sw • R = f(V )distortion sw in • Switch charge injection & clock feedthrough EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 18 Clock Jitter • So far : clock signal controls sampling instants – which we assumed to be precisely equi-distant in time (period T) • Real clock generator some level of variability • Variability in T causes errors – "Aperture Uncertainty" or "Aperture Jitter“ • What is the effect of clock jitter on ADC performance? EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 19 Clock Jitter • Sampling jitter x(t) adds an error actual sampling voltage time t J proportional to the x’(t ) 0 product of (t -t ) J 0 and the derivative of the input signal at the sampling nominal (ideal) instant sampling time t 0 EECS 247 Lecture 16: Data Converters-ADC Design © 2010 Page 20
Description: