HIGH RESOLUTION THERMOPILE INFRARED SENSOR ARRAY HAVING MONOLITHICALLY INTEGRATED SIGNAL PROCESSING

Abstract

A high-resolution thermopile infrared sensor array having monolithically integrated signal processing and a plurality of parallel signal processing channels for the signals from pixels of a sensor array, and a digital port for the serial output of the pixel signals are provided, wherein the sensor array is located on one or more sensor chips. The thermal piled infrared sensor array possesses low power loss, high integration density and high thermal and geometric resolution. Each signal processing channel (K.sub.1 . . . K.sub.N) has at least one analogue/digital converter (ADC), and is assigned a memory region in a memory (RAM) for storing the signals from the pixels (SE).

Claims

1.-15. (canceled)

16. A high-resolution thermopile infrared sensor array having monolithically integrated signal processing and a plurality of parallel signal processing channels for signals from pixels of a sensor array, and also a digital port for serial output of the signals of the pixels, wherein the sensor array is situated on one or more sensor chips, wherein each signal processing channel (K.sub.1 . . . K.sub.N) is arranged on ae sensor chip (SP) or in proximity thereto and comprises at least one analog/digital converter (ADC) and each signal processing channel (K.sub.1 . . . K.sub.N) is assigned a memory area in a memory (RAM) for storing results of the analog/digital converters (ADC), wherein a number (a) of at most 16 or 8 pixels (SE) share a signal processing channel (K.sub.1 . . . K.sub.n) and preferably in each case (a)=4, 3 or 2 pixels (SE) of a row share a signal processing channel (K.sub.1 . . . K.sub.n), or each pixel (SE) is assigned a signal processing channel (K.sub.1 . . . K.sub.n), and a center-to-center distance between individual pixels (SE) of the sensor array (TPA) is between less than 300 ?m and less than 100 ?m.

17. The high-resolution thermopile infrared sensor array as claimed in claim 16, wherein each signal processing channel (K.sub.1 . . . K.sub.N) is assigned at least one pixel (SE) and at most a number (a) of 16 pixels (SE).

18. The high-resolution thermopile infrared sensor array as claimed in claim 17, wherein for each signal processing channel (K.sub.1 . . . K.sub.N) provision is made of a signal multiplexer (MUX) for selecting pixels (SP) of the sensor array (TPA) that are assigned to the signal processing channel (K.sub.1 . . . K.sub.N).

19. The high-resolution thermopile infrared sensor array as claimed in claim 16, wherein a preamplifier (VV) is connected upstream of the analog/digital converter (ADC) in each signal processing channel (K.sub.1 . . . K.sub.N).

20. The high-resolution thermopile infrared sensor array as claimed in claim 19, wherein the preamplifier (VV) has a gain factor that is between less than 500 and less than 100.

21. The high-resolution thermopile infrared sensor array as claimed in claim 16, wherein the analog/digital converter (ADC) has a resolution of between at least 10 bits and at least 16 bits.

22. The high-resolution thermopile infrared sensor array as claimed in claim 16, wherein the analog/digital converter (ADC) operates according to a charge balancing or delta-sigma method.

23. The high-resolution thermopile infrared sensor array as claimed in claim 16, wherein each signal processing channel (K.sub.1 . . . K.sub.N) contains a low-pass filter (TPF) for limiting noise bandwidth, a cut-off frequency of said low-pass filter corresponding to at least a product of the frame rate of the thermopile infrared sensor array (TPA) and the number of pixels (SE) per signal processing channel (K.sub.1 . . . K.sub.N), but at most to triple an absolute value of the product.

24. The high-resolution thermopile infrared sensor array as claimed in claim 23, wherein the low-pass filter (TPF) is part of an integrating analog/digital converter (ADC).

25. The high-resolution thermopile infrared sensor array as claimed in claim 16, wherein a noise bandwidth of each signal processing channel (K.sub.1 . . . K.sub.N) is determined depending on a respective frame rate by use of an integrating analog/digital converter (ADC) and an externally predefined or internally generated master clock and a predefined conversion rate of the analog/digital converter (ADC).

26. The high-resolution thermopile infrared sensor array as claimed in claim 16, wherein a portion of the signal processing channels (K.sub.1 . . . K.sub.N) is arranged in each case in an interspace between the pixels (SE).

27. The high-resolution thermopile infrared sensor array as claimed in claim 16, wherein at least one portion of the signal processing channels (K.sub.1 . . . K.sub.N) is arranged together with further electronics in a region surrounding the sensor array (TPA) in an outer edge region of the sensor chip (SP).

28. The high-resolution thermopile infrared sensor array as claimed in claim 18, wherein a reference voltage (VREF), a voltage supply (VDD, VSS), a temperature reference, a reference data memory (REF/PTAT), a reference data memory, a clock generator (CLK), an I/O digital port (DIO) and also signal processing are integrated on the sensor chip (SP) alongside the signal processing channels (K.sub.1 . . . K.sub.N) and the signal multiplexers (MUX), wherein the thermopile infrared sensor chip (SP) and the separate chip (ROIC) are then fixedly connected to one another.

29. The high-resolution thermopile infrared sensor array as claimed in claim 28, wherein at least one portion of the signal processing channels (K.sub.1 . . . K.sub.N) is arranged below the sensor chip (SP) with the sensor elements (SE) on a separate chip, wherein the electrical connection of the pixels (SE) to the signal processing channels (K.sub.1 . . . K.sub.N) or/and further electronic components is effected by vias (VIA) through the sensor chip (SP) or redistribution wirings, wherein the thermopile infrared sensor chip (SP) and the separate chip (ROIC) are fixedly connected to one another.

Description

[0040] The invention is explained in greater detail below on the basis of exemplary embodiments. In the associated figures of the drawings:

[0041] FIG. 1: shows the basic construction of a thermopile infrared sensor array according to the invention;

[0042] FIG. 2: shows a block diagram of a circuit arrangement according to the invention for integrated signal processing for thermopile infrared array sensor chips with low-pass filter and respectively one signal processing channel per pixel;

[0043] FIG. 3: shows a block diagram of a circuit arrangement according to the invention for the integrated signal processing for thermopile infrared array sensor chips in a second embodiment, in which the low-pass filter function is implemented in an integrated ADC (AD converter);

[0044] FIG. 4: shows a block diagram of a circuit arrangement according to the invention for integrated signal processing for thermopile infrared array chips in a third embodiment, in which a plurality of pixels share a signal processing channel, with

[0045] FIG. 4a: low-pass filter downstream of the preamplifier and

[0046] FIG. 4b: an integrating AD converter that performs the low-pass filter function;

[0047] FIG. 5: shows a block diagram of a further circuit arrangement according to the invention for the integrated signal processing for thermopile infrared sensor arrays, in which the low-pass filter function is implemented in an integrating AD converter;

[0048] FIG. 6a: shows a schematic sectional illustration of a sensor chip having vias and, arranged below it, a separate chip having signal processing channels and further memory and signal processing electronics integrated therein; and

[0049] FIG. 6b: shows the arrangement according to FIG. 6a, but supplemented by a cap wafer having a radiation entrance window and additional electrical connection means such as bond wires for connection to further functional assemblies for the operation of the thermopile infrared sensor array.

[0050] FIGS. 1 and 2 reveal the basic construction of a thermopile infrared sensor array according to the invention comprising, arranged in the center of the sensor chip, a thermopile infrared sensor array TPA in matrix form having m?n pixels SE 1.1 . . . SE 1.n?SE m1.1 . . . SE m, nAround the pixels SE or around the pixel array, preferably on two sides, m?(n/2)/a signal processing channels K1 . . . KN are situated on the same sensor chip SP, which signal processing channels amplify and filter the signals of the individual pixels SE and convert them into digital signals. Here a signal processing channel K1 . . . KN is shared by a respective number of a pixels.

[0051] In principle, all the signal processing channels may also lie on one side of the pixel array, wherein the thermopile infrared sensor array then has an asymmetrical heat distribution.

[0052] What is essential for the function of thermopiles is that they have hot and cold contacts which are connected to one another via the longest possible conductive tracks, i.e. are arranged as far away from one another as possible, and wherein the hot contacts are arranged on a radiation receiver (not illustrated) and the cold contacts are arranged on a heat sink at the edge of the pixel in order to generate an evaluatable signal voltage depending on the temperature difference between the hot and cold contacts.

[0053] Each pixel SE of the thermopile infrared sensor array TPA contains a miniaturized thermopile cell, known per se, and above each thermopile cell optionally a radiation entrance window with a suitable optical unit. The individual thermopile cells have a center-to-center distance (so-called pixel pitch) of at most 400 ?m, preferably less than 200 ?m and particularly preferably of less than 100 ?m. The smaller the pixel pitch, the smaller the entire thermopile infrared sensor array chip becomes and the dimensions of the required optical unit for imaging the infrared radiation on the pixel SE also decrease for the same number of pixels. A reduction of the chip and optical unit dimensions usually also leads to lower manufacturing costs.

[0054] Optionally, a smaller pitch allows more pixels SE to be accommodated on a sensor chip of predefined size, in order thus to achieve a higher optional resolution capability.

[0055] FIG. 2 shows a block diagram of a circuit arrangement according to the invention for the integrated signal processing for thermopile infrared sensor arrays TPA having a mirror-inverted construction, having a central thermopile infrared sensor array TPA having a preamplifier VV, a low-pass filter TPF connected downstream and an analog/digital converter ADC for each signal processing channel K1 . . . KN on two sides of the sensor array TPA per pixel SE.

[0056] The outputs of the analog/digital converters ADC are connected to a memory array RAM, which can be read via a control circuit CRTL, such that the digital output signals are available for further processing at a digital input and output port DIO.

[0057] Furthermore, there are situated on each chip SP the assemblies necessary for operation such as clock generator CLK and for the necessary voltage supply VDD, VSS and one of more reference voltages VREF or REF/PTAT and also additional ESD circuit blocks are available.

[0058] According to the invention, a large number of individual signal processing channels K1 . . . KN are integrated on or below the same sensor chip SP, wherein a number a of at most 16 or 8 pixels SE share a signal processing channel K1 . . . KN, but preferably only in each case a=4, 3 or 2 pixels SE share a signal processing channel.

[0059] The corresponding number a of pixels SE is connected to the respectively assigned signal processing channel K1 . . . KN via multiplexers MUX, or regions of a multiplexer MUX (FIG. 4a). Particularly preferably, each pixel has a dedicated signal processing channel K1 . . . KN (i.e. a=1; see FIG. 3 and FIG. 4b). The smallest noise bandwidth and thus the lowest noise and the best temperature resolution are then achieved. Moreover, it is then possible to dispense with the multiplexers upstream of the signal processing channel K1 . . . KN (FIG. 2).

[0060] However, since space requirement and power loss also increase as a result of more signal processing channels K1 . . . KN, it may also be expedient, primarily in the case of sensor arrays TPA having very many pixels taking account of thermal resolution and space requirement, to choose a>1.

[0061] In order to accommodate as many signal processing channels K1 . . . KN as possible on the chip, both the area requirement and the power loss of the individual channels must be very small in order to keep down chip size and costs, but also thermal crosstalk between the thermopile pixels SE of the sensor array TPA.

[0062] In order to achieve that, signal processing channels K1 . . . KN with in each case only one small low-noise preamplifier VV having a comparatively low gain factor, i.e. less than 500-fold, and a slow power-saving analog/digital converter ADC having a high resolution, i.e. having at least 10 bits, are used.

[0063] Preferably, preamplifiers VV having a gain factor of less than 100-fold are involved and the resolution of the analog/digital converter ADC should preferably be 16 to 24 bits.

[0064] The combination of a preamplifier VV having a low gain factor and a slow analog/digital converter ADC having a high resolution ensure a small area requirement owing to the low gain factor. Moreover, a low current consumption is ensured owing to the analog/digital converter ADC which, although it has a high resolution, operates comparatively with a low transmission rate.

[0065] Furthermore, the use of a high-resolution analog/digital converter ADC without the use of a preamplifier VV is conceivable. A small difference in the positive and negative reference voltage VREF of the analog/digital converters ADC is advantageous because the temperature resolution is increased as a result.

[0066] By way of example, so-called auto-zero (switched chopper) amplifiers, which are distinguished by low offset voltages and offset voltage drifts, are suitable as preamplifiers VV. Given a gain factor of<100 . . . 500, a chopper amplifier can be constructed with a single stage and thus in a manner that saves a particularly large amount of space and power.

[0067] By way of example, the sigma/delta method or the charge-balancing method is suitable for slow analog/digital converters ADC having a high resolution. Since very many analog/digital converters ADC operate in parallel on the sensor chip SP, in comparison with conventional thermopile infrared sensor arrays having only one analog/digital converter ADC at the output a low conversion rate arises, which leads to the required low power loss and the small space requirement, as can be explained on the basis of a 64?64 sensor array TPA.

[0068] Analog/digital converters operating according to the sigma/delta method or the charge-balancing method are converters that are routine and known among those skilled in the art.

[0069] A sensor array constructed according to the prior art and having 64?64 pixels within only one analog/digital converter at a frame rate of 15 Hz requires a conversion rate of the analog/digital converter ADC of 64?64 pixels?15 Hz=61.440 Hz.

[0070] In the case of the analog/digital converters ADC operating in parallel according to the invention, a conversion rate of just 15 Hz (given a=1) or 60 Hz (given a=4) is required. That makes it possible to realize analog/digital converters ADC having a high resolution (e.g. 16 bits or more) with a very small current and space consumption.

[0071] The digitized signals of each signal processing channel K1 . . . KN can be buffer-stored in a memory array of a memory RAM before they are relayed to the serial output data stream of an I/O digital port DIO. As a result, the time regime for reading out the data via the digital port DIO can be chosen such that the entire time of a frame is available for the integration and low-pass filtering of the pixel signals.

[0072] The noise bandwidth of the signal processing channels K1 . . . KN should preferably be reduced to the minimum necessary, which results from the product of the number of pixels SE per signal processing channel K1 . . . KN and the frame rate of the sensor array TPA.

[0073] That can be realized in a simple manner by a low-pass filter TPF being integrated upstream of the analog/digital converter ADC, which is possible for example as part of the preamplifier VV or as an additional low-pass filter TPF.

[0074] FIG. 3 and FIG. 4b illustrate a particularly space-saving variant in which the reduction of the noise bandwidth is achieved in a particularly preferred manner by means of the integrator behavior of a suitable analog/digital converter ADC, e.g. of an analog/digital converter ADC operating according to the charge balancing method.

[0075] A significant improvement in performance overall is achieved with the novel signal processing described above.

[0076] In the case of white noise, as is known, the noise increases with the root of the signal or noise bandwidth of the preamplifier VV. With the use of a 64?64 sensor array according to the prior art, the noise bandwidth would increase by 64?64 times the frame rate in the case of only one preamplifier VV and still by 64-fold in the case of one signal amplifier per column.

[0077] Consequently, the total noise and the temperature resolution of a 64?64 sensor array with one preamplifier VV would be higher by 64-fold, and in the case of 64 column amplifiers would still be higher by approximately 8-fold, compared with an array in which each pixel has a dedicated signal channel.

[0078] Therefore, by way of example, a 64?64 sensor array TPA according to the invention can achieve a thermal resolution capability up to 8 times higher than that of known sensor arrays constructed according to the prior art. Following the same mode of consideration, this results in a possible improvement in the thermal resolution capability by 4-fold in the case of a 16?16 sensor array TPA according to the invention, by 5.5-fold in the case of a 32?32 sensor array TPA and by 11-fold in the case of a 128?128 sensor array TPA.

[0079] If, in the case of a 128?128 sensor array TPA, the number of signal processing channels were reduced and e.g. a=16 pixels shared a signal processing channel, instead of an 11-fold improvement in the thermal resolution capability this would still achieve a 3-fold improvement in the signal/noise ratio by comparison with WO 2006/122529 A1 and by comparison with the rest of the prior art having only one signal transmission channel by 32-fold.

[0080] The signal processing channels K1 . . . KN configured according to the invention can be arranged in the edge region of the individual pixels SE, as illustrated in FIG. 5, in the peripheral edge region of the sensor chip SP, i.e. outside the pixels, or in a manner distributed between both regions.

[0081] In order to achieve a good thermal balance and thus a homogeneous thermal image over the entire sensor chip SE, the power losses of the various assemblies should be distributed as homogeneously and symmetrically as possible on the sensor chip SE.

[0082] Alongside the actual signal processing channels K1 . . . KN and the multiplexers MUX, further electronic components can be concomitantly integrated on the sensor chip and be connected to the digital interface via the multiplexers MUX (cf. FIG. 1 and FIG. 5). Said further electronic components may be temperature references, voltage references, memory means (e.g. EEPROMs for storing calibration data), possibly also small ?Controllers, e.g. for further signal conditioning or temperature calculation.

[0083] Furthermore, it may be advantageous to have additional information REF/PTAT, such as, for example, the signals from image pixels or pitch elements, inserted via the same signal processing channel with the serial data string, in order to compensate for drift effects and thereby to increase the measurement accuracy.

[0084] The setting of the most favorable noise bandwidth for the respective frame rate can be predefined during the use of an integrating analog/digital converter ADC by means of the internally generated master clock and the conversion rate predefined by the clock regime.

[0085] For the sake of completeness, it should be mentioned that the signal processing channels K1 . . . KN can also be arranged on a separate chip ROIC, below the actual sensor chip SP (FIGS. 6a, 6b).

[0086] FIG. 6a shows a schematic sectional illustration of a sensor chip SP having vias TSV and, arranged below it, a separate chip ROIC having, integrated therein, signal processing channels K1 . . . KN and the further memory and signal processing electronics as described above. The vias TSV are through openings through the sensor chip SP which are filled with a conductive material, which are insulated from the sensor chip SP and which are connected at the ends in each case to conductive tracks (not illustrated) on the sensor chip SP and respectively the separate chip RIOC. It goes without saying that there must also be a mechanically fixed connection between the sensor chip SP and the separate chip ROIC.

[0087] FIG. 6b reveals the same arrangement as in FIG. 6a, but supplemented by a cap wafer CAP on the sensor chip SP with a radiation entrance window SEF. The cap wafer CAP may consist in its entirety of an infrared-transmissive material, or merely comprise a region of this type above the sensor array TPA. Furthermore, provision can be made of additional electrical connection means such as bond wires BD for connecting the sensor chip SP to further functional assemblies on a printed circuit board, or the like, which are required for the operation of the thermopile infrared sensor array TPA.

[0088] Instead of the vias TSV for electrical connection between the sensor chip SP and the separate chip ROIC, redistribution wirings are also suitable, wherein conductive tracks are led around the side edge from the sensor chip SP to the separate chip ROIC.