Skip navigation

Gravity Probe B

Testing Einstein's Universe

Relativity mission requires custom-built detector mounts

Low-temperature system provides thermal insulation between detector electronics operating at 80 K and a quartz telescope operating at 2.5 K.

Reprinted from Laser Focus World March, 1998
Mark Sullivan and Matthew Bye

Gravity Probe B (GP-B) is a relativity experiment to measure the geodetic and frame-dragging precessions of a gyroscope placed in a 650-km-altitude polar orbit about the Earth. The measurements are to be made relative to an inertial reference frame provided by the fixed stars. For Einstein`s general relativity, the precessions are calculated to be 6.6 arc sec/year for the geodetic precession and 0.042 arc sec/year for the frame-dragging precession. The goal of the GP-B experiment is to measure these precessions to better than 0.01% and 1%, respectively.1

FIGURE 1. In the Gravity Probe B Science Instrument, four gyroscopes are positioned along an axis in a fused-quartz block. The fused-quartz telescope orients the GP-B spacecraft by tracking a guide star.

To perform these measurements, the GP-B science instrument consists of four quartz gyroscopes aligned in a fused-quartz block (see Fig. 1). A fused-quartz, star-tracking telescope is attached to the block and orients the instrument by tracking a guide star. A sensitive, superconducting sensor is used to accurately measure the axis of rotation of each gyroscope with respect to the quartz block.2 This information is compared with the absolute angle reference provided by the telescope and its read-out electronics. The science instrument is operated in a cryogenic, low-noise environment provided by a low-vacuum (<10-13 torr); a superfluid, liquid-helium (lhe) dewar operating at 2.5 k; superconducting shielding; and a drag-free satellite. a pure gravitational orbit for the satellite is determined by using displacement signals derived from the gyroscope suspension system. this gyroscope acts as a drag-free proof mass.

The star-tracking telescope is a modified Cassegrain design with a 144-mm aperture, an overall length of 505 mm, and a focal length of 3810 mm (f/27). The telescope has an ellipsoidal primary mirror, a spherical secondary mirror, and a spherical tertiary mirror. Starlight enters the telescope through a forward plate, proceeds through the imaging optics, and is sent to an image divider. A beamsplitter splits the photon beam. These two beams fall onto two sharp, roof-edge prisms that split each beam again.

The two sets of divided light, now four separate beams, proceed to two dual-photodiode detectors. By measuring the relative intensities of each of the two beam pairs, the telescope read-out electronics tell the spacecraft attitude-control system how much to adjust the spacecraft orientation. One detector set senses light intensity for pitch position. The other set senses light intensity and provides yaw position. The GP-B star-tracking telescope will provide an inertial reference for the science instrument with accuracy approaching 100 �arc sec/year.3

FIGURE 2. In the detector packages installed on the GP-B telescope, starlight is divided into four beams in the image divider. Two sets of beams are sent to the detector packages to control the pitch and yaw of the spacecraft.

GP-B detector packages are 90� apart to receive light from the image divider (see Fig. 2). The detector packages must be compatible with the all-quartz, cryogenic telescope and not obscure the clear aperture. Detector electrical signals flow through cables and connectors to the Dewar probe (not shown).

Design considerations

The essential design problem is to thermally isolate a warm electronics assembly from the rest of a cryogenic telescope and instrument. The telescope read-out electronics use silicon junction field-effect transistors (JFETs) because of their extremely low-noise characteristics. Unfortun ately, silicon JFETs do not operate reliably at temperatures below 60 K. This is not welcome in a 2.5 K instrument for which superconducting critical temperatures and cryogenic lifetimes are very important.

The table below lists some of the detector-package requirements. The principal tasks are limiting the power dissipation to less than 1 mW per detector set while also achieving an adequately stiff support structure. Because the JFET amplification is temperature-dependent, the temperature stability of the detector platform must be better than 2 mK. These requirements, and others, are made more difficult by the relatively narrow, annular volume allowed to the detector packages.

The first resonance of the detector package and its subassemblies must be at least 100 Hz (see table). Because the entire instrument will be launched on a Delta II rocket, it must also survive a random vibration environment of 5.0 grms.

Alignment repeatability requirements for the detector package are 10 arc min in pitch and yaw and 30 arc min in roll. Defocus and decenter requirements are 500 �m. These alignment requirements must be maintained after the instrument is cooled to its operating temperature, an excursion of 296 K.

Because of the extreme sensitivity of the GP-B gyroscope read-out sensor, the detector package also provides electromagnetic interference (EMI) shielding and has a low-remanent magnetic moment.

The evaluation of candidate materials for the isolator started with commonly used, low-temperature, low-conductivity materials. Altex graphite/epoxy composite has a thermal conductivity greater than fused quartz at room temperature. However, at 4 K its thermal conductivity is less than that of Teflon. Kapton, while no better than commonly used G-10 graphite/ epoxy at room temperatures, is 5.5 times better than G-10 at 4 K. Kapton was ultimately chosen for its low thermal conductivity and acceptable dimensional change.

The power-dissipation requirement of 1 mW coupled with a vibration requirement of a 100 Hz fundamental frequency implies a light, stiff structure. We initially considered straight-legged, rod-type standoffs and then made some attempts at creating an isolator based on the Stewart Platform. Ultimately, fabrication and structural complications led us toward cylindrical shells.

Design approach

FIGURE 3. Exploded view of the detector package shows the optical subassembly and detector mounts attached to a titanium housing. Various EMI shields and electrical isolators are also shown.

FIGURE 4. Side/section view of the detector package shows light entering from underneath the package through the plano-convex lens, which focuses the telescope light onto the detectors. Each package has a redundant detector mount, achieved through use of a beamsplitter.

FIGURE 5. On the detector mount, the dual- photodiode detectors and JFETs are thermally sunk to a single-crystal sapphire substrate. The substrate is mounted to a larger, alumina platform. The platform is thermally isolated from the titanium base by a Kapton thermal isolator.

In the design approach taken for the GP-B detector package, a titanium housing encloses an optical subassembly consisting of a plano-convex lens and a beamsplitter (see Fig. 3). The plano-convex lens takes the two f/22 beams coming from the image divider and focuses them on the detector sets 12 mm away. The beamsplitter takes the f/5 light from the plano-convex lens and sends half to the primary detector set and half to a redundant detector set. Each detector mount is individually shielded to limit EMI from propagating to other sensitive subsystems in the instrument.

As previously described, the light gets refocused by the plano-convex lens and then proceeds to the dual-photodiode detector sets (see Fig. 4). The beamsplitter simply allows a redundant detector set in each package. Also, there is a semikinematic attachment scheme for the detector package to the telescope. A relatively broad flat, relieved in the center, constrains pitch, yaw, and defocus. A second, narrower flat constrains roll and x-decenter. Y-decenter is limited by the fit between the shoulder screw and the hole in the quartz post.

The detector mount provides the thermal isolation necessary between the detector circuit and the rest of the telescope (see Fig. 5). The circuit substrate is single-crystal sapphire to achieve the thermal stability requirement of less than 2 mK. The platform on which it sits is alumina.

The isolator is 50-�m-thick Kapton polyimide film with titanium/gold traces connecting the circuit to the pins on the base. The 12 mm diameter and 11 mm height are compatible with the telescope packaging and vibration constraints.

The base is titanium for thermal compatibility with the housing. It is the interface to the rest of the detector package and the thermal ground for the detector mount.

The thermal story is obviously not finished at the detector mount base. To bring the heat from the detector mount to an appropriate thermal sink, a multifunction detector cable is used. The cable has three simultaneous functions: electrical connection between the 24 pins on the detector mount and the connector on the Dewar probe, thermal connection between the detector mount base and the LHe Dewar, and EMI shielding between the detector signals and other extremely sensitive components in the science instrument. Experimental results We made use of simple 1-D conduction and radiation models to help choose geometry and materials. The conduction model was set up in a spreadsheet and used to verify our initial design. Finite element analyses have predicted a fundamental frequency of the detector mount in excess of 400 Hz. This has been qualitatively verified by tapping the completed assemblies. As long as the isolator is not buckled, the cylindrical structure is remarkably stiff.

FIGURE 6. Plot shows detector circuit temperature vs. power input. We currently reach 61 K for 1 mW of power dissipation.

To date, we have built four functioning units using a previous four-leg design. A plot of the measured platform temperature as a function of total power dissipation is shown in Fig. 6. The total power dissipation on the circuit is the sum of the photodetection circuit and a heater resistor. The circuit dissipates 0.8 mW of power, and the balance is provided by the heater resistor. During flight, the heater can be operated from 0 to 0.2 mW. Because the JFET noise characteristics and gate leakage current are temperature-dependent, it would be desirable to have a wider temperature range in which to find an optimum operating point. From the graph, it is evident that we can operate between 55 K and 61 K and stay within our 1-mW requirement. The current three-leg design should show improved isolation and an increased operating range. o


The research team for this project included D. DeBra, P. Ehrensberger Jr, E. Romero, P. Limtiaco, D. Gill, and J. Grant from Stanford University (Stanford, CA); H. Demroff, S. Fletcher, L. Huff, A. Jefferson, and K. Tribes from Lockheed Martin Missiles and Space (Palo Alto, CA); J. Goebel of NASA Ames Research Center (Moffett Field, CA); D. Davidson of Optical Instrument Design (Etiwanda, CA); and A. Kashani of Atlas Scientific (Sunnyvale, CA). Additional help and encouragement were also provided by B. Farley, K. Bower, J. Turneaure, K. Coleman, J. Gwo, B. Taller, L. Novak, N. Scott, and J. Wade. This work is supported by NASA Marshall Space Flight Center under contract NAS8-39225.


  1. J. P. Turneaure et al., "Development of the Gravity Probe B Flight Mission," Adv. Space Res. (1996).
  2. B. Muhlfelder, J. M. Lockhart, and G. M. Gutt, "The Gravity Probe B Gyroscope Readout System," Adv. Space Res. (1996).
  3. D.-H. Gwo et al., "The Gravity Probe B Star-Tracking Telescope," Adv. Space Res. (1996).