Adaptive sampling-based on-chip integrating-type analog-to-digital converter
Patent Information
- Application Number
- IN202511129796
- Authority / Receiving Office
- IN · IN
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2025-12-22
- Publication Date
- 2026-10-13
- Estimated Expiration
- 2045-12-22
AI Technical Summary
Conventional high-resolution analog-to-digital converters face inefficiencies such as extended conversion times, high dynamic power consumption, and susceptibility to process-voltage-temperature variations, making them unsuitable for power-autonomous systems and resource-constrained environments.
An adaptive sampling-based on-chip integrating-type data converter that initiates each conversion from an intermediate reference derived from the prior cycle, utilizing a feedback-driven architecture with a comparator-directed selector and ramp integration to minimize switching cycles and adapt to signal variations.
Achieves high-resolution digital outputs with reduced conversion times and power consumption, ensuring robust operation across environmental extremes by leveraging signal correlations and minimizing redundant processing.
Abstract
Description
FIELD OF THE INVENTIONThe present subject matter described herein relates to the field of analog-to-digitalconverters for high-resolution conversion in low-speed applications, such assensor interfaces and instrumentation systems, in general, and more particularly toan adaptive sampling-based on-chip integrating-type data converter for high resolution analog-to-digital conversion with dynamically adjustable conversiontimes and reduced dynamic power consumption in low-speed precisionapplications such as sensor interfaces and medical instrumentation.BACKGROUND ARTAnalog-to-digital converters (ADCs) play a pivotal role in modern electronics bybridging the gap between continuous analog signals and discrete digitalrepresentations, enabling the processing, storage, and transmission of real-worlddata in digital systems. These devices are indispensable in a wide array ofapplications, including sensor interfaces, medical equipment, scientificinstruments, metrology systems, Internet of Things (IoT) devices, audioprocessing, industrial automation, and defense technologies. In such domains, thefidelity of signal conversion directly impacts system performance, where evensubtle variations in analog inputs-such as those from biomedical sensors orenvironmental monitors-must be captured with exceptional precision to ensurereliable decision-making and data integrity.High-resolution ADCs, capable of distinguishing minute changes in input signals(often on the order of microvolts or less), are particularly valued for their ability todeliver a granular digital output that closely mirrors the original analog waveform.Resolution in an ADC is fundamentally tied to the smallest detectable incrementin the input voltage that corresponds to a single least significant bit (LSB) in theoutput code. As resolution increases-typically measured in bits (e.g., 16-bit orhigher)-the voltage span per bit decreases, allowing for finer detail extraction.This characteristic makes high-resolution ADCs essential for noise-sensitiveenvironments, where low signal-to-noise ratios demand converters that suppressquantization errors and maintain signal integrity across a broad dynamic range.The evolution of ADC architectures has been driven by the need to balanceresolution, speed, power efficiency, and integration complexity. Early designs,such as flash ADCs, offered rapid conversion through parallel comparisons but atthe expense of exponential power and area scaling with bit depth, rendering themimpractical for high resolutions beyond 8-10 bits. Successive approximationregister (SAR) ADCs emerged as a more scalable alternative, employing a binarysearch algorithm to iteratively refine the digital output over N clock cycles for anN-bit resolution. SAR converters sample the input via a hold circuit, then use adigital-to-analog converter (DAC) and comparator to test bit decisionssequentially, starting from the most significant bit (MSB). This approach achievesa favorable compromise in medium-resolution applications, such as portablemedical devices and wireless sensors, by minimizing power through capacitiveDACs and asynchronous logic.Integrating-type ADCs, including single-slope and dual-slope variants, representanother cornerstone of precision conversion. In single-slope designs, the inputvoltage charges a capacitor or integrator, and the time required for a referenceramp to reach the integrated level is measured by a counter, directly yielding thedigital code. Dual-slope architectures enhance this by first integrating theunknown input over a fixed period and then de-integrating against a reference,inherently rejecting common-mode noise like 50 / 60 Hz line interference. Thesemethods excel in metrology and instrumentation due to their inherent linearity andlow offset, often achieving resolutions up to 20 bits or more. However, theirreliance on precise timing and integration periods limits throughput, as conversiontimes scale linearly with resolution and input amplitude.Charge-balancing ADCs build on integrating principles by using switched capacitor networks to accumulate and equilibrate charge packets corresponding tothe input signal. The process involves transferring charge from an input samplingcapacitor to an integrator, which is periodically reset by reference charges untilbalance is achieved, with the number of balancing cycles forming the digitaloutput. This architecture provides robust noise rejection and high accuracy overextended periods, making it suitable for low-power sensor nodes and biomedicalimplants. Despite these strengths, the sequential charge transfers introducelatency, as the full conversion may require multiple cycles proportional to theinput value, constraining applicability in time-critical scenarios.Oversampling architectures, such as sigma-delta (Σ-Δ) modulators, leveragenoise-shaping and decimation to push quantization noise outside the signal band,enabling effective resolutions exceeding 20 bits at low frequencies. A first-orderΣ-Δ converter, for instance, includes an integrator, quantizer, and feedback DAC,where the modulator runs at a multiple of the Nyquist rate to spread noise,followed by digital filtering to recover the high-resolution signal. Higher-orderdesigns amplify this effect but introduce stability challenges. Σ-Δ ADCs dominateaudio and vibration sensing due to their anti-aliasing properties and tolerance toanalog imperfections. Yet, the oversampling ratio (OSR) directly inflates digitalprocessing overhead, increasing latency and power draw in resource-constrainedsystems.Hybrid approaches, like zoom ADCs, combine coarse and fine stages to mitigateindividual weaknesses. A low-resolution SAR stage provides a preliminaryestimate, subtracting it from the input to generate a residue for a high-resolutionΣ-Δ fine stage, effectively "zooming" into the error range. This pipelined structureaccelerates overall conversion while preserving precision, with applications inbridge transducers and precision readouts. Nonetheless, the dual-path designamplifies mismatch errors between stages, necessitating calibration, and the fixedcoarse-fine partitioning limits adaptability to varying signal dynamics.Despite these advancements, conventional high-resolution ADC architecturesgrapple with inherent trade-offs that undermine efficiency in emergingapplications. SAR and integrating designs, while power-efficient at moderatespeeds, demand N to 2N clock cycles per conversion, escalating dynamic powerthrough repeated switching of capacitors and comparators-often consuming tensto hundreds of microjoules per sample. Σ-Δ and zoom converters, though superiorin noise performance, impose even greater latency via oversampling and multi stage processing, rendering them unsuitable for real-time IoT edge nodes orbattery-limited wearables. Charge-balancing methods, prized for accuracy, sufferfrom amplitude-dependent conversion times that can span thousands of cycles forfull-scale inputs, exacerbating power budgets in continuous monitoring. Acrossthese paradigms, the fixed reset-to-zero paradigm at each cycle's start ignorestemporal correlations in signals, such as those in slowly varying sensor data,forcing redundant traversals of the full code range and inflating energy perconversion.The cumulative effect of these limitations manifests in elevated dynamic powerdissipation, prolonged latency, and suboptimal area utilization, particularly inPVT-sensitive on-chip environments where process variations, voltagefluctuations, and temperature shifts degrade performance. High-resolutiondemands amplify comparator offset, DAC linearity errors, and clock jittersensitivities, often requiring elaborate calibration that further burdens power andcomplexity. In power-autonomous systems like implantable medical devices orremote environmental sensors, where sub-microwatt operation is imperative, theseinefficiencies curtail battery life and deployment feasibility. Moreover, theescalating integration density in mixed-signal SoCs demands ADCs that not onlyresolve fine details but also adapt to signal sparsity without excessive overhead.The limitations of conventional state-of-the-art technologies, such as extendedconversion times scaling with resolution, high dynamic power from repetitive full range switching, susceptibility to PVT variations, and inflexibility to signaldynamics, highlight the need for a more advanced on-chip integrating-type dataconverter that can deliver high-resolution digital codes with dynamicallyadjustable conversion durations, minimized switching activity, and robustoperation across environmental extremes.Thus, there exists a need for an adaptive sampling mechanism in high-resolutionADCs that leverages prior conversions to initiate subsequent ones fromintermediate states, enabling conversion times as low as a single clock cycle whilepreserving precision and slashing power consumption, as provided by the presentinvention.SUMMARY OF INVENTIONIn view of the foregoing disadvantages inherent in the conventional techniques foranalog-to-digital conversion in the prior art, the present invention provides animproved on-chip integrating-type data converter utilizing adaptive sampling anddynamic conversion control to achieve high-resolution digital output withsignificantly reduced conversion times and power consumption.The present invention relates to an adaptive sampling-based on-chip integrating type data converter configured to generate high-resolution digital codes for analoginput signals with dynamically adjustable conversion times as low as a singleclock cycle. This converter employs a feedback-driven architecture that initiateseach successive conversion from an intermediate reference derived from the priorcycle's latched digital output, reconstructed via an integrated digital-to-analogconverter, rather than a full-range reset. The adaptive mechanism, informed by acomparator-directed selector and ramp integration path, enables bidirectionalincremental counting-upward or downward-proportional to the input signal'sdeviation from the previous sample, thereby minimizing switching cycles andachieving PVT-insensitive operation. This design optimizes the resolution-powertrade-off by exploiting signal correlations in low-variation environments, such assensor interfaces and biomedical monitoring, while ensuring precise digitalrepresentation without compromising throughput or linearity.According to a first aspect, the present invention provides an adaptive sampling based on-chip integrating-type data converter. The data converter comprising: asample and hold circuit configured to capture an input analog signal (Vin) andprovide a held sample (Vin*) in response to a start-of-conversion (SOC) signal; acomparator configured to compare the held sample (Vin*) with a previousreconstructed analog value (VPREV) and generate a comparator output (Comp_out)indicating a direction of deviation; a selector circuit configured to select a positivereference voltage (VREF+) or a negative reference voltage (VREF-) based on thecomparator output (Comp_out); a ramp generator configured to generate a rampsignal (VRAMP) with a slope polarity determined by the selected reference voltage,starting from a level corresponding to the previous reconstructed analog value(VPREV); a decision making circuit configured to monitor the ramp signal (VRAMP)relative to the held sample (Vin*) and propagate a clock signal (Clk_dmc) duringactive ramping until intersection; a synchronizer and edge detector configured togenerate an end-of-conversion (EOC) signal upon deassertion of the clock signal(Clk_dmc) and produce the start-of-conversion (SOC) signal after a delay; acounter and latch configured to count bidirectionally based on the comparatoroutput (Comp_out) and the clock signal (Clk_dmc), initialize to a mid-scale code,and latch a final count as a digital output code (DOUT) upon the end-of-conversion(EOC) signal; a digital-to-analog converter configured to convert the latcheddigital output code (DOUT) to an analog equivalent (VDAC) synchronized with asystem clock (CLK); and a control switch configured to gate the analog equivalent(VDAC) as the previous reconstructed analog value (VPREV) to the comparator uponthe start-of-conversion (SOC) signal.According to a second aspect, the present invention provides a method for analog to-digital conversion using adaptive sampling in an on-chip integrating-type dataconverter. The method comprises the steps of: capturing an input analog signal(Vin) using a sample and hold circuit to provide a held sample (Vin*) in responseto a start-of-conversion (SOC) signal; comparing the held sample (Vin*) with aprevious reconstructed analog value (VPREV) using a comparator to generate acomparator output (Comp_out) indicating a direction of deviation; selecting apositive reference voltage (VREF+) or a negative reference voltage (VREF-) using a selector circuit based on the comparator output (Comp_out); generating a rampsignal (VRAMP) with a slope polarity determined by the selected reference voltageusing a ramp generator, starting from a level corresponding to the previousreconstructed analog value (VPREV); monitoring the ramp signal (VRAMP) relative tothe held sample (Vin*) using a decision making circuit and propagating a clocksignal (Clk_dmc) during active ramping until intersection; generating an end-of conversion (EOC) signal upon deassertion of the clock signal (Clk_dmc) andproducing the start-of-conversion (SOC) signal after a delay using a synchronizerand edge detector; counting bidirectionally based on the comparator output(Comp_out) and the clock signal (Clk_dmc) using a counter and latch, initializingto a mid-scale code, and latching a final count as a digital output code (DOUT) uponthe end-of-conversion (EOC) signal; converting the latched digital output code(DOUT) to an analog equivalent (VDAC) using a digital-to-analog convertersynchronized with a system clock (CLK); and gating the analog equivalent (VDAC)as the previous reconstructed analog value (VPREV) to the comparator using acontrol switch upon the start-of-conversion (SOC) signal. According to a third aspect, the present invention provides a sensor interfacesystem. The system comprising: an adaptive sampling-based on-chip integrating type data converter of the first aspect; a sensor coupled to provide the input analogsignal (Vin) to the sample and hold circuit; and a processor configured to receivethe digital output code (DOUT) from the counter and latch for further processing.The various embodiments of the present invention provide an economical andeffective solution for deploying high-precision analog-to-digital conversion inresource-constrained environments, such as sensor interfaces, medicalinstrumentation, IoT nodes, and industrial automation systems, where low-power,real-time signal digitization is essential for enhanced reliability and extendedoperational longevity.These aspects or other aspects in the present disclosure may be clearer and moreintelligible in the detail description of the various embodiments. Within the scopeof this patent application, it is expressly envisaged that the various aspects,embodiments, examples and alternatives set out in the preceding paragraphs, inthe claims and / or in the following description and drawings, and in particular theindividual features thereof, may be taken independently or in any combination.Features described in connection with one embodiment are applicable to allembodiments, unless such features are incompatible.BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWINGSThe accompanying drawings are included to provide a further understanding ofthe present disclosure and constitute a part of this specification. The drawingsillustrate exemplary embodiments of the present disclosure and, together with thedescription, serve to elaborate the principles of the present disclosure.Figure 1 illustrates a flowchart depicting the overall workflow of the adaptivesampling-based integrating-type data converter, according to an embodiment ofthe present invention.Figure 2 illustrates a block diagram of the adaptive sampling-based integrating type data converter, according to an embodiment of the present invention.Figure 3 illustrates a circuit-level diagram of the adaptive sampling-basedintegrating-type data converter, according to an embodiment of the presentinvention.Figure 4 illustrates a block diagram of a selector circuit within the adaptivesampling-based integrating-type data converter, according to an embodiment ofthe present invention.Figure 5 illustrates the functionality of a decision-making circuit within theadaptive sampling-based integrating-type data converter, according to anembodiment of the present invention.Figure 6 illustrates the functionality of a synchronizer and edge detector withinthe adaptive sampling-based integrating-type data converter, according to anembodiment of the present invention.Figure 7 illustrates the operation of a counter and latch circuit within the adaptivesampling-based integrating-type data converter, according to an embodiment ofthe present invention.Figure 8 illustrates a timing diagram for the adaptive sampling-based integrating type data converter, exemplified for a 4-bit resolution, according to anembodiment of the present invention.Figure 9 illustrates a block diagram of the sensor interface system, according to anembodiment of the present invention.The diagrams are not to scale, but for the purpose of illustration only, hence arenot a limitation to the present disclosure.DETAILED DESCRIPTIONThe present invention will be described in further detail with reference to theaccompanying drawings and examples. It is to be understood that the specificembodiments described herein are merely illustrative of the invention and are notlimiting to the invention.The following paragraphs describe in detail, the various embodiments of presentinvention with reference to accompanying drawings. Terms used in embodimentsare only used for explaining specific embodiments of the present invention, andare not intended to limit this invention. It is clear that the described embodimentsare merely the most essential rather than all of embodiments of the presentinvention. All other embodiments obtained by a person skilled in the art, throughworkshop modifications on the embodiments described herein below withoutcreative efforts ought to fall within the protection scope of the present invention.Various objects, features, aspects and advantages of the present disclosure willbecome more apparent from the following detailed description of preferredembodiments, along with the accompanying drawing figures in which likenumerals represent like features. The present invention pertains to an adaptive sampling-based on-chip integrating type data converter designed for high-resolution analog-to-digital conversion withdynamically adjustable conversion times. This data converter processes inputanalog signals to produce corresponding digital codes by leveraging incrementaladjustments from prior conversion results, thereby optimizing efficiency in low speed, precision applications.Analog-to-digital converters (ADCs) serve as essential interfaces in electronicsystems, transforming continuous analog signals from real-world sources intodiscrete digital formats for further processing, analysis, and storage. The presentinvention focuses on an adaptive sampling-based on-chip integrating-type dataconverter tailored for high-resolution, low-speed applications. This converteraddresses the challenges of precision signal digitization by dynamically adjustingconversion times based on input signal variations, leveraging prior cycle outputsto minimize redundant processing. By incorporating ramp integration withbidirectional feedback, it achieves fine-grained digital representations suitable forenvironments demanding accuracy and efficiency, such as sensor networks andinstrumentation.The current state of the art in ADC design encompasses a diverse array ofarchitectures optimized for specific performance metrics. Flash ADCs providehigh-speed parallel comparisons but scale poorly with resolution due toexponential hardware demands. Successive approximation register (SAR) ADCsutilize binary search algorithms for balanced speed and power, employing sample and-hold circuits, comparators, and DACs to resolve bits sequentially over Ncycles. Integrating ADCs, including single-slope and dual-slope variants, rely ontime-based charge accumulation against a reference ramp, offering inherentlinearity and noise rejection ideal for precision measurements. Charge-balancingADCs extend this by equilibrating input charges through switched capacitors,enhancing stability in low-power scenarios.Further advancements include oversampling techniques in sigma-delta (Σ-Δ)ADCs, which shape quantization noise outside the signal band via high-ratemodulation and digital decimation, achieving effective resolutions beyond 20 bitsfor audio and biomedical signals. Hybrid zoom ADCs combine a coarse SARstage with a fine Σ-Δ residue converter, accelerating throughput while preservingdetail in transducer interfaces. Recent innovations, such as level-crossing andtime-interleaved designs, incorporate asynchronous elements or multi-coreparallelism to mitigate latency, with publications and patents demonstratingincremental improvements in dynamic range and figure-of-merit for embeddedsystems.Despite these developments, existing ADC solutions face significant drawbacksthat compromise their viability in power-sensitive, high-fidelity applications.Conversion times in SAR and integrating architectures scale directly withresolution, often requiring N to 2N clock cycles and incurring high dynamicpower from repetitive capacitor switching and full-range resets. Σ-Δ and zoomdesigns, while noise-resilient, introduce processing latency through oversamplingratios and multi-stage mismatches, alongside increased digital overhead thatburdens low-voltage operation. Charge-balancing methods, though accurate,exhibit amplitude-dependent durations that prolong full-scale conversions,exacerbating energy inefficiency in continuous monitoring. Collectively, theselimitations amplify susceptibility to process-voltage-temperature (PVT)variations, offset errors, and jitter, necessitating costly calibrations that inflate areaand complexity.The limitations of conventional state-of-the-art technologies, such as protractedconversion durations, excessive switching-induced power dissipation, PVTvulnerabilities, and rigidity to signal dynamics, underscore the need for anadvanced on-chip integrating-type data converter that delivers high-resolutionoutputs with adaptive, minimal-cycle processing from intermediate states. Thus,there exists a need for a mechanism that exploits temporal input correlations toenable single-cycle conversions for small variations, curtail overall energy use,and ensure robust, calibration-free performance across diverse operationalconditions, as addressed by the present invention.The present invention encompasses several embodiments to accommodate variousapplications and configurations:The detailed description provided herein is sufficient to enable a person skilled inthe art of analog and mixed-signal integrated circuit design to make and use theinvention, including simulating the architecture in SPICE or Cadence Virtuosotools, implementing the ramp generator and counter logic in Verilog-AMS forbehavioral modeling, fabricating prototypes in standard CMOS processes such as180 nm or 65 nm nodes, and evaluating performance via standard ADCbenchmarks like signal-to-noise-and-distortion ratio (SNDR), effective number ofbits (ENOB), and power efficiency figures-of-merit. Variations, such asincorporating alternative DAC topologies (e.g., current-steering instead ofcapacitive) or extending resolutions beyond 12 bits with multi-stage counters, maybe implemented without departing from the scope of the invention.In accordance with an embodiment of the present invention, there is provided anadaptive sampling-based integrating-type data converter configured to generateequivalent digital codes for input analog signals by initiating each conversioncycle from an intermediate state derived from the prior cycle, rather than resettingto a baseline reference. This architecture leverages a comparison between thecurrent sampled input and a reconstructed analog representation of the previousdigital output to determine the direction and extent of adjustment, enabling theconversion process to proceed incrementally in an upward or downward manner.By employing a ramp-based integration mechanism synchronized with decisionlogic, the converter dynamically adjusts the number of clock cycles required forcompletion, ranging from a minimum of one cycle to a maximum dependent onsignal variation, thereby minimizing unnecessary switching activity across theconversion path.In accordance with another embodiment of the present invention, the dataconverter incorporates feedback elements that ensure precise synchronizationbetween sampling, integration, and digital encoding stages, while maintaininginsensitivity to process-voltage-temperature variations. The adaptive nature of thesampling allows the system to exploit temporal correlations in input signals, suchas those prevalent in low-frequency sensor data, to curtail the overall energyexpenditure associated with repetitive full-range traversals. This results in arobust, PVT-independent operation that preserves high resolution without thelatency penalties of oversampling or multi-stage hybrids, addressing the coreinefficiencies of fixed-cycle architectures.At a high level, as illustrated in Fig. 2, the adaptive sampling-based on-chipintegrating-type data converter (100) comprises a sample and hold circuit (101), acomparator (102), a selector circuit (103), a ramp generator (104), a decisionmaking circuit (105), a synchronizer and edge detector (106), a counter and latch(107), a digital-to-analog converter (108), and a control switch (109). The overallarchitecture operates on a clock signal (CLK) to process an input analog signal(Vin), sampling it discretely and generating a digital output code (DOUT) thatrepresents the signal with high fidelity.The conversion begins with sampling (Vin) at the assertion of a start-of conversion (SOC) signal, compares it against a reconstructed prior value, andadjusts a ramp-integrated counter bidirectionally until alignment, yielding DOUT in1 to 2(N-1) clock cycles for an N-bit resolution. This flow, depicted in the flowchartof Fig. 1, ensures that each cycle builds on the previous, adapting to signaldirectionality for minimal processing.The sample and hold circuit (101) captures the continuous-time input analogsignal (Vin) and converts it to a discrete-time held value (Vin*) upon receivingthe SOC signal from the synchronizer and edge detector (106). This circuittypically employs a switched-capacitor configuration with a holding capacitor tomaintain Vin* stable during subsequent comparisons, preventing droop ordistortion. As shown in the circuit-level diagram of Fig. 3, the sample and holdcircuit (101) interfaces directly with the comparator (102) and is clockedsynchronously with the system CLK to align with conversion phases.The comparator (102) receives the held sample (Vin*) and compares it against theprevious conversion's reconstructed analog value (VPREV), which is gated throughthe control switch (109). The comparator output (Comp_out) indicates whetherVin* exceeds VPREV (positive direction) or falls below it (negative direction),serving as a directional control signal. This binary decision drives subsequentmodules, ensuring the conversion path aligns with the signal's deviation. In Fig. 3,the comparator (102) is depicted as a differential amplifier with offset cancellationto enhance precision across PVT conditions.Based on Comp_out, the selector circuit (103) routes either a positive referencevoltage (VREF+) or negative reference voltage (VREF-) to the ramp generator (104).As detailed in the block diagram of Fig. 4, the selector circuit (103) comprisesswitches controlled by Comp_out, selecting VREF- for upward ramps when Vin* >VPREV or VREF+ for downward ramps otherwise. This selection determines theramp slope polarity, enabling bidirectional integration without full-range resets.The ramp generator (104) produces a linear ramp signal (VRAMP) with a slopeproportional to the selected reference voltage, starting from an initial levelcorresponding to VPREV. The ramp continues until VRAMP intersects Vin*, at whichpoint the decision making circuit (105) halts it. Fig. 3 illustrates the rampgenerator (104) as an integrator whose slope is proportional to the integrationtime, independent of RC time constants for PVT robustness. The generatoroutputs clock pulses (Clk_dmc) during active ramping, gated by the intersectiondetection.The decision making circuit (105) monitors the relationship between VRAMP andVin*, generating Clk_dmc to propagate CLK only while VRAMP undershoots orovershoots Vin* based on direction. Upon intersection (VRAMP = Vin*), itdeasserts Clk_dmc, terminating the ramp and signaling conversion completion. Asshown in Fig. 5, this circuit includes a secondary comparator and AND / OR logicto AND Clk_dmc with the non-intersection condition, ensuring precise cyclecounting without jitter.The synchronizer and edge detector (106) processes Clk_dmc to generate end-of conversion (EOC) upon its deassertion and delays EOC to produce SOC for thenext cycle. This module synchronizes edges to CLK, preventing metastability, andenforces a delay (e.g., one CLK period) between EOC and SOC for settling. Fig. 6depicts the synchronizer as a flip-flop and the edge detector as a combination offlip-flop and logic gates, outputting EOC as soon as Clk_dmc becomes low andSOC after propagation delay.The counter and latch (107) initializes to a mid-scale code (e.g., 2(N-1) for N bits,equivalent to VDD / 2) and increments or decrements on each Clk_dmc pulse,directed by Comp_out (up for positive, down for negative). Upon EOC, the finalcount latches as DOUT and holds for the next cycle. Fig. 7 illustrates the counter(107_1) as an up / down ripple counter with N D-flip-flops (i.e (N-1) 107_3 andone 107_4) and (N-1) multiplexers (105_3) for mode selection, while the latch(107_2) uses N D-flip-flops (106_1) clocked on Clk_dmc.The digital-to-analog converter (108) converts the latched DOUT to an analogequivalent (VDAC) synchronized with CLK, initializing to VDD / 2 and updatingcontinuously with the latched data. This VDAC becomes VPREV for the subsequentcomparison based on SOC. The DAC may employ resistive, capacitive, current steering, or charge-based topologies, as shown in Fig. 3 interfacing with thecontrol switch (109).The control switch (109) gates VDAC to the comparator (102) only on SOCassertion, isolating prior values during ramping. This ensures clean handoffbetween cycles, as visible in the timing alignment of Fig. 3.Data flow commences with SOC triggering sample and hold (101) to capture Vinas Vin*, which comparator (102) assesses against VPREV (from DAC (108) viacontrol switch (109)). Comp_out routes through selector circuit (103) to set ramppolarity in generator (104), whose VRAMP feeds decision making circuit (105) forClk_dmc gating to counter and latch (107) and synchronizer (106). Synchronizerand edge detector (106) yields EOC on Clk_dmc cessation, latching DOUT andupdating DAC (108) for the next SOC-initiated cycle. This closed-loopinterconnection exploits prior DOUT for adaptive starting points, as flowcharted inFig. 1.In an optional embodiment, for resolutions beyond 12 bits, the counter (107_1)may incorporate asynchronous elements to reduce power, or the ramp generator(104) could use a multi-slope profile for nonlinear inputs. For example, in a 10-bitconfiguration with CLK at 1 MHz, a small signal variation (e.g., 1 LSB)completes in one cycle, while full-scale shifts approach 29 cycles, as exemplified in the 4-bit timing diagram of Fig. 8, where traces show SOC / EOC pulses,Comp_out toggling, ramp slopes, and DOUT evolution over cycles.In one embodiment, an adaptive sampling-based on-chip integrating-type dataconverter (100), as illustrated in Figure 2, is disclosed. The data converter (100)comprising: a sample and hold circuit (101) configured to capture an input analogsignal (Vin) and provide a held sample (Vin*) in response to a start-of-conversion(SOC) signal; a comparator (102) configured to compare the held sample (Vin*) with aprevious reconstructed analog value (VPREV) and generate a comparator output(Comp_out) indicating a direction of deviation;a selector circuit (103) configured to select a positive reference voltage(VREF+) or a negative reference voltage (VREF-) based on the comparator output(Comp_out);a ramp generator (104) configured to generate a ramp signal (VRAMP) witha slope polarity determined by the selected reference voltage, starting from a levelcorresponding to the previous reconstructed analog value (VPREV);a decision making circuit (105) configured to monitor the ramp signal(VRAMP) relative to the held sample (Vin*) and propagate a clock signal(Clk_dmc) during active ramping until intersection;a synchronizer and edge detector (106) configured to generate an end-of conversion (EOC) signal upon deassertion of the clock signal (Clk_dmc) andproduce the start-of-conversion (SOC) signal after a delay;a counter and latch (107) configured to count bidirectionally based on thecomparator output (Comp_out) and the clock signal (Clk_dmc), initialize to amid-scale code, and latch a final count as a digital output code (DOUT) upon theend-of-conversion (EOC) signal;a digital-to-analog converter (108) configured to convert the latched digitaloutput code (DOUT) to an analog equivalent (VDAC) synchronized with a systemclock (CLK); and a control switch (109) configured to gate the analog equivalent (VDAC) asthe previous reconstructed analog value (VPREV) to the comparator (102) upon thestart-of-conversion (SOC) signal.In one embodiment, the counter and latch (107) further comprises: a counter(107_1) with N flip-flops and (N-1) multiplexers for bidirectional operation in anN-bit configuration; and a latch (107_2) with N flip-flops to hold the digital outputcode (DOUT) for subsequent cycles.In one embodiment, the ramp generator (104) generates the ramp signal (VRAMP)independent of RC time constants for process-voltage-temperature (PVT)robustness.In one embodiment, the digital-to-analog converter (108) is selected fromresistive, capacitive, current-steering, or charge-based topologies and initializes toan equivalent of VDD / 2.In one embodiment, conversion cycles range dynamically from 1 to 2(N-1) based ondeviation between the held sample (Vin*) and the previous reconstructed analogvalue (VPREV) for an N-bit resolution.In one embodiment, a method (S300) for analog-to-digital conversion usingadaptive sampling in an on-chip integrating-type data converter (100), as illustratedin Figure 1, is disclosed. The method comprising the steps of:capturing (S301) an input analog signal (Vin) using a sample and holdcircuit (101) to provide a held sample (Vin*) in response to a start-of-conversion(SOC) signal;comparing (S302) the held sample (Vin*) with a previous reconstructedanalog value (VPREV) using a comparator (102) to generate a comparator output(Comp_out) indicating a direction of deviation;selecting (S303) a positive reference voltage (VREF+) or a negativereference voltage (VREF-) using a selector circuit (103) based on the comparatoroutput (Comp_out); generating (S304) a ramp signal (VRAMP) with a slope polarity determinedby the selected reference voltage using a ramp generator (104), starting from alevel corresponding to the previous reconstructed analog value (VPREV);monitoring (S305) the ramp signal (VRAMP) relative to the held sample(Vin*) using a decision making circuit (105) and propagating a clock signal(Clk_dmc) during active ramping until intersection;generating (S306) an end-of-conversion (EOC) signal upon deassertion ofthe clock signal (Clk_dmc) and producing the start-of-conversion (SOC) signalafter a delay using a synchronizer and edge detector (106);counting bidirectionally (S307) based on the comparator output(Comp_out) and the clock signal (Clk_dmc) using a counter and latch (107),initializing to a mid-scale code, and latching a final count as a digital output code(DOUT) upon the end-of-conversion (EOC) signal;converting (S308) the latched digital output code (DOUT) to an analogequivalent (VDAC) using a digital-to-analog converter (108) synchronized with asystem clock (CLK); and gating (S309) the analog equivalent (VDAC) as the previous reconstructedanalog value (VPREV) to the comparator (102) using a control switch (109) uponthe start-of-conversion (SOC) signal.In one embodiment, the counting step uses a counter (107_1) configured for up ordown operation with N flip-flops and (N-1) multiplexers in an N-bitconfiguration, and latching with N flip-flop to hold the digital output code (DOUT)for feedback in subsequent conversions.In one embodiment, generating the ramp signal (VRAMP) is independent of RCtime constants to ensure process-voltage-temperature (PVT) insensitivity.In one embodiment, the conversion completes in a dynamic number of cycles from1 to 2(N-1) proportional to the deviation between the held sample (Vin*) and theprevious reconstructed analog value (VPREV) for an N-bit resolution.In one embodiment, the method further comprises: initializing the digital-toanalog converter (108) to an equivalent of VDD / 2 and selecting its topology fromresistive, capacitive, current-steering, or charge-based configurations.In one embodiment, a sensor interface system (200), as illustrated in Figure 9, isdisclosed. The sensor interface system (200) comprises:an adaptive sampling-based on-chip integrating-type data converter (100);a sensor (201) coupled to provide the input analog signal (Vin) to thesample and hold circuit (101); and a processor (202) configured to receive the digital output code (DOUT)from the counter and latch (107) for further processing.In one embodiment, the adaptive sampling-based on-chip integrating-type dataconverter (100) dynamically adjusts conversion cycles from 1 to 2(N-1) to minimizepower consumption based on signal variations from the sensor.In one embodiment, the counter and latch (107) initializes to a mid-scale codeequivalent to VDD / 2 for bidirectional counting.In one embodiment, the digital-to-analog converter (108) provides feedback byconverting the latched digital output code (DOUT) to the previous reconstructedanalog value (VPREV) for adaptive starting points in subsequent conversions.The technical effect of the data converter of the present invention lies in its abilityto reduce average conversion time and dynamic power by up to 50% compared tofixed-cycle integrating ADCs, achieved through the distinguishing features ofbidirectional ramp integration from prior-cycle intermediates and comparator directed selective referencing. Unlike prior art that mandates full-range resets, thisarchitecture exploits input signal correlations via latched feedback, enablingsingle-cycle completions for minimal variations while scaling gracefully to largerdeviations, thereby enhancing energy efficiency in PVT-variable environmentswithout additional calibration circuitry.The present invention provides a significant technical advancement over the priorart by introducing an adaptive sampling architecture for on-chip integrating-typedata converters with dynamic ramp-based feedback, which eliminates the relianceon fixed-cycle resets, full-range traversals, and multi-stage calibrations through ahardware-integrated bidirectional integration path that reconfigures conversionduration from intermediate prior-cycle states based on real-time signal deviationanalysis. This distinguishing comparator-directed selector and latched DACmechanism enhances variation-aware adaptability and reduces unnecessaryswitching cycles and ramp excursions, leading to markedly improved energyefficiency-evidenced by up to 50% dynamic power savings in correlated low frequency workloads-and superior conversion performance, includingminimized latency, enhanced PVT robustness, and preserved linearity, comparedto conventional SAR, Σ-Δ, or zoom ADCs that suffer from protracted N-to-2Ncycle overheads, elevated dissipation, and environmental sensitivities. Byachieving these improvements without auxiliary filters or asynchronousinterleaving, the invention overcomes the throughput and integration barriers ofexisting solutions, enabling sustained high-fidelity digitization under resource-limited conditions and extended operational cycles in precision-constrainedenvironments.Some non-limiting advantages of the present invention are as follows:- Dynamic conversion cycles (1 to 2(N-1)) minimize switching activity, loweringdynamic power dissipation relative to SAR or Σ-Δ designs that requireconsistent N or oversampled cycles.- Bidirectional counting from mid-scale initialization (VDD / 2 equivalent)ensures PVT-independent accuracy, independent of ramp RC constants,improving noise immunity over charge-balancing ADCs.- Integrated feedback via DAC reconstruction of prior outputs adapts to signalsparsity in sensor applications, reducing latency versus zoom hybrids withoutmulti-stage mismatches.- On-chip implementation with minimal components (e.g., 2N flip-flops and N1 muxes) optimizes area for IoT embedding, contrasting the digital filteroverhead in Σ-Δ converters.- Autonomous EOC / SOC generation via edge detection supports continuousoperation, enhancing throughput in real-time systems over manually clockedintegrating types.The industrial applicability of the present data converter extends to sectorsdemanding precise, low-power analog signal digitization, such as healthcare forimplantable sensors monitoring vital signs with minimal battery drain, industrialautomation for metrology instruments ensuring sub-LSB accuracy in harshenvironments, and IoT devices for environmental monitoring where adaptiveefficiency extends deployment longevity. Implementation involves integrating theconverter as an analog front-end in CMOS processes (e.g., 180 nm or below),interfacing with microcontrollers via standard digital buses, and scalingresolutions from 10-20 bits to suit applications like audio processing or defensetelemetry, thereby facilitating cost-effective upgrades in existing mixed-signalsystems.In summary, the present invention provides an adaptive sampling-based on-chipintegrating-type data converter with dynamic ramp-based feedback comprising asample and hold circuit, comparator, selector circuit, ramp generator, decisionmaking circuit, synchronizer and edge detector, counter and latch, digital-to analog converter, and control switch. This enables precision analog-to-digitalconversion in low-power embedded systems for sensor signal processing,biomedical monitoring, and industrial instrumentation, offering a compact, PVT robust, and efficient solution for high-resolution data acquisition in resource limited environments. By addressing the drawbacks of prior art technologies, suchas extended fixed-cycle conversions, high switching-induced power dissipation,PVT sensitivities, and limited adaptability to signal variations, the inventionprovides enhanced dynamic efficiency, minimized latency, superior energysavings, and maintained linearity, enhancing system reliability in applicationsranging from IoT edge nodes to medical diagnostic devices.It is to be noted that the foregoing description is only illustrative of the preferredembodiments of the present invention and the technical principles employed. Itwill be understood by those skilled in the art that the present invention is notlimited to the particular embodiments described herein, but is capable of variousobvious changes, rearrangements and substitutions as will now become apparentto those skilled in the art without departing from the scope of the invention.Therefore, although the present invention has been described in greater detail bythe above embodiments, the present invention is not limited to the aboveembodiments, and may include other equivalent embodiments without departingfrom the spirit of the present invention, and the scope of the present invention isdetermined by the scope of the appended claims.
Claims
1. An adaptive sampling-based on-chip integrating-type data converter (100) comprising: a sample and hold circuit (101) configured to capture an input analog signal (Vin) and provide a held sample (Vin*) in response to a start-of-conversion (SOC) signal; a comparator (102) configured to compare the held sample (Vin*) with a previous reconstructed analog value (VPREV) and generate a comparator output (Comp_out) indicating a direction of deviation; a selector circuit (103) configured to select a positive reference voltage (VREF+) or a negative reference voltage (VREF- ) based on the comparator output (Comp_out); a ramp generator (104) configured to generate a ramp signal (VRAMP) with a slope polarity determined by the selected reference voltage, starting from a level corresponding to the previous reconstructed analog value (VPREV); a decision making circuit (105) configured to monitor the ramp signal (VRAMP) relative to the held sample (Vin*) and propagate a clock signal (Clk_dmc) during active ramping until intersection; a synchronizer and edge detector (106) configured to generate an end-of conversion (EOC) signal upon deassertion of the clock signal (Clk_dmc) and produce the start-of-conversion (SOC) signal after a delay; a counter and latch (107) configured to count bidirectionally based on the comparator output (Comp_out) and the clock signal (Clk_dmc), initialize to a mid-scale code, and latch a final count as a digital output code (DOUT) upon the end-of-conversion (EOC) signal; a digital-to-analog converter (108) configured to convert the latched digital output code (DOUT) to an analog equivalent (VDAC) synchronized with a system clock (CLK); and a control switch (109) configured to gate the analog equivalent (VDAC) as the previous reconstructed analog value (VPREV) to the comparator (102) upon the start-of-conversion (SOC) signal.
2. The data converter (100) as claimed in claim 1, wherein the counter and latch (107) comprises: a counter (107_1) with N flip-flops and (N-1) multiplexers for bidirectional operation in an N-bit configuration, and a latch (107_2) with N flip-flops to hold the digital output code (DOUT) for subsequent cycles.
3. The data converter (100) as claimed in claim 1, wherein the ramp generator (104) generates the ramp signal (VRAMP) independent of RC time constants for process-voltage-temperature (PVT) robustness.
4. The data converter (100) as claimed in claim 1, wherein the digital-to analog converter (108) is selected from resistive, capacitive, current-steering, or charge-based topologies and initializes to an equivalent of VDD / 2.
5. The data converter (100) as claimed in claim 1, wherein conversion cycles range dynamically from 1 to 2(N-1) based on deviation between the held sample (Vin*) and the previous reconstructed analog value (VPREV) for an N-bit resolution.
6. A method (S300) for analog-to-digital conversion using adaptive sampling in an on-chip integrating-type data converter (100), the method comprising: capturing (S301) an input analog signal (Vin) using a sample and hold circuit (101) to provide a held sample (Vin*) in response to a start-of-conversion (SOC) signal; comparing (S302) the held sample (Vin*) with a previous reconstructed analog value (VPREV) using a comparator (102) to generate a comparator output (Comp_out) indicating a direction of deviation; selecting (S303) a positive reference voltage (VREF+ ) or a negative reference voltage (VREF- ) using a selector circuit (103) based on the comparator output (Comp_out); generating (S304) a ramp signal (VRAMP) with a slope polarity determined by the selected reference voltage using a ramp generator (104), starting from a level corresponding to the previous reconstructed analog value (VPREV); monitoring (S305) the ramp signal (VRAMP) relative to the held sample (Vin*) using a decision-making circuit (105) and propagating a clock signal (Clk_dmc) during active ramping until intersection; generating (S306) an end-of-conversion (EOC) signal upon deassertion of the clock signal (Clk_dmc) and producing the start-of-conversion (SOC) signal after a delay using a synchronizer and edge detector (106); counting bidirectionally (S307) based on the comparator output (Comp_out) and the clock signal (Clk_dmc) using a counter and latch (107), initializing to a mid-scale code, and latching a final count as a digital output code (DOUT) upon the end-of-conversion (EOC) signal; converting (S308) the latched digital output code (DOUT) to an analog equivalent (VDAC) using a digital-to-analog converter (108) synchronized with a system clock (CLK); and gating (S309) the analog equivalent (VDAC) as the previous reconstructed analog value (VPREV) to the comparator (102) using a control switch (109) upon the start-of-conversion (SOC) signal.
7. The method as claimed in claim 6, wherein the counting step uses a counter (107_1) configured for up or down operation with N flip-flops and (N-1) multiplexers in an N-bit configuration, and latching with N flip-flips to holds the digital output code (DOUT) for feedback in subsequent conversions.
8. The method as claimed in claim 6, wherein generating the ramp signal (VRAMP) is independent of RC time constants to ensure process-voltage temperature (PVT) insensitivity.
9. The method as claimed in claim 6, wherein the conversion completes in a dynamic number of cycles from 1 to 2(N-1) proportional to the deviation between the held sample (Vin*) and the previous reconstructed analog value (VPREV) for an N-bit resolution.
10. The method as claimed in claim 6, further comprising initializing the digital-to-analog converter (108) to an equivalent of VDD / 2 and selecting its topology from resistive, capacitive, current-steering, or charge-based configurations.
11. A sensor interface system comprising: an adaptive sampling-based on-chip integrating-type data converter (100) as claimed in claim 1; a sensor coupled to provide the input analog signal (Vin) to the sample and hold circuit (101); and a processor configured to receive the digital output code (DOUT) from the counter and latch (107) for further processing.
12. The sensor interface system as claimed in claim 11, wherein the adaptive sampling-based on-chip integrating-type data converter (100) dynamically adjusts conversion cycles from 1 to 2(N-1) to minimize power consumption based on signal variations from the sensor.
13. The sensor interface system as claimed in claim 11, wherein the counter and latch (107) initializes to a mid-scale code equivalent to VDD / 2 for bidirectional counting.
14. The sensor interface system as claimed in claim 11, wherein the digital-to analog converter (108) provides feedback by converting the latched digital output code (DOUT) to the previous reconstructed analog value (VPREV) for adaptive starting points in subsequent conversions.
