The Space-Based Visible Program: Appendix I SBV Hardware
Introduction
Introduction
The sbv sensor hardware, shown in Figure 1, was designed and integrated at Lincoln Laboratory, and a number of the components were fabricated at the Laboratory in the late 1980s under stringent cleanliness and functionality requirements. This appendix provides a detailed look at the SBV sensor hardware. A companion appendix on system integration and testing provides details on how the SBV sensor was tested and calibrated to achieve the impressive capability demonstrated on orbit. (Note: these two appendices contain updated versions of material that was originally published in the Johns Hopkins APL Technical Digest [1].)
FIGURE 1. SBV sensor hardware. The telescope assembly was built by SSG Inc., Waltham, Massachusetts, and the focal-plane array, the CCD camera, the signal processor, and other components of the SBV sensor were fabricated at Lincoln Laboratory. Integration, testing, and calibration of the SBV sensor were performed in the Engineering division and the Aerospace division at Lincoln Laboratory.
Figure 2 shows a diagram of the components of the SBV sensor [1]. The sensor is located in two zones-the instrument section and the electronics side assembly. The telescope and focal-plane assembly, the telescope aperture cover and controller, and the CCD camera are in the instrument section, which is located on the forward section of the spacecraft. The signal processor, the experiment controller, the power conditioner, and the telemetry interface to the spacecraft are in a single assembly called the electronics side assembly, which is located in the aft section of the spacecraft. Figure 3 illustrates the relative locations of the components of the SBV sensor [1-3].
FIGURE 2. Components of the SBV sensor hardware are separated into two zones. The telescope and analog processing electronics are in the instrument section on the top of the MSX satellite, along with the other sensors, while the signal processor, experiment controller, telemetry interface module, and power conditioner are in the electronics side assembly in the rear of the MSX. Reprinted from the Johns Hopkins APL Technical Digest by permission [1].
FIGURE 3. Relative location of the SBV sensor elements. The electronics of the SBV sensor and those of the other sensors were separated from the telescope and detector assemblies to prevent excessive thermal loading on the SPIRIT III infrared detectors. The SPIRIT III, an infrared imaging telescope, was the primary sensor on board the MSX satellite, and its focal planes had to be maintained at a temperature of 8�K. Reprinted from the Johns Hopkins APL Technical Digest by permission [1].
Telescope
The telescope was developed by SSG, Inc., Waltham, Massachusetts. It employs a three-mirror anastigmat, off-axis reimaging design to maximize stray-light rejection of bright sources such as the sunlit earth, which may be just outside the field of view of the telescope. This design configuration introduces well-defined spatial distortions, which are considered a tradeoff for good focusing characteristics [4].
The total field of view of the telescope, including distortions, is around 1.4� x 6.6� with four CCDs in the focal plane. Each pixel has a near-square field of view of approximately 60 mrad on a side; local distortions vary the size of the pixel field of view from the nominal. An aperture cover, opened and closed on command, is used for contamination control, thus preventing dirt accumulation during launch and on orbit when the SBV sensor is not being used. The cover has a secondary mechanism that can open the cover permanently if the normal mechanism fails.
Focus quality and stability became an issue early in the development and construction of the SBV sensor. Initial results describing problems with the Hubble telescope were published just before the SBV sensor's Critical Design Review. To maintain focus independent of operating temperature, the telescope housing and mirrors were made of the same type of aluminum, resulting in an athermal system. All elements expand or shrink at the same rate, keeping the focal plane in a fixed location. Plate scale and distortion maps are expected to change with temperature, and computer modeling indicated that it was important to keep temperature gradients low, thus making athermal assumptions valid. Thick walls, thick mirrors, thermal isolation from the spacecraft mounting surface, and a multilayer insulation blanket were incorporated into the design to keep gradients well below the design allowable limits of 3.5�C. Focus is specified as an ensquared energy percentage, which is the percent of total energy in the central pixel after optimum centering. The ensquared energy limit was set to be no less than 50% to maximize detection probability and no more than 80% to allow ground analysis of star centroids for subpixel pointing determination by a slight oversampling of point sources.
A significant challenge for the SBV program is to keep the telescope optics clean up to, during, and after launch. The first two mirrors are superpolished with gold surfaces to maintain low scatter characteristics. Our process to maintain cleanliness covers three areas-design, handling, and operation. The design of the telescope eliminates penetrating holes inside the optical cavity. On the ground, the telescope aperture cover was opened only in a Class-100 clean tent or a vacuum chamber. A continuous dry nitrogen purge was maintained, even during shipping, until just before launch [5].
Focal Plane
The SBV focal plane is comprised of a 1 x 4 array of Lincoln Laboratory-fabricated frame-transfer visible-light CCD imagers, which were specifically developed for space surveillance. The CCD imagers are mounted on a ceramic substrate for electrical interconnects; the substrate is then bonded to a Kovar tray (Kovar is an iron alloy commonly used in semiconductor packages). A thermoelectric cooler controls the temperature of the focal plane. The focal plane is read out in a sequential mode, one imager at a time, at a pixel rate of 0.5 MHz. Each single observation sequence requires that a full frameset be taken with a single CCD before switching to the next CCD for data acquisition. The focal-plane wiring is designed to prevent a single CCD failure from affecting more than half the focal plane; any single failure allows the two imagers on the other half of the focal plane to continue operating normally [6-8].
Figure 4 shows a diagram of a single SBV CCD imager [1]. The CCD layout was designed for three-side butting such that a 2 x N focal-plane array could be fabricated. Each CCD has a 420 x 422-pixel imaging area and a 420 x 422-pixel storage area. The additional two lines in the image and storage areas allow for the possibility of some aluminum light-shield misalignment, and they reduce red diffusion effects of charge into the top of the storage area.
FIGURE 4. Block diagram of the CCD imager. Fabricated in the Solid State division at Lincoln Laboratory, each of the four CCD detectors in the SBV sensor is a 420 x 422-pixel, front-illuminated, frame-transfer imaging device, maintained on orbit at a temperature of -40�C by a thermoelectric cooler. Reprinted from the Johns Hopkins APL Technical Digest by permission [1].
The CCD imagers are cooled actively. Waste heat is radiated to space by a radiator on (but thermally isolated from) the telescope body. The radiator is maintained at a minimum temperature of -43�C by survival heaters. A thermoelectric cooler is attached to the focal-plane Kovar tray through a flexible multilayer aluminum strap. The thermoelectric cooler is driven by an electronics servo system to maintain a maximum temperature of -40�C during operation. At this low temperature, dark current is on the order of eighteen electrons per pixel per second.
CCD Camera
The camera selects the operating CCD, generates CCD focal-plane clocks and biases, coordinates timing with the other SBV elements, and provides a low-noise twelve-bit digitized readout. The camera uses a fully redundant, dual-channel architecture for fault tolerance, only one channel of which is powered at a time. The camera has two commandable gains and up to five different integration times. The gains are nominally set to provide six and twenty-five electrons per least significant bit of the twelve-bit analog-to-digital converters for full-scale responses of about 24,000 and 100,000 electrons, respectively.
Two integration times (0.4 and 1.6 sec) are used during space-surveillance operations, and three other integration times (0.625, 1.0, and 3.125 sec) are used for target-track and background-phenomena measurements, requiring the MSX tape recorder to operate. The camera can gate its output and synchronize to an external pulse, allowing accurate precise time tagging of space-surveillance integration times.
Signal Processor
The signal processor hardware can accept from two to sixteen data frames from the camera; each frame consists of about two megabits of information. Using this data, the signal processor automatically detects targets and reduces the data flow to a few kilobits per frameset, effectively making a data compression ratio of greater than 1000:1. The signal processor has two redundant channels, each of which uses a Motorola DSP56001 digital signal processor operating at 20 MHz as its processing engine. The algorithmic core of the signal processor is centered around an assumed velocity filter that employs maximum-likelihood estimation for clutter rejection and automatic target detection. A one-bit binary implementation of this algorithm reduces the computational load while maintaining as much performance as possible from a full twelve-bit binary assumed velocity filter. The signal processor algorithm normally operates with the spacecraft stabilized in inertial space, with stars stationary, and with the images of moving objects forming streaks in the focal plane. This mode is referred to as sidereal track. The signal processor automatically detects these streaks and generates target reports consisting of position and velocity estimates in focal-plane coordinates. A target signature, using twelve-bit camera information, can also be included in the target report if commanded. The signal processor can also save a commandable number of the brightest stars' twelve-bit data for telemetering to the ground. The star data are used to refine the pointing knowledge of the SBV sensor.
The detection algorithm is run in two stages. While the camera is transmitting data, the signal processor has a real-time stage. Data are stored in three arrays-an average of all frames, the peak value of the frameset, and the second-highest peak value of the frameset. Frame number of occurrence is stored along with the values in the peak value and second-highest peak value arrays. When the data set is completed, the signal processor enters its second stage, in which stars are selected and the main part of the detection algorithm is run. Data from the second-highest value are used to estimate the variance of a pixel. After subtraction of the average frame data from the peak data, the result is divided by the variance estimate array. This new array has higher values for pixels that saw a changing scene during the frameset, corresponding to anything that had motion relative to the stars. The values are thresholded, which provides the single-bit array for the detection process.
Support Electronics
The support electronics comprise an experiment controller, a power subsystem, and a telemetry interface module. The experiment controller is based on a mDACS (micro-packaged data and control system) computer from SCI Inc., of Huntsville, Alabama. This controller is a Harris 80RH86 microprocessor-based computer with a high degree of fault tolerance and error correction. All control busses are triply redundant with majority logic, and the address and data busses have single-bit error-correction syndrome bits. The bus redundancy and error correction allows the mDACS to operate with no loss of speed, functionality, or capacity after the failure of any internal signal line. The experiment controller is used to read commands uplinked from the ground through the MSX spacecraft and to convert them to the set of commands to operate the SBV sensor. Under ground command, the experiment controller is used to configure all the other redundant electrical units of the SBV sensor by power and signal switching, and to provide health and status messages to the telemetry. Memory inside the experiment controller permits storage of many signal processor results, allowing the SBV sensor to gather and store observation results by using neither the MSX tape recorders nor contact with a ground station. When the spacecraft is over a ground receiving site, the MSX telemetry system relays the stored data to the ground.
The power subsystem was built for Lincoln Laboratory by Gulton Data Systems, Albuquerque, New Mexico. It provides conditioned, isolated secondary power for all units except the experiment controller, which has an internal power conditioner. It also switches primary and secondary power under experiment-controller control for fault recovery.
The telemetry interface module, designed and built at Lincoln Laboratory, is a first-in first-out unit capable of running up to the maximum instantaneous bit rate of 25 Mbits/sec. It is used to interleave the SBV telemetry into the MSX telemetry stream. The telemetry interface module interfaces to a redundant MSX telemetry system with a complete independent interface circuit assigned to each spacecraft system. The high bit rates used in the MSX telemetry system would have made a cross-strapping scheme large and power hungry, which was inconsistent with the weight and power budgets for the SBV sensor.
Next Topic: Appendix II - Calibration and Testing
Return to SBV Table of Contents