GB2241849A - A detector array - Google Patents

Abstract

A detector array comprises a plurality of detectors 1, for example pyroelectric detectors, each detector 1 being independently susceptible to offset voltage in dependence upon age and temperature cycling. The differential offset voltages of the detectors 1 are substantially equalised using a filter, for example an infinite impulse response (IIR) filter 7. A memory store 11 and approximation processor means are provided to reduce the dynamic range of differential offset voltages between detectors of the array. <IMAGE>

Description

DEFLECTOR ARRAYS The present invention relates to detector arrays and more particularly to equalisation of detector differential offset across detector arrays.

Detectors, particularly pyroelectric detectors, are arranged into detector arrays with each detector in an array being interrogated in a timed sequence using a multiplexer.

Multiplexed pyroelectric detector arrays suffer from the problem of differential detector offsets. Each of the detector elements is amplified by a separate integrated amplifier, and the variations in DC bias conditions of the amplifier circuitry cause each channel to have a different voltage offset at the output of the device.

Typically, the differential offsets are an order of magnitude greater than the required signal from the detector. A further problem is that the bias conditions of each integrated amplifier drifts with temperature and time causing a corresponding drift in the voltage output of each channel.

It is known to reduce differential offset by "calibrating" each detector channel to determine an associated DC offset for that channel. These calibration DC offset values are then stored for example in a programmable read only memory (PROM). During operation of the detector array, as detector data is multiplexed out of each detector channel the associated calibration DC offset value is subtracted from it thus, the effects of differential offset are equalised across the detector array. This is a "rough" offset removal technique as dynamic offset drift due to age etc. is not compensated for.

There are two well known approaches to detector data calibration offset value subtraction. Firstly, a large word-length analogue to digital converter (ADC) may be used to digitise the data stream and then subtract by digital processing. A second and preferred approach is to use a digital to analogue converter (DAC) to convert stored calibration offset values to analogue signals that can be subtracted from the detector data before digitisation. The second method is preferable as offset removal before digitisation reduces the dynamic range of data signals thus allowing a lower precision, and thus cheaper, ADC to be used.

As stated previously, the rough offset removal approach to equalisation of differential offsets can not compensate for drift in detector data. Consequently, there is an elaboration on the subtraction approach to periodically measure the differential offsets while the equipment is operational. These periodic offset measurements, rather than fixed values stored in a PROM, are used to correct the detector differential offsets. Since the periodic offset measurements are made while the equipment is operational, the method compensates for time and temperature dependent offset drift. The disadvantages of this method are two fold.Firstly, data can not be acquired from the sensor during the periodic offset measurement phases, and secondly the method is only applicable where an opto-mechanical shutter is used to provide detector signal modulation and so prevent radiation from the imaged scene impinging on the detector during offset measurement. The requirement for a mechanical sub-assembly increases the cost and reduces the reliability of the system.

The transfer function of the detector is expressed as the voltage responsivity (Rv), the ratio of output signal voltage to input radiant flux. Voltage responsivity is frequency dependant, with a frequency response of the form shown in figure 1, frequencies fl and f2 are typically between 0.1 and 10 Hertz. The important feature to note is that pyroelectric detectors have zero response at DC and so only respond to changes in radiant flux. Thus, it is usual to modulate the incident radiation with an optomechanical chopper before it reaches the detector and demodulate the detector output to reconstruct the true scene data using an image difference processing (IDP) scheme.

The signal due to the incident radiant flux is superimposed on a large offset voltage. For instance a single element detector looking at a modulated temperature source may produce an output similar to figure 2. The peak to peak voltage (AVt) due to temperature variation in the scene is typically less than a tenth of the offset voltage (Vo). The offset voltage tends to drift with variations in ambient temperature, so cannot effectively be removed by a onceonly calibration such as adjustment of a potentiometer or storage of a correction factor in a look-up table.

Single element detectors and small detector arrays are available as direct readout devices - i.e. each detector output is fed to a different output pin. For these devices a separate analogue filter may be used to process each detector output to remove the offsets and drift.

It is impractical to make direct readout detector arrays with large numbers of elements as the large number of wire bonds reduces the yield and reliability. Larger arrays use an on-chip multiplexer to feed the data from each element in turn out of a single output pin. A typical output waveform is shown in figure 3. The offsets vary from one detector element to another. The average of all the element offsets is referred to as the mean offset and the difference of each elements offset from this value as the element's differential offset. High pass filtering the multiplexed data stream removes the mean offset but the differential offsets remain as multiplexing is a modulation process that shifts the low frequency offsets and drift to higher frequencies that are not separable from the signal frequencies using continuous time analogue filtering.

It is an objective of the present invention to provide a detector array wherein the above problems of offset equalisation across the detector array are substantially reduced.

According to the present invention there is provided a detector arrangement comprising a plurality of self-modulating detectors arranged to form a detector array self-modulated by movement, each detector being coupled to rough offset subtraction means and signal filter means, the rough offset means being arranged to substantially equalise each detector's inherent detector output signal offset whilst the filter means is arranged to homogenise cyclic drift in respective detector output signals with time.

An embodiment of the present invention will now be described by way of example only with reference to the accompanying drawings, in which: Figure 4 is a schematic illustration of a detector array according to the present invention; and, Figure 5 is a schematic illustration of an infinite impulse response (IIR) filter for use in the detector array of Figure 4.

When an imager using a pyroelectric detector is moving relative to the scene the motion may provide sufficient modulation of the incident radiation without using an optomechanical chopper.

When this is the case it is highly desirable to implement the system without a chopper as any mechanical components decrease the system's reliability. When a multiplexed detector array is used the differential offsets can not be removed by analogue filtering, as above. As digital filtering is carried out on sampled data streams it is applicable to multiplexed data. The multiplexed data stream contains samples of each detector element output. By repeating the filter processing for each sample the outputs of all detector elements are individually filtered to remove the offsets. As data from the detector elements occur sequentially in the multiplexed data stream the filtering may be carried out by a single hardware unit operating on each stream in turn, or by an iterative software module.

A simple digital filter structure suitable for offset removal is a first order high pass IIR (infinite impulse response) filter, although other filter types may be used to provide additional signal conditioning.

It is a property of most detectors that their responsivity is dependent upon frequency, for example pyroelectric detectors exhibit a low frequency cut-off, typically between 1 Hz and 50Hz, below which there is negligible signal response. Offset drift due to temperature and time tend to have a very low frequency thus a high pass frequency filter with a cut-off frequency below the minimum signal frequency can be used to remove offset drift from the required data signals.

In the present detector array an infinite impulse response (IIR) digital filter is used. The data from each detector is digitised by an ADC and forwarded to processor apparatus.

Referring to Figure 4, a group of detectors 1 are arranged into a detector array. Signals from the detectors 1 are multiplexed by an analogue multiplexer 5. A rough offset removal arrangement comprising a memory 11 and an analogue-to-digital converter (ADC) 10 is used to provide calibration offset values to a subtractor (difference amplifier) 15. The subtractor 15 subtracts these calibration offset values from each respective detector signal. The memory 11 stores a distinctive offset value for each detector 1 but cannot be adapted to adjust for dynamic drift of that detector's offset value. An ADC 3 converts the output from the subtractor 15 to a sequential stream of digital words.An infinite impulse response (IIR) filter 7 is arranged to receive this stream of digitised data and processes each detectors data sequentially to remove the offsets and drift remaining after rough offset removal. The operation and function of IIR filters is explained in "An introduction to the Analysis and Processing of Signals", Paul A. Lynn (MacMillan Education Limited) 2nd Edition, pages 206 to 212 and 232.

The IIR filter 7 has a frequency response such that time and temperature drift between detector offset differentials are equalised.

Thus, for example, low frequency time and temperature drift with pyroelectric detectors is removed using a high pass frequency IIR filter. The effects of detector amplifier offset drift is thus equalised across the array in the data stream input to a digital processor 9.

Figure 5 illustrates schematically the structure of a high pass frequency IIR filter suitable for a pyroelectric detector array. The values given below are for a detector array sampled at a frequency fs. A cut-off frequency fo is chosen for the filter.

SAMPLING FREQUENCY = fs SAMPLING PERIOD = 1/fs 3dB CUT OFF FREQUENCY = fo T = 1/2 it fo T/Ts (A) 1+(T/Ts) TRANSFER FUNCTION (Z - DOMAIN) = H(Z) =

The IIR filter 7 comprises a first digital adder 21, a second digital adder 23, a multiplier 25 and a delay unit 27. The delay unit 27 is provided by storing the calculated values of U(Z) for each channel in a RAM and reading them back after a pre-defined time period (the array sampling period). The read/write line and address lines of the RAM are generated by a digital control unit. The RAM address increments in synchrony with the multiplexed input data and the number of addresses accessed is equal to the number of channels multiplexed.

The multiplier a is given from the above equation (A) in the embodiment illustrated. The effect of the IIR filter 7 is thus to process a detector digital input word X(Z) by the above equation (B) to provide a respective detector array equalised digital word Y(Z). The filter 7 therefore provides identical high pass filtering on each channel of a detector array multiplexed data stream, processing each channel in turn. The number of channels processed is limited only by the size of the delay unit 27, RAM and the processor speed.

Depending on the speed and accuracy required, the system may be implemented in dedicated hardware, or as software in a microprocessor.

It will be appreciated that because the offset and drift are very low frequency signals and the detector only responds to alternating signals that, according to the present invention, drift may be removed from the required signal by high pass filtering the detector output.

Claims (7)

1. A detector arrangement comprising a plurality of detectors arranged to form a detector array self-modulated by movement, each detector being coupled to rough offset subtraction means and signal filter means, the rough offset means being arranged to substantially equalise each detector's inherent detector output signal offset whilst the filter means is arranged to homogenise cyclic drift in respective detector output signals with time.

2. A detector arrangement as claimed in claim 1 wherein the filter means is an infinite impulse response (IIR) filter arrangement.

3. A detector arrangement as claimed in claims 1 or 2 wherein the filter means is a high pass filter.

4. A detector arrangement as claimed in claims 1 or 2 wherein the filter means is a band pass filter.

5. A detector arrangement is claimed in any preceding claim wherein the rough offset means includes a memory store arranged to store a plurality of digital words indicative of respective detector offset values and, a digital-to-analogue converter to convert each digital word into an analogue signal for subtraction from a respective detector signal.

6. A detector arrangement as claimed in any preceding claim wherein the detectors are pyroelectric detectors.

7. A detector arrangement substantially as hereinbefore described with reference to the accompanying drawings.