IR detector system and method

The invention relates to an Infra Red (IR) detector system and method. More specifically, but not exclusively, it relates to a system and method that changes the configuration of elements within the pixel to reduce the time delay between stare times for successive fields of data, reducing image tear and smear when operating in enhanced sensitivity or binning mode.

The performance of conventional infrared staring array detectors can be enhanced by using the widely known technique of pixel binning. Pixel integration capacitors are shared or binned in a group and are then connected in turn to each photodiode detector element in the group to increase the integrated signal charge storage capacity and therefore improve thermal sensitivity. Thermal sensitivity increases by a factor √n where n is the number of binned pixels in the group and can be seen to be advantageous. A field of image data is acquired and read out for each element that when taken together allows a full image to be composed.

In operation, the integration time increases by factor n which increases the frame period. More significantly, the separation in time between fields of image data by the field readout time leads to undesirable imaging artefacts such as multiple images for moving targets and image blur that degrades image quality and limits performance for such an enhanced sensitivity mode in the sensor. Integration and readout operation for conventional Integrate then Read (ITR) and Binning modes are illustrated in FIGS. 1( a ) and 1 ( b ).

In some cases, usually, but not exclusively, in Medium Waveband Infrared (MWIR) applications, the integration capacitor may be split into a number of smaller capacitor elements with connecting gates to accommodate a range of signal conditions or for low flux applications.

In other cases, a sample and hold (s/h) circuit is used of a form shown in FIG. 2 b . It uses a pass transistor (M 2 ), reset transistor (M 6 ) and hold capacitor (Ch) to allow the pixel circuit to operate in so called Integrate While Read (IWR). The signal is copied onto Ch that has been previously reset by operating the pass transistor M 2 . When used, this allows photodiode signal integration to be accomplished in parallel with the read operation to enable longer integration times to be achieved within the frame period.

In yet other cases, usually, but not exclusively in Long Waveband Infrared (LWIR) applications having large flux, the best sensitivity can be achieved using the largest possible integration capacitor. Therefore LWIR sensors are less likely to use split integration capacitor designs. Since the stare time may be very short for a LWIR sensor, the sensitivity can be enhanced by sharing integration capacitors between pixels. This can be achieved using the binning technique by sharing the integration capacitors for example, between pixels from adjacent odd and even rows by using a transistor pass gate between the capacitors to choose whether or not to share, i.e. switch between conventional and binning modes of operation. Other binning configurations using neighbouring column pixels or other pixel group formations are equally possible. Sample and hold capacitors are not normally required or used for LWIR operation because integration times are generally short enough that when added to the readout time, are still less than the frame period. In fact, providing a s/h capacitor detracts from the maximum possible integration capacitance in a pixel design and is usually avoided.

In MWIR sensors the opposite problem can occur, where the flux is sufficiently low that in order to make full use of the available capacitance the stare time is longer than the frame time less the time required to read out the data from the array. In this case a sample and hold capacitor is added to allow integration to occur whilst the sensor is simultaneously reading out the data from the previous image frame. This approach has the disadvantage that the s/h capacitor detracts from the integration capacitance in the pixel and compromises the ultimately achievable sensitivity.

According to the invention there is provided an infrared detector system comprising a series of pixels and further comprising enhanced sensitivity circuit means, said circuit means being configured so as to operate in an enhanced sensitivity mode comprising a combination of independently operable sample and hold circuit means and binning function means, said combination acting so as to reduce undesirable imaging artefacts.

According to the invention there is further provided a method of reducing undesirable imaging artefacts in IR detector systems comprising the steps of:

a) binning together pixels in pixel groups by asserting respective binning control(s) of said pixels b) Configuring said pixel group so the first pixel in a group sample and hold circuit is turned on (opened) such that when operated will integrate a signal into a combined integration capacitance on a first sample and hold capacitor, and will not integrate into the other pixels in the groups' sample and hold capacitor(s) c) Operating the first pixel in the group so as to reset the combined integration capacitance d) Operating the first pixel in the group such that a signal is acquired onto the combined integration capacitance e) Turning off (closing) the sample and hold circuit for the first pixel in the group f) Configuring the pixel group so the second pixel in the group sample and hold circuit is turned on (opened) such that when operated will integrate a signal into the combined integration capacitance on a second sample and hold capacitor, and will not integrate into the other pixels in the groups' sample and hold capacitor(s) g) Retaining the signal for the first pixel in the group on said first sample and hold capacitor. h) Repeating steps (c) to (g) for the second and all other pixels in the group

such that the time between collecting the signal information for each pixel in the group is minimised thereby reducing the effects of undesirable imaging artefacts on an output image.

In this manner, an Enhanced Binning mode is proposed that combines existing circuit elements with a new configuration and operational technique to minimise the temporal separation between fields to reduce the effects of such undesirable imaging artefacts, to improve the use of enhanced sensitivity mode operation and so improve performance in this mode. In this way it changes the configuration of elements within the pixel to reduce the time delay between stare times for successive fields of data, reducing image tear and smear.

The invention will now be described with reference to the accompanying diagrammatic drawings in which:

FIG. 1 is a schematic drawing showing the frame readout sequence and timing (in ms) for a full TV format array (640×512, 4 outputs); a) top—ITR; b) middle—ITR binning; c) bottom—Enhanced binning, in accordance with one form of the invention. Note that controls applied to configure each operation are omitted for clarity;

FIG. 2 is a schematic drawing showing example prior art pixel circuits (a) Direct Inject Gate (DIG) circuit, and (b) DIG circuit with IWR function; and

FIG. 3 is a schematic drawing showing pixel circuits having a 2-way binning connection between odd and even rows.

In one embodiment of the invention, a 2-way binning case is shown in FIG. 3 and is used to explain the enhanced binning operation in general terms. In this example, integration capacitors in adjacent odd and even rows are binned (n=2).

The Read Out Integrated Circuit (ROIC) is configured for operation in binning mode by asserting BIN to turn transistor M 7 on. Integration capacitor Cint 1 is therefore electrically connected to integration capacitor Cint 2 . In this embodiment and for discussion purposes, Cint 12 =Cint 1 +Cint 2 . It will be appreciated that these capacitance terms include any contribution from parasitic terms present on the respective nets.

The pixel array is configured to acquire image signal data from the array's odd rows. VG 21 is asserted to turn on the sample and hold transistor M 21 to connect the odd pixel s/h capacitor Ch 1 to the conjoined integration capacitance Cint. VG 22 is retracted to turn off the transistor M 22 to isolate the even pixel s/h capacitor Ch 2 . The combined integration capacitance is Cint=Cint 12 +Ch 1 =Cint 1 +Cint 2 +Ch 1 which is reset using any of the reset transistors M 51 , M 52 , M 61 by asserting RS 1 , RS 2 or RS 21 respectively.

The integration period is started for the odd rows by biasing the DIG transistor M 11 . During the integration period the signal is integrated onto the total combined integration capacitance Cint. After the conclusion of the odd row integration period, VG 21 is retracted to turn the s/h transistor M 21 off and the odd pixel signal voltage remains sampled on the s/h capacitor Ch 1 .

The pixel array is then configured to acquire image signal data from the array's even rows.

VG 22 is asserted to turn on the sample and hold transistor M 22 to connect the even pixel s/h capacitor Ch 2 to the conjoined integration capacitance Cint. VG 21 remains retracted to ensure transistor M 21 remains off to isolate the signal on the odd pixel s/h capacitor Ch 1 . The combined integration capacitance is Cint=Cint 12 +Ch 2 =Cint 1 +Cint 2 +Ch 2 which is reset using any of the reset transistors M 52 , M 51 , M 62 by asserting RS 2 , RS 1 or RS 22 respectively.

The integration period is started for the even rows by biasing the DIG transistor M 12 . During the integration period the signal is integrated onto the total combined integration capacitance Cint. After the conclusion of the even row integration period, VG 22 is retracted to turn the s/h transistor M 22 off and the even pixel signal voltage remains sampled on the s/h capacitor Ch 2 .

This approach offers advantages over the conventional approach to operation in binning mode. Most significantly, the image field capture periods are separated only by the stare time for the first (odd rows) field and the reset period required for the second (even rows) field. For comparison see FIG. 1( b ) and FIG. 1( c ). In practice, a long waveband detector operating in enhanced binning mode, could be expected to have an integration separation time of around 0.65 ms (0.6 ms integration time and 0.05 ms, say, reset time). Comparing this to operation in conventional binning mode, the integration separation time is dominated by the readout time which is expected to be around 4.9 ms for a full TV (640×512) array, say, with 4 outputs being clocked at 10 MHz and considerably more at slower clock speeds. Operating in enhanced binning mode is therefore expected to reduce the impact of image artefacts such as image smear or multiple images of fast moving targets by a factor proportional to the reduction in integration separation time of 4.9/0.65≅7.5.

It will be appreciated that the sample and hold circuit is shown in its simplest form. In this form, the transistor gate terminal control signals (i.e. V 21 and V 22 in the case of a 2 way binning pixel group (n=2)) may be implemented using alternative configurations such as two transistors in series. This approach is increasingly advantageous for larger pixel groups allowing a control matrix track structure to be implemented more easily than by separate controls to each pixel in the group. The novelty of the invention is unaffected by the realisation of the sample and hold circuit implementation.

It will be appreciated that another advantage of operation in enhanced binning mode is that the readout sequence becomes identical to normal ITR or blinking mode.

It will also be appreciated, that operation in non binning mode is made possible by retaining all appropriate circuit elements. In this configuration, BIN is retracted turning M 7 off. VG 21 and VG 22 are asserted turning s/h transistors M 21 and M 22 on respectively connecting s/h capacitors Ch 1 and Ch 2 to their respective integration capacitors Cint 1 and Cint 2 thereby maximising the integration capacitance and thermal sensitivity for LWIR use for example.

It will also be appreciated that this technique is valid for all other pixel circuit designs using a binning function and is not limited in any way to the Direct Inject circuit form used herein to describe this innovation.

It will also be appreciated that whereas this technique has been explained in detail for the case of 2 way binning (n=2), the technique is equally applicable to other binning configurations or solutions.

In practice, the imaging artefacts associated with conventional binning limit the amount of binning that can be used. This technique described in one form of the invention enables the user to increase the amount of binning with associated improvement in sensitivity and reduction in unwanted imaging artefacts.