digital compensation of dynamic acquisition errors at the front-end of ...

More documents

Recommendations

Info

18 Chapter 2: Nonlinearity at the ADC’s Front-End One common method that is usually used in the acquisition stage of highperformance ADCs is to use a bottom-plate track-and-hold for minimizing the input dependent charge injection [30], [31]. Figure 2.13 shows the structure of a bottomplate track-and-hold circuit. The way this circuit works is that the bottom switch (M 2 ) is turned off first to freeze the charge at the bottom plate of the sampling capacitor. Therefore, when M 1 turns off node V bot is a floating node and its charge can not change with the charge injection from M 1 . The charge at V bot is then redistributed using a feedback circuit, as the sampled value of the input voltage. In this architecture we only have charge injection from M 2 affecting the sampled value. The charge stored in M 2 channel is defined by (2.1) and is independent of the input voltage. However, as we discussed above, the fraction of transistor charge going to the capacitor depends on the impedance seen at the drain of M 2 and therefore depends on the resistance of M 1 . Due to the dependency of M 1 resistance to the input voltage (as will be discussed in detail in next section), there is still some small nonlinear input dependency in the charge injection from the bottom switch to the capacitor that can distort the sampled value. φ 1d φ 1 V in + V top M 1 M 2 C S Q inj - V C V bot Figure 2.13: A bottom-plate track-and-hold circuit. The switch resistance of M 1 can be kept nearly independent of the input voltage by using the bootstrap circuit discussed in the next section [32]. A bootstrap structure
Chapter 2: Nonlinearity at the ADC’s Front-End 19 keeps V gs of transistor constant during the tracking mode and therefore makes its on resistance constant. However there are still some small input dependencies in the switch resistance due to the backgate effect in the transistor and also parasitic capacitances in the nonideal bootstrap circuitry. As a result, the charge injection to the sampling capacitor can still cause small, but significant distortion on the sampled value. We did several device and mixed-mode simulations, using TCAD simulation tools [33], [34], of the circuit shown in Figure 2.13 (with a bootstrapped M 1 switch) to find the impact of different circuit parameters on the nonlinearity caused by charge injection. Device models were used for NMOS transistors in the circuit together with ideal capacitor and voltage sources. Clock was applied between source and gate terminal of M 1 to model an ideal bootstrapped switch. A channel length of 0.13µm was used for transistors and their doping profiles were optimized to achieve nominal performance for a 0.13µm process. The advantage of simulating the circuit using a TCAD device model for transistors is that all the charge flow and transistor channel variations can be observed and also the simulation results does not rely on the accuracy of the circuit models used for the devices. Appendix A includes an example netlist for modeling the transistor and the netlist of the circuit in Figure 2.13 implemented in the Taurus simulation tool. In a bottom-plate track-and-hold, the input voltage is sampled as the charge on the bottom plate of the capacitor. Therefore, linearity of this charge defines the linearity performance of this block. In order to measure the nonlinearity caused by charge injection, a DC input voltage was used and its value was swept from 0 to 1 V. For each input voltage, the charge injected to the capacitor during M 2 turn-off was measured and was plotted versus the input voltage. A straight line is fitted to this plot and the linearity performance is defined as the ratio of the maximum deviation from the straight line to the total charge on the capacitor at full-scale (see Figure 2.14). This charge linearity is defined in the following equation
Page 1 and 2: DIGITAL COMPENSATION OF DYNAMIC ACQ
Page 3: I certify that I have read this dis
Page 6 and 7: apply the inverse nonlinear functio
Page 8 and 9: Dr. Echere Iroaga, Dr. Yangjin Oh,
Page 10 and 11: Table of Contents Abstract.........
Page 12 and 13: 4.3.3 Algorithm performance on diff
Page 14 and 15: List of Figures Figure 1.1: Block d
Page 16 and 17: Figure 3.1: Block diagram of ADC er
Page 18 and 19: Figure 4.4: Frequency spectrum of t
Page 21 and 22: Chapter 1 Introduction 1.1 Motivati
Page 23 and 24: Chapter 1: Introduction 3 cancel th
Page 25 and 26: Chapter 1: Introduction 5 to correc
Page 27 and 28: Chapter 2 Nonlinearity at the ADC
Page 29 and 30: Chapter 2: Nonlinearity at the ADC
Page 37: Chapter 2: Nonlinearity at the ADC
Page 57 and 58: Chapter 3 Digital Correction of Dyn
Page 59 and 60: Chapter 3: Digital Correction of Dy
Page 85 and 86: Chapter 4 Experimental Results In t
Page 87 and 88: Chapter 4: Experimental Results 67
Page 89 and 90:
Chapter 4: Experimental Results 69
Page 91 and 92:
Page 93 and 94:
Page 95 and 96:
Page 97 and 98:
Page 99 and 100:
Page 101 and 102:
Chapter 5 Impact of Substrate Noise
Page 103 and 104:
Chapter 5: Impact of Substrate Nois
Page 105 and 106:
Page 107 and 108:
Page 109 and 110:
Page 111 and 112:
Page 113 and 114:
Chapter 6 Conclusion 6.1 Summary IF
Page 115 and 116:
Chapter 6: Conclusion 95 very sensi
Page 117 and 118:
Chapter 6: Conclusion 97 the desire
Page 119 and 120:
Appendix A TCAD Simulation of a Tra
Page 121 and 122:
Appendix A: TCAD Simulation of Trac
Page 123 and 124:
Appendix A: TCAD Simulation of Trac
Page 125 and 126:
Appendix B Modeling Signal Derivati
Page 127 and 128:
Appendix B: Modeling Signal Derivat
Page 129 and 130:
Appendix C Coefficient Calibration
Page 131 and 132:
Appendix C: Coefficient Calibration
Page 133 and 134:
Bibliography [1] A.M.A. Ali et al.,
Page 135 and 136:
Bibliography 115 [23] J. Kuo, R. Du
Page 137 and 138:
Bibliography 117 [46] F. H. Irons,
Page 139:
Bibliography 119 [68] B. Razavi, B.
show all

digital compensation of dynamic acquisition errors at the front-end of ...

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?