WO1995020750A1

WO1995020750A1 - Sensor device

Info

Publication number: WO1995020750A1
Application number: PCT/SE1995/000077
Authority: WO
Inventors: Ulf Ringh; Christer Jansson; Christer Svensson
Original assignee: Försvarets Forskningsanstalt
Priority date: 1994-01-26
Filing date: 1995-01-26
Publication date: 1995-08-03
Also published as: SE9400239D0; SE503643C2; SE9400239L

Abstract

The invention relates to a sensor device comprising an array of detectors. In such a detector array, it is difficult to read out weak electric signals, especially if there are many detectors. The invention solves the contemplated problem by using a device in which each detector consists of a radiation-sensitive component included in an oscillator, for example a resistor or a capacitor included in an RC oscillator, whose frequency is used as a measure of the detected quantity.

Description

Sensor device

The present invention relates to a sensor device comprising an array of detectors. It has proved difficult to read out weak electric signals from such an array, especially if it includes a large number of detectors.

The present invention has been developed in connection with work on the read-out of signals from an array of IR detectors (so-called Infrared Focal Plane Array) of the resistance bolometer type, i.e. detectors where the resistance is a function of the quantity to be measured. However, the invention can be used for capacitive detectors as well, and also when only a few detectors are used.

Traditional technology uses different types of analogue circuit solutions. At the low signal levels involved in IR detection, it may however be devastating, in view of the noise, to transport analogue signals in long lines from the detector.

The present invention uses oscillators whose frequency is dependent on the quantity to be measured. Such concepts have been published for small arrays using but a few elements, and cannot easily be scaled up. No one has hitherto been able to present a solution to the problem to read out the frequency of all the elements in a large image-producing array of e.g. 100x100 detectors or more. This problem is solved by the present invention, which is suitable for implementation in CMOS process technology.

The present invention solves the contemplated problem of signal read-out by being designed in the manner defined in the appended independent claim. The other claims define suitable embodiments of the invention.

The invention will be described in more detail hereinafter with reference to the accompanying drawings, in which

Fig. 1 shows a detector array of 100x100 detectors,

Fig. 2 shows an oscillator signal with an observation window according to a first embodiment of the invention, Fig. 3 shows a detector array with an enlarged pixel cell according to a second embodiment of the invention, Fig. 4 shows the oscillator signal during line read-out according to the second embodiment, and Fig. 5 shows an alternative to the RC oscillator, called ring oscillator.

For greater clarity, the invention will be described hereinafter with reference to an array of detectors which are arranged in rows and columns. The invention is however not restricted to a detector array of such a design. The arrangement of rows and columns as described below may readily be exchanged by a person skilled in the art for other groups of detectors which in this context are obvious to him.

The basic concept of the invention is to use an oscillator for integrating and reading out the detector signal. The detector signal is converted to frequency, which normally is thereafter read out in parallel one row or column at a time. The invention reduces the noise on pixel level by the integration during the time elapsing between the read-outs. A suitable way of implementing the invention is using as oscillator an RC oscillator, in which either a resistor, R, or a capacitor, C, constitutes the detector proper. It is however possible to use other types of oscillators. One type of oscillator that can be used in the invention instead of an RC oscillator is a ring oscillator. Fig. 5 shows an example of such an oscillator made up of three inverters and a capacitor as the radiation-sensitive component.

In a first embodiment of the invention, each detector is permanently connected to an oscillator, i.e. there is one oscillator per pixel. Thus, to each detector are permanently allocated one or more non-radiation-sensitive supplementary components, which together with the detector are adapted to form the oscillator.

Fig. 1 shows an example of an array of 100x100 detectors. To explain the function, a numeric example of this array is given below with a line reset each twentieth ms. Since the capacitive, C, or resistive, R, detector is included in an RC oscillator, its frequency will be dependent on the detected quantity. The nominal oscillator frequency is assumed to be 10 kHz, i.e. the period time is 0.1 ms.

In the range 19.8-20.0 ms from reset of the line, an observation window is opened where the time of the about 200 periods of the oscillator signal since reset is observed (the integration resides herein), see Fig. 2. The time from the beginning of the window up to the first positive edge is measured. Moreover, the period time is measured by measuring the time between two consecutive positive edges. The purpose of the latter measurement is to determine the number of pulses since reset. In this example, read-out is carried out at a frequency of 10 MHz. An external counter is started at the beginning of the observation window. At each read-out of the window, no steps are taken if the oscillator signal is low. The first time the signal is high, a memory cell belonging to the pixel concerned is set equal to the value of the counter.

By carrying out the read-out at 10 MHz in a window after about 20 ms, the relative resolution becomes about 1 :200,000. The dynamic range within a window becomes 1:1000 (the number of read-outs in the period time 0.1 ms). By measuring the period time, the number of pulses since reset can be determined, and a dynamic range being a multiple of 1000 is thus obtainable. If, for example, there is a variation of +/-4 pulses since reset, this means a dynamic range of 1:16,000 corresponding to 14 bits.

By using the remaining 99 0.2-ms intervals, it is possible to read out the remainder of the columns or rows in a 100x100 image with the same hardware.

Read-out frequency and read-out window can each be selected depending on the oscillator frequency and the demand on resolution and dynamic range. In large arrays, one may choose to read out several lines at a time.

In the given resistance bolometer example, a capacitor is placed at each pixel as a part of the array. The capacitance, and thus the requirement of space, can be maintained low if the detector resistance is high or if a high oscillator frequency can be used. In addition to the noise reduction which occurs by the integration on pixel level, noise-free read-out from the sensor array is achieved, in that the read-out is digital.

This embodiment of the invention yields a system providing

1) A/D conversion on pixel level, giving a digital, i.e. noise-free, read-out from the pixel, 2) noise reduction by integration on pixel level, and

3) holding down the capacitance of the capacitor and thus the surface required on the chip by using repeated charging and discharging in the oscillator. In a second embodiment of the invention, each group of detectors has allocated to it one or more non-radiation-sensitive supplementary components. A network including switches is adapted to successively connect different detectors in the group to the supplementary component or components so as to form the oscillator. Suitably, such a group consists of one column in the array.

Using only one oscillator per column instead of one oscillator per pixel involves both advantages and disadvantages.

In the following, we assume that use is made of a detector array having NxN resistive or capactive detectors. The electric integration then takes place only during an N:th part of the image period. The noise suppression of the oscillator and detector therefore decreases by a factor /N as compared with the integration during the entire image period. The advantages conferred should be considered in the light hereof.

Using only one oscillator per column instead of one in each pixel, each oscillator can be allowed to occupy a considerably larger chip area and also to consume N times as high a power. This can be used for designing oscillators of lower phase noise, since larger surfaces and bias currents for critical transistors can be used as well as a higher capacitance value in the RC product. Moreover, it is possible to obtain a lower phase noise by integrating in the device a lower sensitivity to noise on the supply voltage generated, inter alia, by other oscillators, but also by integrating more efficient compensation of the 1/f noise of the oscillator.

The effects of the 1/f-noise of the oscillator that is below the image frequency can however also be compensated for along with the thermal compensation of the silicon chip. This takes place if a row of separate radiation-protected resistors at the upper and lower edges of the sensor array is used as references at each image. The frequency change occurring columnwise because of the 1/f-noise of the respective oscillator cannot then be distinguished from temperature variations and will therefore be compensated for in the same way.

Improved separation between the oscillators is also possible, in that the supply and bias lines may be amply provided. Besides, N times as few oscillators will be connected simultaneously, for which reason mutual influence should decrease drastically. One advantage of the frequency-to-digital conversion is achieved by the oscillator being permanently connected to the following logics. Conversion then becomes more reliable, since the risk of miscalculation of the number of cycles is eliminated.

One example of an IR sensor having NxN resistive detectors is shown in Fig. 3. Each pixel contains a resistive detector element, R, and two switches for contacting the detector resistance to the oscillator. Each pixel also comprises two vertical buses and one horizontal address line connecting the detector resistance to the oscillator via the switches . The connected detector resistance is included, together with a capacitance, C, as a frequency-determining component in the RC oscillator. In this embodiment, reference detectors are also included in the upper and lower edges of the sensor array. These are shielded against IR radiation and react only to the inherent temperature of the silicon, such that a change thereof can be compensated off from the signals from the other elements. The location in the upper and the lower edges also enables compensation of a linear temperature gra¬ dient over the detector array.

The RC oscillators are connected for one detector row at a time. Let 1/T_D be the image frequency. The reading time T| for each detector row then becomes T__*JN if the two reference detectors are neglected. With an oscillator frequency, f₀, then about 2f₀T_| half cycles will be recorded per read-out. Normally, only about 100-1000 cycles will be involved, which in itself gives too low a resolution. On the basis of a well-defined starting point by resetting the oscillator at the beginning of the read-out period, a read-out of the final value of the oscillator gives a yet higher resolution. Fig. 4 illustrates this procedure. A dynamic range of about 15 bits should be possible.

To be able to use the entire line time for reading out the detector, pipelining must be used at the A/D conversion of the final value of the oscillator. This can be so carried out that the final value of the oscillator is sampled and held, it being possible to carry out the A D conversion in parallel with the read-out of the next row. It should be noted that since the charging and discharging procedure is non-linear, correction must be made therefor .

Claims

Claims:

1. A sensor device comprising an array of detectors, c h a r a c t e r i z e d in that each detector consists of a radiation-sensitive component included in an oscillator, for example a resistor or a capacitor included in an RC oscillator, whose frequency is used as a measure of the detected quantity.

2. A sensor device as claimed in claim 1, c h a r a c t e r i z e d in that the detector signals are to be read out in parallel within groups of detectors and serially group by group.

3. A sensor device as claimed in claim 2, c h a r a c t e r i z e d in that said groups consist of one or more rows of detectors.

4. A sensor device as claimed in any one of claims 1-3, c h a r a c t e r ¬ i z e d in that to each detector are permanently allocated one or more non- radiation- sensitive supplementary components which, together with the detector, are adapted to form the oscillator.

5. A sensor device as claimed in claim 4, c h a r a c t e r i z e d in that it comprises a control and calculating device which successively calculates and reads out the oscillator signals group by group in parallel by measuring with high resolution, during an observation window which extends to the time of the read-out of the next group, the time to the first positive edge of the oscillator signal and the time between two consecutive edges, and calculating on the basis of the latter the number of pulses since the preceding observation window so as to obtain an integrated measure from the preceding observation window.

6. A sensor device as claimed in any one of claims 1-3, c a r a c t e r- i z e d in that to each group of detectors are allocated one or more non-radiation- sensitive supplementary components and that there is provided a network including switches, which is adapted to successively connect the different detectors in the group to the supplementary component or components so as to form the oscillator.

7. A sensor device as claimed in claim 6, c h a r a c t e r i z e d in that it comprises a control and calculating device which successively connects the detector elements and reads out the oscillator signals group by group in parallel, which is done, during the connecting time, by said control and calculating device first resetting the oscillator, then counting the number of half periods and finally recording the analogue value of the oscillator at the end of the connecting time.

8. A sensor device as claimed in any one of the preceding claims, c h a r a c - t e r i z e d in that there are provided one or more radiation-protected oscillators which are adapted to produce a reference signal which is used for compensating for temperature drift.

9. A sensor device as claimed in claim 8, c h a r a c t e r i z e d in that there are provided one or more radiation-protected oscillators arranged on each side of the detector array, as seen in one direction, and adapted to provide a reference signal used for compensating for a linear temperature gradient over the detector.