Improving Electrical Ground Support Equipment Development for Satellite Testing

James Canoy Groves Resources Solutions Incorporated

The Johns Hopkins University Applied Physics Laboratory 11100 Johns Hopkins Road

Laurel, MD 20723 James.Canoy@jhuapl.edu

Abstract This paper examines the basic test method employed with Electrical Ground Support Equipment; evaluates the technical merits associated with enhanced capabilities and compares the relative cost of these enhancements. In addition, the paper explores the pros and cons of replacing current manual operations with integrating automated test methods. This study is not intended to be a precise cost analysis but is intended to provide a comparison of technical enhancements along with a basic measure of cost. This paper will provide some practical examples to support the premise that automated test methods give rise to a more thoroughly tested spacecraft, reduced test times and potential cost savings.

I. INTRODUCTION The standard spacecraft model has continued to evolve into more complex systems. The mission planners continue to extend the spacecraft mission durations and survive in increasingly hostile environments. These expectations have driven the design engineers to develop more complex and robust systems. In addition to these expectations, spacecraft manufacturers are also constantly evaluating methods to reduce manufacturing costs. This has led engineers to look for creative manufacturing strategies to achieve the technical requirements while streamlining the manufacturing process. Technological innovation and cost requirements have driven the advancements in spacecraft design requirements and techniques. However, Electrical Ground Support Equipment (EGSE) has not kept pace with this development curve. EGSE remains the most essential element necessary to facilitate the manufacturing process. An example of a typical EGSE test rack is shown in Figure 1 and a block diagram is shown in Figure 2. While EGSE continues to

incorporate proven methods, these methods do not always leverage more technical improvements nor are spacecraft manufacturers encouraged to invest in EGSE development. In order for EGSE to achieve technical equity, the EGSE architectural philosophy needs to evolve to incorporate fully autonomous applications built on reconfigurable platforms. By incorporating these features, the EGSE platform can provide the benefits of redundancy to increase robust test and simulation activities, enhanced repeatability, reduced manufacturing labor costs, and reduced manufacturing time. To streamline EGSE development, developers need to modify the current EGSE model to more closely replicate the spacecraft model. In addition to a standardized modular approach, the EGSE suite should contain autonomous components which are controlled by a primary and secondary source for command & control, data collection and data processing. Each autonomous component should be capable of servicing a task to completion without the need for external commanding. This paper presents a new strategy to ensure that telemetry data could be collected and transmitted to the EGSE computers in a real-time manner and accumulated for transmission at the conclusion of the task. By developing autonomous processes, test subsystems would be less dependent on the ground control system commanding. This feature would provide an additional layer of robustness, since test components could contain firmware and software which is unique to a particular component or activity. By incorporating parallel component activity or redundant activity, all activity could be supported by a secondary method of test verification; thus reducing the possibility of a single point failure. This strategy would make the EGSE suite more modular, require less centralized control, make the

978-1-4244-9363-0/11/$26.00 ©2011 IEEE

system more easily reconfigurable and increase the robustness of the actions. This approach could also provide an added benefit of reducing delays inherit with communication limitations between the ground control and the EGSE test suite, since the bulk of the data processing could occur within the EGSE. To design EGSE for current and future needs, the architectural model needs to evolve and strategies need

to change. The tra -needs to be transformed to accommodate a variety of applications and uses. By utilizing a reconfigurable modular approach, EGSE can become a more efficient tool in the spacecraft manufacturing process. Two cases are examined in this paper to illustrate this position: (1) safe-to-mate measurements and (2) register verification.

Figure 1 Typical EGSE Rack

Figure 2 Typical Umbilical GSE Block Diagram

I. JUSTIFICATION FOR AUTOMATED TEST METHODS Before a position can be supported, it is necessary to determine the basis by which the opinions are derived. The following list provides some of the observations and justifications for the implementation of automated test methods:

An automated test typically requires less time to execute than an equivalent manual test.

Manual testing is inherently less repeatable because the testing method involves human resources that are subject to mistakes based on human conditions (hunger, distractions, boredom, etc).

During the test execution, automated testing provides the opportunity for the direct transfer of test data needed to produce test logs.

Because of the time required to perform manual testing, along with manual resources, manual testing is typically a more expensive form of testing. This observation excludes the cost of the test equipment required for either the manual or the automated test.

Automated testing provides a greater opportunity to collect larger amounts of test data; usually for no more cost than a manual test.

In cases where waveforms are generated as part of the test execution, an automated test can be developed to evaluate the waveform. In manual testing, typically a human resource would be required to evaluate the waveform. The manual process would be subject to the objectivity of the manual resource.

Automated test methods generally require a larger expense for the procurement of controllable equipment, the addition of support circuitry only required for automated control and the associated development costs.

II. AUTOMATED SAFE-TO-MATE TEST METHOD Prior to connecting the EGSE test suite to flight hardware, a safe-to-mate procedure is typically performed. This procedure is written, and then executed to ensure that the EGSE is electrically safe to connect to the flight hardware. In many cases, the safe-to-mate procedure will require that hundreds, if not thousands of measurements are made. In many cases, this process is performed manually. In order to streamline the manual process, two human resources are required.

III. EXAMPLE SAFE-TO-MATE CASE This example is developed to analyze the effective cost of performing a safe-to-mate procedure on an I/O configuration. The safe-to-mate procedure is traditionally developed by determining the minimal set of measurements (usually resistance) necessary to establish a satisfactory sense of safety, prior to connecting the device or rack to flight hardware. An example of a typical EGSE rear panel interconnect is shown in Figure 3. This figure does not demonstrate the actual interconnections as defined for the example but is intended to provide a representative view of a typical interconnection panel. The interconnect configuration for this case is assumed to contain 5 connectors with a count of 50 contacts per connector and 5 connectors with a count of 78 contacts per connector. In this example, the cost of human resources required to perform the task in full is defined as $100 per human resource per hour. This number is used for example purposes and is consistent for all cases. For this evaluation, it is assumed that the safe-to-mate would consist of 600 test measurements. In addition, this example assumes that the safe-to-mate process will be performed 10 times during the course of the spacecraft manufacturing process. A representative test chassis is shown in Figure 4. The cost analysis for this examination is provided in Table 1 and was developed for three specific cases:

1. Manual testing of the minimal test set (600 measurements).

2. Automated testing of the minimal test set (600 measurements).

3. Automated testing of the maximum test set. (The maximum test set is computed by determining the max number of measurements per connector and summing the total for all connectors)

Figure 3 Typical EGSE Rear Panel

Case #1 Case #2 (Note 2)

Case #3 (Note 3)

Number of Measurements per occurrence 600 600 21,140

Total Setup Time (minutes) 15 15 15

Time per Step (seconds) 7 0.2 0.2

Total Test Time (seconds) 4,200 120 4,228

Total Test Time (minutes) 70.00 2.00 70.47

Total Time, Setup + Test Time (minutes) 85.00 17.00 85.47

Human Resources Required to Perform Testing 2 1 1

Human Resources Required to Verify Setup (Note 1) 2 2 2

Cost of Resource (hour) $ 100 $ 100 $ 100

Cost of Test (human resources) $ 283.33 $ 53.33 $ 167.44

Typical # of Safe-to-Mate executions during in manufacturing 10 10 10

Total Safe-to-Mate cost (resources) $ 2,833 $ 533 $ 1,674

Total Safe-to-Mate time (hours) 14.2 2.8 14.2

Total Safe-to-Mate time (schedule days @ 8 hrs per day) 1.8 0.4 1.8

Notes:

1. The human resources required to verify the setup can also be used for manual testing. 2. This automated test assumes that the same number of measurements as the manual test. 3. This case assumes (5) connectors with 50 pins / connector and (5) connectors with 78 pins / connector.

4. A full spacecraft safe-to-mate for resistance measurements can take 5-7 days done manually.

Table 1: Analysis of Cost and Time for Manual and Automated Test Methods

Figure 4 Typical Test Chassis

A. Case #1, Manual Testing of the Minimal Test Set In this examination, the cost analysis evaluates the amount of time required to perform resistance measurements on the minimally required set of 600 measurements. The number of human resources required to perform the setup is 2 and the number of human resources required to execute the test is 2. Based on the test case, the total test time required would be 85 minutes per occurrence; with a total of 850 minutes (14.2 hours) over the course of the entire spacecraft manufacturing process.

B. Case #2, Automated Testing of the Minimal Test Set In this examination, the cost analysis evaluates the amount of time required to perform resistance measurements on the minimally set of 600 measurements. The number of human resources required to perform the setup is 2 and the number of human resources required to execute the test is 1. Based

on the test case, the total time required would be 17 minutes per occurrence; with a total of 170 minutes (2.8 hours) over the course of the entire spacecraft manufacturing process.

C. Case #3, Automated Testing of the Maximum Test Set

In this examination, the cost analysis evaluates the amount of time required to perform resistance measurements on the maximum set of 21,140 measurements. The number of human resources required to perform the setup is 2 and the number of human resources required to execute the test is 1. Based on the test case, the total time required would be 85.5 minutes per occurrence; with a total of 855 minutes (14.3 hours) over the course of the entire spacecraft manufacturing process.

D. Conclusion of Test Cases The test analysis demonstrates that utilizing an automated test approach will provide a superior test result. The amount of time required to perform the setup and manually perform the minimal test set (Case #1) is approximately equal to the amount of time required to setup and measure the resistance for all pin combinations within each individual connector (Case #3). An analysis of the costs associated with the same two cases reveals that approximately 35 times more measurements can be made with the automated test system than could be manually executed in the minimal test set; for approximately 60% of the cost associated with manually executing a minimal test set. In other words, an automated script, written to test all pin combinations within a given connector (multiplied times all connectors) would perform 35 times more measurements in approximately the same amount of time as the manual test of a minimal measurement count, would cost approximately 60% less to perform these measurements and would only require 1 human resource to execute the entire test set. This example does assume an initial investment of approximately $50K for hardware and software capable of performing the automated testing. While it is possible to develop this type of test system for each test suite, the more practical solution would be to create one safe-to-mate, portable system with a software configuration and associated cabling assembly for each test requirement. In this way, the cost and benefits could be shared by all spacecraft or subsystem which requires similar safe-to-mate testing.

IV. EXAMPLE REGISTER APPLICATION A register is used to determine a specific parameter within a thruster maneuver. A spacecraft requirement defines that the acceptable value can be written to the register if the value is within the limits of defined acceptable values. If an attempt is made to write an unacceptable value to the register, the register will reject the operation and the register will remain at the previously set value. For the purpose of this example, the register used is an 8-bit register with a maximum bit count of 256 distinct values; 0 to 255. However, the requirement defines the set of acceptable values to be all valu .  

A. Test Method #1 An automated test script was written whereby attempts were made to write to the associated register of the min value, max value, min value -1, max value +1, and by setting the register with values within the required limits. This provided a test of a subset of the acceptable values and a subset of unacceptable values. To further evaluate the test condition, an additional script was written to verify that each of the bits within the associated register could be toggled between the binary states. While evaluating the state of the individual bits in question, all other bits of the associated register were evaluated to ensure that they did not exhibit any state changes. This verified that no bits were electrically stuck.

B. Evaluation of Test Method #1 The test results confirm that the associated register can only be set with values within the required limits, the register cannot be set with values which are not contained in the defined limits, and the individual bits can be set individually. Therefore, the test confirms that that the register behaves as required.

C. Test Method #2 The set of acceptable values is determined to contain 50 possible values. An automated test could be written to determine if all of the acceptable values can be written to the register as defined in the requirement. In addition, the test could be developed to permit attempts to write to the register with values which are not contained in the  set of acceptable values. The unacceptable values were determined to be 0 and 51.  

D. Evaluation of Test Method #2 The test results confirm that the associated register can only be set with values within the required limits and the register cannot be set with values which are not contained in the defined limits. However, this method only evaluates the operation of a small subset of the unacceptable values.  

E. Conclusion of Test Methods In this example, there are 50 acceptable values and 205 unacceptable values. The test in Method #1 attempts to write 4 distinct values to the register; 2 values that are within the defined limits and 2 values that are not within the defined limits. In addition, each of the 8 bits is individually written with an evaluation of the behavior of

the other bits within the register. This method requires that 12 attempts be made to write to the register. In the test for Method #2, an attempt is made to write each of the 50 acceptable values to the register along with 2 unacceptable values for a total of 52 attempts to write to the register. Obviously, more iteration attempts would be required in Test Method #2 than in Test Method #1 and as such, Test Method #2 would also require more time to perform. While the number of iteration attempts may be of consideration if the test were to be performed manually, as part of an automated test suite, the additional time required to execute either test method would be relatively small. Unfortunately, both of the defined methods listed above attempt to only write 2 unacceptable values to the register; min-1 and max+1. Neither method addresses the remaining 203 possible unacceptable values. To develop a truly comprehensive test, the test would need to evaluate the ability of the register to accept all possible acceptable values and reject all unacceptable values. While this philosophy may be acceptable for a small value set, the test could become excessive for a much larger set of values and ultimately make the approach less attractive. Some would argue that the only true approach in determining the behavior of a circuit or system is to test all possible conditions. As stated before, this philosophy could be accepted as long as the sample size is sufficiently small. If the register used in this example had been a 32 bit register, the number of possible attempts would be 4,294,967,296. Obviously, this case would be prohibitively large and would require an alternative method of evaluation. In this case, a statistical determination could be made to determine to most appropriate test sample size. As with all statistical evaluations, the sample size is di

required. Therefore, the test strategy will be dictated by the number of possible test conditions to be evaluated, the method of implementation (Manual vs. Automation) and

within this paper, the most effective method of implementation will usually be Automation.

V. CREATIVE TEST SOLUTIONS AND METHODS THROUGH AUTOMATION

Automated testing supports the concept of a more extensive scope of testing. This can be demonstrated via the previous example presented in Section IV. When an engineer initially begins developing the list of measurements to be made, some consideration is made to the number of measurements. If the measurements are being manually performed, a conscious decision is made to limit the number of measurements to a value which is

test method, the measurements can be made on a much larger scale. In the case of a manual test, this would likely be dismissed as impractical. So far, examples have been provided to justify automated test methodology based on quantity, time and cost. These are obviously important factors in determining the relevance of a test method. However, we cannot overlook another very important factor which is far less easy to evaluate with traditional metrics. When automated test methods are utilized, it is necessary to have a predefined value for every point of comparison. If the test method is determined to be a manual operation, there will likely be fewer measurements to be made. However, if an automated method is employed, the number of measurements will likely increase dramatically because the comparative test time per measurement has typically gone down dramatically. By increasing the number of measurements to be made, it is also necessary to determine the expected results for this larger set of points. Therefore, by utilizing a more extensive test method, the system engineer would be required to consider measurements that might otherwise be overlooked, had a less comprehensive test method been used.

VI. RECONFIGURABLE EGSE TEST SUITES So far, the primary theme discussed in this paper has been the question of whether automated test methods are superior to manual test methods. While arguments and specific test conditions can be shown to support both positions, it seems reasonable that in most cases, an automated approach provides a more suitable test platform. However, one topic which has yet to be discussed is the idea that the entire EGSE suite should be automated with switching matrix to allow the test suite to become more configurable.

Typically, an EGSE suite is developed and built with the mandate of testing a specific spacecraft or system component. Once this test process has concluded, the rack or test suite is maintained for a fixed period of time and then, the test system is disassembled and the recoverable components are salvaged for use in the next test effort. However, what if the EGSE test suite were designed to facilitate the testing of multiple spacecraft, development costs could be shared across multiple budgets. This would dramatically reduce the cost per spacecraft.

A. Method #1 for Reconfigurable EGSE The most critical non-standard component in a reconfigurable EGSE is a switching matrix. In a standard EGSE test configuration, the interconnect points of a test component is connected directly to one of the rear panel interconnects. This is the primary source point of entry and exit for signals to and from the spacecraft. However, for an EGSE suite to be considered reconfigurable, the EGSE must be capable of providing simulation, stimulation and test capability to more than one specific spacecraft or system. Therefore, a method is required to allow for the routing of alternate signal paths. In an automated design approach, the switching matrix component would be inserted between the EGSE component interface and the rear panel interconnects, to allow for control of the switched path. Since most EGSE test suites contain a localized computer for processing and controlling local components, the switched matrix could also be controlled via this same computer.

B. Method #2 for Reconfigurable EGSE The previous method requires the use of some form of automated switch to route the signals to the appropriate destination within the EGSE. However, another less technically sophisticated method would be to use a spacecraft specific custom cabling harness to reroute signals between the spacecraft and the EGSE. Some form of this technique is already in use in most test facilities. Obviously, spacecraft manufactures currently seek the most cost efficient manufacturing methods. Therefore, if a custom cabling harness could be developed and integrated into the EGSE, this would permit a minimal amount of changes to the EGSE, which would be very desirable.

C. Obstacles to Reconfigurable EGSE While these two approaches might seem attractive, the basic obstacle to overcome would be in the technical capabilities that reconfigurable EGSE would need to address. Basic reconfigurable EGSE would be suitable for spacecraft having similar functions and interconnections. Of course, this is rarely the case. For EGSE to be truly full functioning, reconfigurable systems, the EGSE would need to be capable of supporting the full scope of testing for two varied spacecraft mission types. This suggests that a comprehensive reconfigurable EGSE system would basically be the merging of two or more individual test systems. The resulting EGSE would contain spacecraft specific components and share components which service common functions. The overall cost of such a system would likely be slightly less than the cost of the individual racks; however, cost savings in equipment would likely be offset by the additional design costs of the reconfigurable portions of the system.

VII. CONCLUSIONS There are test conditions when the use of manual test methods is a more suitable approach. In these cases, the time and expense do not add value to the test process. However, as has been described in some detail, the use of automation in both test implementation and EGSE reconfiguration can provide multiple benefits. When automation is used to perform tasks of repetitive functions, such as the execution of a safe-to-mate procedure, the time and the human resources required would be reduced. Not only would this result in a cost savings, there is the possibility that overall schedule time could be reduced. Perhaps more importantly, by introducing automation into operations that were previously executed manually, the resulting quantity of test data could prove to be vastly more comprehensive. Some would argue that the been utilized for many years and the basic philosophy as proven to be sufficient. Unfortunately, if we choose to accept this minimalist approach, we choose not to take advantage of the insight that can only be gleaned through more robust test methods. While there are times when additional test data is merely that; more test data, and adds little to our understanding of system performance, few could argue that it is absolutely necessary to acquire as much test data as possible while the spacecraft is in the manufacturing process. By using automated techniques, the acquisition of test data can be simplified.

In addition, the use of automated processes can foster ideas that would be less practical to implement, using less sophisticated test methods. As technologies advance, the spacecraft industry continues to evolve and this evolution places more demands on the spacecraft. The same should be true of the EGSE. However, many are convinced that the same techniques and tools used

for the testing of the spacecraft of today. By embracing the technologies of today and continuing to demand more from the EGSE, we can expect that the spacecraft of tomorrow will be more thoroughly tested and will be more capable of reliably handling the ever increasing demands.

VIII. REFERENCES [1]

pages 425-434, 10.1109/AUTEST.2007.4374250 [2] H. Nguyen and I. Miller, Acceptance Testing of Electrical Ground Support

submission for Autotestconn 2011 [3]

-139, 10.1109/AUTEST.2002.1047883 [4] itectures for the

-247, 10.1109/AUTEST.2002.1047895 [5] M. Levesque, J. LouieExecution Control Tool: Automating Testing in

-7803-5846-5, 2000 IEEE Aerospace Conference