SAR ADC and Related Methods

The SAR ADC addresses SQNR reduction by employing multiple comparisons and statistical noise analysis to determine integer and fractional bits, enhancing resolution and reducing quantization errors.

JP7578224B2Active Publication Date: 2024-11-06XMEMS LABS INC
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
JP2023052926
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2023-02-21
Filing Date
2023-03-29
Publication Date
2024-11-06
Estimated Expiration
2043-03-29

AI Technical Summary

Technical Problem

Conventional SAR ADCs suffer from reduced signal-to-quantization-noise ratio (SQNR) due to comparator noise, particularly affecting the least significant bit (LSB), leading to quantization errors and suboptimal resolution.

Method used

The SAR ADC employs multiple comparisons and utilizes statistical properties of comparator noise to determine both integer and fractional bits, incorporating a controller to record and analyze the results of these comparisons to improve resolution.

Benefits of technology

The improved SAR ADC enhances SQNR and effective number of bits (ENOB) by accurately estimating the LSB and fractional bits, resulting in increased resolution and reduced quantization errors.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007578224000010
    Figure 0007578224000010
  • Figure 0007578224000011
    Figure 0007578224000011
  • Figure 0007578224000012
    Figure 0007578224000012
Patent Text Reader

Abstract

To provide a successive approximation register (SAR) analog-to-digital converter (ADC) having improved signal-to-quantization noise ratio (SQNR).SOLUTION: A successive approximation register (SAR) analog-to-digital converter (ADC) includes a comparator, a threshold generator and a controller. The comparator receives an analog signal, and the SAR ADC outputs an output codeword. The comparator performs a plurality of first comparisons and a plurality of second comparisons. The controller determines a plurality of most significant bits of the output codeword according to a plurality of first comparison results corresponding to the first comparisons. The first comparisons are performed by comparing the analog signal with a plurality of first thresholds. The controller determines a plurality of least significant bits of the output codeword according to a plurality of second comparison results corresponding to the plurality of second comparisons. The second comparisons are performed by comparing the analog signal with a second threshold. The controller controls the threshold generator to produce the plurality of first thresholds and the second threshold according to the first comparison results.SELECTED DRAWING: Figure 2
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to successive approximation register (SAR) analog-to-digital converters (ADCs), and more particularly to SAR ADCs having improved signal to quantization noise ratio (SQNR). [Background technology]

[0002] A conventional successive approximation register (SAR) analog-to-digital converter (ADC) performs analog-to-digital conversion through a method known as successive approximation register. The conversion process derives the bits of a digital output codeword successively, from most significant bit (MSB) to least significant bit (LSB), according to the result of comparing the sampled value of an analog input signal with a threshold voltage set according to the value in the register. Summary of the Invention

[0003] An object of the present invention is to provide a successive approximation register (SAR) analog-to-digital converter (ADC) with improved signal-to-quantization noise ratio (SQNR) by including fractional bits based on the original least significant bit (LSB) comparison result.

[0004] An embodiment of the present invention discloses a SAR ADC including a comparator, a threshold generator, and a controller. The comparator receives an analog signal, and the SAR ADC outputs an output codeword. The comparator performs a plurality of first comparisons and a plurality of second comparisons. The controller determines a plurality of most significant bits of the output codeword according to a plurality of first comparison results corresponding to the plurality of first comparisons. The plurality of first comparisons are performed by comparing the analog signal with a plurality of first thresholds. The controller determines a plurality of least significant bits of the output codeword according to a plurality of second comparison results corresponding to the plurality of second comparisons. The plurality of second comparisons are performed by comparing the analog signal with a second threshold. The controller controls the threshold generator to generate a plurality of first thresholds and a second threshold according to the plurality of first comparison results.

[0005] Another embodiment of the present invention discloses a method applied to a SAR ADC configured to output an output codeword, the method including the steps of receiving an analog signal, performing a plurality of first comparisons on the analog signal with a plurality of first thresholds, determining a plurality of most significant bits of the output codeword according to a plurality of first comparison results corresponding to the plurality of first comparisons, performing a plurality of second comparisons on the analog signal with a second threshold, determining a plurality of least significant bits of the output codeword according to a plurality of second comparison results corresponding to the plurality of second comparisons, and determining a plurality of first and second thresholds according to the plurality of first comparison results.

[0006] These and other objectives of the present invention will no doubt become obvious to those of ordinary skill in the art after reading the following detailed description of the preferred embodiment that is illustrated in the various figures and drawings. [Brief description of the drawings]

[0007] [Figure 1] FIG. 1 is a schematic diagram of an N-bit SAR ADC. [Diagram 2] FIG. 1 is a schematic diagram of a SAR ADC according to an embodiment of the present invention. [Diagram 3]4 is a flow chart of a process according to an embodiment of the present invention. [Figure 4] 1 illustrates a scenario of operation of the comparator. [Diagram 5] The PDF curve of random noise is shown. [Figure 6] We present some examples of Vf / σT values ​​for which the expected values ​​of the comparison results are 0, 1, ..., Q. [Figure 7] FIG. 2 is a schematic diagram of a SAR ADC with a detailed implementation of a controller according to an embodiment of the present invention. [Figure 8] FIG. 2 is a schematic diagram of a SAR ADC with another detailed implementation of a controller according to an embodiment of the present invention. [Figure 9] 4 shows an example content of a LUT for estimating voltage. [Figure 10] FIG. 13 is a waveform diagram of threshold voltages in multi-path operation. [Figure 11] FIG. 2 is a schematic diagram of a SAR ADC with further detailed implementation of a controller according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0008] 1 is a schematic diagram of an N-bit successive approximation register (SAR) analog-to-digital converter (ADC) 10 configured to receive an analog signal adc_in and output an N-bit output codeword adc_out. The SAR ADC 10 includes a sample-and-hold (S / H) circuit 102, a comparator 104, a threshold generator 106, a controller 108, and a register 110. To generate the output codeword adc_out, the comparator 104 compares the sampled analog signal adc_in (i.e., sampled by the S / H circuit 102) with a given threshold (i.e., generated by the threshold generator 106) to generate a bit sequence b N-1 The controller 108 may determine the value of the bit sequence b 0 according to the comparison result. N-1 Generate a value of ~b0 and change the bit value to the most significant bit (MSB) b N-1to the least significant bit b0 into register 110 and correspondingly control threshold generator 106 to generate appropriate thresholds for subsequent comparisons.

[0009] One parameter for measuring the performance of a SAR ADC is the signal-to-quantization noise ratio (SQNR). Theoretically, the SQNR of the SAR ADC 10 is determined by the acceptable level of the analog signal adc_in and the number of bits of the output codeword adc_out. In a practical design of a typical SAR ADC such as the SAR ADC 10, the SQNR is further reduced under the noise generated by the comparator 104. To meet the design requirements of the SAR ADC 10, the comparator 104 may be required to use a high-speed and low-power design, which usually entails non-negligible comparator noise.

[0010] During the conversion process of the SAR ADC10, each bit of the output codeword adc_out, in the direction from MSB to LSB, N-1 ~b0 is V + and V - The result of a single comparison between the two input voltages of the comparator 104, denoted as b0, is determined by the result of a single comparison between the two input voltages of the comparator 104, denoted as b0. Considering the random nature of the noise in the comparator 104, the noise magnitude may exceed a certain level, causing an erroneous decision of this particular bit, resulting in a quantization error larger than that in noise-free conditions, and therefore in a conventional SAR ADC, a multi-iteration-compare may be performed on b0 to reduce the possibility of such a quantization error. In the present invention, the random nature of the noise and the multi-iteration-compare operation are combined to achieve a resolution corresponding to a fraction of b0.

[0011] The operation of the SAR ADC 10 typically proceeds bit by bit until the LSB is derived. The SAR ADC 10 then marks the end of one analog-to-digital conversion at this point without looking back to correct the erroneous bits.

[0012] Note that the comparison of the LSB b0 of the output codeword adc_out is most susceptible to comparator noise. Therefore, when the comparator noise is close to or larger than the signal level corresponding to b0, some designers of SAR ADC 10 try to improve the SQNR by using multiple comparisons and deriving the LSB b0 through majority voting among the results of the multiple comparisons. Referring to FIG. 1, the N-1 bits of the output codeword adc_out, i.e., b N-1 After b1,...,b2 are determined, the conversion process enters the final stage, where the LSBb0 is determined. + (i.e., the sampled input analog signal adc_in) and V - Instead of one single comparison of the LSB (i.e., the threshold voltage generated by the threshold generator 106), the modified approach repeats the comparison multiple times and then uses a majority rule to determine the LSB. As an example, the process repeats the comparison eight times. + >V - If the event occurs five or more times, b0 = 1. On the other hand, if the event occurs more than five times, b0 = 1. + >V - If the event occurs four or less times, b0=0.

[0013] The SAR ADC 10 with multiple comparisons for the LSBb0 provides an improvement to the SQNR. In the SAR ADC 10, the nature of the noise in the comparator 104 is a random process. To a good approximation, the noise sampling values ​​at different times are considered as independent and identically distributed random variables. Performing multiple comparisons and making a decision based on the majority rule is equivalent to reducing the effect of the noise and thus increasing the SQNR. However, the majority rule only exploits the most basic statistical properties of the noise, and the improvement to the SQNR is not optimal.

[0014] The present invention provides a new SAR ADC that offers improvements over existing SAR ADCs by making more comparisons and applying statistical characteristics of the noise generated during the comparison in a more comprehensive manner than majority rule, thereby allowing the SQNR and effective number of bits (ENOB) of the SAR ADC to be improved over that of conventional SAR ADCs.

[0015] 2 is a schematic diagram of a SAR ADC 20 according to an embodiment of the present invention. The SAR ADC 20 includes a S / H circuit 202, a comparator 204, a threshold generator 206, and a controller 208. The operation of the SAR ADC 20 is similar to that of the SAR ADC 10, so signals and elements having similar functions are indicated with the same symbols. The operation of the S / H circuit 202, the comparator 204, and the threshold generator 206 is similar to that of the S / H circuit 102, the comparator 104, and the threshold generator 106 shown in FIG. 1. The difference between the SAR ADC 20 and the SAR ADC 10 is that the controller 208 of the SAR ADC 20 includes a register 210, which stores the integer bits b obtained from the comparison with the corresponding threshold voltages. N-1 . . b1 and also to record the values ​​of some fractional bits f1 to f4 and LSBb0 obtained from a comparison with a threshold voltage for LSBb0.

[0016] Fig. 3 is a flow chart of a process 30 according to an embodiment of the present invention. The process 30 may be implemented in a SAR ADC, such as the SAR ADC 20 shown in Fig. 2. As shown in Fig. 3, the process 30 includes the following steps.

[0017] Step 302: The SAR ADC 20 receives the analog signal adc_in.

[0018] Step 304: The comparator 204 performs a plurality of first comparisons on the analog signal adc_in with a plurality of first thresholds.

[0019] Step 306: The controller 208 determines multiple MSBs of the output codeword adc_out according to multiple first comparison results corresponding to the multiple first comparisons, and each of the multiple first thresholds is determined according to a previous comparison result among the multiple first comparison results.

[0020] Step 308: The comparator 204 performs a plurality of second comparisons on the analog signal adc_in with a second threshold, where the second threshold is determined according to a previous comparison result among the plurality of first comparison results.

[0021] Step 310: The controller 208 determines a plurality of LSBs of the output codeword adc_out according to a plurality of second comparison results corresponding to the plurality of second comparisons.

[0022] 3 in conjunction with FIG. 2, the controller 208 regulates the operation of the comparator 204 to set the threshold voltage required for comparison (i.e., V - The controller 208 may control the threshold generator 206 to generate a value of the integer bit b according to the first comparison result, and enable the comparator 204 to perform the comparison and generate a comparison result. The controller 208 may further record the bit values ​​in the register 210 as a bit sequence of the output codeword adc_out. In particular, the comparator 204 performs a plurality of first comparisons with respective first thresholds to generate a first comparison result, and the controller 208 may control the threshold generator 206 to generate a value of the integer bit b according to the first comparison result. N-1 〜b1 may be determined. MSBb N-1 may have a predetermined level, and the first threshold for MSBb N-1 The comparison result is the second MSBb N-2 may be applied to determine a first threshold for the second MSBb N-2 The comparison result is the third MSBb N-3 may be applied to determine a first threshold for MSBb, and so on. N-1 Each of the first thresholds other than the first threshold for may be determined according to a previous first comparison result.

[0023] Then, the comparator 204 may perform multiple second comparisons with a second threshold value. The second comparisons are used to determine values ​​of the LSB b0 and fractional bits f1-f4, and the combinations of the LSB b0 and fractional bits f1-f4 are considered as multiple LSBs in the process 30. In this step, the second threshold value may be determined according to the previous first comparison result for bit b1.

[0024] Thus, at the end of the sequence of comparison operations, the controller 208 generates the N integer bits b N-1 ~b0 and M fractional bits f1~f M where M and N are positive integers. In the embodiment shown in Figure 2, there are four fractional bits f1 to f4.

[0025] Unlike the above majority rule in which redundant comparison is only applied to determine the value of LSBb0, in the SAR ADC 20 of the present invention, information of multiple comparisons for LSBb0, such as statistical properties, is provided to LSBb0 and also to fractional bits f1-f4. Thus, the comparison results may be reflected by the values ​​of LSBb0 and fractional bits f1-f4. In this situation, the statistical properties of the multiple comparison results may be applied to increase the resolution of the SAR ADC 20 by including these fractional bits f1-f4.

[0026] Figure 4 shows a scenario of comparator operation. The vertical direction shows the levels of the various signals involved in the comparison process. The voltage V in is the input analog signal that is tracked and held for comparison. Correspondingly, a threshold voltage V that indicates the threshold between output code C and output code C+1. th (C,C+1) is set. For completeness, the adjacent threshold voltages V th (C-1, C) and V th (C+1, C+2) is also shown in Figure 4. In Figure 4, b N-1 The bit values ​​from b1 to b2 are generated, and the comparator 204 detects the threshold voltage V thAssume that a comparison is being performed to determine the value of LSBb0 based on (C,C+1). The analog signal V before being contaminated by noise in is the threshold voltage V th Voltage V away from (C,C+1) f When comparator 204 begins to perform the comparison, at this point an amount of noise T is generated. In the following analysis, the analog signal V present at the positive input terminal of comparator 204 is in Therefore, the positive input voltage V + As shown by the dashed line, V in +T. If the effect of noise at the negative input terminal of comparator 204 is taken into account, the analysis leads to the same conclusion.

[0027] Referring to FIG. 4 in conjunction with FIG. 2, the comparator 204 in the SAR ADC 20 is V th (C,C+1)+V f +T>V th (C,C+1) or T>-V f If and only if V + >V - , or equivalently V in +T>V th (C, C+1). In the following explanation, p1 is the first bit of the comparison result that corresponds to V + >V - p0(=1-p1) is the probability that the comparison result is V + ≦V - If the curve represents the probability density function (PDF) of random noise T, as shown in Figure 5, the area of ​​the shaded region under the PDF curve is equal to p0. This shaded region is the probability that the probability of T<-V f Equivalently, a binomial random variable X may be used to describe the statistical behavior of the comparator 204.

[0028]

number

[0029] In many practical systems, the noise T has a standard deviation σ T , i.e., T:N(0,σ T ) Using this model for the noise T generated in the comparator 204 in the SAR ADC 20, the following can be derived:

[0030]

number

[0031] Probability p1 and voltage V f (This is the input analog signal V in and the corresponding threshold voltage V th Because of the one-to-one correspondence between the voltages V f may be used to estimate the voltage V f is the analog signal V in It may further be applied to find

[0032] Since the probability p1 is less than 1, it is preferable to perform multiple comparisons to find an expected value for the comparison result, such as the multiple second comparisons of process 30 used to determine the values ​​of the LSB b0 and fractional bits f1-f4. in and leaving the threshold voltage unchanged (e.g., V th If the comparison is repeated Q times in succession with (C,C+1), the outcome of these Q successive comparisons may be modeled by a random process.

[0033]

number

[0034]

number

[0035] Figure 6 shows the expected values ​​for implementations where the number of comparisons Q is 4, 7, and 8. <Y Q > V for 0, 1..., Q f / σ T Here are some examples of enumerating the values ​​of . <Y Q > may have five possible outcomes: 0, 1, 2, 3, and 4, when Q is equal to 4. Similarly, the expected value <Y Q > may have eight possible outcomes when Q is equal to seven, and nine possible outcomes when Q is equal to eight. In this embodiment, the value V is determined based on the assumption that the noise T is a normal (Gaussian) random process. f / σ T is calculated. V + >V - The more comparison results that generate an output value of 1, the higher the V f / σ T can be large, which is due to the input analog signal V in is the threshold voltage V th This means that there is a possibility that the voltage of the resistor (C, C+1) may be greater than that of the resistor (C, C+1). + >V - The fewer comparison results that show V + ≦V - The more comparison results that show f / σ T can become small or negative, which is the difference between the input analog signal V in is the threshold voltage V th This means that V may be close to or lower than (C,C+1). Therefore, after Q comparisons are performed, V is determined according to the combination of the Q comparison results. f / σ T may be obtained, and if the standard deviation σT is a known value, the voltage V f may be calculated. The voltage V f is the analog signal V in To achieve a better approximation and estimation of the input analog signal V in It may further be used to calculate

[0036] For the SAR ADC20, the value Vf / σ T may be applied to estimate the fractional portion as fractional bits f1-f4 that are stored in register 210. In this regard, the results of the multiple comparisons for LSBb0 may be applied to increase the resolution of SAR ADC 20. That is, the resolution of SAR ADC 20 has four more bits and additional finer information is recorded in the four fractional bits f1-f4.

[0037] 7 is a schematic diagram of the SAR ADC 20 with a detailed implementation of the controller 208 according to an embodiment of the present invention. As shown in FIG. 7, the controller 208 includes a counter 702, an adder 704, and two registers REG_A and REG_B. The counter 702 is + >V - The register REG_A is used to store bit values ​​for controlling the threshold generator 206 to generate the appropriate threshold voltages, and the threshold voltages for the comparison are determined by the most significant N bits b N-1 .about.b0 may be set according to the respective values ​​of .about.b0. Register REG_B is used to store the output codeword adc_out.

[0038] During comparator operation, the controller 208 first calculates b N-1 The comparator 204 is configured to control the comparator 204 to perform one comparison for every N-1 bits from bit b N-1 After the value of .about.b1 is derived, the controller 208 then sets bit b0 in the register REG_A to 1 and decrements the bit pattern [b N-1, ..., b1, 1]. The threshold voltage V th While V is held at this level, comparator 204 performs a sequence of eight comparisons, and counter 702 counts the number of events V out of the eight comparisons. + >V -At the end of these eight comparisons, the cumulative result generated by counter 702, labeled as y in FIG. 7, is calculated based on the random process Y defined in Equation 3. Q If a uniformly distributed statistical model for the noise T is adopted, then the average observation

[0039]

number

[0040] 7, the accumulation results with appropriate offsets may be used to determine the values ​​of some LSBs of the output codeword adc_out, including b0, f1, and f2, where f1 and f2 may be considered as fractional bits to improve the resolution of the SAR ADC 20. For example, the bit values ​​of [b0,f1,f2] are the same as the results V for any of the eight comparisons. + >V - If the accumulated result y indicates that the system does not have a result V, then it may be [0,0,0], and only one of the eight comparisons will have a result V + >V - If the accumulated result y indicates that the user has a result V, then it may be [0,0,1] and all eight comparisons have a result V + >V - If the accumulation result y indicates that the vector has the same length as the vector, then it may be [1,1,1].

[0041] Note that the offset may be optionally included or omitted and may have different values ​​in different implementations. If no offset is required, the accumulated result y may be summed with the bit value stored in register REG_A to generate the output codeword adc_out that is stored in register REG_B. Alternatively, in an embodiment with an offset as shown in FIG. 7, one option for the amount of offset is to set the offset equal to −3 / 4Q. For eight comparisons, Q may be equal to 8 and the offset set to −6. It is an obvious extension for comparator 204 to perform a number of comparisons other than eight. For example, SAR ADC 20 may calculate a voltage V f and the step of estimating the LSB may be configured to perform 16 comparisons, allocate 3 fractional bits in both registers REG_A and REG_B, and apply an offset of -12 to generate the N+3 bit output codeword adc_out.

[0042] In the above embodiment, the accumulated result y is added directly to the bit values ​​b0, f1, and f2 to generate the output codeword adc_out, and the noise T may be assumed to have a uniform distribution. In other embodiments, the noise T may follow other distributions, such as a Gaussian distribution as shown in Figure 5, such that the accumulated result y is transformed into bit values ​​in a non-linear manner.

[0043] 8 is a schematic diagram of the SAR ADC 20 with another detailed implementation of the controller 208 according to an embodiment of the present invention. As shown in FIG. 8, the controller 208 includes a counter 802, a look up table (LUT) 804, a combiner 806, and two registers REG_A and REG_B. Similarly, the counter 802 is + >V - The register REG_A is used to store a bit value for controlling the threshold generator 206 to generate the appropriate threshold voltage, and the register REG_B is used to store the output codeword adc_out. Additionally, the LUT 804 is configured to record a comparison result indicating the number of times when the voltage Vf 8, the LUT 804 may include an entry consisting of the LSB b0 and the values ​​of the four fractional bits f1-f4, and the controller 208 may apply the output result of the counter 802 to index the LUT 804 and use the output of the LUT 804 to fill the bit values ​​of b0 and f1-f4 in the register REG_B to form the output codeword adc_out.

[0044] In this embodiment, the counter 802 may record an accumulated result y of the comparison, and the LUT 804 is used to derive the fractional bits of the output codeword adc_out based on the accumulated result y. This embodiment is valid regardless of whether the statistical characteristics of the noise T follow a uniform distribution, a Gaussian distribution, or any other possible distribution. Zero-mean normally distributed noise N(0, σ T ) corresponds to the expected result of the comparison in the case of V f / σ T Let us refer again to the table in Figure 6(b), which gives examples of values ​​of V. f and <Y Q It can be seen that the relationship between f1, f2, f3, and f4 deviates from linear. This fact leads to the use of a LUT 804 to map the observations obtained from counter 802 to estimates of fractional bits f1 through f4.

[0045] For example, the controller 208 may include a memory configured to store the LUT 804. Upon obtaining the accumulated result y from the counter 802, the controller 208 may obtain a table index according to the accumulated result y. By searching the LUT 804 according to the table index, the controller 208 may obtain values ​​of several LSBs of the output codeword adc_out, including the original LSB b0 and fractional bits f1-f4, which are further combined with other bit values ​​stored in the register REG_A (through the combiner 806) to generate the output codeword adc_out, which is stored in the register REG_B.

[0046] Figure 9 shows the voltage V f 9(b) shows an exemplary content of the LUT 804 for estimating f. If the comparator 204 is characterized as having zero-mean Gaussian noise with standard deviation σT=0.5 LSB, and the SAR ADC 20 is configured with four fractional bits f1-f4, and seven comparisons are performed to find the fractional bits f1-f4 as shown in FIG. 8, then based on the content of the table in FIG. 6(b), the LUT 804 may be configured with the entries shown in FIG. 9(b). As an example, in a transformation in which seven comparison steps produce a result where the accumulated result y is 3, the voltage V f is estimated to have a value of −1 / 16, which is converted to a bit value that is incorporated into fractional bits f1 through f4. As another example, if counter 802 records that accumulated result y is 6, then voltage V f is estimated to be 9 / 16. It is noteworthy that the table entries in FIG. 9(b) are restricted to the range [-15 / 16, 15 / 16] to accommodate the values ​​of the four fractional bits f1 to f4. Depending on the characteristics of the noise and the number of fractional bits, the minimum and maximum values ​​of the table entries may be selected appropriately. FIG. 9(a) shows that σ T =0.5LSB, voltage V f 9(c) lists the table entries for the different cases of 4-bit estimates for σ and 4 comparisons in the step of obtaining the LSB. T =0.8LSB, voltage V f 8 lists table entries for yet another case of a 5-bit estimate for , and 8 comparisons in the step of obtaining the LSB.

[0047] Thus, if Q comparisons are performed under the same threshold voltage, there may be Q+1 possible outcomes (0 to Q) of the accumulated result y, and the LUT 804 stores the voltages V corresponding to these Q+1 outcomes. f The values ​​of the entries in the table shown in FIG. 9 are determined by the noise T having a given standard deviation σ TNote that the table is set up based on the assumption that the noise T has a Gaussian distribution with standard deviation σ, but the contents of the table should not be limited to those provided in this disclosure. In other embodiments, the noise T may follow other types of distributions and / or may have a standard deviation σ T If different, the contents of LUT 804 may be modified accordingly.

[0048] In an embodiment, the controller 208 may further include a multi-path design. Where the above comparison to generate the LSB is considered as the first path processing, the controller 208 may be configured to add further paths, each of which is comprised of multiple comparisons of other sets. The multi-path scheme uses the results of a previous path to determine a threshold voltage V for comparison in a subsequent path. th and accordingly determine the appropriate threshold voltage V th , which is used in the subsequent paths. th may be the same as or different from that used in the previous route.

[0049] Figure 10 shows the threshold voltage V th , and for easy reference, the sampled and held input analog signal V in The voltage V f In addition to the Q1 comparisons in path 1 to estimate V f To improve the accuracy of the estimation of Q, a second pass of Q comparisons may be added. Based on the results of the comparison in pass 1, the controller 208 may determine and set an optimal threshold level for the comparison in pass 2.

[0050] In an embodiment, the SAR ADC 20 compares the analog signal V in During the conversion operation, the SAR ADC 20 starts at step 1 and converts the MSBb N-1 Then, the conversion process proceeds towards step N-1, and bit b N-1Find the value of ~b1. Bit b N-1 After the value of .about.b1 is found, in step N, the controller 208 adjusts the threshold voltage V th,N 2. According to the results of path 1, controller 208 instructs threshold generator 206 to generate a threshold voltage V th,N+1 The threshold voltage of step N may be V th,N =V th (C, C+1), when the cumulative result generated from the Q1 comparisons is represented by y1, the controller 208 sets the threshold voltage V th,N+1 One exemplary criterion for determining is as follows:

[0051]

number

[0052] The threshold voltage V used in path 2 th,N+1 Once determined, the conversion process enters step N+1 and performs Q2 comparisons. The accumulated result of the Q2 comparisons in path 2 is then denoted as y2 and stored in another counter.

[0053] Returning to FIG. 3, the comparator 204 performs a series of first comparisons with respective thresholds to generate a first comparison result for determining the value of the MSB, and then performs a series of second comparisons with particular thresholds to determine the value of b0 and f1 through f2. M In consideration of multi-path operation, the comparator 204 may further perform a series of third comparisons to generate a plurality of third comparison results, and the input analog signal V in is the threshold voltage V determined according to the second comparison result. th,N+1 Thus, the controller 208 may generate the LSB of the output codeword adc_out according to the accumulated result of the third comparison, or further according to the accumulated result of the second comparison.

[0054] FIG. 11 is a schematic diagram of a SAR ADC 20 with a more detailed implementation of a controller 208 according to an embodiment of the present invention. The SAR ADC 20 outputs N integer bits b N-1 ~b0 and M fractional bits f1~f M 11, the controller 208 may include two counters 1102_1 and 1102_2, a LUT 1104, a combiner 1106, and two registers REG_A and REG_B. The counter 1102_1 is configured to record the comparison results of Q1 comparisons in the path 1 to generate an accumulated result y1, and the counter 1102_2 is configured to record the comparison results of Q2 comparisons in the path 2 to generate an accumulated result y2. The operations of the other circuit blocks are similar to those shown in the above paragraphs and will not be repeated here.

[0055] At the end of the comparison of path 2, the combiner 1106 may implement the operation of deriving an N+M-bit output codeword adc_out from the value stored in the register REG_A and the values ​​of the accumulated results y1 and y2. th,N+1 ≠V th,N , the combiner 1106 uses the accumulated result y2 to generate table indexes to access the corresponding Q2 entries in the LUT 1104, and the voltage V f to obtain an estimate of the voltage V f The estimate of and the value in register REG_A (as the MSB) may be combined to derive the output codeword adc_out. As described in the above embodiment, an offset is optionally used in the derivation of the output codeword adc_out.

[0056] In another embodiment, V th,N+1 =V th,NIf so, the combiner 1106 calculates the sum of the accumulated results as y=y1+y2 and uses the sum result y to index the corresponding Q1+Q2 entries in the LUT 1104 to obtain the voltage V f Then, the value in register REG_A (as the MSB) and the above voltage V (as the LSB) are calculated. f may be combined with the estimate of , optionally with the offset to the output codeword adc_out as described above.

[0057] 11, the controller 208 includes two counters 1102_1 and 1102_2 for generating accumulation results y1 and y2 for comparison of path 1 and path 2, respectively. Note that the multi-path operation of the SAR ADC 20 may include more than two paths with the same or different threshold voltages, and thus the controller 208 may have more than two counters for recording the corresponding accumulation results.

[0058] In addition, the threshold voltage V th,N+1 It should be noted that the above criteria for determining the threshold voltage V on path 2 are merely examples and that the threshold levels 0.2Q1 and / or 0.8Q1 may be varied. th,N+1 Another exemplary rule for determining is given as follows:

[0059]

number

[0060] Alternatively, the threshold voltage V used in path 2 th,N+1 The rule for determining y may be implemented by using a LUT indexed by the accumulated result y1. The entries in this LUT may be programmed according to the statistical properties of the noise and the number of comparisons in path 1.

[0061] Simulations are performed for various combinations of parameters including analog-front-end (AFE) noise, comparator noise, number of comparisons in path 1, and number of comparisons in path 2 used in embodiments of the present invention. The results are listed in Tables 1 and 2 below.

[0062] [Table 1]

[0063] [Table 2] In these embodiments, the SAR ADC scales the full-scale sinusoidal input signal to 14 integer bits (i.e., b 13 1 to 4, the SAR ADC may be converted to a digital output codeword containing a digital input signal (i.e., 1 to 4) and four fractional bits (i.e., f1 to f4) with a sample rate of 768 kilosamples per second (ksps). The comparator noise is the noise T described in the paragraph above, and the AFE noise is the noise generated in the AFE circuitry and applied to the input analog signal before it enters the SAR ADC. The AFE noise and the comparator noise are determined by the least significant bit level (i.e., the LSB ADC ) Table 1 shows the simulation results when a 1 kHz single-tone input signal is applied, and Table 2 shows the simulation results when a 300 kHz single-tone input signal is applied.

[0064] Row 3 in Tables 1 and 2, labeled "SQNR of Conventional SAR ADC," shows the signal to quantization noise ratio of a conventional SAR ADC without multiple comparisons for LSB(b0). Row 5, labeled "SQNR (1 path, by majority rule)," lists the results of a variation of the conventional SAR ADC that performs multiple comparisons in one path and applies a majority rule to derive LSB(b0). Results corresponding to embodiments of the present invention are provided in rows 6 and 8 of the tables. Row 6, labeled "SQNR (1 path, with 4-bit LUT)," shows the signal to quantization noise ratio of a conventional SAR ADC without multiple comparisons for LSB(b0). f The following lists simulation results for an embodiment configured with a LUT to evaluate the estimation of V. Line 8, labeled "SQNR (2-path, with 4-bit LUT)", shows the comparison of the two paths and the voltage V f 13 shows simulation results for an embodiment configured with a LUT to evaluate the estimate of

[0065] Tables 1 and 2 clearly show the SQNR improvement provided by the embodiment of the present invention. For example, as shown in Table 1, when AFE noise is ignored and the comparator noise is 0.8 LSB, the data in column 3 shows that the SAR ADC with one path and six comparisons outperforms the conventional SAR ADC by about 3.5 dB of SQNR improvement, and the SAR ADC with two paths, six comparisons on path 1 and six comparisons on path 2 outperforms the conventional SAR ADC by about 5.9 dB, which corresponds to an increase of nearly 1 bit of ENOB (effective number of bits). If we assume that the AFE noise has a value of 0.5 LSB, the improvement achieved in the SAR ADC is slightly lower due to the non-zero AFE noise. For example, according to the data in column 6, the SAR ADC with one path and six comparisons achieves a 2.4 dB improvement, and the SAR ADC with two paths, six comparisons on path 1 and six comparisons on path 2 achieves a 3.6 dB improvement, which corresponds to an increase of about 0.6 bits of ENOB over the conventional SAR ADC. Table 2 shows the simulation results for a 300kHz single-tone test signal, and the SQNR improvement over a conventional SAR ADC is consistent with that for a 1kHz single-tone test signal.

[0066] In summary, the present invention provides a novel SAR ADC that can achieve an improvement in SQNR by generating fractional bits based on the statistical characteristics of noise under multiple comparisons on the LSB to increase the resolution of the SAR ADC. In an embodiment, the SAR ADC applies a counter to record the accumulated results of multiple comparisons, and generates the original LSB b0 and fractional bits f1 to f M may be added as an LSB to generate an output codeword. In an embodiment, a controller of the SAR ADC may take the accumulated results of multiple comparisons and index the LUT to find a bit value for estimating a voltage difference and a threshold voltage of the input analog signal to generate an output codeword by referencing the LUT. In an embodiment, a multi-path scheme is applied to use the result of the comparison in a previous path to determine a threshold voltage for the comparison in a subsequent path and control the threshold generator to generate an appropriate threshold voltage accordingly to find a more favorable threshold voltage for the comparison. The implementation and operation provided in the embodiment of the present invention may improve the SQNR of the SAR ADC, thereby improving the overall performance of a control system using the SAR ADC.

[0067] Those skilled in the art will readily observe that numerous modifications and variations of the devices and methods may be made while retaining the teachings of the present invention. Accordingly, the above disclosure should be construed as limited only by the content and scope of the appended claims.

Claims

1. 1. A successive approximation register (SAR) analog-to-digital converter (ADC) including a comparator, a threshold generator, and a controller, The comparator receives an analog signal and the SAR ADC outputs an output codeword; the comparator performs a plurality of first comparisons and a plurality of second comparisons; the controller determines a plurality of most significant bits of the output codeword according to a plurality of first comparison results corresponding to the plurality of first comparisons; the plurality of first comparisons are performed by comparing the analog signal to a plurality of first thresholds; the controller determines a combination of least significant bits and fractional bits of the output codeword according to a plurality of second comparison results corresponding to the plurality of second comparisons; the plurality of second comparisons are performed by comparing the analog signal to a second threshold; The controller controls the threshold generator to generate the first thresholds and the second threshold according to the first comparison results.

2. The SAR ADC of claim 1 , wherein the controller obtains an accumulation result of the plurality of second comparison results, and determines the combination of the least significant bits and the fractional bits of the output codeword according to the accumulation result.

3. The SAR ADC of claim 2 , wherein the controller includes a counter configured to generate the accumulated result.

4. The SAR ADC of claim 2 , wherein the controller determines the combination of the least significant bits and the fractional bits of the output codeword according to the accumulation result and an offset.

5. the controller includes a memory configured to store a lookup table; The controller obtains a table index according to the accumulation result; The SAR ADC of claim 2 , wherein the controller obtains the combination of the least significant bits and the fractional bits of the output codeword by searching the lookup table according to the table index.

6. the comparator further performs a plurality of third comparisons; the controller determines the combination of the least significant bits and the fractional bits of the output codeword according to a plurality of third comparison results corresponding to the plurality of third comparisons; the plurality of third comparisons are performed by comparing the analog signal to a third threshold; The SAR ADC of claim 1 , wherein the controller controls the threshold generator to generate the third threshold.

7. The SAR ADC of claim 6 , wherein the controller determines the third threshold value according to the plurality of second comparison results.

8. the controller obtains a first accumulation result of the plurality of second comparison results and a second accumulation result of the plurality of third comparison results; The SAR ADC of claim 6 , wherein the controller determines the combination of the least significant bits and the fractional bits of the output codeword according to the second accumulation result.

9. the controller includes a memory configured to store a lookup table; The controller obtains a table index according to the second accumulation result; The SAR ADC of claim 8 , wherein the controller obtains the combination of the least significant bits and the fractional bits of the output codeword by searching the lookup table according to the table index.

10. the controller includes a memory configured to store a lookup table; The controller obtains a table index according to the first accumulation result and the second accumulation result; The SAR ADC of claim 8 , wherein the controller obtains the combination of the least significant bits and the fractional bits of the output codeword by searching the lookup table according to the table index.

11. The SAR ADC of claim 8 , wherein the controller includes a first counter and a second counter configured to generate the first accumulation result and the second accumulation result.

12. 1. A method applied to a successive approximation register (SAR) analog-to-digital converter (ADC) configured to output an output codeword, comprising: receiving an analog signal; performing a plurality of first comparisons on the analog signal to a plurality of first thresholds; determining a plurality of most significant bits of the output codeword according to a plurality of first comparison results corresponding to the plurality of first comparisons; performing a plurality of second comparisons of the analog signal with a second threshold value; determining a combination of least significant bits and fractional bits of the output codeword according to a plurality of second comparison results corresponding to the plurality of second comparisons; determining the first threshold values ​​and the second threshold value according to the first comparison results; The method includes:

13. determining the combination of the least significant bits and the fractional bits according to the plurality of second comparison results, obtaining an accumulation result of the plurality of second comparison results; determining the combination of the least significant bits and the fractional bits of the output codeword according to the accumulation result; The method of claim 12, comprising:

14. determining the combination of the least significant bits and the fractional bits according to the accumulation result, 14. The method of claim 13, comprising determining the combination of the least significant bits and the fractional bits of the output codeword according to the accumulated result and an offset.

15. the SAR ADC includes a memory configured to store a lookup table; determining the combination of the least significant bits and the fractional bits according to the accumulation result, obtaining a table index according to the accumulation result; obtaining the combination of the least significant bit and the fractional bit of the output codeword by searching the lookup table according to the table index; The method of claim 13, comprising:

16. performing a plurality of third comparisons on the analog signal with a third threshold value; determining the combination of the least significant bits and the fractional bits of the output codeword according to a plurality of third comparison results corresponding to the plurality of third comparisons; The method of claim 12 further comprising:

17. The method of claim 16 , further comprising determining the third threshold value according to the plurality of second comparison results.

18. The step of determining the combination of the least significant bits and the fractional bits comprises: obtaining a first accumulation result of the plurality of second comparison results and a second accumulation result of the plurality of third comparison results; determining the combination of the least significant bits and the fractional bits of the output codeword according to the second accumulation result; 17. The method of claim 16, comprising:

19. the SAR ADC includes a memory configured to store a lookup table; determining the combination of the least significant bit and the fractional bit according to the second accumulation result, obtaining a table index according to the second accumulation result; obtaining the combination of the least significant bit and the fractional bit of the output codeword by searching the lookup table according to the table index; 20. The method of claim 18, comprising:

20. the SAR ADC includes a memory configured to store a lookup table; The step of determining the combination of the least significant bits and the fractional bits comprises: obtaining a table index according to the first accumulation result and the second accumulation result; obtaining the combination of the least significant bit and the fractional bit of the output codeword by searching the lookup table according to the table index; 20. The method of claim 18, comprising:

Citation Information

Patent Citations

  • Successive approximation type a / d converter, successive approximation type a / d converter controlling method, solid-state imaging device, and image capturing apparatus

    JP2010063055A

  • A / d converter

    JP2012244579A

  • A / d conversion circuit and electronic apparatus

    JP2013211611A