DESIGN AND IMPLEMENTATION OF A

HIGH SPEED AND LOW POWER FLASH ADC

WITH FULLY DYNAMIC COMPARATORS

LI TI

NATIONAL UNIVERSITY OF SINGAPORE

2010
 DESIGN AND IMPLEMENTATION OF A
HIGH SPEED AND LOW POWER FLASH ADC
WITH FULLY DYNAMIC COMPARATORS

LI TI

A THESIS SUBMITTED

FOR THE DEGREE OF MASTER OF ENGINEERING

DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING

NATIONAL UNIVERSITY OF SINGAPORE

2010
 Acknowledgements

First, I would like to express my most sincere gratitude to my supervisors Dr. Yao

Libin and Dr. Lian Yong. For almost two and a half years I have worked with Dr. Yao, he

has never failed to guide and inspire me with his profound knowledge and experience,

without which, I cannot imagine how much harder this journey would have been.

Although I did not have the privilege to also work directly under Dr. Lian’s supervision,

still I have learned a lot, and in a very profound way, just by observing his way of doing

things.

I feel most fortunate to have met and got to know all the wonderful personalities in

VLSI and Signal Processing Lab. Not only have they always been good friends and

pleasant companions, but also they have helped me tremendously in coping with my

study and research. Time spent with them has been nothing but full of joy and amusement,

which most certainly has made the experience more precious and memorable.

Last but most importantly, I would like to thank both of my parents, my father Li Wei

and my mother Li Shufang, for everything that I have ever achieved in my life.
I
 Contents
Acknowledgements ..................................................................................... I

Summary .................................................................................................. IV

List of Tables ............................................................................................ VI

List of Figures .........................................................................................VII

List of Abbreviations ............................................................................... XI

Chapter 1 Introduction.............................................................................. 1

Chapter 2 Overview of Flash ADC Designs ............................................. 4

2.1 Conventional Flash ADC....................................................................................... 4

2.2 Flash ADC Designs with Resistive Averaging and Interpolation ......................... 7

2.3 Flash ADC Designs with Calibration Techniques ............................................... 14

Chapter 3 A Background Comparator Offset Calibration Technique . 22

3.1 System Overview of the Proposed Background Comparator Offset Calibration

Technique .................................................................................................................. 22

3.2 Circuit Implementation ........................................................................................ 31

3.3 Measurement Results ........................................................................................... 35

Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully

Dynamic Comparators ............................................................................ 39

4.1 System Model and Simulation ............................................................................. 41

4.2 Track & Hold Circuit ........................................................................................... 48

4.3 Fully Dynamic Comparator and Calibration ....................................................... 55

II
 4.4 Calibration Control Circuit .................................................................................. 66

4.5 Encoder ................................................................................................................ 68

4.6 Clock Driver ........................................................................................................ 74

4.7 Power Consumption of the Flash ADC ............................................................... 78

4.8 Flash ADC Layout Design .................................................................................. 80

Chapter 5 Conclusion .............................................................................. 83

References ................................................................................................................. 85

III
 Summary

This work primarily focuses on design and implementation of a high speed low power

flash ADC with fully dynamic comparators. For flash ADC design, fully dynamic

comparator offers several very desirable attributes, like high speed and low power

consumption. As a result, a significant improvement of overall performance is expected.

However, the application of fully dynamic comparator is restricted by a few things,

among which large offset variation is a primary issue.

To capitalize the vast potential promised by fully dynamic comparator, we have first

developed a background comparator offset calibration technique, which has provided a

foundation for fully dynamic comparator’s use in a flash ADC design. We use chopper to

isolate offset from input signal, so that it can later be extracted by a LPF and we have

also proposed a mechanism to adjust the comparator’s offset. A proto type chip fabricated

in AMS 0.35um CMOS technology has demonstrated its effectiveness. With 23

comparators tested, all of their offset voltages are brought down to below 0.8mV, while

its initial value can be as high as above 20mV.

IV
 This technique is further developed and applied to a 6 bit 500MHz flash ADC design,

which has been implemented in IBM 0.13um CMOS technology. The 63 fully dynamic

comparators used in this 6 bit flash ADC are background calibrated in a serial manner,

where a general control scheme is proposed. To optimize the calibration technique for use

in such a system, SAR search algorithm is adopted for calibration of each comparator,

instead of the linear search algorithm used initially. Simulation result has shown that the

flash ADC, including T&H circuit, resistor ladder and encoder, consumes only 9.5mW of

power running at 500MHz.

V
 List of Tables

Table 2.1 Summary of 6 bit flash ADC designs ............................................................... 14

Table 2.2 Summary of flash ADC with foreground calibration techniques ..................... 18

VI
 List of Figures

Figure 2.1 Typical conventional flash ADC design............................................................ 5

Figure 2.2 Resistive averaging [2]. ..................................................................................... 8

Figure 2.3 Interpolation [3]. .............................................................................................. 10

Figure 2.4 Preamplifier array with resistive averaging combined with interpolation [6]. 11

Figure 2.5 Preamplifier schematic. ................................................................................... 13

Figure 2.6 A comparator offset calibration scheme [9]. ................................................... 15

Figure 2.7 Imbalanced fully dynamic comparator [10]. ................................................... 17

Figure 2.8 Power consumption versus clock frequency of fully dynamic flash ADC [10].

........................................................................................................................................... 17

Figure 2.9 ENOB degradation of flash ADC with respect to supply voltage and input

common mode voltage [10]. ............................................................................................. 19

Figure 2.10 Comparator with random chopping [11]. ...................................................... 20

Figure 3.1 Comparator with proposed background calibration technique. ....................... 23

Figure 3.2 Effect of chopping for the input signal in the frequency domain. ................... 25

Figure 3.3 Amplitude spectrum of chopped comparator outputs with and without input

offset voltage. .................................................................................................................... 26

VII
Figure 3.4 Impulse response of accumulator with Ns=1000. ........................................... 28

Figure 3.5 Convergence behavior of comparator offset with the calibration. .................. 30

Figure 3.6 Two stage comparator with preamplifier followed by fully dynamic latch. ... 31

Figure 3.7 Block diagrams of the feedback loop. ............................................................. 33

Figure 3.8 Configuration to measure comparator input referred offset voltage. .............. 35

Figure 3.9 Die photo of the test chip................................................................................. 37

Figure 3.10 Histogram of comparator input referred offset voltage (absolute value)

before and after calibration. ............................................................................................. 38

Figure 4.1 Comparator with calibration block .................................................................. 41

Figure 4.2 Convergence behavior of comparator offset with SAR algorithm. ................. 43

Figure 4.3 Convergence behavior of comparator offset with linear search algorithm. .... 44

Figure 4.4 DNL and INL without calibration ................................................................... 46

Figure 4.5 Output PSD without calibration ...................................................................... 47

Figure 4.6 DNL and INL with calibration ........................................................................ 47

Figure 4.7 Output PSD with calibration ........................................................................... 48

Figure 4.8 Pseudo differential PMOS source follower. .................................................... 50

Figure 4.9 Replica biasing for the source follower. .......................................................... 52

Figure 4.10 Source follower with sampling switch and capacitor. ................................... 53
VIII
Figure 4.11Simulated output power spectral density of a 243MHz sine wave sampled by

proposed T&H circuit at 500MHz. ................................................................................... 54

Figure 4.12 Fully dynamic comparator with MOS cap array. .......................................... 55

Figure 4.13 Comparator output voltage at different phases within one clock cycle......... 56

Figure 4.14 Comparator output with a SR latch. .............................................................. 57

Figure 4.15 Schematic of SAR logic [19]. ........................................................................ 61

Figure 4.16 Comparator with circuits for calibration. ...................................................... 62

Figure 4.17 Simulation result of comparator output with chopper placed at different

positions. ........................................................................................................................... 64

Figure 4.18 Metastability converter schematic. ................................................................ 65

Figure 4.19 Calibration control block diagram. ................................................................ 67

Figure 4.20 Illustration of first and second order bubble errors [20]................................ 69

Figure 4.21 3 input AND gate used as encoder to suppress bubble error [3]. .................. 70

Figure 4.22 Binary coded ROM with metastability error [20]. ........................................ 71

Figure 4.23 Gray coded ROM with metastability error [20]. ........................................... 72

Figure 4.24 Schematic of DFF for ROM output. .............................................................. 73

Figure 4.25 Simulation result of ROM output and DFF output. ....................................... 74

Figure 4.26 Simulation result of clock signals.................................................................. 75

IX
Figure 4.27 Percent of power consumption by each block of the flash ADC. ................. 78

Figure 4.28 Flash ADC layout design............................................................................... 81

Figure 4.29 Post layout simulation results of the flash ADC ........................................... 82

X
 List of Abbreviations

ADC Analog to digital converter

CMFB Common mode feedback

CMOS Complementary metal oxide semiconductor

DAC Digital to analog converter

DEMUX Demultiplexer

DFF D flip flop

DNL Differential nonlinearity error

ENOB Effective number of bits

FIR Finite impulse response

INL Integral nonlinearity error

FoM Figure of merit

LPF Low pass filter

LSB Least significant bit

MOS Metal oxide semiconductor

MUX Multiplexer

OTA Over transconductance amplifier

ROM Read-only memory

SAR Successive approximation registers

XI
SNR Signal to noise ratio

SNDR Signal to noise plus distortion ratio

T&H Track and hold

UWB Ultra wide band

XII
 Chapter 1 Introduction

Chapter 1 Introduction

Over the years, development of digital integrated circuit has closely followed

Moore’s Law. As a result, transistor size has greatly shrunk and the speed of digital

circuit has been exponentially increased. This trend, which still continues today, widens

the gap between the digital circuit and its analog counterpart, for which the technology

advance is not as beneficial. On one hand, there exists very high speed digital circuit with

its ever growing processing power and efficiency. On the other hand, analog circuit

struggles and largely fails to keep pace. To make matter worse, most of systems need to

communicate with the real analog world at some point, so that analog interface circuit,

although usually being the limiting factor in the whole system, is still indispensable. It is

thus desirable to push the analog/digital boundary closer to the real world, where the

system can take better advantage of the high speed digital circuit.

This trend puts high pressure on analog circuit designers to develop very high speed

interface circuits, namely, analog to digital and digital to analog converters (ADCs and

DACs) that can keep up with the digital world yet still maintains other desirable attributes

like low power consumption and small chip area. With shrinking of available power
1
 Chapter 1 Introduction

supply voltage and a number of new issues brought about by greatly reduced transistor

size, this task seems to be more daunting than ever.

Particularly, there is a category of applications including disk read channel, UWB

receiver and wired or wireless communication system demanding for high speed (above

500MHz) and comparatively low resolution (4 to 8 bits) ADCs. Among various ADC

architectures, flash ADC suits this purpose favorably because of its inherent parallel thus

very fast structure and low signal latency. Also, the large area overhead that comes with

this structure is less severe when put in a low resolution context. Therefore, it is of great

interest to develop high speed and low power flash ADC that can be integrated in these

systems.

Although requirements for different applications may have emphasis on different

aspects, the ultimate goal is always to push for higher performance at lower power

consumption. To achieve this goal, researchers have come a long way from the

conventional structure and developed various flash ADC designs, some of which will be

discussed in Chapter 2. The most critical component in a flash ADC is the comparator,

where a bunch of techniques are proposed to mitigate or circumvent the inherit tradeoff
2
 Chapter 1 Introduction

between performance and accuracy. Comparator calibration techniques seem to be fairly

effective in this aspect. By leaving the problem to after the chip’s fabrication, where non

idealities are determined and can be measured, rather than in the design phase, where

they can only be described in a statistical sense, these techniques are more effiecient so as

to avoid large overhead that usually results in large power consumption. They have the

potential to give designers more freedom during the circuit design phase.

This work focuses on designing of a flash ADC that utilizes fully dynamic

comparators, which is largely made possible by incorporating a background comparator

offset calibration technique developed earlier. The resulted benefit is much relieved

frontend design and significantly less power consumption, comparing to more

conventional designs. This thesis is organized as follows. Chapter 2 gives an overview of

existing flash ADC designs, where their performances are compared. Chapter 3

introduces a proposed background comparator calibration technique with circuit

implementation in AMS 0.35um CMOS process and corresponding measurement results.

Chapter 4 gives detailed account of a flash ADC design that utilizes fully dynamic

comparators implemented in IBM 0.13um CMOS technology, together with its

calibration and control circuits, while the last chapter concludes this work.
3
 Chapter 2 Overview of Flash ADC Designs

Chapter 2 Overview of Flash ADC

Designs

To boost speed and reduce power consumption, in the mean time adapt to more

advanced technology, the design of flash ADC has evolved from conventional structure

to more complicated structures that incorporate techniques including resistive averaging,

interpolation and sophisticated calibration. This chapter discusses several published flash

ADC designs implemented in various technologies. Comparison of their performances is

made in the hope of revealing the ongoing trend in this aspect.

2.1 Conventional Flash ADC

A conventional flash ADC (figure 2.1) has a track and hold (T&H) frontend, a

comparator array and a digital decoder that converts the thermal meter code produced by

the comparators to valid N bit binary output. Also, a resistor ladder is used to generate

required reference voltage at the input of each comparator. The parallel structure ensures

a high operation speed and minimized conversion delay. The necessity of a T&H circuit

is mandated by the fact that due to clock delay, individual comparator may sample the

4
 Chapter 2 Overview of Flash ADC Designs

same input at different instants, causing severe problems under certain circumstances.

However, adding an additional frontend stage that can hold the input while being sampled

by the comparator array mitigates this problem and relaxes layout requirement of clock

route.

Figure 2.1 Typical conventional flash ADC design.

Evidently, comparator plays a very crucial part in this structure. Not only its speed

determines the highest sampling rate achievable by the ADC, but also, its key

5
 Chapter 2 Overview of Flash ADC Designs

characteristics will largely affect the overall dynamic and static performance, especially

note that the offset of each comparator directly contributes to DNL and INL, two very

important performance indicators.

Assuming the comparator employs preamplifiers, as is often necessary for high speed

flash ADC implemented in deep submicron technology, the overall offset is dominated

by the first stage preamplifier, which can be approximated by the following equation [1],

AVT
σ OFFSET = (2.1)
WL

where WL is the gate size of input transistor and AVT is a process determined factor.

Based on this equation, the only way to reduce offset variation is to increase the input

transistor size. Once the size is determined, load capacitance of the T&H circuit can be

subsequently estimated, which leads to its transconductance and power consumption.

Also, kickback noise and different feedforward and feedback routes from the comparator

to its reference input, along with mismatch considerations, determine the total resistance

of the resistor ladder, which usually contributes a considerable portion of power

consumption. Therefore, the comparator design assumes such a pivotal position in

conventional flash ADC design that its importance can hardly be overstated.

6
 Chapter 2 Overview of Flash ADC Designs

2.2 Flash ADC Designs with Resistive Averaging and Interpolation

The simple tradeoff discussed above in the conventional structure is no longer an

optimal or even viable solution for designs in deep sub micron technologies because of

significantly reduced supply voltage and paramount need of low power design. If one still

attempts to achieve the desired offset by simply increasing the transistor size, the likely

result would be unacceptable power consumption and (or) chip area, as discussed above.

It is easily identified that these tradeoffs primarily originate from input referred offset

voltage prescribed by equation 2.1. Consequently, researchers have put a lot of efforts on

circumventing this offset issue.

Kattmann and Barrow [2] proposed a technique to address this very problem. In the

configuration shown in figure 2.2, all the preamplifiers are connected together by a

resistive network. Thus the originally uncorrelated input offsets contributed by individual

preamplifiers are correlated and their effect is averaged. In other words, their input

referred offset contributions are reduced. The reduction factor (as an indication of its

effectiveness) is determined by the ratio between the unit resistor R2 in the resistive

network and the load resistance R1. This scheme, though proved to be effective, suffered

7
 Chapter 2 Overview of Flash ADC Designs

from several serious drawbacks. First of all, as the edge of the resistive network is not

properly terminated, preamplifiers at both ends tend to cause a large INL. Moreover, the

reduction of offset, i.e. the reduction of input transistor size, is still more or less limited.

Figure 2.2 Resistive averaging [2].

Figure 2.3 shows another technique called interpolation [3], which is conceived from

a totally different point of view. It aims to reduce the number of preamplifiers needed to
8
 Chapter 2 Overview of Flash ADC Designs

achieve the same number of bits at the ADC output. The comparators used in a flash

ADC are essentially a bunch of zero crossing points, the function of which can be

abstracted to comparing of the input voltage to a certain reference. In this aspect, the

physical existence of comparators are not essential, as long as corresponding zero

crossing points can be created. Considering that outputs from preamplifiers are linear,

this can be easily achieved. Figure 2.3 shows two pair of differential outputs from two

preamplifiers [3]. VX1 and VY1 form a zero crossing point at Vr1 while VX2 and VY2 form

another zero crossing point at Vr2, these are two original zero crossing points produced by

two physical preamplifiers A1 and A2. If one takes a look at VY1 and VX2 (or VY2 and VX1),

their difference produce an additional zero cross point at

Vm = (Vr1+Vr2)/2 (2.2)

which is the same as output from another preamplifier inserted in the middle of A1 and

A2 with Vm as its reference. Therefore, by simply applying this technique, the number of

preamplifiers required, which is 2 to the power of the number of bits, can at least be

halved.

9
 Chapter 2 Overview of Flash ADC Designs

Figure 2.3 Interpolation [3].

Combining the two techniques together forms a more effective solution [4-7], as

shown in figure 2.4. In this case, averaging resistors are in the mean time used as voltage

divider so that even more zero crossing points can be created. To provide enough voltage

gain while achieving high speed, there are usually multiple stages of preamplifiers

involved and each stage is interpolated and averaged at its output. As a result of this

powerful combination, the number of first stage preamplifiers needed is reduced by

several times. For example, the 6 bit flash ADC shown in figure 2.4 [6] needs only 9 first

stage preamplifiers, instead of 63 in the conventional structure. Although the resistive

network still needs to be properly terminated so as to cause minimum distortion, it greatly

relieves the requirement for the T&H circuit, whose gm as well as power consumption can

be accordingly minimized due to significantly less number of preamplifiers to drive.

10
 Chapter 2 Overview of Flash ADC Designs

Figure 2.4 Preamplifier array with resistive averaging combined with interpolation [6].

Over the years, this combination of techniques has almost been pushed to its

perfection by researchers around the world, but it is also limited in certain aspects. For

one thing, the interpolation factor can only be so high that the outputs from adjacent

preamplifiers do not exceed their output linear range. This is probably why most of these

works choose a minimal interpolation factor of 2. For another, more importantly, this

comparator structure with multi stage preamplifiers followed by a dynamic latch has its

inherent disadvantages that restrict it from achieving high power efficiency, which will

be discussed in the following paragraph.

11
 Chapter 2 Overview of Flash ADC Designs

One of the main reasons to add multi stage of preamplifiers is to suppress the large

offset variation from the dynamic latch. The combined gain of all the stages has to be

high enough so that when the latch’s offset is referred back to the input, it becomes

insignificant, as shown by equation 2.3,

σ Latch
σi = (2.3)
A1 A2 L An

where σ is latch offset variation referred back to the input of first stage and An is the gain

of the nth preamplifier. For the sake of power consumption, there should not be too many

stages of preamplifiers as each stage consumes a considerable amount of power.

However, if the number of stages is decreased, the gain of each stage has to be boosted so

as to get the same overall gain while speed has to be maintained. This may lead to an

even high power consumption in a single stage than multi stages of preamplifiers.

Considering the simple preamplifier structure shown in figure 2.5, this would end up with

an increase either in gm of input transistor or load resistance R. Both are undesirable as

larger gm corresponds with larger current or W/L ratio, where larger R1 may reduce the

speed achievable and the effectiveness of averaging. In the end, the number of stages

selected is a result of meticulous pondering over these complicated tradeoffs that involve

12
 Chapter 2 Overview of Flash ADC Designs

quite a number of factors. Nevertheless, the existence of several stages of preamplifiers

results in a significant static power consumption.

Figure 2.5 Preamplifier schematic.

Table 2.1 summarizes several published 6 bit flash ADC designs. The first four

designs apply both interpolation and resistive averaging. And the last one [8] is a more

conventional design list here for comparison. It can be seen that interpolation and

resistive averaging do help to achieve a better result. Advance of technology also helps

tremendously to reduce overall power consumption. However, it is worth noting that

while analog part usually consumes more than half of the total power, preamplifiers take

up at least 70% of power consumed by the analog part. In some sense, that huge amount

13
 Chapter 2 Overview of Flash ADC Designs

of power is mostly dedicated to tackle the offset problem, as the size of preamplifiers is

prescribed by offset voltage.

Paper Sampling Supply Process Total Analog Preamplifier

Rate voltage (um) power power power
(MHz) (V) (mw) (mw) (mw)*
(Mic, 01)[4] 1300 3.3 0.35 545 272.5 233
(Aet, 02)[5] 1600 1.95(A) 0.18 340 NA NA
2.35(D)
(Aym, 08)[6] 1600 1.5 0.13 180 126.3 106.1
(Kaz, 08)[7] 3500 0.9 0.09 98 46 32.3
(Koe) [8] 1000 3.3 0.35 700 535 304

Table 2.1 Summary of 6 bit flash ADC designs

*Calculated based on data given in paper

2.3 Flash ADC Designs with Calibration Techniques

Rather than averaging and interpolating to reduce the comparator offset, another

possible solution is to find a way to calibrate it, which may get around the fundamental

tradeoff shown in equation 2.1 in a more complete way, and in turn drastically reduce the

size of preamplifiers [9], which is previously limited by linearity requirement. In certain

cases, the use of preamplifiers can even be completely eliminated [10]. In [9], a

foreground calibration scheme is proposed, which is illustrated in figure 2.6. During the

calibration process, Ref[k], the ideal reference voltage of kth stage, is attached to the
14
 Chapter 2 Overview of Flash ADC Designs

positive input of the preamplifier and the initial negative input is also set to Ref[k]. If the

comparator does not have offset, the output, after passing an alternating amplifier A,

should have a zero mean. If it is not the case, DFF in Digital Calibration Circuit will

overflow and adjust the voltage at the negative input so as to compensate for the offset.

The authors claim that by applying calibration, the required size of preamplifiers is

reduced by 278 times, which means the load of T&H circuit is also reduce by 278 times.

The flash ADC they designed consumes only 12mW of power at 800MHz sampling rate

and though the data is not given directly, the analog part is estimated to have contributed

only 3mW to total power consumption.

Figure 2.6 A comparator offset calibration scheme [9].

15
 Chapter 2 Overview of Flash ADC Designs

In [10], the authors proposed an even more aggressive foreground calibration

technique that enables a 5 bit flash ADC design with almost complete dynamic

components, even the need for reference ladder is eliminated by adopting built in

reference. Also, folding technique is applied to halve the number of comparators needed.

Most of the benefits can be attributed to the fully dynamic comparator structure

shown in figure 2.7. It is purposefully made imbalanced with a PMOS P4 acting as a

MOS cap at the left side. The MOS cap can be sized in such a fashion that it causes an

initial offset coarsely the same as the desired reference voltage. And N2 is used to finely

calibrate that imbalance as well initial offset. An advantage of this structure is that it does

not need a resistor ladder to generate the necessary reference voltage. Consequently,

power consumption from the resistor ladder, which normally takes a considerable portion

from the overall power, is removed.

While working at a sampling frequency of 1.75GS/s, this ADC consumes only

2.2mW of power and has a very low 50fJ/step FoM. Also, due to its dynamic nature, the

total power consumption has a linear relationship with its sampling rate (figure 2.8),

16
 Chapter 2 Overview of Flash ADC Designs

which is similar to pure digital circuit and is highly desirable. From this impressive result

we can see that improvement is huge as long as static power consumption is eliminated.

Figure 2.7 Imbalanced fully dynamic comparator [10].

Figure 2.8 Power consumption versus clock frequency of fully dynamic flash ADC [10].

17
 Chapter 2 Overview of Flash ADC Designs

Paper Sampling Supply Resolution Process Total

Rate voltage (Bit) (nm) power
(MHz) (V) (mw)
(Chu, 09)[9] 800 1.2 6 65 12
(Bob, 09)[10] 1750 1 5 90 2.2

Table 2.2 Summary of flash ADC with foreground calibration techniques

Table 2.2 summarizes performance of two ADC designs that use foreground

calibration techniques. From Table 1.1 and Table 1.2, it can be concluded that calibration

techniques generally achieve a better result comparing to interpolation and resistive

averaging techniques. Even if one takes the more advanced processes into account, the

power consumption is still one or two orders of magnitude less. However, they take

additional calibration steps before the ADC can be put into use and once the calibration is

done, it can no longer track the changes of conditions such as supply voltage, temperature

and clock frequency. As pointed out in [10] and shown in figure 2.9, when the supply

voltage or input common mode voltage changes with respect to the value at which the

ADC is calibrated, its low frequency ENOB degrades. In certain applications, this can

pose as a serious issue. If the operating point drifts far away from the optimal point, at

which the flash ADC is calibrated, the overall performance of the whole system might

deteriorate considerably due to the degradation of flash ADC performance. It is even

18
 Chapter 2 Overview of Flash ADC Designs

worse if the ADC in the mean time needs to run constantly and no time can be spared for

occasionally recalibration. The price paid for maintaining a stable working environment

can be huge.

Figure 2.9 ENOB degradation of flash ADC with respect to supply voltage and input

common mode voltage [10].

On the other hand, background calibration technique does not suffer from this

problem. Due to its background nature, it will be able to adapt to environmental changes,

as long as such changes can be compensated. The challenge lies in how the offset

information can be extracted without interrupting the ADC’s normal job. Moreover,
19
 Chapter 2 Overview of Flash ADC Designs

methods that can precisely calibrate the comparator offset without causing too much

additional trouble should be developed. In [11], the authors resort to random chopping to

achieve an averaging effect upon offset (figure 2.10). The input chopper S1 and output

chopper S2 are controlled by the same random signal r. They claim that this technique,

when used in a flash ADC, will improve dynamic performance because it will reduce the

spurious tones while raising noise floor. However, as the offset is not further dealt with

by calibration, it is unlikely to help with static performances in terms of DNL and INL. In

[12], a background calibration technique also based on random chopping is proposed.

Although a comprehensive analysis and abundant simulation results were given, no actual

ADC was implemented, nor did the authors get into the details as how the comparator can

be calibrated without causing much performance degradation.

Figure 2.10 Comparator with random chopping [11].

20
 Chapter 2 Overview of Flash ADC Designs

Therefore, this work focuses on developing background comparator calibration

technique and more importantly, its application to a low power and high speed flash ADC

design.

21
 Chapter 3 A Background Comparator Offset Calibration Technique

Chapter 3 A Background Comparator

Offset Calibration Technique

In this chapter, a background comparator offset calibration technique will be

discussed in detail. This technique utilizes chopping to extract offset information without

interrupting comparator’s work. Also, the additional circuit developed to calibrate the

comparator has negligible influence on its performance. A prototype chip was fabricated

in AMS 0.35um CMOS technology and measurement result has demonstrated its

effectiveness.

3.1 System Overview of the Proposed Background Comparator Offset

Calibration Technique

To develop a background calibration technique, there are two major issues. One is to

find a mechanism to extract the offset information in the background. The other is to

make the comparator offset adjustable, preferably in a linearly way, within certain

accuracy and range as mandated by the application. The rest is just to interface them and

put them together in a feedback loop, where the offset will be automatically forced to
22
 Chapter 3 A Background Comparator Offset Calibration Technique

converge to a small value. The first issue will be discussed in the following section while

the later will be introduced when it comes to circuit implementation.

Figure 3.1 Comparator with proposed background calibration technique.

Figure 3.1 shows system architecture of the proposed technique. The background

operation is guaranteed by the pair of choppers at both input and output. They are

controlled by the same signal so they either pass the signal directly or invert both input

and output at the same time. As a result, Dout will always be the same as the output of a

23
 Chapter 3 A Background Comparator Offset Calibration Technique

normal comparator. As will be discussed later, the comparator structure is slightly

changed so that its offset can be adjusted. The accumulator is used to pick up the

information of input referred offset voltage from Dq (see figure 3.1) and the trimming

block is the interface between extraction and calibration, so that offset can be calibrated

accordingly. Thus, when the comparator is at work, the input referred offset is forced by

the feedback loop to converge at a low value determined by how precise it can be

calibrated.

The key issue of this scheme is to effectively extract the offset. As in operational

amplifier design, the first chopper at the comparator input moves the input signal Vin,

which we assume originally expands from –fin to fin in the frequency domain, to center

around the chopping frequency fc and its odd harmonicas. The relationship between Vc

and Vin is determined by the following equations [13],

+∞
Vc ( f ) = ∑ AV
k =−∞
k in ( f + k ⋅ fc ) (3.1)

k
Ak = sin c( ) (3.2)
2
where k is odd integer and fc is the chopping frequency. Figure 3.2 shows the effect of

input chopping. It might be of some interest to note that if chopping frequency fc is

24
 Chapter 3 A Background Comparator Offset Calibration Technique

greater than input signal bandwidth fin, the frequency band around dc will be void after

chopping. So the input referred offset Vos, which occupies a very narrow frequency band

near dc, is clearly separated from input signal in the frequency domain. Therefore, the

signal bandwidth must be less than the chopping frequency for the technique to be

effective.

Figure 3.2 Effect of chopping for the input signal in the frequency domain.

Chopped input signal Vc, plus comparator input referred offset voltage Vos, is then

quantized by the comparator. This process does two things. First, the signal is sampled by

the comparator with clock frequency fs that is purposely set to 2fc. From equation 3.1, the

25
 Chapter 3 A Background Comparator Offset Calibration Technique

input signal after chopper is centered around odd integer times of fc, with the original

signal bandwidth fin < fc. By sampling it at 2fc, the resulted signal would still be centered

at odd times of fc. Since fin < fc, it ensures no signal is aliased back to dc. Second, the

nonlinear quantization takes place and transfers the analog input to digital output. From

the amplitude spectrum of the output signal, it is found that the offset information is not

spoiled by this process.

Voltage (V)

Figure 3.3 Amplitude spectrum of chopped comparator outputs with and without input

offset voltage.

26
 Chapter 3 A Background Comparator Offset Calibration Technique

This can clearly be demonstrated in the frequency domain. Figure 3.3 shows

amplitude spectrum of two comparator outputs with the same sine wave as the input, the

only difference being one has and the other does not have input offset voltage. It is

clearly seen that when the input referred offset is present (the upper graph), there is a

spike at dc, which indicates offset voltage that sits around dc. Note that for the purpose of

demonstration, the offset voltage is set to a comparatively large value in this simulation.

To retrieve offset information, the following stage needs to be a low pass filter (LPF),

preferably with very low cut off frequency and high dc gain, because this LPF will

eventually be put in a feedback loop, actual value of the input referred offset voltage is

not important, as long as its sign is known. Based on this observation, a digital

accumulator is selected to accomplish the task. By setting its output limits to N and –N

respectively, it will overflow whenever the output reaches either limit, and the direction it

overflows indicates offset polarity. The accumulator can be regarded as a finite impulse

response (FIR) digital LPF with the following system function

1− e N S
−j ω

H (ω) = (3.3)
1 − e− jω

27
 Chapter 3 A Background Comparator Offset Calibration Technique

where Ns is the number of comparator output (the number of samples) accumulated

before it overflows. From equation (3.3), we can see the dc gain, as well as the order of

the FIR filter is determined by Ns. Figure 3.4 shows impulse response of a LPF with Ns=

1000.

Figure 3.4 Impulse response of accumulator with Ns=1000.

This gives an interesting system property. Smaller offset generally ends up with

larger Ns, as a larger number of samples needs to be averaged in order to pick the dc

value out. For the system function of the FIR filter shown in equation (3.3), it means a

larger dc gain and narrower cutoff frequency, which are both helpful to the signal to noise

ratio (SNR). In that sense, the filter is adaptive so that it can still be effective even

dealing with a small offset value. When Ns becomes very large, it approaches ideal

accumulator that has the following system function

1
H (ω) = (3.4)
1− e− jω
In this extreme case, the gain of the ideal accumulator is infinite at dc.
28
 Chapter 3 A Background Comparator Offset Calibration Technique

Once the polarity of offset is known, the task of the following trimming stage is

simple. As discussed earlier, it is essentially an interface circuit between the extraction

part and calibration part. It does so by recording the number of overflow happened in the

previous stage and according to the direction of each overflow, i.e. the sign of offset at

that moment, it adjusts the comparator to compensate for it in a step by step manner,

assume that the comparator can be adjusted linearly.

The last but not least is to actually make the comparator adjustable. While it is not

going to be discussed in detail here, there are primarily two design issues involved. The

first is how precise the comparator offset needs to be adjusted and the second is the

calibration range. The former directly relates to the ADC that the comparator is applied to

because the smallest calibration step should be a fraction of ADC’s least significant bit

(LSB). The calibration range, however, depends on the comparator’s offset variation in a

certain process, which is determined by its structure and transistor size. Assuming the

variation can be described by Gaussian distribution, the calibration range needs to cover

at least three times of the variation to ensure a high yield.

29
 Chapter 3 A Background Comparator Offset Calibration Technique

Once the components are connected together and feedback loop is formed, the

comparator offset will diminish with the number of samples taken and eventually

converges to the minimum calibration step. Figure 3.5 shows convergence behavior of

the system with an initial offset voltage of 21mV, a calibration step of 1.8 mV and

accumulator limit of 10. Random noise is also added before the input enters the

comparator. The result shows that the comparator offset diminishes in an exponential

way with respect to the number of samples and ends up fluctuating around the smallest

step. From the figure, we can see that the last few steps take much more time than the

first few steps.

Figure 3.5 Convergence behavior of comparator offset with the calibration.

30
 Chapter 3 A Background Comparator Offset Calibration Technique

3.2 Circuit Implementation

A two stage comparator with a preamplifier followed by a fully dynamic latch [14] is

used in this design, whose schematics are shown in figure 3.6. This kind of structure is

also the choice of most flash ADC designs discussed earlier, except that they generally

use multi stages of preamplifiers. The preamplifier has a certain amount of voltage gain

to suppress the large offset from the dynamic latch, while the latch provides high speed

comparison due to its positive feedback configuration. However, in this case, the

preamplifier is not primarily employed to suppress offset, but rather for the convenience

of adjusting its offset.

Figure 3.6 Two stage comparator with preamplifier followed by fully dynamic latch.

31
 Chapter 3 A Background Comparator Offset Calibration Technique

A PMOS transistor M3 acting as a controllable current source is inserted in parallel

with two load resistors, with either S1 or S2 closed, so that the single transistor can

compensate for offset in both directions. Using small signal approximation, the input

referred offset is determined by the following equation

I3
∆VOS = (3.5)
gm1,2

where I3 is current injected by M3. The linear relationship between injected current and

offset voltage shown by this equation is extremely desirable as it can be controlled in a

straightforward and reliable way. The magnitude of this current is decided by a simple

current output DAC, the output LSB of which, together with gm of the input transistor,

determines the LSB of offset calibration. In real application, gm is determined by the

bandwidth requirement for the preamplifier and we can easily set the LSB of DAC to

achieve the desired minimal calibration step, which is set to around 0.5mV in our design.

This method, with only two switches and a PMOS transistor, adds negligible load to the

preamplifier and in the mean time provides linear and wide range offset calibration.

32
 Chapter 3 A Background Comparator Offset Calibration Technique

Figure 3.7 Block diagrams of the feedback loop.

Components that constitute the calibration circuit are shown in figure 3.7. As

discussed earlier, an accumulator is used as a LPF to extract offset from comparator

output. Its limit N is set to 64, 2 to the power of 6, so that it can be easily implemented

with a 6 bit up/down counter. The trimming control part, acting as an interface between

accumulator and the current DAC, records the number of overflow happened as well as

the direction of each overflow and use that information to adjust DAC output current and

switch control signal S1/S2. The DAC current output is binary coded and its output is

controlled by NMOS switches. Each time overflow happens, its output current is

increased or decreased by one LSB, depends on the direction. As the current is mirrored

to M3 in the preamplifier, the offset is then calibrated by one LSB at a time. It is worth

33
 Chapter 3 A Background Comparator Offset Calibration Technique

noting that unlike a DAC used for data conversion, the output of this DAC does not have

to be precise because it is in a feedback loop, the only concern is that its current output

LSB should be restricted to below a certain limit in order to meet the calibration

requirement. As a result, the design effort is greatly relieved and hence it can be realized

with the simple structure as shown in figure 3.7.

For use in an ADC, the minimal calibration step should be set to a fraction of its LSB,

and the calibration range should at least be able to cover 3 times of comparator offset

variation to ensure a high yield. Together they will determine how many bits are needed

for the trimming control circuit and DAC, as well as its LSB. The selection of

accumulator limit involves tradeoff between convergence time and fluctuation after

convergence. A smaller limit will certainly result in a faster convergence, but the residual

offset will be bumpier. In addition, the limit must be large enough to make sure a strong

averaging effect of the comparator output. Otherwise the probability of misjudgment will

increase.

34
 Chapter 3 A Background Comparator Offset Calibration Technique

3.3 Measurement Results

Figure 3.8 Configuration to measure comparator input referred offset voltage.

A prototype design has been fabricated in a 0.35µm standard CMOS process, and the

test configuration shown in figure 3.8 is adapted to measure the comparator input referred

offset voltage. Since output from the comparator is essentially a digital signal, the output

chopper is implemented with a simple 2-input XOR gate.

Intuitively, measurement of input referred offset voltage can be achieved by the

following process. Attach a fixed dc voltage Vdc at the comparator’s negative input and

35
 Chapter 3 A Background Comparator Offset Calibration Technique

vary the voltage at the other input around Vdc. If this is an ideal comparator, there should

be a point below which the output is logic high and above which the output is logic low at

every sample. By definition of input referred offset voltage, it can be calculated by Vdc

minus the input voltage value at that point.

However, due to noise of dc input signal, this point cannot be observed on

oscilloscope. There is actually an input transition range within which output shows

random behavior, bouncing between logic high and logic low with different samples. In

this case, it is still possible to measure offset by using the same method as above. But

instead of finding the point on oscilloscope, probability of output at a given dc input level

is measured [15]. Since noise of dc input signal has a zero mean, what would happen at

the point is that comparator output would have a 50% probability to be high. By taking a

large number of samples (65536) at a given dc input level and calculating its probability,

then varying the input signal accordingly and iterating the same process, the point that is

conceptually equivalent to the point in the ideal case would be identified.

To test the effectiveness of proposed calibration technique, input offset voltage would

be measured before and after calibration for comparison. A sine wave is injected for the
36
 Chapter 3 A Background Comparator Offset Calibration Technique

comparator to calibrate itself first, and then the input offset voltage is measured following

the above method.

The sampling rate of the comparator is 10MHz, which makes chopping frequency

5MHz. The input sine wave has a frequency of 3.43MHz. Die photo of the test chip is

shown in figure 3.9. Totally 23 comparators were tested. The histograms of their input

referred offset voltages before and after calibration are shown in figure 3.10. The

horizontal axis indicates the input referred offset voltage range in absolute terms and the

vertical axis shows the number of comparators falls in that range. It is clearly shown that

while the initial offset is as high as 25mV, after calibration, they are brought down to

below 0.8mV.
COMS

COMS

CAP

COMS

Figure 3.9 Die photo of the test chip.

37
 Chapter 3 A Background Comparator Offset Calibration Technique

Before After Calibration

Calibration 10

NUMBER
10
NUMBER

5
5

0 0
5 10 15 20 25 0.2 0.4 0.6 0.8
Offset (mV) Offset (mV)

Figure 3.10 Histogram of comparator input referred offset voltage (absolute value) before

and after calibration.

38
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Chapter 4 Design of a 6 bit 500MHz Flash

ADC Employing Fully Dynamic
Comparators

As discussed in chapter 2, flash ADC designed with fully dynamic comparators

shows a superior performance especially in terms of power consumption. Without multi

stages of preamplifiers, large portion of analog power consumption can be eliminated.

The limiting factor, however, is its large initial offset. Before it can be applied to a flash

ADC, which has a stringent requirement for comparator offset, calibration techniques

must be utilized to mitigate this problem. The background comparator offset calibration

technique introduced in the previous chapter is tailored to suit the use on fully dynamic

comparators here. With this technique, offset of each comparator is expected to be

reduced to a fraction of the ADC’s LSB and it would enable us to use small input

transistor size that is limited by other issues rather than offset voltage. Thus the power

consumed by comparators can be significantly reduced. The front end T&H circuit design

would also benefit from this because lower capacitance and kickback noise is expected

from the comparator array.

39
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

The major problem is to find a way to linearly adjust the fully dynamic comparator’s

offset voltage, while imposing minimal additional load to the comparator itself. This

proves to be a hard task. Unlike the case with preamplifier that has a linear output range

itself, where linear relation is easily obtained, the fully dynamic comparator operates by

taking advantage of positive feedback and it is difficult to control it in a linear way. This,

as will be shown in section 4.2, is solved by inserting binary coded MOS capacitor arrays

to the critical nodes of the fully dynamic comparator. The use of fully dynamic

comparators with calibration also imposes different requirements and tradeoffs from a

more traditional flash ADC design. Needless to say, with 63 comparators to be calibrated,

a general control scheme also needs to be designed, which involves generating additional

clock signal, disabling and enabling calibration with a certain comparator and sharing

some calibration components. Moreover, certain new issues have to be addressed with the

T&H circuit.

This chapter will first show simulation results obtained from a system model to

demonstrate its effectiveness and the rest will be dedicated to the detailed design of a 6

bit 500MHz flash ADC using IBM 0.13um CMOS technology.

40
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

4.1 System Model and Simulation

A 6 bit flash ADC system with background calibration is built with Matlab Simulink,

which the structure of actual circuit design is based on. Major components of a 6-bit

conventional flash ADC are 63 comparators. However, unlike conventional flash ADC,

here each comparator is equipped with a background offset calibration block, as shown in

figure 4.1 below.

Figure 4.1 Comparator with calibration block

The principle of the background calibration block is essentially the same as that of the

calibration scheme explained in Chapter 3. However, in the previous section, the offset is
41
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

calibrated by a linear search algorithm that change offset voltage in a step by step manner

by the calibration LSB. This process may be optimized by utilizing successive

approximation algorithm (SAR) [18]. By using SAR, the comparator is calibrated in a bit

by bit manner and it works as follows.

At beginning, the most significant bit (MSB) value for the calibration is assumed, and

the accumulator will then judge the polarity of the offset. If this proves to be too much,

i.e. sign of the offset has been reversed, the MSB is unselected and the bit next to it is

assumed. Otherwise if the offset is under compensated, i.e. sign of the offset has not

changed, MSB is asserted and the bit next to it is assumed. This process is iterated to the

last bit.

For an 8 bit calibration, the accumulator only needs to overflow 8 times to reach

convergence with any initial offset value for SAR while the former method may need as

many as 256 steps. At first glance, this may seem to be a great improvement to

convergence time. However, convergence speed has not actually benefitted that much. As

discussed in chapter 3, the convergence speed depends on residual offset and accumulator

needs more time to pick the offset out when the residual is small. Therefore, majority of
42
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

calibration time is spent on the last few steps. Although SAR needs fewer steps to

converge, the last several steps still take a long time because the residual offset is likely

to be small. Under certain circumstances, like the offset is initially small, SAR may even

take more steps than the previous method and converges slower. This point is illustrated

by simulation results shown in figure 4.2 and figure 4.3. Figure 4.2 shows a comparator

calibrated with SAR, the initial offset is 20mV, calibration LSB is 6 mV and it is a 5 bit

calibration. Figure 4.3 shows a comparator calibrated using linear method with

everything else set to be identical. In this case, SAR takes a larger number of samples,

thus more time to finally converge.

Figure 4.2 Convergence behavior of comparator offset with SAR algorithm.

43
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Figure 4.3 Convergence behavior of comparator offset with linear search algorithm.

As will be explained in Section 4.4, the selection of SAR instead of linear calibration

logic here is due to practical circuit design constrains rather than convergence time

considerations. With SAR, the indication of convergence can be simply obtained by

monitoring its last bit where as in a linear search algorithm the convergence time is not

easily estimated as it depends on various factors. With this critical information, all

comparators can be calibrated in a series manner with a single set of accumulator shared

among them.

44
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

As will be explained in Section 4.4, the selection of SAR instead of linear calibration

logic here is due to practical circuit design constrains rather than convergence time

considerations. With SAR, the indication of convergence can be simply obtained by

monitoring its last bit where as in a linear search algorithm the convergence time is not

easily estimated as it depends on various factors. With this critical information, all

comparators can be calibrated in a series manner with a single set of accumulator shared

among them.

The 6 bit flash ADC is modeled with 63 such comparators, each with its own

background calibration block. The input offset is assumed to have a Gaussian (Normal)

distribution, with zero mean and a 70mV variance, in accordance with Monte Carlo

simulation result from comparator circuit implementation in Section 4.3. The calibration

circuit has a LSB of 2mV and the SAR is of 8 bit, which ensures a ±256mV calibration

range that covers ±3∆ of input offset distribution. The accumulator overflow limit is set

to ±64, or 6 bit. Input signal is a 243MHz sine wave and the sampling rate is 500MHz,

which makes the chopping frequency 250MHz. Simulation is first run with calibration

blocks disabled so that offset at each comparator’s input is not being calibrated. Figure

45
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

4.4 shows its static performance with measured DNL and INL. Dynamic performance is

measured by PSD from its output. Resulted SNDR is 20.7687dB and ENOB is 3.1576.

From figure 4.5, it can be seen that 3rd harmonica distortion is at -28.11dBFS.

When calibration blocks are enabled, only samples after 300000th are taken to obtain

results, thus it is ensured that all comparators have already been properly calibrated. Not

only drastic improvement is achieved with DNL and INL, as shown by figure 4.6, note

that the scale of Y axis is different from figure 4.4. SNDR and ENOB has increased to

36.37dB and 5.75, respectively, as shown by figure 4.7. Now the 3rd harmonica distortion

is at -53.45 dBFS.

Figure 4.4 DNL and INL without calibration

46
Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Figure 4.5 Output PSD without calibration

Figure 4.6 DNL and INL with calibration

47
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Figure 4.7 Output PSD with calibration

The system simulation result clearly demonstrates effectiveness of this background

calibration scheme applied to a 6 bit flash ADC. The following sections within this

chapter will discuss the implementation of such a system at circuit level in detail.

4.2 Track & Hold Circuit

T&H circuit is necessary for high speed flash ADC to avoid clock dispersion [3]. The

high speed nature and moderate SNR requirement makes operational amplifier connected

48
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

in some sort of feedback configuration a less favorable solution even though they prevail

in higher resolution architectures like pipelined ADC. To reach the desired speed, the

OTA is expected to consume an unacceptable amount of power. Source follower, on the

other hand, has some desirable attributes like low output resistance and high driving

capability and is widely used in flash ADC designs [4], [6] and [7]. Source follower is

therefore chosen for our design. To accommodate differential input, pseudo differential

structure is implemented.

The first choice we have to make is between NMOS source follower and PMOS

source follower. Although NMOS would seem to be a much better choice in terms of

channel mobility, which in turn results in far superior power efficiency, the problem lies

with body effect. In a single well process, using NMOS would introduce a large signal

distortion due to body effect caused by large output swing. Therefore, PMOS source

follower, as shown in figure 4.8, has to be used in spite of its inferior performance.

49
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Figure 4.8 Pseudo differential PMOS source follower.

Another problem to be considered in this ADC design is the output common mode

voltage. From figure 4.9, we can see that it is roughly calculated as the input common

mode voltage plus gate source voltage of the input PMOS transistor. Due to process

variation and mismatch, the output common mode cannot be precisely controlled and is

expected to vary several tens of mVs, causing it to differ from common mode voltage of

the reference. This is less likely an issue if preamplifiers are used in the first stage. As

long as they have a reasonably large input linear range, the difference of common mode

voltage between input signal and reference can be cancelled out and cause negligible

error. However, the dynamic comparator we use here is highly nonlinear and the

50
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

difference will cause unwanted initial offset, which unfortunately cannot be corrected by

calibration.

Common mode feedback circuit is the usual choice to bring this issue under control.

However, conventional continuous common mode feedback (CMFB) technique [16]

inevitably loads the output and simulations have revealed that it is very difficult to find a

balance where output common mode voltage change can be tracked fast enough while

CMFB does not load the circuit in a serious way. Therefore, a replica biasing scheme that

consists of a feedback amplifier and a miniature source follower proportional to the

output source followers is proposed (figure 4.8). Vrefc is reference common mode voltage

and Vinc is input common mode voltage. Because of feedback loop formed by a

transconductance amplifier, the output common mode voltage is biased to be the same as

reference common mode voltage at a given input common mode voltage. The good thing

about this method is that reference common mode and input common mode voltage can

be precisely controlled. The miniature source follower is sized to be 1/20 of the source

followers used to drive the following stage, thus its power consumption is insignificant.

This scheme has the advantage of not intruding the signal path and therefore does not put

51
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

any extra load to the output of source followers. A large capacitor is inserted between

ground to Bias terminal, so as to reduce output feedback that may disturb the bias point.

Figure 4.9 Replica biasing for the source follower.

The transconductance amplifier employs a folded cascode structure to obtain high

gain. The feedback loop is actually formed by both the amplifier and PMOS transistor

M1. Because the application here has virtually no requirement for bandwidth, the

amplifier is designed to consume minimal power and no Miller compensation is needed.

Instead, a large cap is inserted between Bias and ground to make the first pole dominant

so as to solve the stability problem.

52
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Figure 4.10 Source follower with sampling switch and capacitor.

The pseudo source followers are preceded by a pair switches and sampling caps to

form the complete T&H circuit. The single ended version is shown in figure 4.10 for

simplicity. The designed input range is 1V Vpp differential and the common mode is set

close to 250mV. The output range is therefore 1V Vpp as well, because the gain of source

follower is close to one. The output common mode voltage is raised by about a transistor

threshold VT, to 750mV. Each sampling switch needs to accommodate an input that
53
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

ranges from 0 to 500mV and a simple NMOS or transmission gate switch would induce a

large signal dependent distortion. So a constant Vgs bootstrapped switch proposed by

A.M. Abo and P.R. Gray in [17] is adapted here. In simulation, the proposed T&H circuit

achieves a 61.59 dBc 3rd harmonica distortion with a 243 MHz input sine wave sampled

at 500 MHz, as shown in figure 4.11.

Figure 4.11Simulated output power spectral density of a 243MHz sine wave sampled by

proposed T&H circuit at 500MHz.

54
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

4.3 Fully Dynamic Comparator and Calibration

Figure 4.12 Fully dynamic comparator with MOS cap array.

Figure 4.12 demonstrates the comparator topology used in our design. Let us ignore

the MOS cap array on both sides for the moment and focus on operation of the

comparator. When the CLK signal is low, M3 and M4 are off and M5, M6, M7 and M8

are on. The output nodes and node X and Y will be charged to supply voltage. When the

CLK becomes high, M3 and M4 are on, while M5, M6, M7 and M8 are off. Node X and

Y will be discharged. Depending on the value of Vin+ and Vin-, one node will be

55
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

discharged at a faster speed and then the comparator enters regeneration phase, where one

output node will be pushed up to supply voltage and the other pulled down to ground.

Figure 4.13 shows how the voltages at output nodes change at different phases within a

complete clock cycle. Note that the comparator output is only valid during less than half

of a clock period. This is undesirable as it significantly reduces processing time of the

following stage and increases the probability of error. To solve this, a SR latch can be

added to the comparator output so that the valid output is kept for almost one whole clock

period, as shown in figure 4.14.

Figure 4.13 Comparator output voltage at different phases within one clock cycle.

56
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Figure 4.14 Comparator output with a SR latch.

As we do not use preamplifier here, if the same current injection concept as the one

developed in the previous chapter is applied here, we would not have enjoyed the same

linear relation between injected current and offset shift as in the two stage topology.

Because this structure does not have a linear output region, the offset shift would have a

non linear relationship with injected current and that makes control of calibration very

difficult. This point is also proved by simulation results. To obtain the linearity, we have

adopted a technique proposed in [18] for a different fully dynamic comparator structure

but nevertheless shows similar results here.

57
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

The idea is to unbalance the capacitance at node X and Y as shown in figure 4.12 so

that zero cross point is shifted. This can be achieved by inserting an array of PMOS caps

at node X and Y and changing the control signal VRX and VLX. If VRX and VLX are both

high, then node X and Y would see the same amount of parasitic capacitance and the

comparator is balanced. If one signal with VRX or VLX is low, a channel would exist in the

PMOS transistor, thus there is a capacitance difference ∆ C caused by the channel

charge. And shift of offset ∆ V could be described by the following equation

I D ∆C
∆V = (4.1)
g m1,2 CT

where ID and gm1,2 are average current and transconductance of M1 and M2 during

regeneration and CT is the nominal total balanced capacitance at node X and Y. So this is

a roughly linearly relation between the capacitance difference and offset shift, and binary

sized PMOS transistors are placed symmetrically on both sides. The calibration can be

done by digitally changing the control signal VRX and VLX, which is similar to DAC plus

current source method developed in the previous chapter. The smallest MOS cap would

determine the LSB of calibration, while the total capacitance offered would determine the

calibration range.

58
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

We use 8 bit of binary coded MOS cap because of large offset variation shown by this

particular comparator structure. Monte-Carlo simulation has demonstrated that with the

transistor sizes we have chosen, standard variation of its offset is around 70mV. To

ensure a high yield, the calibration range should be able to accommodate at least 3 times

of that variation, roughly 200mV, on both positive and negative sides. Also, by

simulation we have found that the lowest calibration step caused by the smallest PMOS

cap is around 0.8mV. Therefore an 8 bit calibration circuit would have a 200mV

calibration range. The penalty of this technique is that it increases load capacitance at

node X and Y, thus sizes of M1, M2, M3, M4, M5 and M6 are increased accordingly.

Increased sizes of M1 and M2 would cause the load of T&H circuit to increase as well,

hence its power consumption. Also more kick noise is expected from the comparator

array. As transistors M3 to M8 are driven by the clock driver, more power would also be

needed from it.

The most important reason to adapt SAR logic is that the calibration is easily

controlled with SAR. As shown in previous chapter, the total calibration time depends on

initial offset voltage as well as other factors like frequency and amplitude of input signal.

Thus, it is hard to determine whether offset has converged by timing the circuit. This is
59
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

especially true for the linear search method. If we assume the worst case for each

comparator, for an 8 bit calibration we have to wait for 256 overflows. It is extremely

time consuming if offset is initially low and has already converged within the first few

steps. For a 6 bit flash ADC with 63 comparators calibrated in a serial fashion, this

amount of time wastage is simply unacceptable.

However, things are much easier with SAR. As soon as the last bit is judged, we can

move on to the next comparator and do not have to care about various conditions that

could affect calibration time. Implementation of SAR logic needs a few D flip flops

(DFFs). The SAR logic circuit used in the design is shown in figure 4.15 [19]. In this

design, the Q output of last DFF in the upper row could be used as an indication signal

for end of calibration because Q will only become high after the last bit is judged. We use

two sets of this kind of SAR logic for each comparator, which has two sets of MOS cap

arrays.

For the purpose of background offset extraction, two pairs of choppers are needed at

the input of the comparator, as the input is differential. Each comparator in this flash

ADC must have its own dedicated SAR logic circuit, which stores the calibration data
60
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

after it is done. However, the accumulator that used for extraction can be shared by 63

comparators, and this can easily be arranged by using multiplexers and demultiplexers.

The system level design will be discussed more in the following section, while the block

diagram for each comparator unit is shown in figure 4.16, which seems to have added a

number of peripheral circuits, the function which will be discussed in the following

paragraph.

Figure 4.15 Schematic of SAR logic [19].

61
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Figure 4.16 Comparator with circuits for calibration.

SAR logic takes two signals from accumulator. Overflow signal is a pulse that

indicates an overflow has just happened and Sign is a logic input that indicates the

direction of the overflow. Overflow is hooked up with CLK of the SAR logic circuit,

while Sign is connected to COMP. So each time when an Overflow pulse arrives, it

asserts or cancels the previous assumed bit based on Sign and moves on to assume the

next bit. This process will repeat itself until the last bit is reached. In the case, the SAR

logic outputs an End pulse to the control circuit to indicate the end of this comparator’s

calibration.

62
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

SR latch can maintain the output when both of its S and R input are high. Thus it will

keep the output of the comparator valid for almost a fully clock cycle. The first SR latch

is connected to the output before the second chopper. Its output is the input of the

accumulator as it needs to average the chopped output signal rather than normal

comparator output. This can also be understood by observing that the accumulator, acting

as a digital LPF, can only filter out the offset with input signal moved to a higher

frequency.

As can be seen from figure 4.16, output chopper is put between comparator output

and the input of the second latch rather than at the output of the second latch. The reason

to do so can be illustrated by simulation result shown in figure 4.17. The upper pair of

differential output shows what happens when you put the chopper at the output of the SR

latch. During the charge phase, SR latch will maintain the previous output until the next

regeneration phase. If chopper changes direction during that time, the final output will

accordingly be reversed. However, if the position of chopper is arranged as shown in

figure 4.16, the chopper still changes direction during charge phase, but it would hardly

63
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

affect the output because both positive and negative output are high at that time, as can be

seen in figure 4.17.

It is worth noting here that it is better to put a pair of inverters between the

comparator and chopper because the chopper might cause imbalance at the comparator

output, which results in inherent offset. Inverter can serve as a buffer and effectively

solves this problem.

Figure 4.17 Simulation result of comparator output with chopper placed at different

positions.

64
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

There is another block called metastability converter added in the output path and, as

the name suggests, it is used to alleviate metastability error. When the input voltage

happens to be very close to the reference voltage, the comparator output might not have

enough time to regenerate to a valid logic value. This metastable output, when observed

by the decoder, can be interpreted as either logic one or zero. Depending on the decoder

used, this might cause very large errors. To solve this problem, metastability converter

circuit [20] as shown in figure 4.18 can be used. When its input is at metastability state,

both outputs are high so that in a Gray coded ROM decoder, only one LSB error would

have resulted.

Figure 4.18 Metastability converter schematic.

65
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

The comparator, together with all the peripheral circuits shown in figure 4.16,

contains all the components needed for interfacing with shared calibration circuit and

following encoder circuit. The SAR logic, while consumes no power after the calibration

is done, ends up occupying most area in layout. In our design, 63 units of such circuit are

used to produce a 6 bit output.

4.4 Calibration Control Circuit

For a single comparator discussed in the previous section, the calibration control is

easy. Only an accumulator is missing from a complete feedback loop. With 63

comparator calibrated in a serial manner, new issues like generation of end signal for

each comparator’s calibration, transition from one comparator to another and driving of

choppers would certainly need to be addressed.

Accumulator is shared among all the comparators but SAR logic, because it also acts

as registers, cannot be shared. When a single comparator is being calibrated, its output is

hooked up with the accumulator and in turn, the accumulator overflow signal as well as

66
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

another signal that shows the sign of overflow is sent back to this particular comparator.

At the end of each calibration, SAR logic will give out an end signal to indicate the end

of calibration for this particular comparator. All the signals involved are digital, so that

digital multiplexers and de multiplexers are proper choices to route the signals, as shown

in figure 4.19. The channel selection block selects which comparator is being calibrated

and controls all the MUXs and DEMUXs to establish a proper channel to and from the

accumulator.

Figure 4.19 Calibration control block diagram.

The Channel Selection block shown in figure 4.19 is essentially a 6 bit counter with

its clock input connected to a multiplexer, whose inputs are connected to comparators’

end outputs. Depending on Sel signal, a certain end signal is select as the Channel
67
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Selection’s clock input. Sel is the 6 bit output from that counter and is initial set to zero.

At the end of each calibration, a pulse will be sent from the comparator to Channel

Selection block. It will cause the counter to count one time and the Sel will accordingly

select the next channel. Thus calibration will automatically be done from the first channel

to the last, after which the whole process would stop because no more pulse would be

sent to the counter. Another DEMUX is actually employed to enable the selected

comparator’s choppers, while all the choppers from the other comparators would be

disabled. At any given time, only one set of input choppers is working. The disturbance

they may cause to T&H circuit and reference ladder is therefore minimized.

4.5 Encoder

Outputs from comparators form what is known as thermometer code. And it has to be

converted by encoder to a binary code. Under ideal circumstance, in the thermometer

code there should be a level below which all comparator outputs are ones while above

which are all zeros. In reality, this assumption is challenged by two major problems,

bubble error and metastability error [3].

68
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Bubble error refers to the fact that sometimes there are zeros below a one. The

number of zeros encircled by ones is nominated as the order of bubble error. Most of the

time, only first order bubble error needs to be taken care of, as higher order bubble error

is much less likely to occur. The probable causes of this error include different clock

delay seen by different comparators and coparator offset [20], as shown in figure 4.20.

While the first is mitigated by using a dedicated T&H circuit and the later is minimized

by offset calibration, careful design of decoder is still necessary to bring this issue under

control.

Figure 4.20 Illustration of first and second order bubble errors [20].

A simple way to avoid first order bubble error is to use 3 input AND gates to encode

comparator outputs, as shown in figure 4.20. Only when the output of the current

69
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

comparator is one and the outputs of two above comparators are zero, the output of the

AND gate can be set to one. Thus, first order bubble error is suppressed. Single ended

output version is shown for simplicity here, but this also works well from differential

comparator output.

Figure 4.21 3 input AND gate used as encoder to suppress bubble error [3].

Metastability error happens when the input is very close to reference voltage. In that

case, the comparator output may be balanced for a short period of time and may not be

perceived as a valid logic. This output could be randomly interpreted as one or zero. This

may cause the final output to have large errors. Figure 4.22 shows encoder implemented
70
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

as a binary coded ROM with comparator metastability error [20]. If the two undecided

output Xs are both selected as one, then the ROM output is all zero which results in a half

full scale error, as the correct output should be 100000 or 011111. Similar things would

happen if both Xs are zero. The ROM output is all one and again causes a half scale error.

Although these are two worst cases in this scenario that are only likely to happen when

the input is close to half scale, less severe but still serious errors would occur in other

cases.

Figure 4.22 Binary coded ROM with metastability error [20].

71
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Figure 4.23 Gray coded ROM with metastability error [20].

Probability of this error can be greatly reduced if comparator regeneration time is

decreased. However, the penalty is increased power consumption and thus it is not

desirable to do so. Another effective way is to use Gray coded ROM as the encoder, as

shown in figure 4.23. In Gray code, the adjacent two bit lines only differ in one bit.

Therefore, if both bit lines are selected, ROM output would still be valid. As the input is

close to the reference, it does not matter which of the two adjacent bit lines is selected.

However, a large error would still occur if both bit lines are not selected. But this has

already been taken care of by the metastability circuit shown in figure 4.18. The encoder

in our design is implemented with 3 input AND gates and a Gray coded ROM.

72
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Data lines of the ROM need to be charged at the first half of every clock cycle, then

they would be selectively discharged by the 1 out of n code generated by AND gates.

This, like the case with fully dynamic comparator output earlier, means ROM outputs are

only valid for half a clock cycle. A bunch of specially designed DFFs are used as the

solution to this problem [8], whose schematic is shown in figure 4.24 and the effect of

DFF can be demonstrated by figure 4.25. By adding this DFF stage, the valid time for an

output is increased from half a cycle to a full cycle.

Figure 4.24 Schematic of DFF for ROM output.

73
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Figure 4.25 Simulation result of ROM output and DFF output.

4.6 Clock Driver

There are two major concerns involved in designing the clock driver. The first is

timing and the second is driving capability. Timing means the clock of each block,

including T&H circuit, comparator array, choppers and the encoder, needs to be properly

arranged with respect to each other, so that the whole flash ADC would function in a

reliable way. The demand for driving capability comes from the load it is driving. The

74
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

clock for the comparator array, for example, needs to have a very strong driving power so

as to drive 63 comparators. Primarily, inverter chains and specifically designed delay

blocks are used to manipulate the clock signals. Figure 4.26 shows simulation result of

the clock driver outputs.

Figure 4.26 Simulation result of clock signals.

The timing relationship between various clock signals can be understood by first

observing the clock of T&H circuit. In the track phase, T&H’s output is not valid and it

75
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

needs some time to settle to the valid output after entering the hold phase. So the clock of

comparators is arranged in such a way that regeneration happens near the end of hold

phase, when the signal from T&H circuit has already stabilized. Regeneration only

requires a certain amount of time, after which the change at comparator input would no

longer affect its output. So T&H circuit can enter the next track phase, as long as enough

time is left for the comparator to settle before the T&H output changes. Due to existence

of SR Latch at the output of comparator, previous regeneration output is maintained

during comparator charge phase and that is when ROM will discharge based on outputs

from 3 input AND gates, as discussed in the previous section. The slight delay between

ROM discharge and the end of comparator regeneration is kept for time needed by the

AND gates to settle.

As for the choppers, input chopper should toggle while T&H circuit is entering track

phase, causing minimal disturbance to T&H output. The output chopper, on the other

hand, should change direction near the end of comparator’s charge phase, when both

outputs from comparator are high. However, input chopper and output chopper should be

the same when the comparator starts to regenerate, otherwise the final output from

76
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

comparator would be reversed. As can be seen from figure 4.26, arrangement of input

chopper and output chopper clocks satisfies all these requirements.

All the signals are generated by the clock driver block with a single clock input.

Several issues are involved in designing this block. Chopper clock is produced by using a

JK flip flop to double the input signal period. At a relative high speed like 500MHz,

delay of the JK flip flop can no longer be ignored. So a delay block has been designed to

match the delay of JK flip flop by identifying and copying the signal path of the JK flip

flop. Layout of the delay block is also designed in an almost identical way. The simplest

way to design the drivers, which are essentially inverter chains, is to make them identical.

As long as it can meet the highest demand, it would certainly be adequate for all the other

blocks. However, this is far from an optimized solution. As the load of each clock signal

varies in order of magnitudes, it is better to design each inverter chain specifically suited

for the block that needs to be driven, so that power is not wasted. In the end, layout of

critical clock routes requires extra care because otherwise, they can cause serious signal

delay.

77
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

4.7 Power Consumption of the Flash ADC

The whole flash ADC consumes 9.5mW of power running at 500MHz according to

simulation result. Power consumption by calibration circuit is excluded as this part

consumes no power after calibration is done. It is a very competitive result if we look

back on chapter 2. The only two designs that have better performance in this aspect are

both implemented in more advanced technologies and both of them adopt foreground

calibration that cannot track ambient changes. Figure 4.27 shows the percent of power

consumed by each block of the flash ADC.

Power Consumption
Resistor Ladder Analog Front End
and Encoder 12%
5%

Clock Driver
17%

Comparator
66%

Figure 4.27 Percent of power consumption by each block of the flash ADC.
78
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Obviously, majority of power is consumed by comparator array. As discussed in

Section 4.3, MOS cap arrays are inserted to internal nodes of the fully dynamic

comparator as a way to linearly adjust its offset. To cover the whole range dictated by

comparator offset variation, total amount of capacitance added by the MOS caps must be

large, which inevitably puts extra load on the comparator. To meet speed requirement,

sizes of transistors have to be increased which in turn increases its power consumption.

Therefore, the size of comparator is actually restricted by the calibration circuit.

Moreover, a more powerful clock driver must be used to drive the enlarged transistors

that are used to charge the output and internal nodes.

Figure 2.7 in chapter 2 shows an alternative fully dynamic comparator structure

together with its calibration circuit [10]. It has an additional pair of NMOS transistors N2,

which are used to fine tune the comparator’s offset. So, instead of inserting MOS caps, it

relies on an additional pair of input transistors to adjust the speed of discharging.

However, this method is not applicable in our flash ADC design, as a very precise DAC

is mandatory to generate the Cal+ and Cal- signal.

79
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

The design of resistor ladder and analog frontend has greatly benefited from

calibration. As the transistor size, though limited by the calibration circuit, is still much

smaller than it would have been if no calibration is utilized. This fact greatly relieves

demands for resistor ladder and frontend, allowing significant power saving in those

blocks.

4.8 Flash ADC Layout Design

The flash ADC has been implemented in IBM 0.13um standard CMOS technology

and the whole system occupies a total area of 2.5 mm2, the majority of which, again, is

taken up by comparators. Special care is dedicated to route and distribute critical signals

like various clocks. The principle is to minimize the signal length and make everything as

symmetric as possible. The layout is shown in figure 4.28

Post layout simulation for the entire flash ADC was run to ensure the system is

operational, as shown in Figure 4.29, where the first row is flash ADC outputs Q0 to Q5 in

different colors. The second row Ia and the third row Id represent currents from the analog

80
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

and digital source, respectively. The spike in Id at 2ns is caused by the reset signal

transition from 0 to 1. The bottom row shows the system clock signal running at 500

MHz.

Figure 4.28 Flash ADC layout design

81
 Chapter 4 Design of a 6 bit 500MHz Flash ADC Employing Fully Dynamic Comparators

Figure 4.29
29 Post layout simulation results of the flash ADC

For the calibration technique to take effect, the circuit needs to run a large number

(>100000) of samples. Given the complexity of the circuit and available

availabl computational

power, the simulation time is too long to render it practical.

practical Thus, schematic and post

layout simulation results cannot be obtained to demonstrate the ADC’s

ADC performance

before and after calibration

calibration.

82
 Chapter 5 Conclusion

Chapter 5 Conclusion

Calibration technique shows great potential in high speed and low power flash ADC

design. Instead of dedicating large amount of power and chip area to suppress offset, it

solves the problem in a more fundamental and efficient way, which not only significantly

reduces power consumption, but also greatly simplifies the design. Between foreground

and background calibration technique, the later is probably a more universal solution as it

is capable of tracking environmental changes. The challenging problem, however, is to

obtain offset information without interrupting ADC from doing its normal job.

Major contribution of this work is the proposed comparator offset calibration

technique, which is proven to be effective by prototype design in AMS 0.35um CMOS

technology. This method helps designer to circumvent costly offset/area tradeoff so as to

achieve a much better power efficiency. It also has virtue in its simplicity, as only a few

additional components are used and most of them are digital blocks. The result is

negligible additional power consumption by these circuits. Subsequent application of this

technique to a 6 bit 500MHz flash ADC design in IBM 0.13um CMOS technology has

83
 Chapter 5 Conclusion

enabled the use of fully dynamic comparators. The benefits include complete elimination

of static power consumption from comparators and much relieved front end design.

Operating at a 1.2-V supply and running at 500MHz, the proposed flash ADC consumes

only 9.5mW of power.

To further optimize power consumption and speed of this flash ADC, improvements

should be made to the design of comparator. On one hand, comparator array is the

dominant factor in terms of power consumption and the limiting factor in terms of speed.

On the other hand, it plays such a pivotal role in the design process that everything else is

largely determined by the comparator design. Particularly in this work, sizing of

comparator is limited by extra loading from MOS caps used for calibration. It indicates

that by either altering the structure of the comparator with reduced offset variation or

developing more sophisticated calibration circuit that adds less loading, comparator size

can be further decreased. It not only means the comparator array itself would demand less

power, but also analog front end and clock driver can be expected to consume less power

as well. Future work can focus in this direction to achieve the fullest potential.

84
 References

[1] M. Pelgrom, A. Duinmaijer, and A. Welbers, “Matching properties of MOS

transistors,” IEEE J. Solid-State Circuits, vol. 24, no. 5 pp. 1433–1440, Oct. 1989.

[2] K. Kattmann and J. Barrow, “A technique for reducing differential non-linearity

errors in flash A/D converters,” in IEEE ISSCC Dig. Tech. Papers, Feb. 1991, pp. 170–

171.

[3] B. Razavi, Principles of Data Conversion System Design. IEEE Press, 1995

[4] M. Choi and A.A. Abidi, “A 6-b 1.3-Gsamples/s AD converter in 0.35-um CMOS ”

IEEE J. Solid-State Circuits, vol. 36, no. 12, pp. 1847–1857, Dec. 2001.

[5] P. C. S. Scholtens and M. Vertregt, “A 6-b 1.6-Gsamples/s flash ADC in 0.18 um

CMOS using averaging termination,” IEEE J. Solid-State Circuits, vol. 37, no. 12, pp.

1599–1609, Mar. 2002.

[6] A. Ismail and M. Elmasry, “A 6-Bit 1.6-GS/s low-Power wideband flash ADC

converter in 0.13-um CMOS technology,” IEEE J. Solid-State Circuits, vol. 43, no. 9, pp.

1982–1990, Sep. 2008.

85
[7] K. Deguchi, N. Suwa, M. Ito, T. Kumamoto, “A 6-bit 3.5-GS/s 0.9-V 98-mW flash

ADC in 90-nm CMOS,” IEEE J. Solid-State Circuits, vol. 43, no. 10, pp. 2303–2310, Oct.

2008.

[8] K. Uyttenhove, High speed CMOS analog to digital converters. PhD Thesis, Leuven,

Apr. 2003.

[9] [Link], M.Q. Le and K.Y. Kim, “A low power 6-bit flash ADC with reference

voltage and common-mode calibration,” IEEE J. Solid-State Circuits, vol. 44, no. 4, pp.

1041–1046, Apr. 2009.

[10] B. Verbruggen, J. Craninckx, M. Kuijk, P. Wambacq and G. Van der Plas, “A 2.2

mW 1.75 GS/s 5 Bit Folding Flash ADC in 90 nm Digital CMOS,” IEEE J. Solid-State

Circuits, vol. 44, no. 3, pp. 874–882, Mar. 2009.

[11] N. Stefanou, and [Link], “An average low offset comparator for 1.25

Gsample/s ADC in 0.18 µm CMOS” in Proc. Int. Conf. Electron., Circuits, Syst., Dec.

2004, pp. 246-249

[12] C.C. Huang and J.T. Wu, “A background comparator calibration technique for

flash analog to digital converters,” IEEE Trans. Circuits Syst. I, vol.52, no.9, pp. 1732-

1740, Sep. 2005.

86
[13] C.C. Enz, E.A. Vittoz, and F. Krummenacher, “A CMOS chopper amplifier,”

IEEE J. Solid-State Circuits, vol. 22, no.3, pp. 335-342, Jun. 1987.

[14] T. Kobayashi, K. Nogami, [Link], and Y. Fujimoto, “A current controlled

latch sense amplifier and a static power-saving input buffer for low-power architectures,”

IEEE J. Solid-State Circuits, vol. 28, no. 4 pp. 523–527, Apr. 1993.

[15] D. F. Hoeschele, Analog-to-Digital and Digital-to-Analog Conversion Techniques.

2nd ed. New York, U.S.: John Wiley & Sons, 1994.

[16] B. Razavi, Design of Analog CMOS Integrated Circuits. pp.314-325, McGraw-

Hill Companies, 2001.

[17] A.M. Abo and P.R. Gray “A 1.5-V, 10-bit, 14.3-MS/s CMOS Pipeline Analog-to-

Digital Converter,” IEEE J. Solid-State Circuits, vol. 34, no. 5 pp. 599–606, May, 1999.

[18] P. Nuzzo, F.D Bernardinis, P. Terreni and G. Van der Plas, “Noise Analysis of

Regenerative Comparators for Reconfigurable ADC Architectures,” IEEE Trans. Circuits

Syst. I, vol.55, no.6, pp. 1441-1454, Jul. 2008.

[19] M.D. Scott, B.E. Boser and K.S.J Pister, “An Ultralow-Energy ADC for Smart

Dust,” IEEE J. Solid-State Circuits, vol. 38, no. 7 pp. 1123–1129, May, 2003.

[20] K. Uyttenhove and M.S.J. Steyaert, “A 1.8-V 6-Bit 1.3-GHz Flash ADC in 0.25-

um CMOS,” IEEE J. Solid-State Circuits, vol. 38, no. 7 pp. 1115–1122, July, 2003.
87