Randomized data weighted average algorithm circuit applied to Sigma-Delta ADC

Abstract

The invention discloses a randomized data weighted average algorithm circuit applied to a Sigma-Delta ADC, and relates to the technical field of integrated circuits, and the randomized data weighted average algorithm circuit applied to the Sigma-Delta ADC comprises a thermometer code-to-binary code circuit, a random number generation circuit, a full adder circuit and a shifter circuit. The output result of the quantizer is selected through the thermometer code-to-binary code module and the result of the random number generation module to judge whether a pointer needs to be subjected to randomized definition, and then the result of the previous moment is summed through the full adder circuit. And finally, the work of converting the thermometer code into the DWA code is completed through the barrel-shaped shifter circuit, so that the influence caused by the mismatch of the DAC unit can be reduced, the nonlinear problem caused by multi-bit quantization can be solved, the algorithm circuit is simple in structure, and the performance of the Sigma-Delta analog-to-digital converter circuit is greatly improved.

Description

Technical Field

[0001] The invention relates to the technical field of integrated circuits, and in particular to a randomized data weighted average algorithm circuit applied to a Sigma-Delta ADC. Background Art

[0002] Sigma-Delta analog-to-digital converter (SD-ADC) is widely used in the fields of sensors, audio, data acquisition and medical fields. It is a high-precision analog-to-digital converter. Its core principle is to improve the conversion accuracy through oversampling and noise shaping technology. There are two types of quantization methods for Sigma-Delta analog-to-digital converters: single-bit quantization and multi-bit quantization, each with its own advantages and disadvantages. Compared with single-bit quantization, the advantage of multi-bit quantization is that each sample value can be expressed with high accuracy, which can better capture the subtle changes of the input signal. In addition, multi-bit quantization has better dynamic range and noise performance, and is widely used in applications with high precision requirements. However, compared with single-bit quantization, its linearity is poor, mainly because there is a mismatch between the feedback digital-to-analog converter (DAC) units. Therefore, a randomized data weighted average algorithm circuit applied to Sigma-Delta ADC is proposed. Summary of the invention

[0003] In view of the deficiencies in the prior art, the present invention provides a randomized data weighted average algorithm circuit applied to a Sigma-Delta ADC, which solves the problems raised in the above background technology.

[0004] To achieve the above objectives, the present invention is implemented through the following technical solutions: a randomized data weighted average algorithm circuit applied to Sigma-Delta ADC, including a thermometer code to binary code circuit, a random number generation circuit, a full adder circuit and a shifter circuit.

[0005] Optionally, the randomized data weighted average algorithm circuit adopts a randomized DWA algorithm. The DWA algorithm is used for 16 DAC units to be cyclically used. After enough conversion cycles, the probability of using each unit is approximately equal. If the first quantized number is 3, the DAC of units 1-3 is selected. When the next quantized output is 4, it is necessary to add the DAC unit with the largest bit selected once, that is, select 4-7, and so on. When the quantizer output plus the highest bit selected by the DAC once is greater than 16, 16 can be followed by 1. Through this principle of cyclic selection, after a long period of circulation, the probability of using each unit in the DAC is almost equal.

[0006] Optionally, the randomized DWA combines the DWA algorithm and the randomization algorithm. The DWA algorithm is used to add the mismatch errors to zero, including but not limited to selecting cells by rotation.

[0007] Optionally, in the traditional DWA selection, if the DAC input is 3, then C is selected. 1~ C3. For the randomized DWA, any 3 cells can be selected from 8 cell DACs, and C2, C5, C can be selected. 8， When the DAC input is 2, the traditional DWA selects C. 4~ C5. For the randomized DWA, any two of the remaining 5 capacitors can be selected, such as C4 and C7. When the DAC input is 5, the traditional DWA algorithm selects C. 6~ C8, C. 1~ C2. The randomized DWA first selects the remaining C1, C3, and C6, and then randomly selects two from C2, C4, C5, C7, and C8 for the remaining two, such as C5 and C8. And so on for other cases.

[0008] Optionally, if there are p DAC units, and the gain weight of each unit is u. i , with the mean value being u, and the gain error of each DAC unit being ∆u. i Conforming to a random distribution, the following representation is obtained: (1) (2) Define S as the sum of the inputs of all DACs before the Nth cycle, then: (3) Where x(k) is the DAC input in the kth (1 < k ≤ N) cycle. Define r as the remainder after S / N: (4) Due to the reason of DAC mismatch, define ∆S: (5) For both DWA and randomized DWA, the following representation holds: (6) The above formula is the cumulative error of the DAC output. For DWA, due to the reason of cyclic shift, a. i Has the distribution property of thermometer code. For the randomized DWA, a. i Is randomly distributed. Therefore, the cumulative error of the randomized DWA can be regarded as white noise, and the standard deviation of the DAC output is: (7) in, is the standard deviation of the unit DAC mismatch, p is the number of DAC levels, then the average power of the mismatch noise of the randomized DWA is: (8) The signal-to-noise ratio of the DAC output after the random DWA algorithm is: . (9) Optionally, a thermometer code-to-binary code module is developed based on the thermometer code-to-binary code circuit, a random number generation module is developed based on the random number generation circuit, and the shifter circuit adopts a barrel shifter circuit.

[0009] Optionally, the output result of the quantizer in the randomized DWA circuit is passed through a thermometer code to binary code module, and is selected with the result of the random number generation module to determine whether the pointer needs to be randomly defined. Then, the result of the previous moment is summed through a full adder circuit, and finally completed through a barrel shifter circuit, and the thermometer code is converted into a DWA code.

[0010] The present invention provides a randomized data weighted average algorithm circuit applied to Sigma-Delta ADC, which has the following beneficial effects: The randomized data weighted average algorithm circuit applied to Sigma-Delta ADC, the output result of the quantizer is selected through the thermometer code to binary code module and the result of the random number generation module to determine whether the pointer needs to be randomly defined, then the result of the previous moment is summed through the full adder circuit, and finally the thermometer code is converted into DWA code through the barrel shifter circuit, which can reduce the influence of DAC unit mismatch, and further solve the nonlinear problem caused by multi-bit quantization. The algorithm circuit has a simple structure and greatly improves the performance of the Sigma-Delta analog-to-digital converter circuit. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 Schematic diagram of the DWA principle of the present invention, wherein (a) is a schematic diagram of the unit DAC selection logic; (b) is a schematic diagram of the unit DAC selection effect; Figure 2 The DWA algorithm of the present invention is cycle-selected, wherein (a) is the traditional DWA; (b) is the randomized DWA; Figure 3 The randomized DWA block diagram of the present invention; Figure 4 This is the SD-ADC spectrum diagram without using randomized DWA in the present invention; Figure 5 SD-ADC spectrum plot using randomized DWA for the present invention. DETAILED DESCRIPTION

[0012] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0013] See also Figures 1 to 5 ,The present invention provides a technical solution: a randomized data weighted average algorithm circuit applied to Sigma-Delta ADC, comprising a thermometer code to binary code circuit, a random number generation circuit, a full adder circuit and a shifter circuit; Since DAC units will inevitably produce mismatches during the manufacturing process, these mismatches are the source of nonlinearity. If the quantizer output is represented by a thermometer code, the probability of selecting the DAC unit representing the low bit will be high, and the probability of selecting the DAC unit representing the high bit will be low, which will lead to different charging and discharging times between the DAC unit capacitors. The errors accumulated on each capacitor unit will also be different. These errors will affect the accuracy of the Sigma Delta ADC. The purpose of the DWA algorithm is to change the starting pointer position each time so that after a long time, the probability of selecting each DAC unit is approximately equal, so that the errors accumulated on each capacitor unit are approximately the same, and the nonlinearity caused by the mismatch error is converted into random noise distributed over the entire bandwidth, thereby increasing the effective number of bits of the Sigma Delta ADC.

[0014] like Figure 1 As shown, the specific principle of the DWA algorithm can be known. Assuming 4-bits quantization, the main idea of ​​DWA is that the 16 DAC units will be used cyclically. After enough conversion cycles, the probability of using each unit is approximately equal. If the first quantized number is 3, the DAC units of 1-3 are selected. When the next quantized output is 4, it is necessary to add the DAC unit with the largest bit selected once, that is, select 4-7, and so on. It should be noted that when the quantizer output plus the highest bit selected by the DAC once is greater than 16, 16 can be followed by 1. Figure 1 (b) is the effect diagram of the quantizer output selection DAC after DWA algorithm processing, where C 1~ C 16 For the capacitors in the DAC, through this cyclic selection principle, after a long period of circulation, the probability of using each unit in the DAC is almost equal; Randomized DWA combines the DWA algorithm and the randomized algorithm. The purpose of the DWA algorithm is to add the mismatch errors to zero, including but not limited to selecting cells by rotation. The differences between randomized DWA and traditional DWA are as Figure 2 shown. In the traditional DWA selection, if the DAC input is 3, then C is selected 1~ C 3， For randomized DWA, any 3 cells can be selected from 8 cell DACs. For example, C2, C5, and C8 are selected. When the DAC input is 2, the traditional DWA selects C 4~ C5, while randomized DWA can select any two of the remaining 5 capacitors, such as C4 and C7. When the DAC input is 5, the traditional DWA algorithm selects C 6~ C8, C 1~ C2. Randomized DWA first selects the remaining C1, C3, and C6, and then randomly selects two from C2, C4, C5, C7, and C8, such as C5 and C8. The same applies to other cases; If there are p DAC cells, and the gain weight of each cell is u i , with its mean value being u, and the gain error of each DAC cell being ∆u i conforms to a random distribution, then there is the following representation: (1) (2) Define S as the sum of the inputs of all DACs before the Nth cycle, then there is: (3) where x(k) is the DAC input in the kth (1 < k ≤ N) cycle. Define r as the remainder after S / N: (4) Due to the reason of DAC mismatch, define ∆S: (5) For both DWA and randomized DWA, there is the following representation: (6) The above formula is the cumulative error of the DAC output. For DWA, due to the reason of cyclic shift, a i has the distribution property of thermometer code, while for randomized DWA, a i is randomly distributed. Therefore, the cumulative error of randomized DWA can be regarded as a white noise, and the standard deviation of the DAC output is: (7) where is the standard deviation of the unit DAC mismatch, p is the number of DAC levels, then the average power of the mismatch noise of the randomized DWA is: (8) The signal-to-noise ratio of the DAC output after the random DWA algorithm is: (9) Figure 4 The spectrum diagram of the SD-ADC circuit without the randomized DWA circuit is shown for the second-order discrete CIFF Sigma-Delta 4-bit quantized ADC circuit. Figure 5 The spectrum diagram of the SD-ADC circuit using the randomized DWA circuit is shown. It can be seen that the use of the randomized DWA circuit greatly improves the performance of the Sigma-Delta ADC circuit; The output result of the quantizer in the randomized DWA circuit is passed through the thermometer code to binary code module, and is selected with the result of the random number generation module to determine whether the pointer needs to be randomly defined. After that, the full adder circuit is used to sum the result of the previous moment, and finally the barrel shifter circuit is used to complete the work of converting the thermometer code into the DWA code.

[0015] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A randomized data weighted average algorithm circuit applied to Sigma-Delta ADC, characterized in that: It includes a thermometer code to binary code circuit, a random number generation circuit, a full adder circuit and a shifter circuit.

2. The randomized data weighted average algorithm circuit applied to Sigma-Delta ADC according to claim 1, characterized in that: The randomized data weighted averaging algorithm circuit adopts the randomized DWA algorithm. The DWA algorithm is used such that 16 DAC units are cyclically used, and after a sufficient number of conversion cycles, the probability of each unit being used is approximately equal. If the first quantized number is 3, then the DACs of units 1 - 3 are selected. When the next quantized output is 4, the DAC unit of the maximum bit selected last time needs to be added, that is, units 4 - 7 are selected, and so on. When the sum of the highest bit of the DAC selected when the quantizer output is added once is greater than 16, 1 follows immediately after 16. Through this principle of cyclic selection, after a long - term cycle, the probability of each unit in the DAC being used is nearly equal.

3. The randomized data weighted average algorithm circuit applied to Sigma-Delta ADC according to claim 1, characterized in that: The randomized DWA combines the DWA algorithm and the randomized algorithm. The DWA algorithm is used to make the mismatch error sum to zero, including but not limited to selecting units in a rotating manner.

4. The randomized data weighted average algorithm circuit applied to Sigma-Delta ADC according to claim 1, characterized in that: In the traditional DWA selection, if the DAC input is 3, then select C 1~ C3, while for randomized DWA, any 3 units can be selected from the 8 unit DACs, including C2, C5, C 8， When the DAC input is 2, the traditional DWA selects C 4~ C5, and the randomized DWA can select any two of the remaining 5 capacitors, such as C4 and C7. When the DAC input is 5, the traditional DWA algorithm selects C 6~ C8, C 1~ C2, randomized DWA first selects the remaining C1, C3, and C6, and the remaining two are randomly selected from C2, C4, C5, C7, and C8, such as C5 and C8, and so on for other cases.

5. The randomized data weighted average algorithm circuit applied to Sigma-Delta ADC according to claim 1, characterized in that: If there are p DAC units, the gain weight of each unit is u i , whose mean is u, and the gain error of each DAC unit is ∆u i If it conforms to the random distribution, it can be expressed as follows: （1） （2） Define S as the sum of the inputs of all DACs before the Nth cycle, then there is: （3） Where x(k) is the input of the DAC in the kth (1 < k ≤ N) cycle. Define r as the remainder after S / N: （4） Due to the reason of DAC mismatch, define ∆S: （5） For DWA or randomized DWA, there will be the following representation: （6） The above formula is the cumulative error of the DAC output. For DWA, due to the cyclic shift, a i has the distribution property of thermometer code, and for randomized DWA, a i is randomly distributed, so the cumulative error of the randomized DWA can be regarded as a white noise, and the standard deviation of the DAC output is: （7） in, is the standard deviation of the unit DAC mismatch, p is the number of DAC levels, then the average power of the mismatch noise of the randomized DWA is: （8） The signal - to - noise ratio of the DAC output after the random DWA algorithm is: 　　　　　　　　　　（9）。 6. The randomized data weighted average algorithm circuit applied to Sigma-Delta ADC according to claim 1, characterized in that: Based on the thermometer code to binary code circuit, a thermometer code to binary code module is developed. Based on the random number generation circuit, a random number generation module is developed. The shifter circuit adopts a barrel shifter circuit.

7. The randomized data weighted average algorithm circuit applied to Sigma-Delta ADC according to claim 1, characterized in that: The output result of the quantizer in the randomized DWA circuit passes through the thermometer code to binary code module, and is selected with the result of the random number generation module to determine whether to randomize the definition of the pointer. Then, the result of the previous moment is summed through the full adder circuit, and finally, the work of converting the thermometer code into the DWA code is completed through the barrel shifter circuit.

Images