Replacing the STEP Design with DC or AC Cancellation

The PDF papers on this page may be downloaded directly from this website with the Adobe Acrobat Reader.

Testing the Equivalence Principle (EP)

A high-precision (10^-19 to 10^-23 g) EP test is one of the most important experiments left to be done in Gravitational Physics. The EP has already been tested to about 10^-13 g, and one would think that we should simply call it zero and forget it. There are many reasons why this is not true and why a satellite test of the EP should be done. See the EP page in this website or Damour, gr-qc/9606080.

STEP

Around 1970 Paul Worden investigated the low temperature version of the STEP experiment for his Ph.D. thesis in Physics at Stanford. In order to solve the problem of semimajor-axis errors he turned to the single-axis differential accelerometer. By spinning the satellite perpendicular to the accelerometer axis, he was able to distinguish between an EP violation (which gave a signal at spin frequency) and a centering error (which gave a signal at twice spin frequency). By servoing the difference in the positions of the two proof masses using the twice-spin signal, he was able to reduce the centering error to almost zero. This was an excellent solution to an important problem.

The STEP design as it is presently conceived consists of several single-axis force-rebalance differential accelerometers in a spinning Drag-Free satellite. Each differential accelerometer uses two nested cylindrical proof masses. In order to avoid a large Radiometer error (to which cylindrically shaped proof masses are especially sensitive), read out the positions of the proof masses, and implement the single-axis support; the entire accelerometer assembly is cooled to liquid Helium temperatures. See the STEP webpage.

Problems With the STEP Design

STEP as it is presently conceived has a number of problems compared to an approach which would use two free-falling concentric spheres in a Drag-Free satellite at ambient temperatures, DC Cancellation.

Some of these are:

1. The design of the STEP experiment looks more complicated than GP-B. If this turns out to be true, it bodes ill for the project. If GP-B flies in 2002, it will have taken 42 years and cost upwards of $ 800 million. Furthermore complexity severely impacts reliability. By contrast the DC-Cancellation design is maybe 2 or 3 times as complicated as DISCOS which required 3 years to construct and cost only a few million dollars. In addition, DISCOS worked perfectly the first time and exceeded its design specification by a factor of two.

   Part of STEP's complexity arises because there are several accelerometers of different materials in one satellite. Multiple measurements with different materials is important, but not until the size of the effect is first known. And there is probably a better and cheaper way to do this. See below.

2. The single-axis design requires that forces and torques be applied to the two cylindrical proof masses to guide them in the single axis and implement the force-rebalance positioning. In contrast, nested spherical proof masses in a three-axis Drag-Free satellite are completely free floating with no forces or torques deliberately applied to the proof masses.

3. With single-axis bearings the required disturbance specification necessary for 10^-18 g cannot be guaranteed in advance of the flight. For free-floating proof masses, much lower differential disturbances can be guaranteed by the DISCOS flight experience, the expected centering errors of the two spheres, and the expected performance of modern Drag-Free satellites.

4. Prior single-axis Drag-Free experience has demonstrated that a single-axis system is more difficult to implement and less reliable than a three-axis free-floating design. The engineers who worked on the single-axis satellites believed in the beginning that they would be simpler and easier. It turned out that they were more difficult than the three-axis version. Today they say that in retrospect, it would have been better to stick with the three-axis version. (See the NOVA satellites).

5. There is previous experience with the STEP differential accelerometer design which suggests that the problems are much more difficult than has been estimated. A laboratory version of the STEP instrument was built with the goal of achieving a measurement at 10^-12 g. The actual experiment yielded an accuracy of 10^-4 g, a difference of 8 orders of magnitude. (P. W. Worden, Jr., Equivalence principle tests in earth orbit, Acta Astronautica, Vol. 5, pp 27-42. Pergamon Press 1978.)

6. If a Helium-temperature system is compared with a Helium-temperature system, DC-Cancellation is four or five orders of magnitude more accurate than STEP. Even a room-temperature implementation of DC Cancellation, however, is about 3 times as accurate as STEP.

Comparing Different Materials

In order to correctly interpret an EP experiment, it is important that several measurements with different materials be compared (Damour, gr-qc/9606080). A two-sphere free-falling experiment cannot do this in one satellite, but there is a better way.

It is possible to make multiple measurements in several satellites rather than putting several instruments in one satellite. The experiment can be repeated in several flights with different proof-mass materials in each one. At first thought it might seem that this is a more expensive approach than multiple accelerometers in one satellite, but prior experience with satellite programs suggests that this is not the case.

About 40 percent of GP-B's development difficulties arose because the experiment used multiple gyroscopes in one satellite requiring space-qualified Electrically Supported Gyroscopes (ESG's). One of the gyroscopes in the current GP-B design is Unsupported, i.e. kept from contacting the cavity walls by the Drag-Free controller. Since only one mass can be Drag-Free, there can be only one Unsupported Gyroscope; and the rest must be electrically supported. The difficulty of developing the extra ESG's would imply that about $ 300 million extra was spent to do this. Instead of putting everything in one vehicle, four satellites could have been flown without any ESG's. Assuming approximately $ 40 million each for three extra satellites, a multiple-satellite approach could have reduced the program costs by about $ 180 million. If the program had decided to pay for extra launches, this way could have greatly increased the reliability of the GP-B mission; since any design error could have been corrected before the next launch. Alternately the full savings could be realized by launching the four simpler satellites on one launcher as was done with Globalstar.

There is actually prior experience with satellites which supports these contentions. Globalstar, for example, launched 52 identical satellites costing about $ 15 million each. Although all 52 satellites are now fully functional and operational, a number of serious design errors were discovered in the first 4 satellites after the first launch; and it was not until the third launch that all design problems were identified and solved. This was in spite of the fact that the Globalstar satellites were classical vehicles using known and tested technology. Globalstar managed to find very clever work-arounds to make the first eight satellites operationally successful, but such solutions might not exist for other programs. The Globalstar program has demonstrated that it is very easy to launch multiple satellites on one launcher and to put up a large array of satellites.

It makes good sense to first discover the size of any EP violation, and then to investigate its nature rather than to put a lot of effort into a complex multi-mass design without knowing if an EP violation lies within the range of its sensitivity. If the first flight of a simple two-mass EP experiment failed to detect any effect at all, future flights could be modified to take this into account. For example, if a room-temperature DC-Cancellation experiment saw no EP violation at 10^-19 g or higher, it would then be reasonable to spend the extra money and effort to fly a Helium-temperature version which could then measure down to about 10^-23 g.

In the case of an EP experiment using DC Cancellation another possibility exists. The experiment could be launched from the Shuttle or the Space Station. This would open the possibility of flying a mission, retrieving the satellite, changing the proof-mass materials and launching it again. Missions of this type have routinely been done with the Shuttle using the SPAS series of satellites. These were launched overboard from the Shuttle and retrieved at the end of the mission. The breathtaking IMAX movies of the Shuttle in space were taken from ASTRO-SPAS (PDF). See page 6 for a view of the Shuttle from ASTRO-SPAS. The caption on the page says: "ASTRO-SPAS, Reusable Scientific Satellite for Short-Duration Missions with the U. S. Space Shuttle".