NRL Mark III Interferometer

Figure 1: The NRL Mark III interferometer

The Mark III interferometer was built with support from the Naval Research Lab (NRL), the Smithsonian Astrophysical Observatory, MIT, and the U.S. Naval Observatory. From the start, the Mark III was intended as an operating prototype for an even larger facility. Design, construction, and assembly of the Mark III began in 1982, with the team of Shao, Colavita and Simon, joined by Peter Cheimets, John Hershey, Brad Mines, J. Hughes, Don Hutter, George Kaplan, C. Papa, J. Pohlman, then later David Mozurkewich and Tom Armstrong, plus several graduate students.

``First fringes'' occurred on September 20th, 1986. However, it would be another year before actual observations were begun on a regular basis. Initially, the scientific focus was on astrometry - the precise mapping of the locations of stars in the sky. Once operations, Xiao Pei Pan, Craig Denison, M. Vivekanand, Nicholas Elias, David Buscher, Andreas Quirenbach, Christian Hummel, and others began to operate other observing programs.

Figure 2: Layout of the Mark III

Fig. 3 and 4: The North astrometric hut open and closed.

Fig. 5 and 6: The 10" flat mirror and the entry to the evacuated pipe.

These pictures (Fig. 5 and 6) show the south astrometric siderostat. Through an alt-azimuth mount, the main axis is not vertical but instead leans westwards at a 15-degree angle from horizontal. This arrangement permits the use of a second (fixed) ``feed'' mirror to direct the starbeam steadily into the lab building regardless of the star's position in the sky. This setup also eliminated the polarization problems encountered earlier with the 50-foot Stellar Interferometer. In the photo's center, is one of the flat, ``pick-off'' mirrors.

Figure 7: Optical schematic for the Mark III

The small (9 cm diameter) round window end of the optical pipe is at the top. The huts are lined with rigid foam insulation faced with aluminum foil. Daytime hut temperatures are kept near nighttime levels by air conditioning the hut.

This is an optical schematic of the Mark III. Arrows show the starlight pathways as the go through the system. Initially picked off by the siderostats [(1) and (2)], the two beams proceed through their individual delay lines [(3) and (4)] and are combined at the beamsplitted mirror (5), and subsequently the fringe detector (6).

Figure 8: The Mark III Optical Table

The lab building has two major areas - the optical table and the delay lines. Figure 8 shows the optical table. Upon a plate of Super-Invar steel alloy (with zero temperature coefficient) are mounted many of the Mark III's optical parts. At the lower left, is the window end of the vacuum line bringing light from either the east siderostat or the north arm of the variable baseline.

Figure 9: The delay vacuum tanks

At the lower left of Figure 8 can be seen the window end of the vacuum line bringing starbeams from either the east astrometric siderostat or the north arm of the variable baseline. At right foreground under the black cloth is the photon camera (described below). Above is part of the table cover which is lowered down onto the table during operation. Its purpose is to block out stray light, and to keep air currents away. Each tank has a flat window at one end for the lightbeam to enter and exit. These window ends overlook the optical table and can be seen at the right of Figure 8. Using both delay lines the pathlength can be varied up to 18 meters as needed. Figure 9 shows Richard Simon and Xiao Pei standing next to the delay line vacuum tanks. The shelving to the right holds the photomultiplier units, power supplies, and other monitoring equipment.

Figure 10: The delay line cart

Each delay line basically consists of a moveable platform which mounted optics and the track it rides upon, all enclosed in a 10-meter long vacuum tank. The delay line moving platform itself consists of a cart upon which is mounted a pivoted (back and forth) optical platform. This in turn, holds a parabolic mirror at the far end and a small, flat secondary mirror at the front end, which itself is mounted on the end of a piezoelectric crystal stack (Figure 10). Instead of wheels, the cart rides on a roller-bearing track. Figure 11 shows one of these moving platform/cart units outside of the vacuum tank for maintenance and alignment. The parabolic mirror can be seen at the far end.

Figure 11: A delay line cart

The delay line path control has three degrees of precision: coarse, intermediate, and fine. The coarse is set by the position of the cart along its track inside the vacuum chamber. Intermediate control is provided by the pivoted optical platform holding the parabolic mirror and secondary flat. This platform can be moced along the optica axis about 2-3 millimeters and is driven and controlled by a loudspeaker voice coil mounted on the cart and which pushes against the platform. Fine control (in steps of about 1/100th of a fringe, or about 6 nanometers) is achieved by movement of the secondary flat from electrical currents passing through the piezoelectric crystal stack.

The light beam (which is about 8 cm in diameter) enters the lab through a vacuum pipe coming in from the siderostat. Because the latter is computer controlled, the beam is held steady an on one axis regardless of the star's position in the sky. At the optical table the beam is first directed into the delay line vacuum chamber along one side and is reflected and focused by one side of the parabolic mirror to a point on the secondary flat. There, the beam again spreads out to the other side of the parabolic mirror where is converted back into a parallel light beam and leaves the vacuum chamber. In this way, the beam comes in and leaves the delay line essentially unchanged, except that the length of its path is being precisely controlled.

The computer knows where the cart assembly is positioned inside the vacuum chamber by use of a modulated He-Ne laser metrology beam. This beam is sent into the delay line parallel but offset from the starbeams so that it follows the same optical path. This laser beam is then interfered with itself so that interference fringes are formed. Since the laser only emits at one wavelength, the fringes form throughout the entire length of the light beam. A photoelectric cell mounted at the beams exit fromt he vacuum chamber, monitors and counts the fringes as the metrology beam's pathlength varies.

At the beginning of an observing session, both delay line carts are sent to the far end of the vacuum chamber and stopped at a known fiduciary point. The computer's fringer counter is reset, and it begins counting fringes, thereby knowing the cart's position at all times.

Upon exiting the delay line, the starbeam was sent through an optical (glass) wedge with a 5 cm hole in its center. That part of the beam passing through the hole went on to the beamsplitter and eventual interference. The remaining ring or annulus of starlight around the hole was diverted to one side of the wedge and passed on to the Photon Camera. This allowed the control system to see and track the star under observation. The star images were also displayed as a sparkling blob on an oscilloscope in the control room, with a field of view of approximately one arcminute, enabling the observer to monitor the images, and to gage how well the system was working.

Computer Control

Figure 12

Five computers were involved to achieve automated observations with the Mark III. One computer kept track of star catalogs and observing lists and was responsible for slewing and orienting the siderostat mirrors. Once the star was roughly pointed, two computers took over to finely center the star's light on the optical axis and to keep it there during the entire observation. Next, two more computers controlled the delay lines and detected the interferometric fringes. Often, the primary fringe would not be sensed immediately and the system would tweak one delay line in order to search for it, at the rate of 2 cm per hour! Figure 12, taken in the control trailer shows (from L to R) shows Tom Armstrong, Craig Denison and Xiao Pei Pan at the three control computers. The three monitors are displaying values of the various operating parameters. Figure 13 shows Edward Kim (an MIT Electrical Engineering graduate student) in front of the data acquisition computers. A typical night's run would collect about 150 Mbytes of data.

Figure 13

Making Observations

Depending on atmospheric seeing conditions, the Mark III could operate with stars down to fourth magnitude, fifth under exceptional circumstances. At the beginning of the night, the optical system would need to be aligned using a laser beam which would reflect through the entire optical system. After that a model of the system would be built. This was required on a nightly basis because thermal changes during the day would significantly change from the previous night. To build the model, several stars with known coordinates would be centered on. Typically only 8-10 stars would be required and about 45 minutes of observing time. Under poor conditions, it might take twice as long with several more stars.

At this point, the observational program could begin. Stars would be observed by the system with all the pertinent data recorded and a sky reading would be taken (for calibration). The telescope would then slew to the next star on the list and repeat. After each star on the entire list was observed, the observations would be repeated. The purpose for the repetition was to observe each star in as many position from east to west as it moved across the sky. The wider the arc observed, the more accurate the resolution and measurement of a star's position, diameter, limb darkening, etc.

Scientific Results

Initially, the Mark III was used for astrometry - the measurement of stars' position in the sky. The precision of the measurements from the Mark III gave an improvement of nearly two orders of magnitude. This portion of the project was directed by George Kaplan of the USNO.

Figure 14

In 1988, an additional siderostat pair was constructed. It consisted of a north-south line of twelve concrete piers upon which the siderostats could be moved to as needed. By using different pairing of piers, baselines of 10 to 103 feet (3 to 31.5 meters) were now available. The siderostats were recycled from the Mark II interferometer. Vacuum pipes were brought out from the lab matching the inter-pier distance, so that the evacuated lines could match the different baselines. Figures 14 and 15 show David Mozurkewich positioning one siderostat on a North Arm pier of the variable baseline. The vacuum pipes can be seen above the baseline track.

Figure 15

The addition of a variable baseline to the system now permitted the Mark III to take on greater duties. It was now possible to make observations of stellar diameters, limb darkening, and to determine orbital elements of very close binary stars. An example of one very close binary star orbit measured with the Mark III is in Figure 16. When these elements have been calculated, it is possible to accurately determine the distance to the object to a few percent, generally a very difficult task in astronomy.

Figure 16: The apparent orbit of Alpha Andromedae.

Another pioneering program at the Mark III was the probing of stellar atmospheres to learn how they are structured. This uses the instrument's ability to measure stellar diameters, plus limb-darkening. Using molecular lines, like TiO, which only form at specific temperatures, the Mark III could measure different diameters for different molecular species, thereby determining the temperature at different depths of the stellar atmosphere. This effort began in 1992.

Closer to home, the Mark III was used to model atmospheric turbulence at different levels of the Earth's atmosphere. These data were compared with established models. These results are very important to other projects on the mountain, for example adaptive optics and the new CHARA interferometer which need to cope with variable atmospheric seeing.

Perhaps one of the biggest achievements of the Mark III was its measurement of the angular diameter of a nova, Nova Cygni 1992. Although observations were limited, the distance of 9,200 light years was determined, the first direct measurement of a nova's distance.