CN117770826A - Wearable electrocardiosignal reading circuit - Google Patents

Abstract

The invention belongs to the technical field of medical equipment, and particularly relates to a wearable electrocardiosignal reading circuit. The electrocardiosignal reading circuit comprises an analog front end and a motion artifact elimination loop; the analog front end comprises a multichannel output chopper current balance instrument amplifier, a two-electrode common mode bias loop, an ETI excitation current source, a low-pass filter, a program-controlled gain amplifier, a bias circuit and a clock generation circuit; the motion artifact cancellation loop comprises an analog-to-digital converter, an analog-to-digital converter and an off-chip adaptive filter; the multi-channel output CBIA is used as a first stage of the analog front end, so that the requirements of the system on power consumption, input impedance and common mode rejection ratio are met; meanwhile, the multichannel output CBIA realizes multichannel output by multiplexing the transimpedance stages of the CBIA, so that synchronous acquisition of the ECG signal and the skin-electrode impedance signal is realized at lower power consumption cost. The electrocardiosignal reading circuit can effectively eliminate the interference of large fluctuation in signals.

Description

Wearable electrocardiosignal reading circuit

Technical Field

The invention belongs to the technical field of medical equipment, and particularly relates to a wearable electrocardiosignal reading circuit.

Background

According to the world health organization data, about 1790 ten thousand people are deprived of life by cardiovascular disease each year, and thus heart-related physiological indicators are of particular concern. Electrocardiographic (ECG) signal acquisition records the electrical activity of the heart in a non-invasive manner, an important means of assessing and diagnosing cardiovascular disease. Because of the uncertainty and intermittence of symptoms associated with cardiovascular disease, it is necessary to measure ECG signals continuously around the clock in order to avoid the occurrence of missed symptoms. Long-term ECG signal acquisition can be used for arrhythmia recognition, sleep apnea detection, cardiac arrest prediction, emotion recognition, and biometric identification, in addition to helping control and prevent cardiovascular disease.

The wearable medical device can monitor various physiological parameters of the human body for a long period of time, such as acquiring ECG signals to obtain physiological indicators related to the heart. Application to outpatient care or home healthcare (e.g., in some clinical and post-operative rehabilitation settings) can help doctors to remotely manage the patient's disease and health, thereby reducing patient hospitalization time, improving medical experience, and reducing medical costs, and therefore wearable devices are expected to be accepted by an increasing number of consumers.

However, in order to detect the health of an individual in a continuous and comfortable manner, the volume and weight of the wearable medical device should be as small as possible, but the miniaturization of the device also generally means smaller battery capacity, so a low power design is required to enhance the cruising ability of the device. In addition, the medical application has very high requirements on the quality of the acquired physiological signals, and further the acquisition circuit is required to have higher measurement precision. Meanwhile, various types of interference and noise in the environment can influence the measurement of signals, so that the anti-interference performance and the robustness of the circuit are also necessary to be improved. Thus, in order to achieve high quality dynamic ECG signal monitoring, the acquisition system needs to overcome numerous circuit design challenges.

When acquiring such signals, the acquisition system requires electrodes as a medium to convert physiological signals in the human body into electrical signals that can be processed by the electrical circuit. Because of the non-ideal nature of the electrodes, the signals acquired through the electrodes contain direct current electrode offset (DEO) voltages, which can be up to several hundred millivolts in magnitude. To avoid the gain versus noise tradeoff caused by DEO, the analog front end needs to have ac coupling characteristics. Considering that the lower limit frequency of a part of the biopotential signal may be up to 0.5Hz, the cut-off frequency for ac coupling is lower than this frequency. The filter with large time constant is realized by integrating passive resistor and capacitor in the chip, which causes excessive chip area cost, so that the traditional biopotential signal acquisition circuit needs to use off-chip capacitor. In 2003 Harrison et al proposed the use of a modified Capacitively Coupled Instrumentation Amplifier (CCIA) to acquire EEG and AP signals. The design utilizes the ultra-high impedance of Metal Oxide Semiconductor (MOS) field effect transistors operating in the off-region, such that the high-pass cutoff frequency inherent in CCIA reaches 0.025Hz without the use of off-chip components. Although this design achieves rail-to-rail DEO suppression with 16 μA power consumption, it uses two 3200 μm in order to reduce the effect of flicker noise ² And thus occupies a larger chip area as the transistor of the amplifier is input.

Compared with the simple increase of the transistor area, the chopper technology can restrain the influence of flicker noise in a smaller area in the Complementary Metal Oxide Semiconductor (CMOS) process, so that the chopper technology is widely applied to occasions sensitive to low-frequency noise. However, the chopper technology can eliminate the alternating current coupling characteristic of the original circuit, so that Yazicioglu and Denison sequentially introduce direct current servo loops (DSL) of current domains into chopper amplifiers of different types, thereby achieving the purpose of filtering DEO. However, the design of Yazicioglu requires a compromise between the DEO suppression range and the circuit power consumption, whereas the increase of the DEO suppression range in the design of Denison leads to an increase in circuit area and a deterioration in noise performance. In 2014, heleputte et al proposed a new voltage domain DSL that eliminates the trade-off between the DEO suppression range and circuit power consumption, and the circuit can handle 400mV of DEO. However, in this design, the DSL is fed back directly to the input of the amplifier, so that it is desirable to reduce the noise level of the DSL itself as much as possible during design to avoid deteriorating the overall circuit noise performance. In 2020, xu et al propose a direct ac coupling technique that optimizes the input stage load of a chopper-Current Balanced Instrumentation Amplifier (CBIA) so that 500mV DEO rejection can be achieved without affecting circuit noise and power consumption.

In addition to DEO, the non-ideal characteristics of an electrode also include electrode-tissue impedance (ETI). Although wet electrodes can provide smaller ETI to alleviate signal attenuation and distortion problems caused by source impedance and analog front end input impedance voltage division, dry electrodes are typically used in all-weather dynamic ECG signal acquisition as acquisition media because new electrodes need to be frequently replaced when wet electrodes are applied to avoid signal quality degradation caused by drying of conductive gel. A typical size of 1mΩ resistor for the dry electrode ETI is connected in parallel with a 10nF capacitor, and the input impedance of the analog front end needs to be much higher than this value to reduce the signal acquisition error. The input end of a typical CMOS amplifier is the gate of a MOS transistor, so its input impedance is generally greater than 1gΩ at low frequency, but the use of chopping techniques can reduce the impedance to several hundred to several thousandths of the original. In response to this problem Xu et al propose to use a Positive Feedback Loop (PFL) technique to boost the input impedance of a chopper amplifier, but since PFL is very sensitive to process, voltage, temperature (PVT) and parasitic variations, the impedance boosting effect and robustness of this technique is not ideal. In 2016, chandrakumar et al used an auxiliary path chopping technique to alleviate the problem of chopper-induced input impedance drop, and the feedback loop of this technique operated independently of the main amplifier, thereby eliminating the trade-off between impedance rise times and stability in PFL. Although the auxiliary path chopping may achieve a better impedance boosting effect than PFL, it may deteriorate the noise performance of the circuit. In 2017, butteti et al proposed to perform a dynamic element matching operation while chopping switching, so as to reduce the charge-discharge current required for parasitic capacitance during chopping switch switching from the source, thereby increasing the input impedance of the chopper amplifier, but the technology is only applicable to chopper Current Feedback Instrumentation Amplifier (CFIA). 2022, qu et al proposed to use local PFL to reduce the effect of chopping on the input impedance, thereby avoiding the problem that conventional PFL can only be applied to Instrumentation Amplifier (IA) but not to operational amplifier, but its impedance boosting effect is still limited by PVT and parasitic variations.

In order to better detect the health status of the human body, the user needs to wear the wearable medical device as continuously as possible in all weather, so reducing the number of electrodes to achieve miniaturization of the device is an attractive design solution. The traditional ECG signal acquisition requires at least three electrodes to operate simultaneously, wherein one electrode does not directly participate in the acquisition of the signal, and the main function of the traditional ECG signal acquisition is to weaken the influence of power line interference in the environment. If the problem of power line interference is eliminated by the circuit technology, the two electrodes can also realize the function of ECG signal acquisition. In 2019 Koo et al proposed a power line interference cancellation technique that could be used for two electrode ECG signal acquisition. The technology realizes power line interference suppression through a common-mode charge pump, and can process common-mode voltage fluctuation of 30Vpp at maximum under the condition of 220pF coupling capacitance. But the input impedance of the analog front end, the Common Mode Rejection Ratio (CMRR), and the noise performance in this scheme are significantly degraded by the introduction of the charge pump. In comparison to the above-described technique, xu et al propose that better CMRR and noise index can be obtained using a resistor-based common mode feedback loop for two-electrode ECG signal acquisition, however this technique can only handle 500mVpp of common mode disturbance. In 2020, shu et al propose to use a dummy right leg driver circuit (RLD) for common mode biasing a two electrode ECG signal acquisition circuit, which technique achieves 130Vpp common mode rejection while guaranteeing a higher input impedance analog front end. But the driver amplifier in its pseudo RLD uses an adaptive bias structure that is less power efficient. 2021, koo et al optimized a conventional common-mode charge pump based common-mode interference cancellation loop with a minimum root mean square algorithm embedded in the original circuit to improve the CMRR of the circuit. However, the implementation of this scheme requires a complete set of digital-analog hybrid circuits, thereby significantly increasing the complexity of the system.

In the industry, many analog chip tap enterprises have also introduced analog front-end circuits that can be used for bio-signal acquisition. The ADAS1000 product of Adenox (ADI) semiconductor company can measure ECG signal and detect human breath and pulse, and has integrated pacing signal detecting algorithm, so that it may be used in dynamic electrocardiograph monitor and in emergency fields, such as automatic external defibrillation. The AFE 4900 product proposed by Texas Instruments (TI) semiconductor company is an analog front end for synchronizing the collection of ECG signals and photo-blood volume signals, and is suitable for the application fields of heart rate detection and capillary oxygen saturation detection. The mode gear within AFE 4900 is selectable by the designer with 1Hz to 150Hz input reference integrated noise of 1.25 μvrms when switching to 4kHz independent ECG signal samples. The MAX30004 product of the meixin company provides a complete set of heart rate detection analog front-end solutions optimized for wearable devices. The analog front end has an input impedance of 500mΩ, a CMRR of 100dB, an alternating current dynamic input range of 65mVpp in a three electrode ECG signal acquisition configuration. The characteristics of high input impedance and high CMRR reduce errors caused by large ETI of a dry electrode in long-term dynamic ECG signal acquisition, and the large alternating current dynamic input range avoids the problem of analog front end saturation caused by motion artifact. But when the product is configured in a two-electrode ECG signal acquisition mode, the input impedance of its overall circuit and CMRR index will drop significantly.

The circuitry that constitutes the measurement system itself can affect the acquisition of the ECG signal due to its inherent non-ideal characteristics. On one hand, devices such as a resistor and an MOS tube can generate thermal noise spontaneously so as to introduce measurement errors, and on the other hand, a frequency range from 0.5Hz to 150Hz is focused when an ECG signal is measured, so that the flicker noise of the MOS tube can obviously interfere with the measurement of the ECG signal.

In order to acquire ECG signals of the human body, the measurement system needs to convert physiological signals inside the human body into electrical signals that can be processed by an electrical circuit through electrodes. There are many non-ideal characteristics of the interface formed by the electrode contacting the human body, such as dc electrode imbalance, electrode-tissue impedance, etc. The circuit is therefore required to have a wide dc input range and a high input impedance.

The traditional ECG signal acquisition requires three electrodes to work simultaneously, but one electrode does not directly participate in the signal acquisition, and the main function of the traditional ECG signal acquisition is to weaken power frequency interference in the environment. From the viewpoint of improving portability, two-electrode ECG signal acquisition is more attractive than three-electrode ECG signal acquisition, however, the absence of the bias electrode makes the ECG signal acquisition very susceptible to power frequency interference. The circuit therefore needs to be able to handle the large amplitude of power frequency interference in two electrode ECG acquisitions.

In a clinical ECG signal monitoring scene of a hospital, a human body is in a resting state, and the measuring environment is stable and controllable. When the wearable device is used for monitoring, real-time dynamic measurement is faced, and the behaviors and the surrounding environment of a human body cannot be predicted, so that additional design challenges, especially motion artifacts generated during the motion of the human body, are caused.

Fig. 1 shows the amplitude and frequency distribution ranges of various signals in ECG signal acquisition, and how to eliminate the influence of the red interference signal is a technical problem to be solved by the present invention.

Disclosure of Invention

The invention aims to provide a high-performance wearable electrocardiosignal reading circuit capable of eliminating power frequency interference.

The high-performance wearable electrocardiosignal reading circuit provided by the invention supports two-electrode electrocardiosignal acquisition and eliminates motion artifacts, and the performance index is very suitable for wearable electrocardiosignal acquisition equipment. The system architecture diagram is shown in fig. 2, and can be functionally divided into two parts: analog front end and motion artifact cancellation loop; the analog front end comprises a multichannel output chopper Current Balance Instrument Amplifier (CBIA) (for short, multichannel output chopper CBIA), a two-electrode common mode bias loop, an ETI excitation current source, a Low Pass Filter (LPF), a program controlled gain amplifier (PGA), a bias generation circuit and a clock generation circuit; the motion artifact cancellation loop comprises an analog-to-digital converter (ADC), a digital-to-analog converter (DAC), and an off-chip adaptive filter; wherein:

the multichannel output chopper CBIA is used for amplifying Electrocardiosignals (ECG) of a human body and skin-electrode impedance signals (ETI) of acquisition electrodes, the input end of the multichannel output chopper CBIA is connected with an ETI excitation current source, a two-electrode common mode bias loop and the acquisition electrodes, and the output end of the multichannel output chopper CBIA is connected with the input end of the LPF; the multi-channel output chopper CBIA circuit is composed of a Transconductance (TC) stage and three Transimpedance (TI) stages; three channels: ECG (processing of the human electrocardiographic signal), ETII (processing of the real part of the ETI signal) and ETIQ (processing of the imaginary part of the ETI signal) correspond to the outputs of three TI stages;

specifically, as shown in fig. 3, the input of the TC stage serves as the input end of the whole multi-channel output chopper CBIA, and the output is simultaneously connected with the inputs of three TI stages, so that the purposes of saving power consumption and area are achieved. The difference between the different TI stages is the clock frequency and phase of the chopper CH_UM, and the ECG channel, the ETII channel and the ETIQ channel respectively correspond to the output of one TI stage; the TC stage is formed by connecting 8 transistors M1-M8, 2 choppers CH_ M, CH _BS, a capacitor CIN and a resistor RIN through a circuit, and the TI stage is formed by connecting 4M 9-M12, choppers CH_UM and a resistor ROUT through a circuit; transistors M1, M2 in the TC-stage buffer the signal (VIP-VIN) at the input of the multi-channel output chopper CBIA to both ends of resistor RIN and capacitor CIN; wherein, the transistors M1 and M2 are composed of composite transistors MA, MB and MC, and are used for increasing the input impedance and CMRR of the circuit; the capacitor CIN and the resistor RIN can filter out the direct current component in the input signal, and the alternating current signal in the input signal can be mirrored to the transistors M9 and M10 in the TI stage through the current mirrors M3 and M4 and flows through the resistor ROUT to form an amplified input signal (VOP-VON); chopper ch_ M, CH _bs in TC stage and chopper ch_um in TI stage are used to eliminate the influence of circuit disorder and flicker noise; transistors M5, M6, M7, M8 in the TC stage are used to extend the common mode input range of the circuit; transistors M11, M12 in the TI stage are used to bias the output stage.

The two-electrode common mode bias loop is used for eliminating power frequency interference introduced by electrical equipment in the process of collecting human electrocardiosignals by the electrodes, and the input end and the output end of the two-electrode common mode bias loop are connected with the ETI excitation current source, the multichannel output chopper CBIA and the collecting electrodes.

And the ETI excitation current source is used for generating ETI signals of the acquisition electrodes, the input end of the ETI excitation current source is connected with the internal bias generation circuit, and the output end of the ETI excitation current source is connected with the multichannel output chopper CBIA, the two-electrode common mode bias loop and the acquisition electrodes.

The Low Pass Filter (LPF) is used for filtering high-frequency ETI signals in the ECG channels and high-frequency ECG signals in the ETII and ETIQ channels in the multi-channel output chopper CBIA, the input end of the low pass filter is connected with the multi-channel output chopper CBIA, and the output end of the low pass filter is connected with the PGA.

Specifically, as shown in FIG. 2, the ECG signal energy at A (input of the multi-channel output chopper CBIA) is distributed at low frequency, and the ETI signal energy is distributed at fCH/2, through electrode acquisition and ETI excitation current source excitation. The ECG signal energy is shifted to fCH/2 and the ETI signal energy is shifted to fCH at B (the output of TC stage in the multi-channel output chopper CBIA) by the chopper process of TI stage chopper frequency fCH. The TC stage adopts a chopper demodulation with the chopping frequency of fCH, and can see the ECG signal with the energy distributed at the low frequency and the ETI signal with the energy distributed at fCH/2 at the C (the output end of the ECG channel TI stage in the multi-channel output chopping CBIA), and adopts a chopper demodulation with the chopping frequency of fCH/2, and can see the ETI signal with the energy distributed at the low frequency and the ECG signal with the energy distributed at fCH/2 at the D, E (the output ends of the ETII channel and the ETIQ channel TI stage in the multi-channel output chopping CBIA), so that the effect of separating the ECG signals in different TC stages from the ECG signals in the frequency where the ETI signals are located is realized, and the ETI signals in the ECG channels can be filtered based on the low-pass filter (LPF).

The Programmable Gain Amplifier (PGA) for further amplifying the acquired ECG and ETI signals; the PGA input in the ECG channel is connected to the Low Pass Filter (LPF) and DAC, the PGA input in the ETII and ETIQ channels is connected to LPF, and the PGA outputs in the ECG, ETII and ETIQ channels are all connected to ADC.

The bias circuit is used for providing bias signals required by the circuit, the input end of the bias circuit is connected with the off-chip adjustable resistor, and the output end of the bias circuit is connected with the circuit which is required to be biased in the chip.

The clock generating circuit is used for providing clock signals required by the circuit, the input end of the clock generating circuit is connected with an off-chip clock source, and the output end of the clock generating circuit is connected with the circuit requiring clock in the chip.

The ADC is used for quantizing the acquired ECG signal and the acquired ETI signal into digital signals, so as to provide input signals for off-chip adaptive filtering, the input end of the ADC is connected with the LPF, and the output end of the ADC is connected with the off-chip adaptive filtering module.

The DAC is used for converting the error feedback digital signal generated by the self-adaptive filtering module into an analog signal, further eliminating motion artifacts in the analog domain, wherein the input end of the DAC is connected with the self-adaptive filter, and the output end of the DAC is connected with the PGA. The off-chip adaptive filter is used for generating an error feedback digital signal related to motion artifact, the input end of the off-chip adaptive filter is connected with the ADC, and the output end of the off-chip adaptive filter is connected with the DAC.

The working flow (namely, the signal flow direction) of the wearable electrocardiosignal reading circuit is as follows:

the wearable electrocardiosignal reading circuit is connected with the electrode and is connected with the human body, and then the electrocardiosignal of the human body can be seen at the input end of the reading circuit; in addition, common mode interference signals introduced by external electromagnetic interference, direct current electrode offset voltage signals introduced by electrodes and ETI signals generated by ETI excitation current sources are also superimposed on the electrocardiosignal. The common mode interference signal is absorbed by the two-electrode common mode bias loop, so that the signal fed into the input end of the multi-channel output chopper CBIA is superposition of an electrocardiosignal, a direct-current electrode offset voltage signal introduced by an electrode and an ETI signal, the electrocardiosignal energy is concentrated at a low frequency which is 0.5Hz higher than direct current, the direct-current electrode offset voltage signal energy is concentrated at a direct current, and the ETI signal energy is concentrated at a high frequency. Through the processing of TC stage in the multichannel output chopper CBIA, the direct current electrode offset voltage signal can be directly filtered, and the signal appearing at the output end of TC stage is superposition of electrocardiosignal and ETI signal. Then through the TI-stage processing in the multi-channel output chopper CBIA, the central electric signal energy of the ECG channel is still maintained at low frequency, and the ETI signal is still maintained at high frequency; the ETII channel center electrical signal is modulated to a high frequency, and the ETI signal is modulated to a low frequency; the ETIQ channel center electrical signal is modulated to a high frequency, the ETI signal is modulated to a low frequency and a 90 ° phase shift occurs. The LPF in the ECG channel will then filter out the ETI signal in that channel while preserving the ECG signal; the LPF in the ETII channel will filter out the ECG signal in that channel while preserving the ETI signal; the LPF in the ETIQ channel will filter out the ECG signal in that channel while preserving the ETI signal phase shifted by 90 °.

Further, the PGA in the ECG channel amplifies the ECG signal and subtracts the DAC-processed motion artifact feedback signal calculated by the off-chip adaptive filter; the PGA in the ETII channel amplifies the ETI signal; the PGA in the ETIQ channel amplifies the ETI signal phase shifted by 90 °.

Further, the ADC will time-division multiplex the ECG, ETI signals phase shifted by 90 ° in the ECG, ETII, ETIQ channels from analog to digital domain, respectively, for off-chip adaptive filtering for algorithmic processing. The off-chip adaptive filter sends the calculated motion artifact feedback signal to the DAC input end, and the DAC converts the obtained motion artifact feedback signal from the digital domain to the analog domain and sends the motion artifact feedback signal to the PGA in the ECG channel, so that the motion artifact elimination is realized.

In the invention, in order to avoid that flicker noise interferes with an ECG signal, chopping is carried out at a CBIA to eliminate the influence of the flicker noise, and the influence of the flicker noise of a CBIA post-stage circuit is inhibited by the gain of the CBIA so as to be ignored; also, since CBIA is the first stage of the analog front end, chopping here can also increase the CMRR of the system, but chopping can result in a drop in the input impedance of the system. Therefore, the CBIA input transistor is optimized, and the parasitic capacitance of the gate end of the input transistor to the ground is reduced, so that the problem of input impedance reduction is relieved. In addition, the invention uses direct ac coupling technology to achieve highly robust DEO suppression. Meanwhile, in order to overcome the obvious power frequency interference during the acquisition of the ECG signals of the two electrodes, the invention adopts a common mode interference elimination loop based on a current source to be connected in parallel with the input end of the circuit, and proposes a class-AB structure as a driving amplifier in the loop to improve the driving-power consumption efficiency of the circuit. The above will be described in more detail below.

(A) Flicker noise cancellation and input impedance boosting

Fig. 3 shows a schematic circuit diagram of a multi-channel output chopper CBIA designed according to the present invention. FIG. 3 (a) is a Transconductance (TC) stage of a chopper CBIA, FIG. 3 (b) is a TI stage, the multi-channel output is through an input V of three TI stages in parallel _MP 、V _MN Output V to TC stage _MP 、V _MN The different TI stages differ in the control clock of the chopper ch_um.

The inverting voltage follower around transistors M1, M2 in the TC stage can be equivalently a unity gain amplifier, whose function is to divide the input signal V _IP And V _IN Copy to TC-stage load R _IN And C _IN In turn generating an input current:

the current is transmitted to the TI stage load R with a gain of 1:N via a current mirror composed of M3 and M9, M4 and M10 _OUT The transfer function of the overall circuit is therefore:

in order to alleviate the problem that the chopper technology can reduce the input impedance of the amplifier, the invention adopts a composite tube with offset tracking as an input tube of TC stage in chopper CBIA; as shown in fig. 4. The scheme suppresses the influence of chopper switch capacitance resistance by reducing the size of parasitic capacitance at the rear end of the chopper, and the specific analysis is as follows: because the input tube of the TC stage forms a turnover voltage follower structure, the current flowing through the MOS tubes MA and MB is a constant value, and meanwhile, the current on the MC is also determined to be a constant value by the current source IC.

According to the current relation, the gate-source voltages of the MOS transistors MA, MB and MC are fixed values, so that when the voltage fluctuation occurs at the input end of the chopping CBIA, the voltages at the source end, the drain end and the substrate end of the MOS transistor MA also experience the fluctuation of the same swing amplitude, and parasitic capacitances CGS, CGD and CGB of the MOS transistor MA are equivalently eliminated. The chopper switch capacitance resistance is inversely proportional to the parasitic capacitance at the rear side of the chopper, so that the problem of input impedance reduction caused by chopping is relieved from the source by taking the offset tracking composite tube as an input tube. When the bias-tracked composite tube is applied to the present invention, the additional noise caused by the composite tube is mainly derived from the bias current source IC, so the present invention sets the current magnitude of the current source IC to 0.1 μa to minimize the noise degradation thereof.

(B) Motion artifact cancellation

The invention adopts an adaptive filtering algorithm to eliminate motion artifacts generated during human body movement. To achieve adaptive filtering, a signal related to motion artifacts is required as a reference signal. The invention can acquire the reference signals required by the algorithm by synchronizing ECG and ETI signal acquisition. When the circuit works, the process that the ETI excitation current source injects current to the input end to measure ETI can interfere with the measurement of the original ECG signal. The above problem is caused by the fact that the ETI and ECG signal measurements of the system share a single channel, so the ETI signal generated by the ETI injection current is superimposed on the ECG signal, and thus produces a disturbance in the ECG signal. In order to solve the problems, the invention separates the frequency band of the ETI and the ECG by a modulation and demodulation method by using the idea of frequency division multiplexing, and further obtains the required signals in different channels.

As shown in FIG. 2, the ECG signal energy at A (input to the multi-channel output chopper CBIA) is distributed at low frequency, and the ETI signal energy is distributed at fCH/2, through electrode acquisition and excitation by the ETI excitation current source. The ECG signal energy is shifted to fCH/2 and the ETI signal energy is shifted to fCH at B (the output of TC stage in the multi-channel output chopper CBIA) by the chopper process of TI stage chopper frequency fCH. The TC stage adopts a chopper demodulation with the chopping frequency of fCH, and can see the ECG signal with the energy distributed at the low frequency and the ETI signal with the energy distributed at fCH/2 at the C (the output end of the ECG channel TI stage in the multi-channel output chopping CBIA), and adopts a chopper demodulation with the chopping frequency of fCH/2, and can see the ETI signal with the energy distributed at the low frequency and the ECG signal with the energy distributed at fCH/2 at the D, E (the output ends of the ETII channel and the ETIQ channel TI stage in the multi-channel output chopping CBIA), so that the effect of separating the ECG signals in different TC stages from the ECG signals in the frequency where the ETI signals are located is realized, and the ETI signals in the ECG channels can be filtered based on the low-pass filter (LPF).

In the system block diagram shown in fig. 2, the ETI has two channel outputs of synchronous demodulation and quadrature demodulation, the purpose of which is to obtain both real and imaginary parts of the ETI.

In the invention, the multichannel output chopper CBIA is used as the first stage of the analog front end, so that the noise-power consumption efficiency and the input impedance of the system can be obviously improved; meanwhile, the TI level of the multichannel output chopper CBIA can realize multichannel output by multiplexing the TC level, so that the synchronous acquisition of the ECG signal and the ETI signal can be realized at lower power consumption cost.

Drawings

Fig. 1 is a graphical representation of the amplitude and frequency distribution ranges of various signals in a conventional ECG signal acquisition.

Fig. 2 is a schematic diagram of an electrocardiograph signal readout circuit according to the present invention.

Fig. 3 is a schematic circuit diagram of a multi-channel output chopper CBIA in accordance with the present invention. Wherein, (a) TC stage; (b) TI stage.

Fig. 4 is a schematic diagram of the composite transistors M1, M2 in the multi-channel output chopper CBIA according to the present invention.

FIG. 5 is a graph showing the input reference noise test results of the design circuit of the present invention.

FIG. 6 shows the input impedance test results of the design circuit of the present invention.

FIG. 7 shows the gain test results of the design circuit of the present invention at different DEOs.

Fig. 8 shows the result of electrocardiosignal acquisition of the human body for developing physical activities by the design circuit of the invention.

Detailed Description

The invention designs a chip based on TSMC 0.18 mu m CMOS technology to verify the feasibility and practical superiority of the invention. The structural framework diagram of the wearable electrocardiosignal reading circuit designed by the invention is shown in figure 2, and comprises a multichannel output chopper CBIA, a two-electrode common mode bias loop, an ETI excitation current source, LPF, PGA, a bias circuit, a clock generation circuit, ADC, DAC and an off-chip self-adaptive filter. The multichannel output chopper CBIA is used for separating frequencies of the ECG signal and the ETI signal; the two-electrode common mode bias loop is used for eliminating power frequency interference introduced by electrical equipment in the process of collecting human electrocardiosignals by the electrodes; the ETI excitation current source is used for generating an ETI signal of the acquisition electrode; the LPF is used for filtering high-frequency ETI signals in the ECG channels in the multi-channel output chopper CBIA, and high-frequency ECG signals in the ETII and ETIQ channels; the PGA is used to further amplify the acquired ECG and ETI signals; the bias circuit is used for providing bias signals required by the circuit; the clock generation circuit is used for providing clock signals required by the circuit; the ADC is used for quantizing the acquired ECG signal and ETI signal into digital signals; the off-chip adaptive filter is used for generating an error feedback digital signal related to the motion artifact; the DAC is used for converting the error feedback digital signal generated by the adaptive filtering module into an analog signal.

In the present invention, the multi-channel output chopper CBIA in fig. 3 is a specific implementation circuit implementation manner of the multi-channel output chopper CBIA in fig. 2, which is composed of a Transconductance (TC) stage and three Transimpedance (TI) stages, wherein the input of the TC stage is used as the input end of the whole multi-channel output chopper CBIA, the output of the TC stage is simultaneously connected with the inputs of the three TI stages, and the outputs of the three TI stages are respectively used as the inputs of the ECG channel, the ETII channel and the ETIQ channel. The input end of a chopper CH_M in the TC stage of the multi-channel output chopper CBIA is used as the input end of the whole multi-channel output chopper CBIA, the output end of the CH_M is connected with the gate ends of the compound transistors M1 and M2 in the TC stage, the source ends of the compound transistors M1 and M2 are respectively connected with the positive ends of two resistors RIN/2, the negative ends of the two resistors RIN/2 are connected with the input end of a chopper CH_BS, and the output end of the chopper CH_BS is connected with a capacitor CIN. Through the processing of the choppers CH_M, CH_BS and the composite transistors M1, M2, the voltage signal (VIP-VIN) at the input end of the multi-channel output chop CBIA is buffered to the source ends of the composite transistors M1, M2, the voltage signal is converted into current through the two resistors RIN/2, the chopper CH_BS and the capacitor CIN, and further, the direct current electrode offset voltage signal in the input signal is eliminated, and the ECG signal and the ETI signal in the input signal are reserved. Drain terminals of the composite transistors M1, M2 are connected to drain terminals of the transistors M5, M6, respectively, source terminals of the transistors M5, M6 are connected to Ground (GND), and gate terminals of the transistors M5, M6 are connected to bias voltages. The composite transistors M1 and M2 are composed of transistors MA, MB, MC and a current source IC, the gate terminal of MA is the gate terminal of the composite transistors M1, M2, the source terminals of the transistors MA and MC are connected as the source terminals of the composite transistors M1, M2, the drain terminal of the transistor MB is the drain terminal of the composite transistors M1, M2, the source terminal of the transistor MB is connected with the drain terminal of the transistor MA, the gate terminal of the transistor MB and the gate terminal of the transistor MC are connected with the drain terminal, and further connected with the positive terminal of the current source IC, and the negative terminal of the current source IC is connected with the Ground (GND). The drain terminals of the transistors M5, M6 are further connected to the source terminals of the transistors M7, M8, respectively, the gate terminals of the transistors M7, M8 are connected to a bias voltage, the drain terminals of the transistors M7, M8 are connected to the negative terminals of the current sources I1, I2, respectively, and the positive terminals of the current sources I1, I2 are connected to a power supply (VDD). The negative terminals of the current sources I1, I2 are further connected to the gate terminals of the transistors M3, M4, respectively, the gate terminals of the transistors M3, M4 serve as the output terminals of the TC stage of the multi-channel output chopper CBIA, the source terminals of the transistors M3, M4 are connected to the power supply (VDD), and the drain terminals of the transistors M3, M4 are further connected to the source terminals of the complex transistors M1, M2, respectively. Through the processing of the choppers ch_m, ch_bs and the complex transistors M1, M2, the ECG signal and ETI signal in the input voltage signal are converted into current signals, which are transmitted from the TC stage of the multi-channel output chopper CBIA to the TI stage of the multi-channel output chopper CBIA through the transistors M3 and M9, M4 and M10. The gate terminals of the transistors M9 and M10 in the TI stage of the multi-channel output chopper CBIA are used as the input terminals of the TI stage and are connected to the gate terminals of the transistors M3, M4 in the TC stage, the source terminals of the transistors M9, M10 are connected to the power supply (VDD), and the drain terminals of the transistors M9, M10 are connected to the input terminal of the chopper ch_um. The output terminal of the chopper ch_um is connected to the positive terminal and the negative terminal of the resistor ROUT, and is used as the output terminal of the TI stage, and further, the positive terminal and the negative terminal of the resistor ROUT are connected to the drain terminals of the transistors M11, M12, respectively. The gate terminals of the transistors M11, M12 are connected to a bias voltage, and the source terminals of the transistors M11, M12 are connected to Ground (GND). The current signal transmitted from the TC stage of the multi-channel output chopper CBIA to the TI stage of the multi-channel output chopper CBIA is formed by a resistor ROUT to generate an output voltage signal (VOP-VON) containing the amplitude-amplified ECG signal and ETI signal.

The work flow of the wearable electrocardiosignal reading circuit designed by the whole invention is as follows:

as shown in fig. 2, the common-mode interference introduced by the external environment can be eliminated by the two-electrode common-mode bias loop, the ETI excitation current source can generate ETI signals related to the electrodes, the offset voltage signals of the direct current electrodes in the input signals are eliminated by the TC-level processing of the multi-channel output chopper CBIA, the superposition of ECG signals and ETI signals appears at the TC-level output end of the multi-channel output chopper CBIA, then the independent ECG signals, ETI signals and ETI signals with phase shift of 90 degrees can be obtained in ECG, ETII, ETIQ channels through the further processing of the PGA in ECG, ETII, ETIQ channels, the ECG signals, ETI signals and ETI signals with phase shift of 90 degrees obtained in each channel are digitally quantized by the ADC and sent to the off-chip adaptive filter, the adaptive filter carries out algorithm processing on the sent ECG signals and ETI signals, and can calculate motion artifact signals introduced by the body motion artifact signals, and the obtained motion artifact signals are sent to the DAC to be converted into analog signals and fed back to the ECG channels PGA, and then ECG signal measurement under the motion state of the human body is realized.

FIG. 5 shows the input reference noise figure of the present invention with a noise floor of about 50 nV/. V Hz for a circuit and an input reference integrated noise of 0.76 μVrms over a bandwidth of 0.5-150 Hz. Fig. 6 shows the input reference noise figure of the present invention, with a dc input impedance of about 3.1gΩ. Fig. 7 shows the gain of the circuit at different DEOs, it can be seen that when DEO is not greater than 600mV, the circuit gain fluctuates by less than 0.5%, and thus the dc input range of the circuit can be up to 600mV.

Fig. 8 shows the result of acquiring an actual human body electrocardiosignal without additional processing, and the acquired original ECG signal has a larger motion artifact interference component as shown in the blue curve of fig. 8 because the human body performs physical activity during acquisition, and even the circuit is saturated. After the processing of the original ECG signal according to the invention, most of the motion artifact interference components are eliminated and saturation is avoided as shown in the yellow curve of FIG. 8.

Claims (5)

1. The wearable electrocardiosignal reading circuit is characterized by being divided into two parts: analog front end and motion artifact cancellation loop; the analog front end comprises a multichannel output chopper CBIA, a two-electrode common mode bias loop, a skin electrode impedance ETI excitation current source, a low-pass filter LPF, a program-controlled gain amplifier PGA, a bias generating circuit and a clock generating circuit; the motion artifact elimination loop comprises an analog-to-digital converter ADC, a digital-to-analog converter DAC and an off-chip adaptive filter; wherein:

the multichannel output chopper CBIA is used for amplifying electrocardiosignal ECG of a human body and skin-electrode impedance signal ETI of an acquisition electrode, the input end of the multichannel output chopper CBIA is connected with an ETI excitation current source, a two-electrode common mode bias loop and the acquisition electrode, and the output end of the multichannel output chopper CBIA is connected with the input end of the low-pass filter LPF; the multi-channel output chopper CBIA circuit is composed of a transconductance TC stage and three transimpedance TI stages; three channels: the output of three TI stages corresponds to the processing of the human body electrocardiosignal ECG, the processing of the real part ETII of the ETI signal and the processing of the imaginary part ETIQ of the ETI signal;

the two-electrode common mode bias loop is used for eliminating power frequency interference introduced by electrical equipment in the process of collecting human electrocardiosignals by the electrodes, and the input end and the output end of the two-electrode common mode bias loop are connected with an ETI excitation current source, a multichannel output chopper CBIA and the collecting electrodes;

the ETI excitation current source is used for generating an ETI signal of the acquisition electrode, the input end of the ETI excitation current source is connected with the internal bias generation circuit, and the output end of the ETI excitation current source is connected with the multichannel output chopper CBIA, the two-electrode common mode bias loop and the acquisition electrode;

the low-pass filter LPF is used for filtering high-frequency ETI signals in ECG channels in the multi-channel output chopper CBIA and high-frequency ECG signals in ETII and ETIQ channels, the input end of the low-pass filter LPF is connected with the multi-channel output chopper CBIA, and the output end of the low-pass filter LPF is connected with the program-controlled gain amplifier PGA;

the programmable gain amplifier PGA is used for further amplifying the acquired ECG signal and ETI signal; the input ends of the programmable gain amplifiers PGA in the ECG channels are connected with the low-pass filter LPF and the digital-to-analog converter DAC, the input ends of the programmable gain amplifiers PGA in the ETII and ETIQ channels are connected with the LPF, and the output ends of the programmable gain amplifiers PGA in the ECG, ETII and ETIQ channels are connected with the analog-to-digital converter ADC;

the bias circuit is used for providing bias signals required by the circuit, the input end of the bias circuit is connected with the off-chip adjustable resistor, and the output end of the bias circuit is connected with the circuit which is required to be biased in the chip;

the clock generation circuit is used for providing clock signals required by the circuit, the input end of the clock generation circuit is connected with an off-chip clock source, and the output end of the clock generation circuit is connected with the circuit requiring clocks in the chip;

the analog-to-digital converter ADC is used for quantizing the acquired ECG signal and the acquired ETI signal into digital signals so as to provide input signals for off-chip adaptive filtering, the input end of the analog-to-digital converter ADC is connected with the low-pass filter LPF, and the output end of the analog-to-digital converter ADC is connected with the off-chip adaptive filtering module;

the digital-to-analog converter DAC is used for converting the error feedback digital signal generated by the adaptive filtering module into an analog signal, further eliminating motion artifacts in the analog domain, wherein the input end of the digital-to-analog converter DAC is connected with the adaptive filter, and the output end of the digital-to-analog converter DAC is connected with the programmable gain amplifier PGA; the off-chip adaptive filter is used for generating an error feedback digital signal related to motion artifact, the input end of the off-chip adaptive filter is connected with the analog-to-digital converter ADC, and the output end of the off-chip adaptive filter is connected with the digital-to-analog converter DAC.

2. The wearable electrocardiosignal reading circuit of claim 1, wherein in the multichannel output chopper CBIA circuit, the input of the TC stage is used as the input end of the whole multichannel output chopper CBIA, and the output is simultaneously connected with the inputs of three TI stages, thereby achieving the purpose of saving power consumption and area; the difference between the different TI stages is the clock frequency and phase of the chopper CH_UM, and the outputs of the three TI stages are corresponding to the ECG channel, the ETII channel and the ETIQ channel;

the TC stage is formed by connecting 8 transistors M1-M8, 2 choppers CH_ M, CH _BS, a capacitor CIN and a resistor RIN through a circuit, and the TI stage is formed by connecting 4M 9-M12, choppers CH_UM and a resistor ROUT through a circuit; transistors M1, M2 in the TC-stage buffer the signals VIP-VIN at the input of the multi-channel output chopper CBIA to both ends of the resistor RIN and the capacitor CIN; wherein, the transistors M1 and M2 are composed of composite transistors MA, MB and MC, and are used for increasing the input impedance and CMRR of the circuit; the capacitor CIN and the resistor RIN can filter out the direct current component in the input signal, and the alternating current signal in the input signal can be mirrored to the transistors M9 and M10 in the TI stage through the current mirrors M3 and M4 and flows through the resistor ROUT to form an amplified input signal VOP-VON; chopper ch_ M, CH _bs in TC stage and chopper ch_um in TI stage are used to eliminate the influence of circuit disorder and flicker noise; transistors M5, M6, M7, M8 in the TC stage are used to extend the common mode input range of the circuit; transistors M11, M12 in the TI stage are used to bias the output stage.

3. The wearable electrocardiosignal readout circuit of claim 1 wherein ECG signal energy at input a of the multi-channel output chopper CBIA is distributed at low frequency and ETI signal energy is distributed at fCH/2 through electrode acquisition and excitation by an ETI excitation current source; the ECG signal energy at the output end B of the TC stage in the multichannel output chopper CBIA is shifted to fCH/2 and the ETI signal energy is shifted to fCH after being processed by the chopper with the TI stage chopping frequency of fCH; the TC stage adopts a chopper with the chopping frequency of fCH to demodulate, an ECG signal with energy distributed at a low frequency and an ETI signal with energy distributed at fCH/2 are seen at an output end C of an ECG channel TI stage in the multi-channel output chopping CBIA, an ETI signal with energy distributed at a low frequency and an ECG signal with energy distributed at fCH/2 are seen at an output end D, E of an ETII channel and an ETIQ channel TI stage in the multi-channel output chopping CBIA by adopting a chopper with the chopping frequency of fCH/2, so that the effect of separating the frequencies of the ECG signals and the ETI signals in different TC stages is achieved, and the ETI signals in the ECG channels can be filtered based on the low-pass filter LPF.

4. A wearable electrocardiographic signal readout circuit according to any one of claims 1-3, characterized in that the workflow is as follows:

the wearable electrocardiosignal reading circuit is connected with the electrode, and after the electrode is connected to a human body, the electrocardiosignal of the human body is seen at the input end of the reading circuit, and in addition, a common mode interference signal introduced by external electromagnetic interference and an ETI signal generated by an ETI excitation current source are superimposed on the electrocardiosignal; the common mode interference signal is absorbed by the two-electrode common mode bias loop, so that the signal fed into the input end of the multi-channel output chopper CBIA is superposition of an electrocardiosignal and an ETI signal, the electrocardiosignal energy is concentrated at low frequency, and the ETI signal energy is concentrated at high frequency; after the multi-channel output chopper CBIA processing, the central electric signal energy of the ECG channel is still maintained at low frequency, and the ETI signal is still maintained at high frequency; the ETII channel center electrical signal is modulated to a high frequency, and the ETI signal is modulated to a low frequency; the ETIQ channel center electrical signal is modulated to a high frequency, and the ETI signal is modulated to a low frequency and subjected to 90 DEG phase shift; the LPF in the ECG channel then filters out the ETI signal in that channel while preserving the ECG signal; the LPF in the ETII channel filters out the ECG signal in that channel while retaining the ETI signal; the LPF in the ETIQ channel filters out the ECG signal in that channel while preserving the ETI signal phase shifted by 90 °.

5. The wearable electrocardiograph signal readout circuit according to claim 4, wherein:

the PGA in the ECG channel amplifies the ECG signal and subtracts the DAC processed motion artifact feedback signal calculated by the off-chip adaptive filter; the PGA in the ETII channel amplifies the ETI signal; the PGA in the ETIQ channel amplifies the ETI signal phase-shifted by 90 degrees;

the ADC converts ECG, ETI and ETI signals with phase shift of 90 degrees in a ECG, ETII, ETIQ channel from an analog domain to a digital domain through time division multiplexing respectively, and the analog domain and the digital domain are subjected to algorithm processing by off-chip adaptive filtering; the off-chip adaptive filter sends the calculated motion artifact feedback signal to the DAC input end, and the DAC converts the obtained motion artifact feedback signal from the digital domain to the analog domain and sends the motion artifact feedback signal to the PGA in the ECG channel, so that the motion artifact elimination is realized.