A dynamic SAR ADC capacitor array structure
By splitting the high-level capacitor into two split capacitors in the SAR ADC capacitor array and floating the lower plate after switching, the power consumption problem caused by the switching of the high-level capacitor is solved, and the power consumption is significantly reduced while the performance is maintained.
Patent Information
- Application Number
- CN202311123552.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-09-01
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2043-09-01
AI Technical Summary
As the resolution of SAR ADCs increases, the number of capacitors in traditional SAR ADCs increases exponentially, resulting in a significant power consumption contribution from switching high-order capacitors during quantization. Existing capacitor switching strategies have failed to effectively reduce dynamic power consumption.
A dynamic SAR ADC capacitor array structure is adopted, which splits the high-level capacitor and some capacitors into two split capacitors. During the quantization process, only one capacitor is switched, and after the switch, its lower plate is made to float, avoiding charge extraction and discharge, and reducing the number of capacitors for active switching.
It significantly reduces the switching power consumption of the capacitor array while maintaining the successive approximation effect, without affecting the function and performance of the ADC.
Smart Images

Figure CN117176150B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of analog integrated circuit technology, specifically a dynamic SAR ADC capacitor array structure to reduce the power consumption of SAR ADC capacitor array switching, and is suitable for successive approximation analog-to-digital converters. Background Technology
[0002] Compared to Pipeline ADCs and Σ-Δ ADCs, SAR ADCs offer moderate speed and moderate accuracy. Furthermore, due to their larger digital module proportion, they have relatively smaller power consumption and area. Another advantage of a larger digital module proportion is that it is easier to improve accuracy and speed using advanced processes, thus offering greater performance potential. SAR ADCs have gained widespread application due to these advantages. However, as SAR ADC resolution continues to increase, the number of capacitors in traditional SAR ADCs increases exponentially. Even with a segmented capacitor structure, the number of capacitors remains high at high resolutions, resulting in significant power consumption during high-order capacitor switching in the quantization process.
[0003] Researchers have proposed several strategies to reduce the dynamic power consumption of SAR ADCs during quantization, including energy-saving capacitor switching strategies, monotonic capacitor switching strategies, and common-mode voltage-based capacitor switching strategies. In these different capacitor switching strategies, the lower capacitor plate continuously switches between reference voltage, ground, and common-mode voltage, resulting in continuous charge extraction and discharge. This active connection is the direct cause of the switching power consumption. Summary of the Invention
[0004] To address the aforementioned problems or shortcomings, this invention provides a dynamic SAR ADC capacitor array structure. By continuously floating the lower plate of the high-position capacitor during the quantization process, the number of capacitors that perform active switching on the lower plate is continuously reduced, thereby further reducing the switching power consumption of the capacitor array.
[0005] A dynamic SAR ADC capacitor array structure is provided, wherein the highest-order capacitor and the terminal capacitor are non-split capacitors, and the remaining capacitors are set as split capacitors divided into two.
[0006] In the split capacitor, the second-highest capacitor is uniformly split into two identical capacitors, while the split capacitors outside the second-highest capacitor are non-uniformly split. During the quantization process, only one of the split capacitors is cut up or down.
[0007] During quantization in a SAR ADC, the weight of the i-th capacitor can be expressed as: W i =C i / C tot The C of traditional SAR ADCtot Keep constant, C i The process involves continuously reducing the capacitance by 1 / 2 to complete successive approximation operations. However, after floating the lower plate of a high-level capacitor, that capacitor no longer participates in the charge redistribution process. Therefore, during quantization, the ADC... tot The number of digits is also continuously reduced by half, which means that even if Ci is continuously reduced by half, it cannot complete the successive approximation operation like a traditional SAR ADC. Therefore, this invention addresses the issue of C... i Processing is performed so that the C that was switched last time... i Is it the next time C will be switched? i+1 1 / 4, that is, C i+1 =1 / 4C i Therefore, W i+1 =1 / 2W i Therefore, the capacitor array of this invention, except for the highest-order capacitor, the second-highest-order capacitor, and the terminal capacitor, needs to be non-uniformly split, while the second-highest-order capacitor needs to be uniformly split. By splitting, two capacitors are obtained, and only one of the capacitors is overcut or undercut during the quantization process.
[0008] Taking a SAR ADC with plate sampling and a capacitor switching strategy based on common-mode voltage as an example, this paper illustrates a method to further reduce the switching power consumption of the SAR ADC capacitor array. The specific steps are as follows:
[0009] Step 1: During the sampling phase, the two input terminals of CDAC are connected to Vip and Vin respectively, and the lower plates of all capacitors are connected to the common-mode voltage Vcm. At this time, the voltages at the two input terminals of the comparator are VDACP1 = Vip and VDACN1 = Vin, respectively, and the comparator input is VDACP1 - VDACN1 = (Vip - Vin) - 0.
[0010] Step 2: If the sampled VDACP1 is greater than VDACN1, the comparator outputs 1, the lower plate of the highest-order capacitor at the P terminal needs to be switched down to ground GND, and the lower plate of the highest-order capacitor at the N terminal needs to be switched up to the reference voltage VREF; if the sampled VDACP1 is less than VDACN1, the comparator outputs 0, the lower plate of the highest-order capacitor at the P terminal needs to be switched up to the reference voltage VREF, and the lower plate of the highest-order capacitor at the N terminal needs to be switched down to ground GND.
[0011] Then, after CDAC is established, the lower plates of the highest-order capacitors at both ends P and N are simultaneously floating. At this time, the voltages at the two input terminals of the comparator are VDACP2 = Vip - 1 / 2 Vcm and VDACN2 = Vin + 1 / 2 Vcm, respectively, and the comparator input is VDACP2 - VDACN2 = (Vip - Vin) - Vcm = (Vip - Vin) - 1 / 2 VREF.
[0012] Step 3: If VDACP2 is greater than VDACN2, the comparator continues to output 1. The lower plate of one of the second-highest split capacitors at the P terminal needs to be switched to ground GND, while the lower plate of the other capacitor remains connected to the common-mode voltage Vcm. The lower plate of one of the second-highest split capacitors at the N terminal needs to be switched to the reference voltage VREF, while the lower plate of the other capacitor remains connected to the common-mode voltage Vcm. If VDACP2 is less than VDACN2, the comparator outputs 0, and the switching method of the P and N terminal capacitors is reversed.
[0013] Then, after CDAC is established, the lower plates of the two capacitors split at the second-highest bit of P and N are both floating. At this time, the voltages at the two input terminals of the comparator are VDACP3 = Vip - 3 / 4 Vcm and VDACN3 = Vin + 3 / 4 Vcm, respectively, and the comparator input is VDACP3 - VDACN3 = (Vip - Vin) - 3 / 2 Vcm = (Vip - Vin) - 3 / 4 VREF.
[0014] Step 4: If VDACP3 is less than VDACN3, the comparator outputs 0. In the third capacitor split at the P terminal, the lower plate of the smaller capacitor needs to be switched to the reference voltage VREF, while the lower plate of the other larger capacitor remains connected to the common-mode voltage Vcm. In the third capacitor split at the N terminal, the lower plate of the smaller capacitor needs to be switched to ground GND, while the lower plate of the other larger capacitor remains connected to the common-mode voltage Vcm. If VDACP3 is greater than VDACN3, the comparator outputs 1, and the switching method of the P and N terminals is reversed.
[0015] Then, after CDAC is established, the lower plate of the third capacitor, which is split at both ends P and N, is floating. At this time, the voltages at the two input terminals of the comparator are VDACP4 = Vip - 5 / 8Vcm and VDACN4 = Vin + 5 / 8Vcm, respectively. The comparator input is VDACP4 - VDACN4 = (Vip - Vin) - 10 / 8Vcm = (Vip - Vin) - 5 / 8VREF.
[0016] Step 5: The switching method for the remaining capacitors is the same as the capacitor switching method in the above steps. During the quantization process, the lower plate of the high-level capacitor is continuously floated until the last bit is quantized.
[0017] In this invention, the second capacitor (second highest bit) is split into two equal capacitors. One of them is either over- or under-switched, while the lower plate of the other capacitor remains connected to the common-mode voltage Vcm. Starting from the third capacitor, the smaller split capacitor is either over- or under-switched, while the lower plate of the larger capacitor remains connected to the common-mode voltage Vcm. After CDAC is established, the lower plates of both capacitors are floated. During subsequent quantization, the voltage difference across the capacitor with its floated lower plate does not change, and the capacitor no longer draws or discharges charge, thus no longer contributing dynamic power consumption. This further reduces the dynamic power consumption caused by capacitor array switching.
[0018] In summary, unlike the switching method of traditional SAR ADCs, this invention reduces the dynamic power consumption contributed by the high-order capacitors during subsequent capacitor switching by continuously floating the lower plate of the switched high-order capacitors. Simultaneously, by splitting all quantization capacitors except the highest-order and terminal capacitors into two capacitors and only switching one capacitor up or down, it achieves the same successive approximation effect as traditional SAR ADC switching methods. Without affecting its functionality and performance, the ADC's switching power consumption is significantly reduced. Attached Figure Description
[0019] Figure 1 This is a schematic diagram of the traditional SAR ADC switching process based on a common-mode voltage switching strategy;
[0020] Figure 2 This is a schematic diagram of the switching process of the present invention;
[0021] Figure 3 This is a schematic diagram of the switching process in an embodiment. Detailed Implementation
[0022] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.
[0023] Figure 1 This is a schematic diagram of the traditional SAR ADC switching process based on the common-mode voltage switching strategy. During the subsequent capacitor switching process, the lower plate of the high-level capacitor continuously draws and discharges charge from VREF and GND, generating dynamic power consumption.
[0024] Figure 2 This is a schematic diagram of the dynamic SAR ADC switching process based on the common-mode voltage switching strategy of the present invention. In the subsequent capacitor switching process, the lower plates of the capacitors before the capacitor being switched are all floating, unable to extract or discharge charge, and no more dynamic power consumption will be generated.
[0025] Figure 3The diagram shown is a switching schematic of a 5-bit dynamic SAR ADC based on a common-mode voltage switching strategy in this embodiment. As mentioned above, starting from the second capacitor, some capacitors in the capacitor array are split sequentially into C2 = 2Cu, C2′ = 2Cu, C3 = 0.5Cu, C3′ = 1.5Cu, C4 = 0.125Cu, C4′ = 0.875Cu. The highest-bit capacitor C1 and the terminal capacitor CT are not split.
[0026] Step 1: During the sampling phase, the two input terminals of CDAC are connected to Vip and Vin respectively, and the lower plates of all capacitors are connected to the common-mode voltage Vcm. At this time, the voltages at the two input terminals of the comparator are VDACP1 = Vip and VDACN1 = Vin, respectively, and the comparator input is VDACP1 - VDACN1 = (Vip - Vin) - 0.
[0027] Step 2: If the sampled VDACP1 is greater than VDACN1, the comparator outputs 1, the lower plate of the highest-order capacitor C1_P at the P terminal needs to be switched down to ground GND, and the lower plate of the highest-order capacitor C1_N at the N terminal needs to be switched up to the reference voltage VREF; if the sampled VDACP1 is less than VDACN1, the comparator outputs 0, the lower plate of the highest-order capacitor at the P terminal needs to be switched up to the reference voltage VREF, and the lower plate of the highest-order capacitor at the N terminal needs to be switched down to ground GND.
[0028] Then, after CDAC is established, the lower plates of the highest-order capacitors C1_P and C1_N at terminals P and N are both floating. At this time, the voltages at the two input terminals of the comparator are VDACP2 = Vip - 1 / 2 Vcm and VDACN2 = Vin + 1 / 2 Vcm, respectively, and the comparator input is VDACP2 - VDACN2 = (Vip - Vin) - 1 / 2 VREF.
[0029] Step 3: If VDACP2 is greater than VDACN2, the comparator continues to output 1. The lower plate of capacitor C2_P, which is split at the P terminal, needs to be switched to ground GND, while the lower plate of the other capacitor C2′_P remains connected to the common-mode voltage Vcm. The lower plate of capacitor C2_N, which is split at the N terminal, needs to be switched to the reference voltage VREF, while the lower plate of the other capacitor C2′_N remains connected to the common-mode voltage Vcm. If VDACP2 is less than VDACN2, the comparator outputs 0, and the switching method of the P and N terminals is reversed.
[0030] Then, after CDAC is established, the lower plates of the second-highest capacitors C2_P, C2′_P and C2_N, C2′_N at terminals P and N are all floating. At this time, the voltages at the two input terminals of the comparator are VDACP3 = Vip - 3 / 4 Vcm and VDACN3 = Vin + 3 / 4 Vcm, respectively, and the comparator input is VDACP3 - VDACN3 = (Vip - Vin) - 3 / 4 VREF.
[0031] Step 4: If VDACP3 is less than VDACN3, the comparator outputs 0. The lower plate of the third capacitor split at the P terminal needs to be switched to the reference voltage VREF, while the lower plate of the other capacitor C3′_P remains connected to the common-mode voltage Vcm. The lower plate of the third capacitor split at the N terminal needs to be switched to ground GND, while the lower plate of the other capacitor C3′_P remains connected to the common-mode voltage Vcm. If VDACP3 is greater than VDACN3, the comparator outputs 1, and the switching method of the P and N terminals capacitors is reversed.
[0032] Then, after CDAC is established, the lower plates of the third capacitors C3_P, C3′_P and C3_N, C3′_N at terminals P and N are all floating. At this time, the voltages at the two input terminals of the comparator are VDACP4 = Vip - 5 / 8Vcm and VDACN4 = Vin + 5 / 8Vcm, respectively, and the comparator input is VDACP4 - VDACN4 = (Vip - Vin) - 5 / 8VREF.
[0033] Step 5: If VDACP4 is less than VDACN4, the comparator outputs 0. The lower plate of capacitor C4_P, which is split at the P terminal, needs to be switched to the reference voltage VREF, while the lower plate of the other capacitor C4′_P remains connected to the common-mode voltage Vcm. The lower plate of capacitor C4_N, which is split at the N terminal, needs to be switched to ground GND, while the lower plate of the other capacitor C4′_N remains connected to the common-mode voltage Vcm. If VDACP4 is greater than VDACN4, the comparator outputs 1, and the switching method of the P and N terminals capacitors is reversed.
[0034] Then, after CDAC is established, the voltages at the two input terminals of the comparator are VDACP5 = Vip-9 / 16Vcm and VDACN5 = Vin+9 / 16Vcm, respectively.
[0035] Step 6: If VDACP5 is greater than VDACN5, the comparator outputs 1; if VDACP5 is less than VDACN5, the comparator outputs 0. This completes the quantization process.
[0036] The residual voltage on the CDAC is Vres=VDACP5-VDACN5=(Vip-Vin)-9 / 16VREF, and its value is less than 1LSB.
[0037] As can be seen from the above embodiments, the present invention sets the remaining capacitors other than the highest-order capacitor and the terminal capacitor as split capacitors, wherein the second-highest-order capacitor is uniformly split into two identical capacitors, and the split capacitors other than the second-highest-order capacitor are non-uniformly split. During quantization, only one capacitor of the split capacitors is over-cut or under-cut (for non-uniformly split capacitors, the smaller capacitor is over-cut or under-cut), while the lower plate of the other capacitor remains connected to the common-mode voltage Vcm. After CDAC is established, the lower plates of both capacitors are floated. By continuously floating the lower plates of the high-order capacitors during quantization, the number of capacitors with active switching of the lower plates is continuously reduced. The voltage difference across the capacitors whose lower plates have been floated does not change, and the capacitors no longer draw or discharge charge, thus no longer contributing dynamic power consumption. Without affecting its function and performance, the switching power consumption of the ADC is significantly reduced.
Claims
1. A dynamic SAR ADC capacitor array, characterized in that: Its highest-order capacitor and terminal capacitor are non-split capacitors, while the remaining capacitors are set as split capacitors that are split into two. In the split capacitor, the second highest capacitor is uniformly split into two identical capacitors, while the split capacitors outside the second highest capacitor are non-uniformly split. During the quantization process, only one of the split capacitors is cut up or down. The specific capacitor array switching steps are as follows: Step 1: During the sampling phase, the two input terminals of CDAC are connected to Vip and Vin respectively, and the lower plates of all capacitors are connected to the common-mode voltage Vcm; at this time, the voltages at the two input terminals of the comparator are VDACP1=Vip and VDACN1=Vin respectively, and the comparator input is VDACP1-VDACN1=(Vip-Vin)-0. Step 2: If the sampled VDACP1 is greater than VDACN1, the comparator outputs 1, the lower plate of the most significant capacitor at the P terminal needs to be switched down to ground GND, and the lower plate of the most significant capacitor at the N terminal needs to be switched up to the reference voltage VREF; if the sampled VDACP1 is less than VDACN1, the comparator outputs 0, the lower plate of the most significant capacitor at the P terminal needs to be switched up to the reference voltage VREF, and the lower plate of the most significant capacitor at the N terminal needs to be switched down to ground GND. Then, after CDAC is established, the lower plates of the highest-order capacitors at both ends of P and N are simultaneously floating. At this time, the voltages at the two input terminals of the comparator are VDACP2=Vip-1 / 2Vcm and VDACN2=Vin+1 / 2Vcm, respectively. The comparator input is VDACP2-VDACN2=(Vip-Vin)-Vcm=(Vip-Vin)-1 / 2VREF. Step 3: If VDACP2 is greater than VDACN2, the comparator continues to output 1. The lower plate of one of the second-highest split capacitors at the P terminal needs to be switched to ground GND, while the lower plate of the other capacitor remains connected to the common-mode voltage Vcm. The lower plate of one of the second-highest split capacitors at the N terminal needs to be switched to the reference voltage VREF, while the lower plate of the other capacitor remains connected to the common-mode voltage Vcm. If VDACP2 is less than VDACN2, the comparator outputs 0, and the switching method of the P and N terminals capacitors is reversed. Then, after CDAC is established, the lower plates of the two capacitors that were split at the second highest bit of P and N are both floating; at this time, the voltages at the two input terminals of the comparator are VDACP3 = Vip - 3 / 4Vcm and VDACN3 = Vin + 3 / 4Vcm, respectively, and the comparator input is VDACP3 - VDACN3 = (Vip - Vin) - 3 / 2Vcm = (Vip - Vin) - 3 / 4VREF; Step 4: If VDACP3 is less than VDACN3, the comparator outputs 0. The lower plate of the smaller capacitor in the split third capacitor at the P terminal is switched to the reference voltage VREF, while the lower plate of the other larger capacitor remains connected to the common-mode voltage Vcm. In the split third capacitor at the N terminal, the lower plate of the smaller capacitor needs to be switched to ground GND, while the lower plate of the other larger capacitor remains connected to the common-mode voltage Vcm. If VDACP3 is greater than VDACN3, the comparator outputs 1, and the switching method of the P and N terminals is reversed. Then, after CDAC is established, the lower plate of the third capacitor split at both ends P and N is floating. At this time, the voltages at the two input terminals of the comparator are VDACP4=Vip-5 / 8Vcm and VDACN4=Vin+5 / 8Vcm, respectively. The comparator input is VDACP4-VDACN4=(Vip-Vin)-10 / 8Vcm=(Vip-Vin)-5 / 8VREF. Step 5: The switching method for the remaining capacitors is the same as the capacitor switching method in the above steps. During the quantization process, the lower plate of the high-level capacitor is continuously floated until the last bit is quantized.
Citation Information
Patent Citations
Capacitor array switching method of high-energy-efficiency SAR ADC
CN111641413A
SAR ADC and sampling method based on single-channel time-interleaved-sampling
US20220247423A1
