Patent Grant
11612330
This application is a § 371 National Phase Application of International Application No. PCT/EP2018/069358, filed on Jul. 17, 2018, now International Publication No. WO 2019/016191 A1, published on Jan. 24, 2019, which claims priority to International Application No. PCT/EP2017/068502, filed on Jul. 21, 2017, both of which are incorporated herein by reference in their entirety.
The present invention relates to the domain of CMOS sensors for biomedical applications, especially health monitoring.
Nowadays, health monitoring is becoming increasingly important, given both the ageing of population and the combined action of an increase in obesity level and cardio-related pathologies, i.e. cardiovascular diseases. The healthcare industry is becoming more dependent on new methods to monitor patients.
This, along with an increased interest in fitness and wellness, is calling for more affordable and precise health monitoring devices, especially when wearable.
In such a context, photoplethysmography (PPG) is known and appears to be a key technology allowing non-invasive monitoring of vital biological indicators such as heart rate (HR), blood oxygen saturation (SpO2), respiration rate (RR), and arterial pressure (AP).
A standard PPG system comprises pulsed LEDs synchronized with a photo-sensor and a processing chain. The LEDs diffuse light in the human skin. Processing the signal held by the reflected diffused light allows the extraction of some vital parameters.
A standard PPG measures some cardiac parameters by simply shining light at specific wavelengths, on a tissue and reading out either the reflected light or the transmitted light through the tissue. A big part of the light is absorbed by the tissue whilst a small amount of it reaches the detector. Once at the detector the light is converted into a photo-generated current, eventually filtered and processed by an acquisition chain, i.e. an electronic circuit. To achieve this, a standard PPG comprises at least two light emitting diodes (LEDs), and one broadband “Photodiode” (PD).
Typically, two LEDs shine light in time division multiplexing, i.e. in out-of-phase, at two different wavelengths, e.g. visible and mid-infrared, taking advantage of the different absorption properties of the molecules circulating in the blood and particularly of the oxygenated haemoglobin (HbO2), and the deoxygenated haemoglobin (Hb). The “counting” of the HbO2 molecules over the overall haemoglobin ones
gives the oxygen saturation level (SpO2), said parameter reporting the amount of oxygen flowing in the blood. The tissue being a quite complex environment with a lot of absorbing materials, the power of the emitted light should be large enough to allow enough photons impinging on the detector (the majority of them will get absorbed by the tissue). The biggest bottlenecks of PPG based system is then the LED power consumption.
This is why current medical PPG systems based on conventional PN or PIN diodes, such as those disclosed in US 2016/183813 are incompatible with portable or wearable solutions.
And although recently introduced research works present system-on-board (SOB) solutions relying on photodiodes, such SOB present relatively complex circuitry. Moreover, the commercially available smartwatches and other connected clothes fall short of meeting customer requirements in terms of reliability, precision and battery lifetime.
The present invention aims to address these drawbacks.
More precisely, the present invention relates to a photoplethysmography (PPG) sensing device configured to output a signal comprising a DC component and an AC component, said device comprising:
As a result of using a pinned photodiode, which is a very different device to a conventional PN or PIN photodiode, it is possible to use much shorter illumination pulses, in the order of microseconds. This reduces power consumption significantly over prior art PN/PIN diode-based PPG devices, rendering them compatible with small and lightweight devices which can be carried on the person. Furthermore, and as explained further below, by exploiting a particular property unique to PPD's and not present in conventional PN/PIN diodes, the signal to noise ratio can be massively improved at the level of the output from the PPD, by eliminating the DC portion of the signal at the point of readout of the PPD itself, entirely without any need for power-hungry signal processing downstream. The combination of these aspects surprisingly results in massive reduction of power consumption compared to existing PPG systems, rendering them suitable to be carried and worn permanently.
In one embodiment, each pixel further comprises:
In one embodiment, each of said pixels comprises at least one further pinned photodiode (PPD), arranged in parallel with said pinned photodiode (PPD) and connected to said sense node (SN) by means of a respective Transfer Gate (TGtransfer) transistor, each of said Transfer Gate (TGtransfer) transistors being configured to be operated synchronously, for instance by means of having their gates electrically connected together. Advantageously, each of said further pinned photodiodes (PPD) is connected to said constant DC power source or said capacitance, as appropriate, by means of a respective Sink (TGsink) transistor, the gate of each of said Sink (TGsink) transistors being configured to be operated synchronously. Such so-called macropixel structures increases the sensitivity by increasing the amount of photogenerated charge available while retaining simple circuitry.
Preferable, said pixels, whether incorporating single PPD's or multiples in macropixels, are disposed in an array, increasing the sensitivity of the device.
In one embodiment, the photoplethysmography (PPG) sensing device further comprises a processor (DSP) configured to average spatially outputs of the macro-pixels.
In one embodiment, the photoplethysmography (PPG) sensing device further comprises:
In one further embodiment, the photoplethysmography (PPG) sensing device comprises one single CDS block comprising:
In a yet further embodiment, the photoplethysmography (PPG) sensing device may comprise a single CDS block comprising:
All of these variants including one or more CDS blocks serve to filter out undesired DC components of the PPD signal, improving signal to noise ratio, with very few active components other than transistors acting as switches. This results in an excellent signal to noise ratio at an extremely low power consumption.
In one embodiment, the photoplethysmography (PPG) sensing device is built in CMOS technology, which is embedded in a system on chip (SOC).
According to another of its objects, the present invention relates to a method of operating a photoplethysmography (PPG) sensing device according to the present invention, comprising a step of:
This results in being able to eliminate a large proportion, if not all, of the unwanted DC component of the PPD's output at the level of the PPD itself, which massively improves the signal to noise ratio downstream, without requiring any power-hungry processing circuitry. Such operation is impossible with a conventional PN or PIN photodiode, which does not have a charge well, and in essence permit a rejection of the DC component of the signal, which is undesired and contains no exploitable information.
In one embodiment, the Transfer Gate (TGtransfer) transistor voltage is dynamically adapted.
In one embodiment, the method comprises a calibration step comprising the steps of:
This allows optimisation of the value of V_TGtransfer to maximise the signal to noise ratio without complex circuitry or power-hungry processing.
In one embodiment, the method comprises the steps of:
Alternatively the method may, comprise the steps of:
Alternatively, the method may comprise the steps of:
These variants significantly improve the signal-to-noise ratio by eliminating the DC component to a minimum, and the variant relating to the derivative provides further exploitable information relating to the desired signal.
Advantageously, said subtraction is carried out by connecting said first capacitor (CSH1) and second capacitor (CHS2) in parallel such that the polarity of one capacitor is inverted with respect to the other, a point of connection common to both capacitors (CSH1; CSH2) giving a voltage value corresponding to half of the difference between the voltage previously across each capacitor (CSH1; CSH2). This is extremely simple and passive, maintaining power consumption at a bare minimum.
Preferably, the photoplethysmography (PPG) sensing device comprises a plurality of pixels, which may be macro-pixels, said method further comprising a step of averaging spatially the outputs of each pixel, for instance macro-pixel in analog domain by means of passive switch-capacitor network.
Advantageously, the present invention is non-invasive.
Further characteristics and advantages of the present invention will be described in the detailed description, with reference to the drawings.
In the foregoing description and in the figures, whenever a switch is mentioned or illustrated, it typically is implemented as a transistor such as a FET, MOSFET or similar.
PPD Photo-Generated Current
This photo-generated current comprises mostly a large direct current (DC) component, due to the combined action of the tissue (DC tissue), the venous blood layer (DC ven.) and the non-pulsatile arterial blood layer absorption (DC art.). Just a small amount of this photo-generated current (
Typically, depending on the LED wavelength, the total DC component (DC tissue+DC ven.+DC art. with reference to
Such a discrepancy in values gives rise to:
Indeed for PPG applications, the AC is the component of the photo-generated current which is needed to enable health monitoring. For instance, the AC component comes out of the pulsation of the blood in the arteries and is used to determine the heart rate by measuring the distance between two consecutive AC peaks. It is also used to determine the oxygen saturation level, said level being proportional to the logarithmic ratio between the absolute maximum and the absolute minimum value of the full photo-generated signal.
One of the challenges of a PPG signal relates to its dynamic range (DR), which is the ratio between the largest and the smallest values that said signal can assume.
Different solutions, each with a corresponding electronic circuit, exist but none of them is satisfactory enough.
For instance:
To address this, the present invention proposes a new photonic sensing technology to read-out a PPG signal, especially by means of a system-on-chip (SOC) which can be fabricated in a standard CMOS technology, and having low-noise and low-power signal processing chain, a higher sensitivity and a reliable measurement at lower light levels from the LEDs.
The proposed solution, based on a PPG sensor comprising at least a pinned photodiode (PPD) instead of conventional single PD, enables to remove most of the DC component of a PPG signal as explained here after. The use of a PPD rather than a convention PN or PIN photodiode enables elimination of the DC component of the signal without complex and power-hungry circuitry, signal processing or similar, by exploiting the unique properties of PPD's which are not exhibited by PN or PIN photodiodes.
PPD Based Image Sensor
A first embodiment relates to a single transfer gate pixel arrangement.
A PPD comprises a p+-n-p junction (n-p junction buried under a shallow highly doped p+), as shown in
According to this first embodiment, a PPD is electronically connected to the source of a transistor, called Transfer Gate transistor, or TGtransfer transistor, which acts as a transfer gate (TG) between said PPD and a Sense Node, described later. As can be seen from
The grid of the TGtransfer transistor is electronically connected to a DC power source called V_TGtransfer, of which the value can be dynamically adapted.
The TGtransfer transistor is used to control the potential barrier at one edge of the PPD, the other edge being electronically connected to the ground (
Adjusting the TG potential thanks to the TGtransfer transistor enables adjusting a DC Offset (
With reference to
When this potential barrier surrounding the PPD, is lower than the well potential Vwell of the PPD, the photogenerated electrons are kept within that well. Typically, the potential barrier surrounding the PPD is kept at a slightly negative value by means of the TGtransfer transistor, at least in the integration phase described here under.
The other side of the TGtransfer transistor, in this case the drain, is electronically connected to a Sense Node (SN), see
As illustrated in
Switch S1 on
Operating the pixel of
In an integration phase:
The potential barrier surrounding the PPD is kept lower than the well potential Vwell of the PPD. Accordingly the PPD generates photo-generated electrons, which are kept and accumulated within the PPD well.
Preferably the LEDs emit synchronously (in phase). This way, the LEDs consume power only when really needed. The integration phase typically lasts a hundreds of ns to several microseconds, i.e. from about 200 ns to 3 μs, preferably from 300 ns to 2 μs. Such a low integration time is possible thanks to the complete decoupling between the electronic readout chain and the PPD. In a standard PPG sensor based on a conventional PN or PIN photodiode, such a small integration time would require a much faster readout circuitry and hence much more power consumption. Note that a typical pulse time in state of the art commercially available products is 400 μs.
Successively to the integration phase, in a reset phase:
This way, the voltage of the sense node SN is increased and set to V_reset value, draining all electrons (if any) within the SN well to the V_reset DC power source, but photo-generated electrons are still kept within the PPD well during the reset phase.
This enables the SN node to act as a capacitance as mentioned above, thus capable of storing photo-generated electrons that will be transferred in the transfer phase.
Successively to the reset phase, in a transfer phase,
In the transfer phase, the TG potential is increased to a value comprised between the well potential V_well of the PPD and V_reset.
This way, the photo-generated electrons (e−) filling the PPD well diffuse to the SN through the transfer gate TG.
This charge diffusion causes the potential of the SN to drop from the reset level V_reset to a V_transfer value, said V_transfer value being proportional to the number of transferred charges, that is to the number of photons which reached the PPD.
Preferably, the transfer phase does not last more than a 1 μs.
Successively to the transfer phase, in a readout phase:
The value of the voltage at the sense node is proportional to the number of electrons sunk from the PPD, that is to say the number of photon which reached the PPD.
A second embodiment relates to a dual transfer gates pixel arrangement.
Further to the first embodiment, it is proposed here a pixel arrangement where, instead of having the PPD electronically connected to the TGtransfer transistor and the ground like in the previous embodiment, at each pixel level, the PPD is electronically connected between the TGtransfer transistor and another transistor, called TGsink transistor, that is to say two transfer gates TGt and TGs corresponding to TGtransfer transistor and TGsink transistor respectively. This is illustrated in
To this effect, the TGsink transistor works the same way as the TGtransfer transistor works: depending on the value of the voltage V_TGsink applied to the grid of the TGsink transistor, with regard to the value of the voltage V_TGtransfer applied to the grid of the TGtransfer transistor, it is possible to modify the potential barrier surrounding the PPD, meaning transferring electrons from the PPD well through the TGtransfer transistor for sensing or through the TGsink transistor for sinking to VDD. Both transistors TGtransfer and TGsink are independent, although both are arranged to be able to sink charge from the PPD.
In this embodiment, the PPD electrons can be transferred either to one side of the PPD to the Sink Node through the TGsink transistor in a sink phase, or to the Sense Node SN on the other side of the PPD through the TGtransfer transistor in a transfer phase.
In the sink phase:
V_TG transfer is lower than V_well which is lower than V_TGsink. Accordingly, photo-generated electrons that were within the PPD well cannot go through the TGtransfer transistor and can only go through the TGsink transistor.
Electrons of the PPD well go through the TGsink transistor, e.g. to a constant voltage DC power source. In such a case, because the drain of the TGsink transistor is connected to a constant voltage and not to a capacitor, when proper voltage V_TGsink is applied to the gate of the TGsink transistor, all the electrons of the PPD well are lost.
Accordingly, none of these electrons reaches the acquisition chain (which is located downstream of the Sense Node through the TGtransfer transistor). Accordingly, ambient light does not alter the sensing, the PPD well is emptied.
In the integration phase, similar to the integration phase of the first embodiment:
The potential barrier surrounding the PPD is kept lower than the well potential Vwell of the PPD. Accordingly the PPD generates photo-generated electrons, which are kept within the PPD well.
However, most of the integrated signal contains a DC component. Transferring a small fraction of the signal exceeding a predetermined threshold (offset) enables the readout performed after the sense node to subtract said offset from said DC component.
In the reset phase:
This way, the voltage of the sense node SN is increased and set to V_reset value, draining all electrons (if any) within the SN well to the V_reset DC power source, but photo-generated electrons are still kept within the PPD well during the reset phase.
This enables the SN node to act as a capacitance, thus capable of storing photo-generated electrons that will be transferred in the transfer phase.
In the transfer phase,
This way, the photo-generated electrons (e−) filling the PPD well diffuse to the SN through the transfer gate TGt of the TGtransfer transistor, setting the sense node SN to a V_transfer voltage value.
The difference between V_well and V_TGtransfer is the offset (or V_offset) which corresponds to the DC component which is removed from the signal at the pixel level. It should be noted that the minimum value of V_offset can be zero, although it is typically a larger value.
In the readout phase,
Photo-generated electrons (e−) on the sense node SN side are read; and photo-generated electrons within the PPD are kept within its potential well.
After the readout of the sense node voltage, the latter can be reset again and a new loop starting from the sink phase can be implemented.
The n layer of a PPD device is a kind of well: once the well is fully filled, because of the barrier of potential on the TGsink transistor side, the additional photo-generated electrons overflow to the SN depletion through the TGtransfer transistor.
Thanks to the V_TGtransfer dynamic value, only the photo-generated electrons overflowing the offset to the SN depletion will be read, and if the V_TGtransfer is set appropriately, this results in only the AC component of the signal (see
The barrier of potential level chosen depends on the ambient light and the AC/DC ratio of the PPG signal in a calibration phase, as will be made clear below.
Typically, this calibration phase comprises setting the V_TGtransfer to a predetermined value then checking if the corresponding pixel is saturated. If the pixel response is saturated, then a loop comprising increasing said predetermined value of V_TGtransfer by a predetermined pitch (i.e. amount) is performed until the pixel response is no longer saturated.
For instance, the calibration phase enables a same device to be used on different users having different skin pigmentations.
After a calibration phase, it is possible to set a predetermined offset level, corresponding to a predetermined amount of the DC which does not overflow through the TGtransfer transistor, by setting a corresponding predetermined value to V_TGtransfer.
For instance, in the calibration phase it is possible to measure the grand total comprising the DC component+the AC component of the signal. It is known that the AC component represents a few percent of such signal. Accordingly, it is possible to set V_TGtransfer such that the predetermined offset level equals 90% of the signal. Reducing the offset by 90% DC enhances the AC/DC ratio by one order of magnitude, leading to less strict dynamic range constraints at the input. In fact, it should be noted that in PPG measurement, PPG signal typically has a very high DC/AC ratio of 20 to e.g. 500, which is not addressable with classical image sensing solutions.
This embodiment further reduces the DC component of the PPD signal; it thus behaves as a DC remover or as a DR enhancer. This comes without extra cost on power and circuit complexity. Such system does this at the earliest stage of the read-out chain without any additional circuitry.
Advantageously, V_TGtransfer is part of an analog device. Accordingly, it is possible to adjust precisely the level of the potential barrier, i.e. the level of the electrons remaining in the PPD vs. the number of the exceeding electrons being transferred to the sense node (overflow). Such adaptation of the potential barrier level can be done in real time.
Adjusting the value of V_TGtransfer enables adjusting the value of a predetermined part of the DC which is not sensed, which in turn, enables avoiding saturation of the sensor, leading to a better signal to noise ratio.
When the value of V_TGsink equals the value of V_TGtransfer,
When the value of V_TGsink is greater than the value of V_TGtransfer,
When the value of V_TGsink is smaller than the value of V_TGtransfer,
V_TGsink is a constant value which is selectively applied or not to the grid of the TGsink transistor, thanks e.g. to a switch or to a DSP.
Accordingly, the value of V_TGtransfer and of V_TGtransfer only, can be used to control the value of the potential barrier at the edge of the PPD. Such value of V_TGtransfer is set, and can be continuously adapted thanks to a calibration phase, so that the majority of the photo-generated electrons representing the constant component of the PPG signal (offset electrons) are not transferred to the sense node and, instead, remain within the PPD well. Eventually, these electrons can be sunk out in a next phase.
It should be underlined at this stage that, although PPD's are known in image sensors (charge coupled device, CCD's), they are not exploited in the same way as in the present invention. In an image sensor, the aim is to capture and exploit the entirety of the impinging light, since the entire amount of this light contains critical information regarding light intensity. The intensity recorded by each PPD of an image sensor is then used to create the image. As a result, in an image sensor, the entire charge stored in the PPD's charge well is transferred to the remainder of the circuit for signal processing. In the specific case of a PPG sensor, the majority of the light received is simply DC noise, hiding the signal. As described above, some of this comes from ambient light, and some from non-varying reflection from tissue, arteries and veins, the desired signal being simply the varying AC component on top of this noise. When using a conventional PN or PIN photodiode, there is no way to remove this at the level of the photodiode itself, and the entire current generated in the conventional diode must be processed. By exploiting the properties of PPD diodes as described above, it is possible to reject most, if not all, of this DC noise at the level of the PPD, only transferring the small portion of the received light that corresponds to the desired AC component.
A number of the PPD arrangements of
Although two PPD's are arranged in parallel in
The macropixel structure is optimized for the specific properties of a PPG signal which is completely different from the properties of the signal sensed in a classical image sensing. Fundamentally, a PPG output is a one-dimensional signal, and does not require any information regarding the spatial relationship between the individual PPD's, which is mandatory in conventional imaging. Furthermore, use of such macropixels would reduce the image resolution significatively in conventional imaging since they cover a larger area, whereas here, since there is no concept of “resolution”, multiple PPD's can be used in parallel to increase the amount of charge available for sampling at extremely short illumination and sampling times.
The reset voltage, the V_well of the PPD device and the transfer gate voltage used during the transfer are all chosen specifically in order to:
The choice of the number of PPDs per macropixel is defined to optimize the signal to noise ratio specifically for the PPG signal characteristics which is a function involving the electronic read noise, photon shot noise, PPDs dark current noise, quantization noise, saturation level, AC/DC ratio or perfusion index, the number of pixels macropixel, the number of macro pixels per array. Such a function is specific to the PPG signal processing and is not a consideration in image sensor design, as discussed above.
Array of Pixels
Regardless of whether the first or the second embodiment of pixels, or the macropixel structure described above is used, it is advantageous to use a photo detector comprising a plurality of pixels, especially an array of pixels, each pixel comprising one PPD one macropixel.
This leads to significant improvements. Indeed, it has been shown that PPD based imagers achieve outstanding sensitivity and noise performance.
The PPG device proposed in this embodiment bases its functionalities on the distribution of the input light on an array of PPD based pixels and averaging their outputs, this reducing significantly the read-out noise, photonic shot noise and spurious signals.
As mentioned earlier, most of the power consumption of a PPG device is spent in the LEDs emission. The better sensitivity allowed by the proposed device enables to significantly reduce the duty cycle and the illumination of the LEDs, significantly reducing the average bias current and the power consumption.
In addition to this, this new PPG device can also be fully integrated into a single chip, a SOC, which represents a breakthrough with respect to the state-of-the-art-solutions, mostly showing SOBs (discrete electronics).
Replacing a single PD or PIN diode by an array of pixels results at least in the distribution of the input light on the pixels which reduces the dynamic range constraint on the read-out chain. In addition, the averaging of the array pixels output allows a reduction of the read noise variance, which is proportional to the number of averaged outputs.
The fact that there is a reader circuitry at each pixel generates electronic read noise. Averaging the value of all pixels reduces the read noise by a number which is a factor of the total number of pixels of the array. This enables to eradicate the read noise.
The averaging of the outputs of the array pixels is performed in the charge domain.
The timing diagram related to this schematic (
In the array structure presented in different variants in
The array structure allows pulse capturing, ambient light cancellation, noise reduction, DC cancellation with respect to a given perfusion index in less than 10 us for a full array of more than 100,000 PPDs and allows a power consumption at least two orders of magnitude below that achieved by a conventional array. This demonstrates the substantial difference between a pixel array and a conventional image sensor.
Multiple Sampling
The purpose here is to further improve the signal-to-noise ratio SNR.
In a first step, the LEDs are off. Ambient light is sampled for a predetermined length of time thanks to a first capacitor. The signal which is read and recorded only corresponds to ambient light.
In a second step, the LEDs are on for a predetermined length of time, ideally the same as the predetermined length of time for which ambient light is sampled. The light from the ambient light and from the LED is sampled thanks to a second capacitor. The signal which is read and recorded corresponds to LEDs light and ambient light.
This operation is an intrinsic subtraction which eliminates both the constant reset level and also reduces the flicker noise, which is called “Correlated-Double-Sampling” (CDS), using corresponding CDS block or stage.
It is then possible to subtract the signal of the first step from the signal of the second step to get a signal which only corresponds to LEDs light by subtracting the value from the first capacitor from the value of the second capacitor.
The two CDS can be operated within 20 μs (pulse of LED light of 10 μs), it is assumed that ambient light does not change in a significant way within that period of time. This removes any artefacts that don't change within that period of time, e.g. motion artefacts.
It has been previously discussed that the reading-out action is performed in several steps, starting from the reset level sensing and finishing with the CDS.
This mechanism can be enhanced for PPG applications to perform ambient light subtraction and output averaging, i.e. filtering, in an almost fully passive manner, which means inherently zero power.
With reference to
In the first phase (“Ambient Light” with reference to
Subsequently, switches SH1 and SH2 are opened, and switches SH3 are closed, resulting in both capacitors CHS1 and CSH2 being connected in parallel, with the polarity inverted. In other words, the pole of the capacitor CSH2 into which the charge from the PPD flowed is connected to ground, its other pole, which had previously been grounded by the lower of the two SH2 switches, being connected to the input pole of capacitor CSH 1. Thanks to the inversion potential principle, at the end of the first CDS1 stage, the corresponding voltage level results to be 0.5(V_CSH1−V_CSH2), which equals to 0.5(V_amb+V_reset−V_reset)=0.5V_amb. This removes any artefact from a preceding reset. Such result is stored in a CSH4 capacitance (block CDS2 on
In a second phase (“LED Light” with reference to
In this case, at the end of the first CDS1 stage, similarly to the first phase, the corresponding voltage level results to be approximately equal to: 0.5(V_LED+V_amb+V_reset−V_reset)=0.5(V_LED+V_amb).
As for the first phase, V_LED+V_amb is stored in a capacitance CSH3 (block CDS2 on
Since the PPG information of interest is contained in VLED only, the signal can be “cleaned” of ambient light to improve the signal-to-noise ratio.
By construction of block CDS2, it is possible to differentiate CSH4 with CSH3, then to subtract the voltage 0.5(V_amb) related to the ambient light from the voltage 0.5(V_amb+V_LED) related to the ambient light+the signal resulting from LED illumination, thus to obtain the voltage 0.25(V_LED) related to the LEDs only, by operating the various switches of the block CDS2 as described above in the context of CDS1, mutatis mutandis. This final signal is then passed to the ADC and converted to a digital signal.
As a result, the signal to noise ratio can be massively improved by subtracting the DC voltage component from the PPD output which corresponds to the received ambient light measured with the LED off can easily be subtracted from the total signal measured with the LED on.
One other embodiment and operation of the presented photoplethysmography (PPG) sensing device allowing ambient light cancellation with one single CDS is described using
This ambient light cancellation with one CDS is described as follows based on
One other embodiment and method operation of the presented photoplethysmography (PPG) sensing device allowing sensing the derivative of the PPG signal instead of the PPG signal itself is described using
The PPG derivative sensing is described as follows based on
This technique has also the advantage of computing the PPG signal derivative with one single readout scheme and one analog to digital conversion.
Multiple Sampling+Array of Pixels
The multiple sampling described here above can be performed simultaneously for all the pixels in a same column, as shown on
In a same column, pixels share the same sense node SN connected to one voltage buffer, as shown in
The inversion polarity principle of
The averaging all the column reduces the read noise variance by a factor equal to the number of columns, this leading to low-noise, low-power performances.
In a first embodiment,
The averaging involves a fully passive circuit made of switches and sampling capacitances, as shown on
As shown in
After this reset operation, the TGtransfer transistor is turned off and the LED is pulsed on for a predetermined period of time to let the integrated charge cumulate in the PPD and overflow to the sense node. The overflowing charge changes linearly the value of the sense node SN voltage.
The outputs of each voltage follower, SFx with respect to
In a further embodiment,
The two CDS stages perform independently, per each column, the ambient and reset voltage compensation and only in the end the LED signal voltages are averaged among the m-columns.
In a further embodiment illustrated in
The process of averaging and a system level representation is depicted in
As far as the input noise (in e−rms) is concerned, the shot noise is a limiting factor. Indeed, for a 28.5 dB SNR, accounting for the shot noise only, the shot noise accounts for more than 10 ke−rms in terms of input referred noise. This means that the solutions of
This operation is an intrinsic subtraction which eliminates both the constant reset level and also reduces the flicker noise and spurious signals.
Thanks to the present invention, it is possible to remove the DC component of a signal at the pixel level.
The PPD acts as a capacitor whose capacitance value is known. The amount of charges in the PPD can then be calculated by multiplying said capacitance value by V_TGtransfer.
The PPD based image sensor according to the present invention can be a CMOS imager, which is advantageously used in a PPG application. As explained later, a CMOS image sensor can be driven to solve the issues of LED power consumption, dynamic range limitations of a PPG signal which make the signal processing and filtering difficult otherwise.
The device according to the present invention is advantageously built as a system-on-chip (SOC), integrating a different photonic sensing technology with low-noise and low-power signal processing chain, this allowing higher sensitivity, reliable vital parameters measurement at lower active light (LEDs) levels and at the same time fabricated in a standard CMOS process.
As can be seen from the foregoing, with respect to conventional PPG sensing devices, the PPG sensor of the invention has the advantage of complete decoupling between the photoelectrons light integration in the pixel array and the next stages of the readout chain e.g. amplification and analog to digital conversion. This special feature is enabled thanks to the fact that the capacitors can hold an analog value without any external driving trigger. This allows switching-on the power supply to the array only during the light integration time which is typically less than 10 us and switching-off that supply during the rest of readout operation. In this way the overall power consumption of the PPG sensor is even more reduced.
| Number | Date | Country | Kind |
|---|---|---|---|
| PCT/EP2017/068502 | Jul 2017 | WO | international |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2018/069358 | 7/17/2018 | WO |
| Publishing Document | Publishing Date | Country | Kind |
|---|---|---|---|
| WO2019/016191 | 1/24/2019 | WO | A |
| Number | Name | Date | Kind |
|---|---|---|---|
| 9686485 | Agranov | Jun 2017 | B2 |
| 20060146158 | Toros et al. | Jul 2006 | A1 |
| 20090090844 | Yan et al. | Apr 2009 | A1 |
| 20090207284 | Johnson | Aug 2009 | A1 |
| 20120056079 | Levine et al. | Mar 2012 | A1 |
| 20120056080 | Levine et al. | Mar 2012 | A1 |
| 20120193516 | Bogaerts | Aug 2012 | A1 |
| 20120193743 | Kawahito et al. | Aug 2012 | A1 |
| 20150076323 | Mabuchi | Mar 2015 | A1 |
| 20160183813 | Naima | Jun 2016 | A1 |
| 20170007138 | Kim et al. | Jan 2017 | A1 |
| 20170127988 | Tao et al. | May 2017 | A1 |
| 20170231500 | Rothberg | Aug 2017 | A1 |
| Number | Date | Country |
|---|---|---|
| 2001036128 | Feb 2001 | JP |
| 2013104839 | May 2013 | JP |
| 2013211615 | Oct 2013 | JP |
| 2014060631 | Apr 2014 | JP |
| 2017018569 | Jan 2017 | JP |
| WO 2011043339 | Apr 2011 | WO |
| Entry |
|---|
| International Preliminary Report on Patentability, dated Jan. 30, 2020, from International Application No. PCT/EP2018/069358, filed on Jul. 17, 2018. 11 pages. |
| International Search Report and Written Opinion of the International Searching Authority, dated Sep. 17, 2018, from International Application No. PCT/EP2018/069358, filed on Jul. 17, 2018. 13 pages. |
| Number | Date | Country | |
|---|---|---|---|
| 20200205680 A1 | Jul 2020 | US |
