2.1.

Overall Instrument Performance

The SXS x-ray spectrometer uses an array of 36 spatially and spectrally independent detector channels to measure x-ray photon energies with high-spectral resolution from across a broad x-ray band (0.1 to 12 keV). Spectral resolution is the primary performance characteristic of the spectrometer and most performance requirements derive from this requirement: the SXS shall measure x-rays to a spectral resolution of $\le 7\mathrm{eV}$ FWHM. The peak temperature of the detectors corresponds to the energy of the x-ray photons they absorb.¹⁷ Since the deposited energy of individual photons is tiny, the detectors must be extremely cold to measure the temperature response, and the reference temperature from which the response is measured must be extremely stable to obtain a precise measurement of the peak value. The requirements summarized below are imposed on the cooling system to enable the SXS to achieve these exquisitely sensitive measurements.

2.2.

Detector Heat Sink Temperature and Stability

Measuring the temperature response to the absorption of individual x-ray photons requires an extremely cold detector. During science observations, the cooling system shall maintain the detector heat sink at 50 mK. Accurate determination of the peak value of that temperature response requires precision temperature control. During science observations, the cooling system shall stabilize the heat sink temperature to $\le 2.5\text{-}\mu \mathrm{K}$ RMS at 50 mK.

2.3.

He Tank Temperature and Stability

The detectors are packaged inside their heat sink known as the calorimeter thermal sink (CTS). The CTS is mechanically connected to the detector assembly (DA) by a Kevlar suspension system and electrically connected to the DA through fine signal wires. The DA is mounted to the helium tank assembly, making the CTS sensitive to variability in the helium tank temperature through parasitic conductive effects. To keep these effects acceptably small, a requirement is imposed to stabilize the DA interface (helium tank) temperature to $\le 1\text{-}\mathrm{mK}$ RMS during science observations.

2.4.

Lifetime and Operational Duty Cycle

The mission requires a design lifetime $\ge 3$ years. This time is sufficient to accomplish a specific set of baseline science measurements, after which the observatory is made available to the broader x-ray astronomy community. The baseline science measurements are nominally completed within the first six months following completion of the in-orbit checkout activities. An overall mission efficiency of 40% is assumed to account for periods of Earth occultation, satellite maneuvers, data loss, and other activities unrelated to the SXS. From this, the following instrument duty cycle requirement is derived: the SXS shall be available for science observation $\ge 90\%$ of the time. With this requirement, “available for science observation” means the temperature and stability requirements above are simultaneously satisfied. Conservatively, the net observing efficiency would be $40\%\times 90\%=36\%$.

2.5.

Stable Autonomous Operation

The mission is designed for a low-Earth circular orbit using a single primary ground station nominally staffed six days per week. This means that the SXS operation must be autonomous and stable for periods longer than 24 h.

2.6.

Cryogen-Free Operation

The SXS must be able to accomplish its science operations, even without the presence of liquid helium. In other words, the requirements above must be satisfied with or without the liquid helium cryogen. This requirement was adopted for SXS in response to the premature loss of cryogen from its predecessor instrument, the X-Ray Spectrometer,¹⁸ flown aboard Suzaku.

7.1.

Basic Operation

In cryogen-free mode, the ADR performs two separate functions. The first is continuous cooling of the helium tank, and the second is single-shot cooling of the detector array. Cooling of the helium tank is accomplished by rapidly cycling S3 between the JT and helium tank temperatures to absorb heat from the tank and reject heat to the JT. During the S3 cycle, HS4 and HS3 are alternately actuated to connect or isolate the JT or the helium tank to S3. Meanwhile HS2 remains powered, except during recycling of S1, connecting S2 with the helium tank. This allows S2 to function as an active heat capacity and stabilize the tank temperature at $\sim 1.5\mathrm{K}$. It does this by absorbing the tank’s parasitic heat load when S3 is isolated by HS3 and pushing that heat out to S3 when HS3 is on and the two stages are connected. The S1 operates in the usual manner for a single-shot stage; it provides detector cooling for a period of time, then recycles by rejecting heat to S2 (which stores and rejects that heat over time to S3).

The main difference for S1 in the cryogen-free mode, compared with the helium mode, is that S2 is warmer in the cryogen-free mode during the S1 hold time ($\sim 1.5$ versus 0.5 K). This increases the parasitic load on S1 through HS1 ($\sim 1.3$ versus $0.05\mu \mathrm{W}$), which reduces the S1 hold time to about 15 h compared with 48 h in the helium mode. To satisfy the observing efficiency requirement (90%) with a much shorter hold time, the S1 recycle time must to be kept to a fraction of an hour.

Unlike the helium mode, in the cryogen-free mode, all of the heat resulting from the ADR operation is absorbed by cryocoolers and rejected to space by radiators. The heat generated by the ADR itself (heat from the salt pills, hysteresis heat from the magnets, and dissipation in the heat switch getter heaters) is absorbed by the JT cooler and passed on to the precoolers, and the ohmic heat generated in the resistive portion of the magnet leads during ADR operation is absorbed by the shield coolers (see Fig. 3).

Because S1 is operated at a lower average current than the other stages, its magnet leads were designed with higher electrical resistance than the others to minimize the total (conducted and generated) heat of the ADR magnet leads. As a result, the operation of S1 produces more ohmic heating than S2 or S3 at equivalent current. The instantaneous worst case occurs during the S1 recycle while S1 and S2 are both operated at elevated current. This was not initially anticipated to have a significant effect on the ADR operation or its heat sink, the JT cooler. But the cryocoolers are operated at fixed voltages; so additional loads raise the shield temperatures, thereby increasing the heat load onto the JT cooler and, through the CSI, onto the helium tank. This circumstance sets up the following critical point in the operation.

Having just absorbed S1’s heat during its recycle, the stored cooling capacity in S2 after the S1 recycle is reduced, yet it needs to continue absorbing the helium tank’s parasitic heat (which is also at its highest in this timeframe). Excess load on the JT raises its temperature and slows the S3 heat transfer prolonging the time before S3 can again extract heat from the helium tank and begin to charge S2 with its excess cooling capacity. If the S2 current ever becomes too low ($<\sim 50\mathrm{mA}$), temperature control will be lost and the helium tank will warm—a condition from which the ADRs cannot automatically recover. The problem is that after the S1 recycle, the system requires the highest cooling power from the S2 ADR to regain temperature control of the helium tank at the same time as its cooling power is at its lowest ebb.

The original approach to cryogen-free operation²² was to gradually build up cooling capacity in S2 over the 50-mK hold and then fully absorb the heat stored in S1 in a single-shot recycle. This can be achieved if the combined S2 and S3 cooling powers exceed the parasitic heat load to the helium tank with enough margin (a few tenths of a mW). Early ground tests confirmed this; when the JT and shield coolers are operated at their nominal voltages, they maintained the IVCS and JT shields at sufficiently low temperature ($\sim 30$ and 4.3 K). The characteristic profile of this cycling process can be seen in the top half of Fig. 6 where the increasing cooling capacity stored in S2 over multiple S3 cycles is manifest as a rising S2 current.

Fig. 6

Magnet current profiles and shield temperature responses with the original and revised ADR control algorithms.

Unfortunately, the early tests of cryogen-free mode were conducted with more favorable conditions than would exist in flight. The configuration change to set up cryogen-free testing required boiling off significant quantities of liquid helium, which overcooled the shields. This was overcome by operating the cryocoolers at reduced power until the shields warmed to their nominal temperature ranges. The oversight was the effect of the overcooled MVCS, which is isolated from the coolers and recovers with a long time constant of $\sim 10$ days. At equilibrium, the MVCS temperature is over 106 K, and at that temperature, its heat load onto the IVCS represents half of the total IVCS load. With the unrecognized exception of the MVCS temperature, the other initial conditions were quickly met and cryogen-free testing began. The tests proceeded as the MVCS warmed, exposing the colder stages to increasing heat loads during the tests. The evolution of heat loads, masked by the ADRs dynamic loading of the shields, was not noticed. What was observed was an inability to maintain temperature control of the helium tank after S1 recycles because the stored cooling capacity of S2 was too low at that point. Figure 7 shows how much the MVCS temperature changed during the cryogen-free tests.

Fig. 7

MVCS warming during cryogen-free tests.

In subsequent, longer-term tests, it became clear that as the system approached equilibrium, the previously envisioned approach to cryogen-free operation was not stable and a different operating strategy, one that is stable at equilibrium conditions, was needed. But, accomplishing that required new priorities among the performance requirements and agreement on which could be compromised. The conclusion that stable operation was more important than observing efficiency created a pathway to an acceptable solution.

7.2.

Revised Cryogen-Free Mode Operation

Through modeling and simulation, it was determined that the key indicator of stable ADR operation was the IVCS temperature variability, and the ADR’s operation and control algorithm was revised to minimize excursions in the IVCS temperature. The mechanism for doing this was to minimize variations in the ADR’s magnet currents, and the principal change was to limit the maximum current that S2 could build up to (1 A instead of 2 A) as S3 cycled. The penalty was that at this lower current S2 could not store enough cooling capacity to recycle S1 in a single transfer of heat. Instead, when S1 required recycling, S2 would absorb heat until its capacity was depleted, then it would be warmed to a relatively high temperature of 2.5 K where S3 could rapidly rebuild its cooling capacity. Once S2 was charged at 2.5 K, it was cooled down to 0.8 K where it could fully recycle S1, allowing it to demagnetize to 50 mK and begin a new hold time.

The algorithm required a much longer S1 recycle time, on the order of 2.5 h, which reduced the duty cycle to 84%. The gain was in reliability, in the form of an ability to recycle S1 regardless of the prevailing JT and IVCS temperatures associated with their nominal operating voltages.

Before a test campaign could be approved to test this modification, it was necessary to demonstrate through simulations that the algorithm achieved the desired level of stability even if the cryogenic system (OVCS, MVCS, and IVCS) warmed during its approach to equilibrium. A high-fidelity model was constructed,²³ which merged a previous model of the ADR²⁴ with models of how the JT and shield coolers and the vapor cooled shields responded to ohmic heat inputs. The resulting model was able to predict the time evolution of the OVCS, MVCS, and IVCS temperatures and transient response of the JT cryocooler as the ADR model generated instantaneous heat loads produced by its operation. The resulting modeled behavior agreed with measured MVCS, IVCS, and JT temperatures from previous tests.

Based on the success of the integrated modeling effort and good correlation with past test results, as well as a fortuitous unplanned break in the project schedule, an additional cryogen-free mode test campaign was approved by the project. The purpose of the new test was to obtain the data necessary to validate and correlate the updated model, and if time permitted to demonstrate the anticipated stable autonomous operation as the system approached equilibrium. Stable autonomous operation meant that once the setup was complete, no commands would be sent to the instrument. These tests would run as long as possible up to an agreed deadline when control of the spacecraft had to be returned to the project.

Using the modified algorithm, testing began with the initial cooldown from 4.5 K to 50 mK, which completed successfully after 6 h. That was followed by two days of engineering evaluations of the updated algorithms. Then the stability test began after the end of a normally completed cycle. The stability test demonstrated that the performance of the SXS instrument in the final cryogen-free mode was stable and required no human intervention even as the dewar warmed and approached equilibrium. Six ADR cycles were completed during this autonomous operating period lasting 104 h before ending in time to meet the agreed test deadline. Figure 8 compares the modeled prediction with the measured MVCS and IVCS temperatures, and Fig. 9 shows the same comparison for all stages over a longer time scale, demonstrating how the stable operation would continue. Demonstrating the repeatability of the system during cryogen-free operation, Fig. 10 shows an overlay of the current in S1’s magnet during the six hold times of this test. Long-term variations in the heat load to S1 would cause the time base for each hold time to be compressed or elongated. Instead, the exact overlap of all six cycles indicates that the helium tank temperature was maintained at a constant, and the same, value (1.525 K) during all six cycles. Moreover, long-term variations in IVCS and JT temperature did not affect the ability of S1 and S2 to maintain the detectors and helium tank at stable temperatures.

Fig. 8

Modeled versus measured MVCS and IVCS temperatures during the stability test and the calculated ADR generated heat loads on the IVCS and JT. The time constants indicated in this figure represent the cool-down rate of the IVCS among S1 recycles.

Fig. 9

Modeled versus measured stage temperatures and magnet currents during the stability test and predicted results thereafter.

Fig. 10

Overlay plot of the S1 magnet current profile during the six 50-mK hold periods of the stability test.

During the science observation periods of this test, the detector heat sink was controlled at 50 mK with a stability of $<0.8\text{-}\mu \mathrm{K}$ RMS, meeting the requirement of $<2.5\text{-}\mu \mathrm{K}$ RMS, and during the same time periods, the DA thermal gasket (He tank) was controlled at 1.525 K with a stability of $<520\text{-}\mu \mathrm{K}$ RMS, meeting the requirement of $<1\text{-}\mathrm{mK}$ RMS. These results are shown in Fig. 11.

Fig. 11

Temperature stability performance during the stability test, requirements are: detectors $\le 2.5\text{-}\mu \mathrm{K}$ RMS at 50 mK and helium tank $\le 1\text{-}\mathrm{mK}$ RMS at $<1.8\mathrm{K}$.

An operational efficiency of 84% was achieved with consistent 2.5-h-long recycle periods, 18-min detector equilibration durations at 50 mK, and 14.7-h-long periods with stable control at 50 mK. With the temperatures well controlled, the instrument achieved a composite spectral resolution of 4.3 eV FWHM showing that the instrument’s spectral resolution is unaffected by differences between the two operating modes. Table 2 compares requirements with measured performance in the cryogen-free and helium modes.

Table 2

Requirements and measured performance of the SXS cooling system in its two science operating modes. Requirements not met are shown in parentheses.