3.4.1.

Mass design issues

The estimated mass of the observatory went through the typical episodic increases as the design matured, usually at or right before key system level reviews. However, shortly after the early Mission Confirmation Review, the mass margin plunged precipitously, and, for the rest of the program, mass margin was low. Figure 20 shows the mass margin evolution for the project, with the left graph showing the early evolution where margin dove from an estimate above 50% to an estimate lower than 25%. The graph on the right of the figure shows the evolution from SDR through I&T. These graphs show that once the more detailed project engineering started the mass issues became evident, and the project was challenged for the duration as measured by well-established normal margins at each milestone review set by NASA and the AIAA. One of the more significant drops in margin occurred between CDR and SLR when a thermal issue surfaced, as described in Sec. 3.4.2.

Fig. 20

JWST mass margin evolution during the project.

There were multiple system level mass-reduction efforts over the development life of the program, all associated with the major design reviews (systems design review, preliminary design review, CDR, and the systems look-back review). The mass recovery efforts and their associated issues involved:

An example of item 3 above, an “architecture level mass-reduction” trade, is the trade that evaluated the use of a helium dewar against the use of a cryocooler to actively cool the MIRI detector. The selected cryocooler offered a systems level mass savings of over 200 kg while maintaining the capability to cool the MIRI detectors for the life of the mission (not to mention affording safety and I&T advantages as well). A second example is a serendipitous architectural decision for mass margin centered around the first mid-course correction (MCC-1) maneuver. To avoid engineering and contamination issues, no thrusters were on the cold side of the observatory, so JWST did not have the capability to thrust toward the sun. Therefore, the Ariane 5 launcher was required to “aim low” to cover a 3-sigma over-burn launch dispersion to insure only burns away from the sun would be required after separation from the launcher. The MCC-1 maneuver performed by the observatory would make up the energy difference, and in doing so realized a staging efficiency, thus yielding a net mass savings.

3.4.2.

Thermal design issues

Radiative passive cooling of the approximately 3000 kg OTIS was the only practical option, and parasitic heat leaks are notorious risks for even moderately ambitious space borne systems. Therefore, the SET paid very special attention to making sure that the observatory had an adequate area of cryogenic radiators to cover these parasitic heats loads as well as the efficiently uncertainty. At JWST’s launch, the thermal uncertainty would remain large as the verification program never thermally tested the observatory in its fully integrated state. The actual parts of the thermal verification program had the cryogenic side of the observatory tested in the chamber A facility of JSC, the spacecraft bus and stowed sunshield tested in the NG M4 thermal vacuum facility, and the insulating performance of the sunshield computed from thermal models that were validated from tests of one-third scale engineering test unit of the sunshield. Since the region between the spacecraft bus and the OTE, called the “core” region, was an extremely thermally complex region with multiple thermal interfaces, a full-scale engineering model of the core region was tested. The analytical models for these portions of the observatory were validated by these tests and then integrated into an observatory model used to verify that the OTIS would reach its required cryogenic temperatures on-orbit. Aside from the intrinsic uncertainties, such models do not model the possible workmanship errors. Any heat leaks resulting from these errors would represent a threat against the final performance and these were addressed by carrying additional radiator margin.

The SET allocated requirements for the cryogenic margin necessary to cover this piecewise verification and formulated a “burn-down” plan for lowering this margin as more test data were accumulated to lower the uncertainties. The SET tracked this margin carefully over the course of the development as well all the thermal liens and threats against it. As with the mass margin, the decline of the cryogenic margin was episodic, showing decreases as detailed engineering proved that potential threats were in fact highly likely or certain. There were multiple recovery efforts to increase the cryogenic margin to desired levels.

The last of these efforts occurred between the CDR and SLR. Deficiency in cryogenic margin was evident at the CDR and was the source of a significant concern. Following CDR, the margins continued to decrease, and the worst-case margin exhibited by the FGS radiator was estimated to be negative. Figure 21 shows the cryogenic margin evolution during this period. The solution to this was the addition of considerable radiator area on the top of the ISIM enclosure. This additional radiator area was only accommodated by using a deployed assembly, the aft deployed ISIM radiator (ADIR), shown in Fig. 22. This solution came not only with another complex deployable but also with a significant mass cost, as also shown in Fig. 22. In addition, as part of this thermal design update, cryogenic margin requirements and calculation methodologies were updated and margins on heat loads were segregated from the margin on needed passive cooling capability. The radiator changes, in combination with the updated requirements and margin calculation methodology, proved to be very robust to the natural margin degradation experienced during I&T.

Fig. 21

Decrease of cryogenic radiator margin with mass margin.

Fig. 22

Changes to the ISIM cryogenic radiators to address the CDR margin deficiency.

3.4.3.

Stray light design issues

From the beginning, it was difficult to meet the stray light performance for JWST using NASA’s traditionally recommended margin factors of 5 to 10 times lower than required for stray light levels predicted by analyses. The primary sources for stray light, and therefore the primary threats and liens to be addressed by these margins, were different in the NIR and MIR wavelengths.

NIR stray light levels are dominated by scattering of starlight from contaminants on the optical surface and from “sneak paths” of light from the sky through the system to the detectors. Figure 23 illustrates two such sneak paths, called the truant path and the rogue path. The truant path is one where light coming from the sky in the aft direction passes over the perimeter of the primary mirror and scatters off the secondary mirror, entering the optical train. The second, called the rogue path, is a direct path of light from the sky over the secondary mirror into the aft optics subsystem’s Cassegrain aperture and directly impinging the pick-off mirror region of the instruments.

Fig. 23

JWST primary sources of stray light.

MIR stray light levels are dominated by observatory thermal emissions, as heat from observatory components with a clear view to the optics were the primary sources of MIR stray light. Much of the MIR stray light recovery was achieved as a result of the trade studies performed to resolve a baseline design with lower thermal margin, which generally lowered temperatures of primary MIR sources to acceptable levels. The MIR path into the optical train is also shown in Fig. 23.

Stray light control was achieved by three general methods:

A key lesson learned in the area of stray light was the critical importance of setting interfaces properly early on to avoid design issues and integrated modeling difficulties. This can be seen in the on-orbit anomaly section on stray light, Sec. 3.7.4. The prime example was a rogue path design issue that resulted from the telescope development and instrument development teams not working to the same prescription. Once the observatory contractor was selected, it was discovered that the NIRCam and observatory contractor team had different assumptions for the telescope design including the speed of the optical system. The observatory contractor team changed their design to accommodate NIRCam, which had already been started. Eventually, it was determined that the design change had opened a rogue path where light from the sky can enter along the side of the secondary and directly strike instrument pickoff mirrors. Once determined, changing the prescription would have been a major change so, instead, with copious analysis, the team used aperture stops and pupil stops to address it. The new complexity was mostly successful, but the rogue path aspect of the interface did not get sufficiently flowed for analytical verification purposes, which ultimately led to definition of some bright-star keep-out zones during on-orbit characterization and stray light model validation during commissioning. If the interface between the telescope and instruments had been better coordinated and properly specified from the start, these efforts and mitigations would not have been needed. See Sec. 3.7.4 for further details on the stray light investigation and mitigations developed during commissioning.

In other areas, shielding of the various sources and light paths was achieved by stops in the optical trains of the instruments and by use of various “soft-structure” membranes, such as single-layer Kapton sheets use for stray light control, at the observatory level. For NIR truant path, a soft-structure membrane called the “frill” was incorporated around the perimeter of the primary mirror, as shown in Fig. 24. For the rogue path, optical masks in the aft optics subsystem were used to provide the blockage. Specifically, the outer edge of the fine steering mirror pupil stop helped to mitigate truant path.

Fig. 24

Primary stray light shields.

MIR straylight was controlled by two notable shields. The first shields blocked the thermal emissions from the hotter core area of the observatory, emanating from between the bottom of the OTE and the top of the spacecraft, from impinging of the secondary mirror and scattering into the optical paths and this was called the “Bib” for its appearance. During the final stages of detailed design, it was realized that a portion of the Bib itself was heating to temperatures that could present a stray light. So, another, “outer” Bib was added that occulted the hotter regions of the inner Bib. The second shields were constructed of thermal blankets placed on the top and bottoms of the sunshield spreader bars to prevent their thermal emissions from impinging on the secondary mirror or primary mirror. Thermal conductive paths along these spreader bars heat the tips of these spreader bars, which, if unblocked, present an MIR stray light source. These covers were nicknamed the “epaulets” and are shown in Fig. 24 for the mid-boom spreader bars. Similar epaulets are also included on the forward spreader bars.

There were several modifications to these designs that were made relatively late in the design phase and even during I&T. Late changes to thermal multi-layer insulation (MLI) blankets soft-structure, such as these, are common late in I&T. But these soft-structures can interact with precision optics and or deployed interfaces, and any late design changes can have important consequences that can be missed after all other detailed analysis has been already completed. Many of these consequences were identified as issues during the I&T phase.

As part of the stray light control, the JWST contamination control program established cleanliness allocations for particulates and molecular for all optical elements and kept careful track the accumulations during all phases of the I&T program. Particular attention was given to the primary and secondary mirrors (PM, SM), as they were exposed during much of the I&T program due to handling safety concerns with managing a cover. The PM was allocated a total life percent area coverage (PAC) of 1.5% for particulates and the SM was allocated 0.5% PAC. The project leveraged opportunities to perform cleanings of the PM and SM throughout the I&T program and, as a result, the PAC was well below the maximum allowed by the requirements at launch.²⁸

3.4.4.

Thermal distortion

JWST is passively thermally stable where the as-designed thermal time constant was approximately 6 days with a requirement to update the primary mirror alignments no more frequently than every fourteen days. The approach to analyzing thermal stability of the as-designed hardware, two steady state conditions were considered. These were the worst case hot and worst cold conditions after allowing time for stabilization. The magnitude of the wavefront change after worst-case cold condition to worse-case hot condition was allocated from the optical system error budget to be approximately 50 nm root-mean square (RMS) of WFE. The overall stability in the primary optical train’s wavefront was then sub-allocated to multiple contributors by low, mid, and high spatial frequency.²⁹ The main contribution to wavefront instability is the primary mirror segment assembly, which includes the mirror substrate, hexapod, and radius of curvature systems with actuators, delta frame assembly, and the flexures and athermalization systems. The backplane itself is also a major contributor, driven by the coefficient of thermal expansion over the 30 to 55 K operating range of the tubes and due to the less than 100 mK thermal drifts from pointing the telescope within the field of regard affecting the joints and interfaces. In addition, the wing segments of the primary mirror have hinges and latches that are also part of the overall telescope’s stability budget. The predicted change, without model uncertainty, was approximately 18 nm RMS. This value was measured on-orbit and was very close to this model prediction. The on-orbit measured thermally stability performance is described in Ref. 29 and was measured to be approximately 18 nm with a time constant of 1.5 to 2 days. In comparison, the preflight prediction from integrated modeling was 14.6 nm with a time constant of 5 to 6 days predicted at observatory beginning of life, versus requirements of 54 nm rms. This is summarized in Table 1.

Table 1

Comparison of WFE: requirement, prediction, and measurement.

Note that cryogenic bonds were highly complex and difficult to model, and every joint type required special analysis and test validation. In addition, integrated modeling cycles of the backplane were very time consuming due to the nodal density and complexity. The cryogenic passive stability mitigation efforts interacted with the lack of mass margin and at one point in time the entire program had to work to save mass and the backplane had to simplify joint interfaces, which required redesigns. In addition, gravity distortion of the lightweight structures had to be carefully understood as did structural margins for launch.

3.4.5.

Jitter design issues

Jitter refers to high frequency vibration, outside the control bandwidth of the observatory fine guidance control, which disturbs the LOS of the OTE during an observation. Based on the design constraints defined by two requirements, (the overall image quality requirement at an overall WFE of less than 150 nm, and the requirement to be refraction limited at a wavelength of $2\mu \mathrm{m}$) the SET allocated an overall jitter requirement of no more than 0.007 arcsec (69 nm rms WFE equivalent). To achieve this, isolation devices were incorporated between the OTE and the spacecraft bus. These devices isolated the OTE from the two major vibration sources in the spacecraft - the RWAs and the cryocooler compressors for MIRI. Such isolation schemes must avoid vibration transmission paths that can transmit vibrations, and these shunt paths can hide until the final detailed deployed dynamics models of the observatory. During the development of the observatory, many of these paths were identified as more detailed dynamic models of the observatory were developed.

The detailed models of the spacecraft were the pacing item for performing jitter analysis. As cited earlier, the detailed design of the spacecraft was delayed due to funding constraints prior to the CDR. It was therefore between the Observatory-level CDR and the SLR that many of these detailed vibration transmission paths were identified. One path involved the propagation of harmonics of the 30.5 Hz drive frequency of the pulse tube cryocooler. These vibrations were analyzed to travel through the heat pipes connecting the compressor to its radiators on the spacecraft bus, thus shunting past the compressor vibration isolators, through the bus structure, through the deployed tower assembly and into the OTE. Because this vibration transmission path was found late in the design phase, changes to hardware were highly undesirable, and in the end, it was found that tuning of the cooler compressor speed on orbit could be performed to avoid the predicted LOS response. It was fortunate that the MIRI cooler had the thermal lift margin to allow such tuning based on optimizing vibration performance over thermal efficiency. This is another example of how thermal margin was expended to address a late system design issue.

3.4.6.

Deployment design issues

The deployments of the JWST observatory were always recognized as a major challenge, and a separate paper is necessary to fully cover all the issues encountered in their design and implementation.³⁰ The unique design of the sunshield offered most of these design challenges, which can be broadly characterized as follows:

Deployment mechanism design was also an issue for other elements of the observatory, which relied on a total of 178 release mechanisms for its deployments (including those of the sunshield). Each of these release mechanisms contained SPF modes, meaning they were all SPF items as documented and reported by NASA standards. Most of these release devices contained various versions of the EBAD NEA, a device known for its low output shock. These devices come in a variety of sizes, but they all function similarly. They are designed to carry a tensile preload until commanded to release. The preload is applied through a release rod held in place by two separable spool halves, which are, in turn, held together by tight winding of restraining wire. The restraint wire is held in place by redundant electrical fuse wires; actuation of either circuit allows release. Upon firing of the fuse wires, the restraint wire unwinds allowing the spool halves to separate releasing the release rod and the associated preload.

For the most part, the devices were found to perform as expected. However, the largest of the NEA designs used for JWST, the ¾” NEA, named for the size of release rod, which were used to hold the OTIS to the top of the spacecraft bus for launch were found to have issues. These four devices shown in Fig. 25 were in the primary launch load path of the observatory and because of this required very high preload levels. The levels were within the specifications for the devices, but in test firings the output shock of the devices exceeded their required levels. An extensive investigation was conducted that resulted in design modifications and a re-qualification program. In addition to the design modifications, changes were made to the fabrication techniques to make them more repeatable, and therefore more reliable.

Fig. 25

Location of the OTIS to spacecraft launch release devices with ¾” NEA.

3.4.7.

Launch and ascent thermal design issues

A major thermal issue that took over 10 years to fully resolve was Observatory temperatures during launch and specifically temperatures on the stowed telescope. During initial launcher compatibility studies, the lack of a re-startable upper stage on the Ariane 5 was cause for concern regarding the thermal and solar exposure the telescope. Without a re-startable upper stage, the launch would have to occur early in the local morning, thus subjecting the telescope to direct sun after fairing jettison. Without thorough knowledge of the Ariane 5’s capabilities at the time, it was thought that a traditional launch’s thermal management maneuvers of the launcher’s upper stage would moderate temperatures.

Subsequent meetings with Arianespace provided updated trajectory and roll characteristics and thermal analyses prior to PDR indicated that critical temperatures on the telescope’s cryogenic backplane were exceeding the survival limits of the composite’s epoxy system of 50°C. The epoxy in the backplane structure softens when it is warmed to temperatures above 50°C. This softening may not cause a structural failure, but the loads experienced during the time of the elevated temperature may distort the structure, which may result in a deformed structure upon cooling, which can induce stresses into the mirrors. This can cause misalignments and WFE. Subsequent structural and thermal studies investigated the benefit of higher temperature limits that would result from post-curing the backplane during assembly by raising its temperature in a very controlled and mechanically supported manner. Although it could have potentially solved the launch and ascent thermal issue, post curing the backplane was not pursued due to cost, schedule, and complexity concerns.

Prior to CDR, and due to continued predicted temperature violations on the telescope backplane, the traditional thermal management maneuvers were abandoned. In its place, a partial roll, or back and forth rocking motion of the launch upper stage, was proposed. This approach kept the sun on one side of the telescope and better constrained temperature violations to specific areas. After Arianespace agreed to the new proposed profile, a lengthy and multi-year thermal analysis investigated sensitivities to launch date, launch time of day, and roll profile errors. In parallel, a series of tests were conducted on composite samples where distortion was measured versus temperature and structural load. In addition, pre-launch cooling of the telescope structure to below 12°C was implemented and would provide additional margin to the 50°C limit. The combination of the unique roll profile, robust thermal analyses, structural testing, and prelaunch cooling, all resulted in the observatory arriving safely on-orbit.

3.5.1.

Hardware-system test beds

The backplane stability test article (BSTA) was a full ETU of a portion of the primary mirror backplane structure that was used to characterize the cryogenic stability of the structural members and joints. It was also used to characterize the cryogenic damping properties of the structure for line-of-sight stability predictions.

The ¾” NEA verification test unit was a high-fidelity ETU of the modified NEA device used to release the OTE from the spacecraft without imparting output shock. This unit was used to qualify a modified design of the unit. Manufacturing issues were found with the device,³⁰ and this issue is discussed in Sec. 3.4.6.

The test bed telescope (TBT) was a 1:6 subscale replica of the segmented telescope that was used to practice and develop the on-orbit alignment and phasing procedures. The testbed had all the degrees of freedom of the flight telescope with flight-like actuators. The testbed was used to determine the error budget of all wavefront sensing and control algorithms and that budget was then traceable to all flight algorithm requirements.

The pathfinder telescope was a full-scale model of the center section of the telescope with two primary mirror segments (both spares) and a flight spare secondary mirror. This pathfinder telescope was used to practice and validate the test procedures and verify the test equipment for the flight OTIS cryogenic optical tests in the modified JSC Chamber A facility. Three separate tests were performed with the pathfinder. The first test focused on the optical GSE, the second test added the flight aft optics system and cryogenic optical fiber fed sources, and the third test removed the flight aft optics system and demonstrated the thermal test aspects of the test with additional thermal simulators for the mirror segments.

The observatory Core test program consisted of full-scale cryogenic tests of the region of the observatory between the OTE and the top of the spacecraft bus. This was a highly complex thermal region of the observatory and one that needed extra care in the validation of thermal models. This Core test was conducted twice, the first time with an early development model with the goal of proving that this complex region, which contains the IEC, could be adequately modeled and the needed cryogenic temperatures could be achieved. A second test, known as Core 2, was a high-fidelity ETU of the final design and was a key part of the overall thermal verification program.

Existing thermal vacuum chambers were not large enough to accommodate testing of the full-scale sunshield. Therefore, a one-third scale sunshield model was thermally tested to validate the sunshield thermal model.

The sunshield IVA was a full-scale model of the sunshield used to demonstrate the sunshield deployment process and to investigate and practice various folding methods. Much like a parachute, the folding methods for the sunshield were critical to its final success, and this IVA enabled rehearsing of the folding process without adding extra wear and tear from handling to the flight unit. As discussed further in Sec. 3.4.6, the rehearsal of the sunshield folding could have been taken a step further. If it had involved some of the methods required in on the actual flight unit, such as use of manlifts and measures to work near delicate and contamination sensitive hardware, the during of the folding effort would have been realized sooner.

The end-to-end cryo-cooler test was an effective tool to validate the MIRI cryo-cooler performance model. The flight cryo-cooler subsystem is distributed over the entire span of the observatory (compressors in the spacecraft bus, cold heads in the ISIM, and coolant lines that traversed the “core” region between the two) so an end-to-end test of the actual flight cooler after integration into the observatory in flight conditions was not possible. The end-to-end test used flight-spare cooler and a prototype of MIRI that was adapted by the European MIRI team to make sure it would react thermally in the same manner as the fight instrument in the flight environment. The test confirmed that the refrigeration system can successfully cool down MIRI and keep it cold. The end-to-end cooler test also provided an opportunity to fine-tune the operational parameters of MIRI in a simulated environment and to verify overall system performance.

3.5.2.

Electronics and software-system test beds

JWST also relied heavily on electrical and software test beds for the design, validation, and verification at various levels of the systems. This section describes the key test beds used for development, verification, and on-orbit troubleshooting.

Software telemetry simulators (STS): The STS runs the spacecraft and ISIM flight software and was used by the spacecraft and ISIM teams for software development and testing. This system is the quickest of the systems to set up and run for syntax checks on flight products. It is completely software-based and does not require the operator to reside in a test lab to run the simulator. An example of a flight product that was validated on an STS is the formal certification and performance of the deployment system software, including all “negative paths” except for a few paths that required special subsystem flight software loads. Negative software system testing uses invalid input data, or undesired system behaviors, to check for unexpected system errors.

The JWST project used a series of rack-mount simulators called the JWST Certification Labs (JLABs). See the photo of a rack-mounted JLAB system in Fig. 27. The JLABs provided a both a ground system simulator and a spacecraft simulator. They also optionally provided an ISIM simulator. During the peak of hardware and software development, the JWST project maintained 38 JLABs of different fidelity levels spread across the United States, Canada, and Europe.

Fig. 27

Schematic and photograph of a nominal JLAB.

The most basic of the JLABs were called the Science Instrument Development Units, and these were delivered to the four instrument providers and several other subsystem labs for flight software and development of the operational script subsystem.

The science instrument test set (SITS) replicated the JWST ground based interface, spacecraft, and ICDH commands and was used during instrument level integration and test (I&T). Early testing of flight software with flight-like operational systems demonstrated the interactions with ICDH command processing, which reduced risk by identifying command and telemetry interfaces to the ISIM interface with the instrument flight hardware and flight electronics. The SITS was also used for post-shipment checkout of the instrument engineering test units and instrument flight units.

The JLAB systems, their flight-like interface, the JLAB system management, and the configuration control of the flight and non-flight database that supported the JLAB systems were critical to success at the final Observatory level integration of JWST. The planning, installation, maintenance, and effective use of these systems took considerable resources from all the JWST project, partners, and contactors, yet resulted in the success of the Observatory level test campaign, and, more importantly, the successful and on-time in-flight commissioning of the Observatory. A system schematic of a typical JLAB is shown in Fig. 27.

The JWST Project continues to use JLABs with high-fidelity fine guidance subsystem simulators for testing of updates to software and operational scripts. The simulator includes a suite of instrument simulators called the instrument electronics simulator (IES), which substitute for the flight instruments now that the JLABs are not connected to the flight instruments.

The observatory test bed (OTB) is a high-fidelity electrical and software Observatory simulator. It is located at the JWST Mission Operations Center (MOC) at Space Telescope Science Institute (STScI). Besides a spacecraft and ISIM simulator, it includes a vehicle dynamic simulator (VDS), some primary and redundant flight-like hardware, a high-fidelity FGS simulator, and simulators for all the science instruments. This system connects with much of the flight operations system (FOS) located at the MOC. With the VDS and the FGS simulator, the OTB allows the full interaction with the spacecraft flight software, attitude control system, fine guiding system, on-board script subsystem, including functions such as slews, settling of slews, momentum unloads, high gain antenna pointing, and closed loop guidance engagement. Simulation includes solar torque, light-time delay, contact playback, and very realistic science observation plan execution.

However, even this system has limitations, and upgrades or improvements are nearly continuously in work. One of the biggest non-flight like features of the OTB is timing. OTB timing has been optimized but cannot be not fully flight-like due to the simulated systems and non-flight-like components. In addition, timing conflicts can be random in nature and the OTB is not designed for streamlined iterative testing to find corner-case timing conflicts.

The engineering model test bed (EMTB) is another high-fidelity electrical and software Observatory simulator, which is located at Northrop Grumman’s Space Park facility. This system is remotely accessible to FOS, which served as a method to dry run pre-flight ground tests in which the FOS was used to command the actual flight Observatory. In addition, this system was used for deployment product validation and deployment electronics unit (DEU) software validation as it includes an engineering model of the DEU.

The early commissioning activities and configurations were run on this simulator and testing of functions that could not be performed on the STS.

The EMTB also includes an engineering model of the actuator deployment unit, which controls the actuators on the JWST mirror segments. Although extensive testing of these actuators was performed at the test bed telescope, the EMTB put the ADU commanding into a flight-like system. Also, the EMTB includes capabilities to run special tests for deployments with load simulators.

From early software tests to extensive validation and testing of line-of-sight software components and Observatory fault management algorithms, the EMTB has been critical in verifying many of the systems that cannot be sufficiently end-to-end tested in flight-like conditions on the flight Observatory.

Finally, a unique tool was developed for JWST visualization during the on-orbit commissioning called the observatory visualization tool. This tool graphically showed the observatory as it deployed, based on flight telemetry from various on-board sensors. This allowed the operations team deploying the telescope to visualize and readily communicate the flight configuration.

3.6.1.

Sine vibe test issues

The purpose of the OTIS sine vibration test was to qualify the OTIS hardware for the low frequency dynamics launch load environment from the Ariane 5 rocket, provide data for the finite element model verification, and to verify the minimum launch frequency requirement. The test levels were based on the observatory sine vibration launch load provided by the Ariane 5 user’s manual and this load, when transferred to the OTIS hardware, applied an acceleration up to 2 g from 5 to 100 Hz in all three axes. It was critical that the hardware experience the load required to pass the flight environment without greatly over-testing any part of the structure. Therefore, notching of the input test signal was allowed to prevent unrealistic launch inputs but it was a challenge for the input to stay above the sine vibration minimum floor levels provided by Arianespace.

The OTIS hardware consisted of several mechanisms used to deploy the mirror wings and secondary mirror. The mechanisms were qualified through separate random vibration tests, however, they were expected to chatter during low frequency sine vibration. This chattering showed high frequency responses in the test data and made it difficult to control the test to limit structural responses to the mirrors. Therefore, high frequency time history data were reviewed during the test to understand the structural loading on the mirrors and its deployment latches. An anomaly occurred early in testing that showed a high frequency response throughout the hardware due to one of the primary mirror wing mechanical restraints overcoming its preload during the test. While an investigation indicated no hardware damage was found, test limits that were previously set for strength of the hardware were modified to accommodate limits from the preload in the mechanical restraints.

Another lesson learned during the OTIS sine vibration test was the room temperature damping coefficient of the composite mirror backplane structure. The mirror backplane structure was primarily designed to minimize mirror motion due to its cryogenic operational temperatures. Composite material was selected to meet this primary design requirement. However, damping of this composite hardware at room temperature, similar to the launch temperatures, was shown to be lower damping than predicted, which then resulted in higher sine vibration test responses for the aft optics assembly and the secondary mirror support structure. Following the OTIS mechanical testing, dampers were installed for these components. The performance of the dampers was then verified during Observatory-level mechanical testing. These dampers worked as designed. The mirror components were then shown to appropriately respond within design loads and within Arianespace’s required minimum test levels in sine vibration. However, this was a very late change to and had the project had to be confident that the dampers installed at this late stage were the right solution.

In addition to lessons learned that were pertinent to OTIS specific hardware, there were lessons learned from the sine-vibration test configuration. It was shown that determining the test instrumentation and its routing ahead of performing the test was critical to the success of controlling the responses throughout the hardware. The OTIS test instrumentation, which consisted of interface force gages and six hundred response accelerometers, began its design 2 years ahead of testing because it was a very complex ground instrumentation set up. Three things that made it complex included:

Starting the instrumentation design early greatly contributed to the efficiency of running the OTIS sine vibration test.

3.6.2.

Cryogenic test wavefront error instability issues

The OTIS cryogenic tests, conducted in the modified JSC Chamber A Facility, were perhaps the largest and most complex cryogenic tests ever conducted. The design of the optical and thermal GSE as well as the cleaning of the chamber itself to accommodate optical testing was extremely demanding. The OTIS testing took roughly 100 days, which included lengthy controlled cool-down and warm-up periods to avoid contamination on critical surfaces.

Overall, the test went smoothly and as planned, even though the test was conducted through Hurricane Harvey, which hit JSC and the surrounding communities very hard. During OTIS thermal vacuum testing, there were several performance issues that were encountered. The critical issue was unforeseen optical instabilities during the final EE stability tests. One instability was a change in WFE with a time period of minutes, which was correlated to the duty cycles of the IEC heaters. Analysis of the instability showed that there were unaccounted for stress paths between the IEC and the OTE from GSE that varied as heaters cycled. Once accounted for, the bulk of the instability was explained, but there was residual instability that could persist in orbit due to these heaters imparting strain through the harnessing between the IEC and surrounding structure. The residual instability of the WFE and EE was within allocations but was nonetheless unexpected to be identifiable in the telescope WFE once operating on orbit. The residual instabilities due to the heater cycling signature was reduced by defining a narrower bandwidth for the control heaters. An on-orbit commissioning test of this OTIS-level characterization of this instability showed that on-orbit errors are well within acceptable limits.^29,31

The second instability was identified with the stresses from the differential thermal contractions between the OTE frill and the primary mirror backplane structure. Post-test inspections showed sections of the frill did not have the proper amount of slack so that it imparted stress on the PM structure as it cooled down. The backplane had a much larger thermal time constant than the soft-structure frill, due to its large mass. Additions were made to the sections of the frill to increase the slack to its mounting points at room temperatures, but since the OTIS cryogenic test was not to be repeated, the consequence of these corrections was not tested and carried a certain risk, which was reviewed and deemed to be acceptably low. As with the IEC heater instability, these measurements informed the commissioning tests and showed that this instability is well within acceptable limits.^29,31 Ultimately, stability tests in-orbit showed the fixes for these issues worked almost exactly as expected, a key reason for the outstanding stability of the observatory.

3.6.3.

Avionics boxes issues late in integration and test

During the spacecraft and observatory level tests four significant avionics boxes failed and required replacement and repair. The nature of the failures themselves was due to electrical part failures and or workmanship issues. Table 2 lists and describes the failures.

Table 2

Avionics box failures during the spacecraft and observatory I&T program.

It is not uncommon for such failures; in fact, such electrical performance tests are intended to find such issues, and operating hours are required to make sure infant mortality issues reveal themselves and get discovered. The systems level issues and the lessons to be learned are more toward the advantages of box accessibility. This was particularly evident with the failures of the TWTA and CTP. The spacecraft bus was designed to have removable equipment panels to provide late access to the boxes, but the removal and replacement of these was still a very demanding process that involved relatively tight space as shown in Fig. 28.

Fig. 28

Re-installation operations of the TWTA and CTP.

3.6.4.

Soft structure risk of billowing during launch

As described earlier, changes to soft structure on JWST were more consequential than on previous spacecraft and had to be carefully considered. The shape, position, and imparted stresses from these members had much tighter tolerances than on previous spacecraft where soft structure MLI can usually be cut and tailored in the final stages of I&T without concern. Aside from the changes to the primary mirror frill and bib that was already described, other “soft structure” issues were identified late in the I&T program. Most late soft-structure changes were associated with venting behavior during launch and ascent. Soft-structure issues involving deployments are covered in Sec. 3.6.5.

The most significant of the venting issues came as result of late breaking data from other launchers that showed the residual atmospheric pressure within the launcher’s fairing at or near the time of fairing jettison was not as low as thought. Data obtained in late 2016 indicated pressures could be higher than 2 milli-Bars, which raised concerns for the loads that would be imparted to the stowed sunshield membranes from billowing from the rapid depressurization at fairing jettison. This billowing exerts loads not only on the membranes but also on the pins of the MRDs that holds them down. Damage to these pins could result in a failure to release, and thus represented a significant risk to the mission.

Ariane was informed of this concern and agreed to fly US-provided pressure sensors with high enough precision for accurate measurements of these low residual pressures. These sensors showed pressures as high as 0.58 milli-Bars existed just prior to fairing jettison.

In conjunction with Arianespace, the project kicked-off a three-pronged effort to address this risk that consisted of:

Arianespace was successful in improving their venting efficiency via modification of the vent valves, systems engineering passed revised requirements to NG to establish the residual atmosphere at fairing jettison, and NG was successful in qualifying their sunshield hardware to these new requirements. The entire exercise took approximately 2 years. The exercise used data from five Ariane flights prior to JWST, with modified vent valves on the last two launches. The data from the JWST launch showed that the pressure was under the revised pressure requirement and no discernable damage was incurred to any sunshield membrane from the launch depressurization.

Other than depressurization at fairing jettison, the more conventional venting issues associated soft structure was realized early, and design practices were instituted to address them. But as the design matured, and the as-built blanket geometry was examined closely, it became clear that details of the blanket tie downs did not correlate with those that were modeled to adequate level of detail. As late as May 2021, modifications had to be made to the blanket tie downs on the sunshield’s unitized pallet structures (UPS’s) to make sure they conformed with load predicts made by updated venting models. This change is an example of late changes where changes to the tie-downs of the UPS blankets were necessary because of deviations from the configuration in the venting models.

3.6.5.

Deployment, membrane snags

A full-scale engineering model of the sunshield, called the integrated validation assembly, was used for development and practice of sunshield deployments, as shown in Fig. 26. Even with this high-fidelity engineering model, there were still unexpected issues discovered during the deployment tests of the flight sunshield and OTE on the ground. For the sunshield, there were two major deployment issues that surfaced: (1) a significant snag of one of the membrane tensioning subsystem (MTS) cables and (2) an anomaly in the firing of the redundant firing circuit of a 1/8” NEAs in one of the 107 MRDs.

The MTS snag incident occurred during the first deployment test of the flight sunshield in November 2017. During the tensioning of the sunshield layer 1, a spring clip (a curled piece of Kapton) used to control the position of a tensioning cable, held in place longer than expected and negator springs in the path of this cable reached their full stroke. When the clip finally let go, there was a sudden generation of cable slack, which looped around and snagged around the negator spring housing of a neighboring cable. Such a snag on orbit would have been a severe failure for the sunshield deployment. Design modifications were implemented in the form of snag guards in the cables near these negator springs and resizing of the Kapton clips to adjust their release strength. These late changes were simple in nature but since they occurred late in the I&T flow, their actual implementation was challenging. In addition to these changes, the sunshield engineering team conducted audits of the slack management of all the tensioning cables and made changes to their routing.

The second anomaly involved the second deployment test of the flight sunshield. This test was after the sine vibration and thermal-vacuum tests of the system. The 1/8” NEA, which are used to release the 107 MRDs, were fired only on their redundant firing circuits. One of the NEAs failed to release, as shown in Fig. 29. When fired on the primary firing circuit, the NEA did release, and so this was not characterized as a failure since both circuits are fired on orbit. This was nonetheless an important anomaly that triggered an extensive investigation of all the failure modes of these devices, which were considered to have flight heritage.

Fig. 29

1/8” NEA anomaly in the second sunshield deployment test.

Since these NEAs are used in 107 MRDs all of which are SPF Items, it was important to maximize their reliability. The investigation surfaced several previously unknown failure modes, which resulted in minor design changes to the NEA device itself and the implementation of revised assembly and inspection methods.

3.6.6.

Integration and testing anomalies and reporting

Documenting and managing anomalies was a significant effort particularly during the observatory integration and test phase when final deployments and stowing and inspections occurred. The Northrop Grumman mission assurance processes for Anomaly Review Boards (ARB) and Failure Review Boards were used and the Northrop Grumman Mission Assurance Manager oversaw this entire effort. A standing board was formed that reviewed and approved every ARB and FRB activity. This formal board included key technical leads from Northrop Grumman and the NASA mission assurance manager, deputy project manager for technical, and the observatory I&T chief engineer. In addition, the lead mission system engineer, ISIM system engineer, and I&T chief engineer (formerly telescope manager) reported to the Goddard Engineering leadership monthly and the lead MSE had ultimate technical authority and reported to the NASA chief engineer.

3.7.1.

Fine sun sensor glint and fault management

Very early in commissioning, the Observatory entered safe mode, and followed a nominal fault management response where the autonomous software algorithms swap many of the systems to redundant hardware and electronics. The reason for entering safe mode was determined, corrected and on recovery of the primary-side configuration, the JWST Fine Sun Sensor (FSS) located a signal other than the sun while performing a start-up scan of the FSS detector. This “fake sun” was realized by the on-console team and quickly addressed, but an investigation was made into the possible “fake sun” root cause and mitigation to this as an FSS failure mode. This investigation could not conclusively identify the source, but slewing the Observatory to a different sun vector, where it is thought that a glint with a complex path was eliminated or diminished, allowed the FSS to find and track the true sun. For long term mitigation of this issue reoccurring, or to mitigate it happening on the redundant FSS, prime and redundant FSS’s are now always left powered. This allows them to continuously track on the same sun. Frequent checks of the values from the two units are made to ensure that they are both finding the same sun vector.

The lesson learned from this experience extends beyond the failure mode of the FSS and looks at the overlap of the fault management and the hot/cold unit spare philosophy. JWST fault management simplified responses as much as possible to ensure that any possible fault was addressed by the initial response. This means that the algorithm responded as though any fault was the result of the worst possible condition that could cause that fault and it did not require two pieces of telemetry to initiate that full response. Although the JWST fault management responses were extremely well reviewed, a significant reduction in risk during a fault management response, or recovery, was achieved by a small change to the hot/cold spare state of the FSS. Anytime a fault management response, and therefore the recovery, requires a significant number of systems to be powered up on the primary/redundant side, risk is added to the initial fault management response as well as to the recovery.

3.7.2.

Solar array performance in early commissioning

A lesson learned was also realized on thermal modeling. Through the years of design and analysis on JWST, the configuration of “model runs” had to be carefully chosen as they informed and entire modeling cycle where one analysis result fed into others. It was not possible to perform all possible conditions due to the length and cost of modeling different parameters in the complex integrated modeling environment. It is typical to take the worst-case operating extremes to do these models, so for thermal this often winds up as a “worst-case cold” model and a “worst-case hot” model. Add to this the large number of configurations that JWST would go through during deployments and the number of cases, many of which represent transient states, expand considerably.

Even though it was a large number of unique configurations, analyses were done to assess the health and safety of the Observatory while in prolonged partially deployed configurations. These special studies were conducted to confirm that anomalies could be investigated and resolved without the constraint and pressure of time limits. Of specific concern in these partially deployed configurations were excessive momentum build up, off nominal and structurally stressing cool down temperatures, and water ice migration to cooling optics. Although a product of these analyses, solar array temperatures in these partially deployed configuration were never evaluated for their impact on array power output. Although there was no actual thermal concern with the array in the mid-deployment configurations, the assumptions for calibration of the regulators on the solar array assumed a different thermal state than was seen in flight in this time frame. The operations team identified sub-optimal performance in the solar array indicated by a small current being pulled from the battery. Once the issue was identified and a final thermal state of the solar array was achieved, the regulators were calibrated, and all solar array operation has been nominal.

3.7.3.

Deployment indicator switches

Launch-locked, stowed, and deployed items on JWST have telemetry to indicate the state of the subsystem. Some states have a two-condition telemetry point, meaning that an item is either locked or unlocked or stowed or unstowed. In particular, deployed items have multiple telemetry points to designate its current state. Such telemetry may indicate, a locked, unlocked, and deployed state as well as telemetry indicating how many motor steps have been executed. In ground testing, these deployments and the telemetry that indicated the success of those deployments, were tested, trended, and rehearsed. However, testing in the cleanroom, under Earth’s gravity and at room temperature, cannot always fully check out the on-orbit range of conditions. This was seen on-orbit where some of the “deployed” indictor switches did not indicate that hardware had properly or fully deployed. When this occurred, the team fully analyzed the data from the Observatory and ran a parallel case on the full-size sunshield model. From these ground model tests, it was determined that there was good rationale to believe the on-orbit deployment was successful despite the anomalous telemetry condition. For example, the sunshield’s membrane covers were successfully deployed, but, due to on-orbit conditions that were not simulated on the ground, the sensors did not properly indicate a status of “deployed.”

3.7.4.

Stray light in commissioning images

During commissioning, a deficiency in how the stray light model efforts were divided between an effort in the Observatory level model and the instrument level models was found. On such a large Observatory, built by distributed organizations, and where much of the system was designed in parallel, managing design changes in interfaces in the modeling cycles was a challenge and this resulted in a stray light path into the instruments after a design change in the optical prescription of the telescope was made. The changes to mitigate this path were analyzed using the Observatory stray light model. This showed that the stray light path from the telescope did not enter the entrance pupil of the instruments. However, light did enter the ISIM interior and did impact non-optical surfaces of the structure and instruments. The observatory level model used for stray light analysis only included the optical trains of the instrument and not any of the structure of the instruments. The number of surfaces that full geometry models of the instruments would add to the observatory level model would make the observatory level model unwieldy. The instrument teams were asked to take look from their instruments out to the newly proposed geometry in the aft-optics of the telescope and determine, with alignment uncertainties, if the geometry change fully mitigated the issue. What was missed in the inward-outward approach is the low-angle scatter from light impinging on non-optical surfaces that entered through the observatory’s stray light path. In the infrared, even surfaces coated with low emissivity black treatments can still reflect a significant percentage of the photons.

This was first found by the NIRCam team in commissioning images and an extensive investigation lead to the discovery that bright sources located in a narrow field relative to the Observatory stray light path could scatter off instrument baffles and interior surfaces of optical mounts onto the NIRCam detector. There was also a possible path that entered the Observatory’s stray light path only after scattering off a blanket on the telescope’s structure. Although these multiple-scatter paths transmit a very low percentage of the energy from the sky to the detector, a bright source with JWST’s highly sensitive detectors still made the level of the scattered light noticeable. For NIRCam, these features moved with the location of bright sources on the sky and extensive work has been performed to characterize these for calibration purposes. Observation planning can largely avoid placing bright stars in the susceptible regions relative to the NIRCam field. Details can be found in the JWST on-line user documentation.³²

Features were also found within the NIRISS instrument during commissioning and correlated with a low-angle scatter off non-optical structures in the instrument from the stray light path through the telescope. The feature specific to this path has been called the “light-saber” due to its appearance. Further information on the NIRISS stray light features can be found at in the JWST on-line user documentation.³³

3.7.5.

Line-of-sight performance

As mentioned in a prior lesson learned, the effective use of the delays during Observatory level integration and test yielded positive results in several places. A clear positive result was the enhanced simulation capability at the operations center for testing the interactions between the attitude control system, the FGS, and the onboard script subsystem that dealt with the line-of-sight management of the telescope. This enhanced simulator fidelity led to several improvements in the various software and operating scripts to correct for timing or configuration-related potential faults, which were found and corrected prior to launch. During commissioning, additional work could be rapidly conducted on this system to verify operations and correct issues. Because of the issues that were found and resolved early with this enhanced simulator, the line-of-sight closed loop control system, which could not be tested end-to-end on the flight hardware on the ground, worked extremely well. This system was performing well from the point in commissioning where it was critically needed to continue with telescope alignment and its success led directly to the excellent and timely release of JWST’s first images to the public on July 12, 2022.

3.7.6.

Micrometeoroids

The open-architecture of the JWST optics, which is working well from all aspects of the design, was expected to be degraded slowly by micrometeoroid damage. Early design decisions on this architecture were based on the best models of the time on the size, frequency, and speed of micrometeoroids moving through the JWST L2 orbit. Experimental damage testing was carried out on mirror samples to ensure that the mirrors would survive the expected flux of micrometeoroids. The testing on the ground cannot fully simulate the velocities in space and the modeling methods that existed when this was assessed were limited. JWST is actually able to sense micrometeoroids both using pupil images and wavefront sensing (measurable in wavefront) in ways that are more sensitive than has ever been performed. Results to date indicate the number and distribution of micrometeors is consistent with the expected environment for the JWST orbit, but one incident on the C3 mirror created higher damage than was expected for an individual hit over a 5-year life. The most likely explanation for the unexpected nature of damage of this particular micrometeoroid impact is the limitation in being able to model the damage at the early phase of the program combined with the sensitivity of location on mirror and the statistical nature of very high energy micrometeoroids. To minimize the likelihood of future large hits like the one on C3, a micrometeoroid avoidance zone³⁴ has been adopted that reduces telescope pointing in the orbital direction, called the ram direction, where the orbital velocity increases the relative velocity with the micrometeoroid and thereby increases its damage. With this and the significant optical margins, the telescope optical WFE and performance is expected to meet requirements for decades.³¹

3.7.7.

Deep space network availability

The last lesson learned from JWST commissioning extended into the first months of JWST Cycle 1 science as the launch and mission of Artemis 1 overlapped and used the same ground station network as JWST. The overlap in subscription of this asset and low efficiency in the downlinks on this asset due to a variety of compatibility issues resulted in JWST science having to pause to wait for the data to be off-loaded from the flight recorder. The space communications service provided by the DSN is managed by NASA’s Space Communications and Navigation program (SCaN), which serves as the program office for all of NASA’s space communications activities. A joint study was initiated between SCaN and NASA’s science mission directorate (SMD) to understand the issues encountered by JWST and to make recommendations for near- and long-term strategies to address the found issues for JWST, DSN, and for future SMD missions using DSN assets.³⁵

The DSN consists of three facilities spaced equidistant from each other—$\sim 120\mathrm{deg}$ apart in longitude—around the world. These sites are at Goldstone, near Barstow, California; near Madrid, Spain; and near Canberra, Australia. The strategic placement of these sites permits virtually constant communication with spacecraft as Earth rotates.

The study team was co-chaired by JWST and Space Communications and Navigation leadership and consisted of a team of subject matter experts representing different areas of interest. The issues were found to fall into at least one of two major themes. They were either due to a lack of true understanding of the needs, assumptions, or requirement of either side during the mission planning phase or they could be attributed to the DSN’s ongoing challenge to balance limited supply of assets while there was a growing mission demand.

The report detailed the issues identified and laid out recommendation for JWST and DSN to make both near- and long-term improvements to the current situation. Many of the longer-term improvements are geared toward future missions and, specifically, planning around the Artemis use of the system as it is a critical manned mission likely to directly overlap with JWST science data throughput.