Successive approximation analog-to-digital converter and conversion method thereof

Abstract

A successive approximation analog-to-digital converter and conversion method thereof are provided, the successive approximation analog-to-digital converter includes a segmented-multiple-stage capacitor array with redundancy bits, a comparator, a weight-storage circuit, a code reconstruction circuit and a control logic circuit. The successive approximation analog-to-digital converter helps to decrease the complexity of circuit design, featuring small size and low power. Without auxiliary capacitor array, switches and control logic, the circuit can work to precisely measure and correct capacitor mismatch errors.

Description

FIELD

The subject matter herein generally relates to Analog-to-Digital Converters (ADCs). Specifically, the present invention provides a successive approximation Analog-to-Digital Converter and conversion method thereof.

BACKGROUND

A successive approximation analog-to-digital converter generally comprises a capacitor array which has precision thereof determines precision of the successive approximation Analog-to-Digital Converter. Three problems need to be solved in regard to successive approximation Analog-to-Digital Conversion.

1. Structure problem of capacitor array

a. If single-segment architecture is adopt, the number of unit capacitor is expressed as follows:2^Ns  (1)

Wherein, N_s denotes the resolution of A/D converters. The number of unit capacitors increases at an exponential rate as the resolution of A/D converters increases. The size and power consumption of chip increases when the resolution of A/D converters N_s increases. Normally, capacitor mismatch error needs to be corrected when the resolution of A/D converters N_s is larger than 10.

b. If segment architecture is adopted, the number of unit capacitors dramatically decreases. For instance, when N_s is even number, the structure of capacitor arrays is divided into two identical structures, then the number of unit capacitors is expressed as follow:

$\begin{array}{cc}2·2^{\frac{N_s}{2}}+1& \left(2\right)\end{array}$

In doing so, the number of unit capacitors dramatically decreases especially when N_s is large. Multiple-segment structure tends to bring mismatch errors of equivalent capacitors. Therefore, capacitor mismatch error correction is in need for multiple-segment structure which is more complicated than that for single-segment architecture. Traditional capacitor mismatch error correction works to use compensation capacitor array to compensate capacitor mismatch error. That is to say, when a capacitor is involved in the addition of electric charges, a corresponding compensation capacitor array is added to compensate the variation of electric charges caused by mismatch error. The compensation capacitor array features the resolution less than 1 least significant bit (LSB). Therefore, if N_s is large, compensation capacitor array tends to apply a complicated structure to realize high resolution for compensation, which is too complicated to be realized.

A traditional successive approximation analog-to-digital converter works to apply digital correction by recording each capacitor's weight. Even though the weight of each capacitor can be correctly measured, code omission occurs when the weight of the high significant bit is larger than the sum of the weights of the rest bits plus 1 LSB. Take a 4 bit A/D converter as an example. Its actual weights are 9, 3, 2 and 1. The input/output values are correspondingly listed as follows:

In	0	1	2	3	4	5	6	7
Out	0	1	2	3	4	5	6	6
In	8	9	10	11	12	13	14	15
Out	6	9	10	11	12	13	14	15

Wherein, the output code 7 and 8 are missing.

Therefore, an improved structure needs to be made for successive approximation analog-to-digital (A/D) converters.

CONTENTS OF THE INVENTION

It is an object of the invention to provide a successive approximation converter and conversion method thereof to reduce the complexity of circuit design and to have small size and low power consumption. And without auxiliary capacitor array, switches and control logic, the converter can work to precisely measure and correct capacitor mismatch errors.

The foregoing object of the invention is accomplished by using the following techniques.

The invention provides a successive approximation analog-to-digital converter, including a segmented-multiple-stage capacitor array with redundancy bits, comparator, weight-storage circuit, code reconstruction circuit and control logic.

The converter works to sample input voltage and then generates output voltage Vout+ and Vout− under the control of the logic circuit, wherein the segmented-multiple-stage capacitor array with redundancy bits includes the first stage capacitor array, the second stage capacitor array, . . . , the m stage capacitor array, the first segment capacitor C1, the second segment capacitor C2, . . . and the m-1 segment capacitor C(m-1). Each stage capacitor array of the previous m-1 stages comprises at least one redundancy bit. The minimum number of redundancy bit required by the m stage capacitor array depends on the maximum of capacitor mismatch caused by process capability, wherein, m is a positive integer no less than 2.

The comparator is employed to compare the output voltage Vout+ and Vout− and then generate a result.

The weight-storage circuit is employed to record the weight of each unit capacitor of segmented-multiple-stage capacitor array with redundancy bits.

The code reconstruction circuit is employed for successive approximation analog-to-digital converter to calculate an output code according to the output of the comparator and the capacitor's weight from weight-storage circuit.

The control logic circuit is employed to control the foregoing segmented-multiple-stage capacitor array with redundancy bits, comparator, weight-storage circuit and code reconstruction circuit.

Where, the maximum of capacitor mismatch error caused by process and circuit design is N_{mismatch_max} LSB. The number of minimum redundancy bit of the m stage capacitor array is expressed as the following equation: N_{r_min}=log₂(2·N_{mismatch_max})

Wherein, the redundancy bit of segmented-multiple-stage capacitor array starts at the one after the significant bit featuring as same weight as the redundancy bit does.

The first-stage capacitor array comprises positive-end capacitor C₁₁, C₁₂, . . . , C_1i, C_1r, C_1(i+1), . . . , C_1n1, C_1n1′ and negative-end capacitor C′₁₁, C′₁₂, . . . , C′_1i, C′_1r, C′_1(i+1), . . . , C′_1n1, C′_1n1′. C₁₁ and C′₁₁ make up of one group featuring the same weight W₁₁; C₁₂ and C′₁₂ make up of one group featuring the same weight W₁₂; . . . ; C_1n1′ and C′_1n1′ make up of one group featuring the same weight W_1n1′; redundancy capacitor C_1r and C′_1r make up of one group featuring the same weight W_1r;

The second-stage capacitor array comprises positive-end capacitor C₂₁, C₂₂, . . . , C_2i, C_2r, C_2(i+1), . . . , C_2n2 and negative-end capacitor C′₂₁, C′₂₂, . . . , C′_2i, C′_2r, C′_2(i+1), . . . , C′_2n2. C₂₁ and C′₂₁ make up of one group featuring the same weight W₂₁; C₂₂ and C′₂₂ make up of one group featuring the same weight W₂₂; . . . ; C_2n2 and C′_2n2 make up of one group featuring the same weight W_2n2; redundancy capacitor C_2r and C′_2r make up of one group featuring the same weight W_2r;

In the same manner, the m stage capacitor array comprises positive-end capacitor C_m1, C_m2, . . . , C_mi, C_mr1, C_m(i+1), . . . , C_mnm, C_mr2, . . . , C_mrk and negative-end capacitor C′_m1, C′_m2, . . . , C′_mi, C′_mr1, C′_m(i+1), . . . , C′_mnm, C′_mr2, . . . , C′_mrk. Capacitor C_m1 and C′_m1 make up of one group featuring the same weight W_m1; Capacitor C_m2 and C′_m2 make up of one group featuring the same weight W_m2; Capacitor C_mnm and C′_mnm make up of one group featuring the same weight W_mnm; Capacitor C_mr1 and C′_mr1 make up of one group featuring the same weight W_mr1; . . . ; and redundancy capacitor C_mrk and C′_mrk make up of one group featuring the same weight W_mrk.

The first segment capacitor C₁ is twice than the last capacitor C_1n1 of the first-stage capacitor array and capacitor C₁ equals to C′₁; The second segment capacitor C₂ is twice than the last capacitor C_n22 of the second-stage capacitor array and capacitor C₂ equals to C′₂; . . . ; The m-1 segment capacitor C_(m-1) is twice than the last capacitor C_(m-1)n(m-1) of the second-stage capacitor array and capacitor C_(m-1) equals to C′_(m-1).

The segmented-multiple-stage capacitor array with redundancy bits is connected to comparators by using either differential structure or single-end structure.

According to the converter, successive approximation analog-to-digital conversion technique is presented, wherein,

The segmented-multiple-stage capacitor array with redundancy bits samples;

Switches of the first-stage capacitor array are connected to V_cm, and the first-stage capacitor array works to generate the first output of voltage Vout+ and Vout−;

The first output voltage Vout+ and Vout− are compared by comparator which outputs a result R₁₁;

The result R₁₁ is used to control the switch K₁₁, generating the second output voltage Vout+ and Vout−; code reconstruction circuit is employed to obtain the weight W₁₁ of capacitor C₁₁ connected with the switch K₁₁ from weight-storage circuit and calculate an output weight being expressed as the equation: D_out=R₁₁·W₁₁

The second output of voltage Vout+ and Vout− are compared by comparators outputting a result R₁₂;

The result R₁₂ is used to control the switch K₁₂, generating the third output voltage Vout+ and Vout−; code reconstruction circuit is employed to read the weight W₁₂ of capacitor C₁₂ connected with the switch K₁₂ from weight-storage circuit and calculate an output weight being added to the previous result as follows: D_out=R₁₁·W₁₁+R₁₂·W₁₂

In the same manner till the last control switch K_mrk, a result is obtained from code reconstruction circuit being expressed as the following equation: D_out=R₁₁·W₁₁+R₁₂·W₁₂+ . . . +R_mrk·W_mrk

If an output voltage Vout+ is larger than Vout−, the comparator outputs 1, otherwise the comparator outputs 0.

Besides, a method is described to measure capacitor mismatch error, wherein,

Weight-storage circuit is initialized with ideal weight;

Analog-to-digital conversion is carried out, generating a result from comparator and an output code from code reconstruction circuit;

Under the restriction of minimized sum of deviation square, a curve is fit based on output codes from code reconstruction circuit;

The actual weight of capacitors is estimated based on the difference between the true output curve and the fit one;

Weight-storage circuit is initialized with ideal weight;

Wherein, the number of significant bit of the first-stage capacitor array is n₁; the number of significant bit of the second-stage capacitor array is n₂, . . . , the number of significant bit of the m-stage capacitor array is n_m; the ideal weight W₁₁ of the most significant bit of the first-stage capacitor array is W_ideal11, being expressed as equation W_ideal11=2^{n1+n2+ . . . +nm−1}, the ideal weight W₁₂ of the second most significant bit of the first-stage capacitor array is W_ideal12, being expressed as equation W_ideal12=2^{n1+n2+ . . . +nm−2}, . . . , the ideal weight W_m1 of the first significant bit of the m-stage capacitor array is W_idealm1, being expressed as equation W_idealm1=2^2m−1; the ideal weight W_m2 of the second most significant bit of the m-stage capacitor array is W_idealm2, being expressed as equation W_idealm2=2^nm−2, . . . , the ideal weight W_mnm of the n_m significant bit of the m-stage capacitor array is 1;

The weight of the redundancy bit is set, wherein, the ideal weight of redundancy bit of the first stage capacitor array equals to its previous significant bit, namely, W_ideal1r=W_ideal1i; the ideal weight of redundancy bit of the second stage capacitor array equals to that of the previous significant bit, namely, W_ideal2r=W_ideal2i; . . . , the ideal weight of redundancy bit of the m stage capacitor array equals to its previous significant bit. There may exit a plenty of redundancy bits, namelyW_idealmrk=2⁰=1W_idealmr(k-1)=2¹=2W_idealmr(k-2)=2²=4. . .W_idealmr1=2^k-1

Then analog-to-digital conversion is carried out to generate an output from comparators and an output code from code reconstruction circuit;

Wherein, the amplitude of the input signal should be larger than half the Analog-to-Digital Converters (ADC) input signal range.

Analog-to-digital conversion is carried out based on the ideal weights to obtain an output from comparators and an output code from code reconstruction circuit. The number N of sample point must be larger than 2^{n1+n2+ . . . +nm}, (n is a natural number larger than 1).

Under the restriction of minimized sum of deviation square, a curve is fit based on output codes from code reconstruction circuit covering the following steps:

Assume input signal is expressed as function f(i), output signal as function D_out(i) ; i and D_out(i) are given; Under the restriction of minimized sum of deviation square

$\min \sum _i{\left(\hat{f}\left(i\right)-D_{\mathrm{out}}\left(i\right)\right)}^2,$ an input signal curve {circumflex over (f)}(i) is drawn, where i=1, 2, . . . , N.

The weight of capacitors is estimated based on the difference between the true output curve and the fit one.

a. Calculate the difference between the first output and the calculated one, the difference err(1) is expressed as equation:err(1)=D_out(1)−f(1)

b. According to the proportion of weights, the actual weight of capacitor is expressed as follows:

$\begin{array}{c}{W}_{11}\left(1\right)=W_{\mathrm{ideal}11}-\mu \frac{W_{\mathrm{ideal}11}·R_{11}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\\ W_{12}\left(1\right)=W_{\mathrm{ideal}12}-\mu \frac{W_{\mathrm{ideal}12}·R_{12}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\\ \dots \\ W_{\mathrm{mrk}}\left(1\right)=W_{\mathrm{idealmrk}}-\mu \frac{W_{\mathrm{idealmrk}}·R_{\mathrm{mrk}}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\end{array}$

Where, μ is stepwise regression; 0<μ<1

If process deviation can make sure that the weight of the bits after the p bit of the m capacitor array has no influence on the monotonic of capacitor array, then ignore the capacitor mismatch error of which the weight being less than W_mp, and take it as an ideal weight which is expressed as equation:

$\begin{array}{c}{W}_{11}\left(1\right)=W_{\mathrm{ideal}11}-\mu \frac{W_{\mathrm{ideal}11}·R_{11}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\\ W_{12}\left(1\right)=W_{\mathrm{ideal}12}-\mu \frac{W_{\mathrm{ideal}12}·R_{12}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\\ \dots \\ W_{mp}\left(1\right)=W_{\mathrm{idealmp}}-\mu \frac{W_{\mathrm{idealmp}}·R_{mp}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\end{array}$

c. An output is estimated according to the estimated weight:D_{out_est}(1)=R₁₁(2)·W₁₁(1)+R₁₂(2)·W₁₂(1)+ . . . +R_mrk(2)·W_mrk(1)

d. The error is calculated according to the estimated output and the calculated one.err(2)=D_{out_est}(1)−{circumflex over (f)}(2)

e. According to the error and the proportion of weight, the actual weight of capacitors is re-estimated.

$\begin{array}{c}{W}_{11}\left(2\right)=W_{11}\left(1\right)-\mu \frac{W_{\mathrm{ideal}11}·R_{11}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\\ W_{12}\left(2\right)=W_{12}\left(1\right)-\mu \frac{W_{\mathrm{ideal}11}·R_{12}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\\ \dots \\ W_{\mathrm{mrk}}\left(2\right)=W_{\mathrm{mrk}}\left(1\right)-\mu \frac{W_{\mathrm{idealmrk}}·R_{\mathrm{mrk}}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\end{array}$

If process deviation can make sure that the weight of the bits after the p bit of the m capacitor array has no influence on the monotonic of capacitor array, then ignore the capacitor mismatch error of which the weight being less than W_mp, and take it as an ideal weight which is expressed as equation:

$\begin{array}{c}{W}_{11}\left(2\right)=W_{11}\left(1\right)-\mu \frac{W_{\mathrm{ideal}11}·R_{11}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\\ W_{12}\left(2\right)=W_{12}\left(1\right)-\mu \frac{W_{\mathrm{ideal}12}·R_{12}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\\ \dots \\ W_{mp}\left(2\right)=W_{mp}\left(1\right)-\mu \frac{W_{\mathrm{idealmp}}·R_{\mathrm{mrk}}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\end{array}$

f. The cycle of estimation from the step c to step e can be done till the N estimation and there comes out the last estimated weight W₁₁(N), W₁₂(N), . . . , W_mrk(N).

Besides, the capacitor mismatch error is corrected;

Wherein, the estimated capacitor weight is written in weight-storage circuit;

Analog-to-digital conversion is carried out, outputting a code from code reconstruction circuit after the actual capacitor weight being corrected.

The invention has following advantages:

First of all, the segment structure is improved for capacitor array of successive approximation analog-to-digital converters, which greatly decreases the number of unit capacitors and realizes small layout size and low power consumption. For example, a 10-bit A/D converter with a traditional non-segment-structure features unit capacitors of 2¹⁰=1024. By using the segmented-structure of two segments, it features unit capacitors of

$2\times \left(2^{\frac{10}{2}}+1\right)+1=67,$ being reduced by 93.5%.

Second, code omission may happen to the capacitor array with either the traditional single-end or the traditional segment structure when digital correction is employed for capacitor mismatch error. By using the improved segment structure, the digital correction can be employed for capacitor mismatch error of successive approximation analog-to-digital converter.

Third, the invention discloses a successive approximation analog-to-digital converter featuring no auxiliary capacitor array, switches and control logic for measuring mismatch error, which greatly decreases the complexity of circuit design, layout size and power consumption.

Fourth, the invention discloses a successive approximation analog-to-digital converter featuring higher resolution than the traditional one using analog measure manner which is limited by auxiliary measuring capacitor array. By using digital measure manner for capacitor mismatch, the process capability has no effect on the resolution of error-measurement.

Fifth, the compensate capacitance array and related circuit used for capacitance mismatch error correction are eliminated, the design complexity of analog circuit is reduced greatly and the layout area and power consumption of the Analog-to-Digital Converters (ADC) are saved.

Sixth, the correction technique for the capacitor mismatch error features high resolution. By using digital correction technique, the correction technique features as the same resolution as the error measure technique does. As for the traditional analog correction, it features the resolution being limited to the minimum of compensation capacitor which is defined by process capability. As a result, the resolution of capacitor mismatch error compensation is less than that for error measure technique.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram showing a successive approximation analog-to-digital converter with redundancy bits.

FIG. 2 is a schematic diagram showing a segmented-multiple-stage capacitor array with redundancy bits as shown in FIG. 1.

FIG. 3 is a flow chart showing the way to measure capacitor mismatch error for a successive approximation analog-to-digital converter with redundancy bits.

FIG. 4 is a flow chart showing the way to correct capacitor mismatch error for a successive approximation analog-to-digital converter with redundancy bits.

DETAILED DESCRIPTION

According to FIG. 1, according to an embodiment of the invention, the successive approximation analog-to-digital converter includes a segmented-multiple-stage capacitor array 11 with redundancy bits, a comparator 12, a weight-storage circuit 13, a code reconstruction circuit 14 and a control logic circuit 15.

The successive approximation analog-to-digital converter's working principle is described as follow: The signal Vin+ and Vin− are input to segmented-multiple-stage capacitor array with redundancy bits 11. The segmented-multiple-stage capacitor array with redundancy bits 11 is controlled by the control logic circuit 15 to sample the signals and generate output voltage Vout+ and Vout−. Then the output voltage is compared by the comparator 12. The foregoing steps keep in cycle till the switch of the last unit capacitor. The output result from the comparator 12 is sent to the code reconstruction circuit 14. The result from the comparator 12 and the weight data from the weight-storage circuit 13 are recoded by the code reconstruction circuit 14. An output is obtained from the A/D converter, and separately an output from comparator 12 being used to measure the actual weight of capacitors.

The segmented-multiple-stage capacitor array with redundancy bits 11 is mainly employed to sample input voltages and generate output voltages Vout+ and Vout− under the control of logic circuit. The output voltages are compared by comparators. According to the outputs from comparators, the switches are altered one after another under the control of logic circuit. Then the output voltage Vout+ and Vout− are regenerated being sent to comparators for another comparison. The cycle of the steps is carried on till the switch of the last unit capacitor being reset. As shown in FIG. 2, the preferred embodiment shows a segmented-multiple-stage capacitor array with redundancy bits comprising the first stage capacitor array, the second stage capacitor array . . . the m stage capacitor array, the first segment capacitor C₁, the second segment capacitor C₂, . . . , the m-1 segment capacitor C_(m-1), wherein m is an positive integer no less than 2.

The first stage capacitor array comprises positive-end capacitor C₁₁, C₁₂, . . . , C_1i, C_1r, C_1(i+1), . . . , C_1n1, C_1n1′ and negative-end capacitor C′₁₁, C′₁₂, . . . , C′_1i, C′_1r, C′_1(i+1), . . . , C′_1n1, C′_1n1′, positive-end switch K₁₁, K₁₂, . . . , K_1i, K_1r, K_1(i+1), . . . , K_1n1, K_1n1′ and negative-end switch K′₁₁, K′₁₂, . . . , K′_1i, K′_1r, K′_1(i+1), . . . , K′_1n1, K′_1n1′, wherein, C₁₁ is connected to Vout− and the other end of C₁₁ to switch K₁₁. The other end of K₁₁ is connected to one of Vin+, V₊, V₋ and V_cm under the control of logic circuit; C₁₂ is connected to Vout− and the other end of C₁₂ to switch K₁₂. The other end of K₁₂ is connected to one of Vin+, V₊, V₋ and V_cm under the control of logic circuit; . . . ; C_1n1′ is connected to Vout− and the other end of C_1n1′ to switch K_1n1′. The other end of K_1n1′ is connected to one of Vin+, V₊, V₋ and V_cm under the control of logic circuit; redundancy capacitor C_1r is connected to Vout− and the other end of C_1r to switch K_1r. The other end of K_1r is connected to one of V₊, V₋ and V_cm under the control of logic circuit. Negative-end capacitor C′₁₁ is connected to Vout+ and the other end of C′₁₁ to switch K′₁₁. The other end of K′₁₁ is connected to one of Vin+, V₊, V₋ and V_cm; Negative-end capacitor C′₁₂ is connected to Vout+ and the other end of C′₁₂ to switch K′₁₂. The other end of K′₁₂ is connected to one of Vin+, V₊, V₋ and V_cm; . . . ; Negative-end capacitor C′_1n1′ is connected to Vout+ and the other end of C′_1n1′ to switch K′_1n1′. The other end of K′_1n1′ is connected to one of Vin+, V₊, V₋ and V_cm; redundancy capacitor C′_1r is connected to Vout− and the other end of C′_1r to switch K′_1r. The other end of K′_1r is connected to one of V₊, V₋ and V_cm under the control of logic circuit. Given an ideal condition, C₁₁ equals to C′₁₁, C₁₂ equals to C′₁₂, . . . , C_1n1′ equals to C′_1n1′; and C₁₁ equals to two times of C₁₂, C₁₂ is two times of C₁₃, . . . , C_1i is two times of C_1(i+1), C_1(n1−1) is two times of C_1n1, C_1r equals to C_1i, C_1n1 equals to C_1n1′; C′_1(n1−1) is two times of C′_1n1, C′_1r equals to C′_1i and C′_1n1 equals to C′_1n1′.

The capacitor C₁ of the first segment is two times of the last capacitor C_1n1′ of the first-stage capacitor array and C₁ equals to C′₁.

The second stage capacitor array comprises positive-end capacitor C₂₁, C₂₂, . . . , C_2i, C_2r, C_2(i+1), . . . , C_2n2 and negative-end capacitor C′₂₁, C′₂₂, . . . , C′_2i, C′_2r, C′_2(i+1), . . . , C′_2n2, positive-end switch K₂₁, K₂₂, . . . , K_2i, K_2r, K_2(i+1), . . . , K_2n2 and negative-end switch K′₂₁, K′₂₂, . . . , K′_2i, K′_2r, K′_2(i+1), . . . , K′_2n2, wherein, C₂₁ is connected to V2− and the other end of C₂₁ to switch K₂₁. The other end of K₂₁ is connected to one of V₊, V₋ and V_cm under the control of logic circuit; C₂₂ is connected to V2− and the other end of C₂₂ to switch K₂₂. The other end of K₂₂ is connected to one of V₊, V₋ and V_cm under the control of logic circuit; . . . ; C_2n2 is connected to V2− and the other end of C_2n2 to switch K_2n2. The other end of K_2n2 is connected to one of V₊, V₋ and V_cm under the control of logic circuit; redundancy capacitor C_2r is connected to V2− and the other end of C_2r to switch K_2r. The other end of K_2r is connected to one of V₊, V₋ and V_cm under the control of logic circuit. Negative-end capacitor C′₂₁ is connected to V2+ and the other end of C′₂₁ to switch K′₂₁. The other end of K′₂₁ is connected to one of V₊, V₋ and V_cm; Negative-end capacitor C′₂₂ is connected to V2+ and the other end of C′₂₂ to switch K′₂₂. The other end of K′₂₂ is connected to one of V₊, V₋ and V_cm; . . . ; Negative-end capacitor C′_2n2 is connected to V2+ and the other end of C′_2n2 to switch K′_2n2. The other end of K′_2n2 is connected to one of V₊, V₋ and V_cm; redundancy capacitor C′_2r is connected to V2− and the other end of C′_2r to switch K′_2r. The other end of K′_2r is connected to one of V₊, V₋ and V_cm under the control of logic circuit. Given an ideal condition, C₂₁ equals to C′₂₁, C₂₂ equals to C′₂₂, . . . , C_2n2 equals to C′_2n2; and C₂₁ equals to two times of C₂₂, C₂₂ is two times of C₂₃, . . . , C_2i is two times of C_2(i+1), C_2(n2−1) is two times of C_2n2, C_2r equals to C_2i; under an ideal circumstance, C′₂₁ is two times of C′₂₂; C′₂₂ is two times of C′₂₃, . . . , C′_2i is two times of C′_2(i+1); C′_2(n2−1) is two times of C′_2n2; C′_2r equals to C′_2i.

The capacitor C₂ of the second segment is two times of the last capacitor C_2n2 of the second-stage capacitor array and C₂ equals to C′₂.

This can be done in the same manner till the m-stage capacitor array. That is to say, the m-stage capacitor array comprises positive-end capacitors C_m1, C_m2, . . . , C_mi, C_mr1, C_m(i+1), . . . , C_mnm, C_mr2 . . . , C_mrk and negative-end capacitor C′_m1, C′_m2, . . . , C′_mi, C′_mr1, C′_m(i+1), . . . , C′_mnm, C′_mr2 . . . , C′_mrk, positive-end switch K_m1, K_m2, . . . , K_mi, K_mr, K_m(i+1), . . . , K_mnm, K_mr2, . . . , K_mrk and negative-end switch K′_m1, K′_m2, . . . , K′_mi, K′_mr, K′_m(i+1), . . . , K′_mnm, K′_mr2, . . . , K′_mrk, wherein, C_m1 is connected to Vm− and the other end of C_m1 to switch K_m1. The other end of K_m1 is connected to one of V₊, V₋ and V_cm under the control of logic circuit; C_m2 is connected to Vm− and the other end of C_m2 to switch K_m2. The other end of K_m2 is connected to any of V₊, V₋ and V_cm under the control of logic circuit; . . . ; C_mnm is connected to Vm− and the other end of C_mnm to switch K_mnm. The other end of K_mnm is connected to any of V₊, V₋ and V_cm under the control of logic circuit; redundancy capacitor C_mrk is connected to Vm− and the other end of C_mrk to switch K_mrk. The other end of K_mrk is connected to any of V₊, V₋ and V_cm under the control of logic circuit. C′_m1 is connected to Vm+ and the other end of C′_m1 to switch K′_m1. The other end of K′_m1 is connected to any of V₊, V₋ and V_cm under the control of logic circuit; C′_m2 is connected to Vm+ and the other end of C′_m2 to switch K′_m2. The other end of K′_m2 is connected to any of V₊, V₋ and V_cm under the control of logic circuit; . . . ; C′_mnm is connected to Vm+ and the other end of C′_mnm to switch K′_mnm. The other end of K′_mnm is connected to any of V₊, V₋ and V_cm under the control of logic circuit; redundancy capacitor C′_mr1 is connected to Vm+ and the other end of C′_mr1 to switch K′_mr1. The other end of K′_mr1 is connected to any of V₊, V₋ and V_cm under the control of logic circuit, . . . , redundancy capacitor C′_mrk is connected to Vm+ and the other end of C′_mrk to switch K′_mrk. The other end of K′_mrk is connected to any of V₊, V₋ and V_cm under the control of logic circuit.

In the same manner, given an ideal circumstance, the capacitor C_(m-1) of the m-1 segment is two times of the last capacitor C_(m-1)n(m-1) of the m-1 stage capacitor array and C_(m-1) equals to C′_(m-1).

Notice that each of the previous m-1 stage capacitor arrays features at least one redundancy bit being set at the bit right after the significant bit which has the same weight of the redundancy bit. The minimum number of redundancy bits of the last stage capacitor array is defined by the maximum of capacitor mismatch error which depends on process capability. If the maximum of capacitor mismatch error is expressed as N_{mismatch_max} LSB, then the minimum number of redundancy bit is expressed as N_{r_min}=log₂(2·N_{mismatch_max)}. The redundancy bit is set at the bit right after the significant bit featuring the same weight of the redundancy bit.

The comparator 12 is employed to compare the output voltage Vout+ and Vout− of the segmented-multiple-stage capacitor array with redundancy bits 11. If Vout+ is larger than Vout−, then the comparator 12 outputs 1, otherwise 0. The steps are described in details as follows:

1. At sampling phase, sample-switch K_S and K_S′ are switched on. Vout+ and Vout− equal to common-mode sample voltage Vcms; For the first-stage capacitor array, the positive-end switches are connected to input voltage Vin+, except the switch K_1r of redundancy bits and the negative-end switches are connected to input voltage Vin−, except the switch K′_1r of redundancy bits; The switches of the second-stage capacitor array are connected to common-mode voltage Vcm; In the same way, from the third stage to the m stage capacitor array, the switches of the arrays are connected to Vcm; and the sampling phase does not end until the sample-switches K_S and K_S′ being switched off.

2. At comparison phase, Vout+ and Vout− change by the time all switches of the first-stage capacitor array are connected to Vcm. Through comparison, if Vout+ is larger than Vout−, then it outputs R₁₁=1, being sent to logic circuit. The switch K₁₁ is connected to V₊ and K′₁₁ to V₋ under control of logic circuit. Otherwise, the circuit outputs R₁₁=0, and the switch K₁₁ is connected to V₋ and K′₁₁ to V₊. The circuit regenerates Vout+ and Vout−; At the second comparison, if Vout+ is larger than Vout−, the comparator outputs R₁₂=1, being sent to logic circuit. Under control of logic circuit, the switch K₁₂ is connected to V₊ and K′₁₂ to V₋. Otherwise, the comparator outputs R₁₂ =0, which makes K₁₂to be connected to V_— and K′₁₂ to V₊. Vout+ and Vout− are regenerated. The comparison can be done in the same way till the switches of the m stage capacitor array being reset based on the outputs from the comparator of the m stage capacitor array.

And, the weight-storage circuit 13 is employed for storing capacitor's weight. For the first stage capacitor array, C₁₁ and C′₁₁ make up of one group featuring weight W₁₁; C₁₂ and C′₁₂ make up of one group featuring weight W₁₂; . . . ; C_1n1′ and C′_1n1′ make up of one group featuring weight W_1n1′; C′_1r and C′_1r make up of one group featuring weight W_1r. For the second stage capacitor array, C₂₁ and C′₂₁ make up of one group featuring weight W₂₁; C₂₂ and C′₂₂ make up of one group featuring weight W₂₂; . . . ; C_2n2 and C′_2n2 make up of one group featuring weight W_2n2; C_2r and C′_2r make up of one group featuring weight W_2r. This can be done in the same manner till the m stage capacitor. For the m stage capacitor, C_m1 and C′_m1 make up of one group featuring weight W_m1; C_m2 and C′_m2 make up of one group featuring weight W_m2; C_mnm and C′_mnm make up of one group featuring weight W_mnm; C_mr1 and C′_mr1 make up of one group featuring weight W_mr1; . . . ; C_mrk and C′_mrk make up of one group featuring weight W_mrk.

And, the code reconstruction circuit 14 is employed for calculating an output code according to the result from the comparator 12 and the capacitor weight from the weight-storage circuit 13. The steps are described in details as follows:

1) Read output result R₁₁, R₁₂, . . . , R_mrk from the comparator 12;

2) Read capacitors' weight W₁₁, W₁₂, . . . , W_mrk;

3) Calculate the output code by adding up all results from capacitors by weight:D_out=R₁₁·W₁₁+R₁₂·W₁₂+ . . . +R_mrk·W_mrk

The logic control circuit 15 is employed to control the segmented-multiple-stage capacitor array 11, comparator 12, weight-storage circuit 13 and code reconstruction 14. The control of logic control circuit 15 reflects the conversion of successive approximation A/D converter. The conversion flow is described in details as follows:

1) the segmented-multiple-stage capacitor array with redundancy bits samples signals;

2) The logic control circuit works to connect all switches of the first-stage capacitor array to V_cm and Vout+ and Vout− are output for the first time;

3) The comparison of first-output Vout+ and Vout− is carried out by the comparator producing a result R₁₁;

4) According to R₁₁, the switch K₁₁ is controlled by control logic circuit resulting to the second output of Vout+ and Vout−; meanwhile, the weight W₁₁ of capacitor C₁₁, being connected to K₁₁, is obtained from weight-storage circuit by code reconstruction circuit. The corresponding result is calculated as D_out=R₁₁·W₁₁;

5) The comparison of second-output Vout+ and Vout− is carried out by the comparator producing a result R₁₂;

6) According to result R₁₂, the switch K₁₂ is controlled by control logic circuit resulting to the second output of Vout+ and Vout−; meanwhile, the weight W₁₂ of capacitor C₁₂, being connected to K₁₂, is obtained from weight-storage circuit by code reconstruction circuit. By adding the previous one, the result is calculated as D_out=R₁₁·W₁₁+R₁₂·W₁₂;

7) In the same manner till the last switch K_mrk, the circuit generates an output code being expressed as D_out=R₁₁·W₁₁+R₁₂·W₁₂+ . . . +R_mrk·W_mrk.

Still, the result D_out may need to be corrected due to capacitor mismatch error. The invention discloses the way to measure capacitor mismatch error as shown in FIG. 3. The steps are described in details as follow:

1) Set the primitive weight of weight-storage circuit with ideal one.

a. Set the weight of significant bits

The number of significant bits of the first stage capacitor array is n₁, that of the second one is n₂ . . . the number of significant bits of the m stage capacitor array is n_m. So the weight W₁₁ of the most significant bit of the first stage capacitor array is idealized as W_ideal11=2^{n1+n2+ . . . +nm−1}, the weight W₁₂ of the second most significant bit of the first stage capacitor array is idealized as W_ideal12=2^{n1+n2+ . . . +nm−2}, . . . , the weight W_m1 of the most significant bit of the m stage capacitor array is idealized as W_idealm1=2^nm−1; the weight W_m2 of the second most significant bit of the m stage capacitor array is idealized as W_idealm2=2^nm−2, . . . , the weight W_mnm of the n_m significant bit of the m stage capacitor array is idealized as 1.

b. Set the weight of the redundancy bit

The ideal weight of redundancy bits of the first stage capacitor array equals to the ideal weight of its previous significant bit, namely, W_ideal1r=W_ideal1i;

The ideal weight of redundancy bits of the second stage capacitor array equals to the ideal weight of its previous significant bit, namely, W_ideal2r=W_ideal2i;

The ideal weight of redundancy bits of the m stage capacitor array equals to the ideal weight of its previous significant bit which may lead to multiple equations as follows:W_idealmrk=2⁰=1W_idealmr(k-1)=2¹=2W_idealmr(k-2)=2²=4. . .W_idealmr1=2^k-1

2) Analog-to-digital conversion is carried out leading to a result from comparators and an output code from code reconstruction circuit, wherein, the output code from code reconstruction circuit is used to measure and correct the actual weight of capacitors. The following steps are covered for this procedure:

a. Input a signal with known characteristics, such as sine signal and ramp signal. The signal's amplitude is required to be larger than a half of input signal range of A/D converters;

b. According to the ideal weight, A/D conversion is carried out obtaining a compared result and a code from code reconstruction circuit. The number N of sample points must be more than 2^{n1+n2+ . . . +nm}, n is a natural number larger than 1.

3) Under the restriction of minimized sum of deviation square, a curve is fit based on output codes from code reconstruction circuit as being described as follows:

Assume input signal is expressed as function f(i), output signal as function D_out(i); i and D_out(i) are given; based on the minimized sum of deviation square

$\min \sum _i{\left(\hat{f}\left(i\right)-D_{\mathrm{out}}\left(i\right)\right)}^2,$ an input signal curve {circumflex over (f)}(i) is fit, where i=1, 2, . . . , N.

4) The weight of capacitors is estimated based on the difference between the actual output curve and the fit one.

a. Calculate the difference between the first output and the calculated one, the difference err(1) is expressed as follows:err(1)=D_out(1)−f(1);

b. According to the proportion of weights, the actual weight of capacitor is expressed as follows:

$\begin{array}{c}{W}_{11}\left(1\right)=W_{\mathrm{ideal}11}-\mu \frac{W_{\mathrm{ideal}11}·R_{11}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\\ W_{12}\left(1\right)=W_{\mathrm{ideal}12}-\mu \frac{W_{\mathrm{ideal}12}·R_{12}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\\ \dots \\ W_{\mathrm{mrk}}\left(1\right)=W_{\mathrm{idealmrk}}-\mu \frac{W_{\mathrm{idealmrk}}·R_{\mathrm{mrk}}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\end{array}$

Where, μ is stepwise regression, 0<μ<1.

If process deviation can make sure that the weight of the bits after the p bit of the m capacitor array has-no influence on the-monotonic of capacitor array, then ignore the capacitor mismatch error of which the weight being less than W_mp, and take it as an ideal weight which is expressed as equation:

$\begin{array}{c}{W}_{11}\left(1\right)=W_{\mathrm{ideal}11}-\mu \frac{W_{\mathrm{ideal}11}·R_{11}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\\ W_{12}\left(1\right)=W_{\mathrm{ideal}12}-\mu \frac{W_{\mathrm{ideal}12}·R_{12}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\\ \dots \\ W_{mp}\left(1\right)=W_{\mathrm{idealmp}}-\mu \frac{W_{\mathrm{idealmp}}·R_{mp}\left(1\right)·\mathrm{err}\left(1\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\end{array}$

c. An output is estimated according to the estimated weight.D_{out_est}(1)=R₁₁(2)·W₁₁(1)+R₁₂(2)·W₁₂(1)+ . . . +R_mrk(2)·W_mrk(1)

d. The difference between the estimated output and the fit one is calculated as follows:err(2)=D_{out_est}(1)−{circumflex over (f)}(2)

e. According to the error and the proportion of weight, the actual weight of capacitors is re-estimated.

$\begin{array}{c}{W}_{11}\left(2\right)=W_{11}\left(1\right)-\mu \frac{W_{\mathrm{ideal}11}·R_{11}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\\ W_{12}\left(2\right)=W_{12}\left(1\right)-\mu \frac{W_{\mathrm{ideal}11}·R_{12}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\\ \dots \\ W_{\mathrm{mrk}}\left(1\right)=W_{\mathrm{mrk}}-\mu \frac{W_{\mathrm{ideal}\mathrm{mrk}}·R_{\mathrm{mrk}}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmrk}}}\end{array}$

If process deviation can make sure that the weight of the bits after the p bit of the m capacitor array has-no influence on the-monotonic of capacitor array, then ignore the capacitor mismatch error of which the weight being less than W_mp, and take it as an ideal weight which is expressed as equation:

$\begin{array}{c}{W}_{11}\left(2\right)=W_{11}\left(1\right)-\mu \frac{W_{\mathrm{ideal}11}·R_{11}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\\ W_{12}\left(2\right)=W_{12}\left(1\right)-\mu \frac{W_{\mathrm{ideal}12}·R_{12}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\\ \dots \\ W_{mp}\left(2\right)=W_{mp}\left(1\right)-\mu \frac{W_{\mathrm{idealmp}}·R_{\mathrm{mrk}}\left(2\right)·\mathrm{err}\left(2\right)}{\sum W_{\mathrm{ideal}11}+W_{\mathrm{ideal}12}+\dots +W_{\mathrm{idealmp}}}\end{array}$

f. The cycle of estimation from the step c to step e can be done till the N estimation being completed and there comes out the last estimated weights W₁₁(N), W₁₂(N), . . . , W_mrk(N).

Referring to FIG. 4, the capacitor mismatch error is corrected based on the measured weights of capacitors. The steps are described as follows.

1) The estimated weight of capacitors is written in weight-storage circuit as the actual weight of capacitors.

The last estimated weights W₁₁(N) , W₁₂(N) , . . . , W_mrk(N) are taken as the actual weights of capacitors being written in weight-storage circuit.W_act11=W₁₁(N)W_act12=W₁₂(N). . .W_actmrk=W_mrk(N)

2) A/D conversion is carried out. By using the actual weights, the corrected output codes are obtained as expressed as the equation:D_{act_out}=R₁₁·W_act11+R₁₂·W_act12+ . . . +R_mrk·W_actmrk

The final digital result is obtained through the foregoing steps.

The foregoing preferred embodiments are provided to describe, not to limit, technical approaches in the invention. Obviously, bearing the essence and concept of the invention, technologists in this field can make various changes and redesigns to the invention. It should be understood that those changes and redesigns are also covered by claims of the invention, if they are with the same purpose and within the same scope of the present invention.

Claims

1. A successive approximation analog-to-digital converter comprising a segmented-multiple-stage capacitor array with redundancy bits, a comparator, a weight-storage circuit, a code reconstruction circuit and a control logic circuit, wherein: said segmented-multiple-stage capacitor array with redundancy bits is used for sampling input voltage and generating output voltage Vout+ and Vout− under the control of said control logic circuit; the segmented-multiple-stage capacitor array with redundancy bits comprises the first stage capacitor array, the second stage capacitor array, . . . , the m stage capacitor array, the first-segment capacitor C1, the second-segment capacitor C2, . . . , the m-1 segment capacitor C(m-1); each of the previous m-1 stages of capacitor arrays comprises no less than 1 redundancy bit; the minimum number of redundancy bit required by the m stage capacitor array depends on the maximum of capacitor mismatch caused by process capability, wherein m is a positive integer no less than 2;said comparator is employed to compare output voltages Vout+ and Vout− and generate a result;said weight-storage circuit is employed to store the weight of each capacitor unit of segmented-multiple-stage capacitor array with redundancy bits;said code reconstruction circuit is employed for successive approximation analog-to-digital converter to calculate an output code according to the output from the comparator and the capacitor's weight from weight-storage circuit; andsaid control logic circuit is employed to control the foregoing segmented-multiple-stage capacitor array with redundancy bits, comparator, weight-storage circuit and code reconstruction circuit.

2. The successive approximation analog-to-digital converter according to claim 1, wherein the maximum of capacitor mismatch error caused by process and circuit design is expressed as Nmismatch_max LSB (least significant bit); the number of minimum redundancy bit of the m stage capacitor array is expressed as the equation: Nr_min=log2(2·Nmismatch_max).

3. The successive approximation analog-to-digital converter according to claim 1, wherein the redundancy bit of segmented-multiple-stage capacitor array is after the significant bit of the same weight.

4. The successive approximation analog-to-digital converter according to claim 1, wherein the first-stage capacitor array of said segmented-multiple-stage capacitor array with redundancy bits comprises positive-end capacitor C11, C12, . . . , C1i, C1r, C1(i+1), . . . , C1n1, C1n1′ and negative-end capacitor C′11, C′12, . . . , C′1i, C′1r, C′1(i+1), . . . , C′1n1, C′1n1′, wherein C11 and C′11 make up of one group featuring the same weight W11; C12 and C′12 make up of one group featuring the same weight W12; . . . ; C1n1′ and C′1n1′ make up of one group featuring the same weight W1n1′; redundancy capacitor C1r and C′1r make up of one group featuring the same weight W1r, the second stage capacitor array comprises positive-end capacitor C21, C22, . . . , C2i, C2r, C2(i+1), . . . , C2n2 and negative-end capacitor C′21, C′22, . . . , C′2i, C′2r, C′2(i+1), . . . , C′2n2 wherein C21 and C′21 make up of one group featuring the same weight W21; C22 and C′22 make up of one group featuring the same weight W22; . . . ; Cn22 and C′2n2 make up of one group featuring the same weight W2n2; redundancy capacitor C2r and C′2r make up of one group featuring the same weight W2r; In the same manner, the m stage capacitor array comprises positive-end capacitor Cm1, Cm2, . . . , Cmi, Cmr1, Cm(i+1), . . . , Cmnm, Cmr2, . . . , Cmrk and negative-end capacitor C′m1, C′m2, . . . , C′mi, C′mr1, C′m(i+1), . . . , C′mnm, C′mr2, . . . , C′mrk, wherein Capacitor Cm1 and C′m1 make up of one group featuring the same weight Wm1; Capacitor Cm2 and C′m2 make up of one group featuring the same weight Wm2; Capacitor Cmnm and C′mnm make up of one group featuring the same weight Wmnm; Capacitor Cmr1 and C′mr1 make up of one group featuring the same weight Wmr1; . . . ; and redundancy capacitor Cmrk and C′mrk make up of one group featuring the same weight Wmrk.

5. The successive approximation analog-to-digital converter according to claim 4, wherein said first segment capacitor C1 is twice than the last capacitor C1n1 of the first-stage capacitor array and capacitor C1 equals to C′1; the second segment capacitor C2 is twice than the last capacitor C2n2 of the second-stage capacitor array and capacitor C2 equals to C′2; . . . ; The m-1 segment capacitor C(m-1) is twice than the last capacitor C(m-1)n(m-1) of the second-stage capacitor array and capacitor C(m-1) equals to C′(m-1).

6. The successive approximation analog-to-digital converter according to claim 1, wherein said segmented-multiple-stage capacitor array with redundancy bits is connected to said comparator by using either differential structure or single-end structure.

7. A method to realize successive approximation analog-to-digital conversion comprising the following steps: a segmented-multiple-stage capacitor array with redundancy bits samples;switches of the first-stage capacitor array are connected to Vcm, and the first-stage capacitor array works to generate the first output voltage Vout+ and Vout−;the first output voltage Vout+ and Vout− are compared by comparator which outputs a result R11;the result R11 is used to control the switch K11, generating the second output voltage Vout+ and Vout−;a code reconstruction circuit is employed to read the weight W11 of capacitor C11 from weight-storage circuit and calculate a related weight being expressed as the equation Dout=R11·W11;the second output voltage Vout+ and Vout− are compared by comparators outputting a result R12;the result R12 is used to control the switch K12, generating the third output voltage Vout+ and Vout−;a code reconstruction circuit tends to read the weight W12 of capacitor C12 from weight-storage circuit and calculate a related weight being added to the previous result as equation Dout=R11·W11+R12·W12;in the same manner till the last switch Kmrk, an output code is obtained through equation: Dout=R11·W11+R12·W12+ . . . +Rmrk·Wmrk.

8. The method to realize successive approximation analog-to-digital conversion according to claim 7, wherein, if the output voltage Vout+ is larger than Vout−, the comparator outputs 1, otherwise the comparator outputs 0.

9. The method to realize successive approximation analog-to-digital conversion according to claim 7, wherein, the method to measure capacitor mismatch error includes the following steps: the ideal weights of weight-storage circuit is set; analog-to-digital conversion is carried out to generate the comparator output code and the analog-to-digital conversion (ADC) output code; under the restriction of the minimized sum of deviation square, a curve is fit based on output codes from code reconstruction circuit; the actual weight of capacitor is estimated based on the difference between the actual output curve and the fit one.

10. The method to realize successive approximation analog-to-digital conversion according to claim 9, wherein the primitive weight of weight-storage circuit is set with an ideal weight, the method of setting ideal weight comprises the following steps: set the weight of significant bits, wherein the number of significant bits of the first stage capacitor array is n1, that of the second one is n2 . . . the number of significant bits of the m stage capacitor array is nm, so the weight W11 of the most significant bit of the first stage capacitor array is idealized as Wideal11=2n1+n2+ . . . +nm−1, the weight W12 of the second most significant bit of the first stage capacitor array is idealized as Wideal12=2n1+n2+ . . . +nm−2, . . . , the weight Wm1 of the most significant bit of the m stage capacitor array is idealized as Widealm1=2nm−1;the weight Wm2 of the second most significant bit of the m stage capacitor array is idealized as Widealm22nm−2, . . . , the weight Wmnm of the nm significant bit of the m stage capacitor array is idealized as 1;set the weight of the redundancy bit:the ideal weight of redundancy bits of the first stage capacitor array equals to the ideal weight of its previous significant bit, namely, Wideal1r=Wideal1i;the ideal weight of redundancy bits of the second stage capacitor array equals to the ideal weight of its previous significant bit, namely, Wideal2r=Wideal2i;the ideal weight of redundancy bits of the m stage capacitor array equals to the ideal weight of its previous significant bit which may lead to multiple equations as follows: Widealmrk=20=1Widealmr(k-1)=21=2Widealmr(k-2)=22=4

11. The method to realize successive approximation analog-to-digital conversion according to claim 9, wherein analog-to-digital conversion is carried out leading to a result of comparators and an output code of code reconstruction circuit, the method of carrying out the result of comparators and the output code of code reconstruction circuit comprises the following steps: input a signal with known characteristics, such as sine signal and ramp signal, the signal's amplitude is required to be larger than a half of input signal range of A/D converters;according to the ideal weight, A/D conversion is carried out obtaining a compared result and a code from code reconstruction circuit, the number N of sample points must be more than 2n1+n2+ . . . +nm, n is a natural number larger than 1.

12. The method to realize successive approximation analog-to-digital conversion according to claim 9, wherein under the restriction of minimized sum of deviation square, a curve is fit based on output codes from code reconstruction circuit, the method of fitting the curve comprises the following steps: assume input signal is expressed as function f(i), output signal as function Dout(i);i and Dout(i) are given;based on the minimized sum of deviation square

13. The method to realize successive approximation analog-to-digital conversion according to claim 12, wherein the actual weight of capacitors is estimated based on the difference between the actual output curve and the fit curve, the method of estimating the actual weight of capacitors comprises the following steps: a. calculate the difference between the first output and the calculated one, the difference err(1) is expressed as equation err(1)=Dout(1)−f(1);b. according to the proportion of weights, the actual weight of capacitor is expressed as follow:

14. The method to realize successive approximation analog-to-digital conversion according to claim 9, further comprising a step of correcting the capacitor mismatch error, wherein the step of the correcting the capacitor mismatch error comprises the following steps: the actual weight is written into weight-storage circuit;analog-to-digital conversion is carried out generating an output code after the correction of the actual capacitor weight.