Photo: Ron Laub

Introduction

The CSO and the JCMT were  linked together to form the first astronomical interferometer operating at submillimeter wavelengths. The telescopes are located at an altitude of 4100m on the summit of Mauna Kea, Hawaii, separated by a baseline of length 164m, giving a fringe-spacing of 1.1 arcseconds at 345 GHz.

Two element interferometry with the CSO  and the JCMT is now no longer performed since submillimeter interferometry on Mauna Kea can now be accomplished with the Smithsonian Submillimeter Array.  Eventually we expect that the CSO and the JCMT will be integrated into the SMA for performing observations when increased collection area and longer baselines are scientifically useful.

For historical reference here follows a brief discussion on the technical design, the baseline geometry, the correlator, the system temperature, sensitivity, etc, relating to the now obsolete two-element CSO-JCMT interferometer.

Technical outline

The interferometer used the existing heterodyne receivers at the two telescopes to down-convert signals to an Intermediate Frequency (IF) in the range 1-2 GHz. The two local oscillators were highly coherent for interferometry to be possible, so they were phase-locked to a common reference frequency that was transmitted from the CSO to the JCMT over a wideband fiber-optic link. Optical fibers were also used to deliver the two IF signals from the receivers to the digital correlator at the JCMT. In general, astronomical signals arrive at one antenna before the other, and delay compensation is thus needed before correlation. Coarse delay compensation was realized by switching appropriate lengths of optical fiber into the signal paths. Finer corrections were made in the software during post-processing of the data. The rotation of the Earth causes the sources to move through the fringe pattern of the interferometer producing a fringe rate of up to ~15Hz for observations at 345GHz. Numerically-controlled oscillators were used to generate offset frequencies which are introduced into the LO and IF chains in such a way that this fringe rate was removed. Phase-switching at a fixed rate was also introduced which helped to remove any offsets in the correlator.

This arrangement naturally ensured that only one of the two sidebands generated by the receivers was correlated. The fringe rate can only be removed precisely for one frequency (chosen to be the center of the IF band) and this sets a limit on the maximum integration time for individual records of around 10 seconds.

The interferometer used heterodyne receivers in the 220-270GHz and 318-360GHz bands. Observations around 460-500GHz were also possible, but the performance in this band was very uncertain and was  extremely weather dependent.

Baseline geometry, resolution and applications

The components of the baseline (going from CSO to JCMT) are +43.38m in the polar direction, -157.91m in an east-west direction and -1.93m toward the celestial equator. The fringe spacing is given by lambda divided by the length of the baseline projected perpendicular to the source. The minimum fringe spacing (1.1" at 345GHz) therefore occurs when the source passes through the plane perpendicular to the baseline.

With a single baseline, the imaging ability of the interferometer is of course very limited. However the visibility of the fringes as a function of the length and orientation of the projected baseline can be measured and this can provide a good estimate of the size of the source and some information on the shape. By measuring the phase of the fringe, it may  be possible to measure positions of objects relative to known sources like quasars, to an accuracy of perhaps 0.1", but this has not yet been achieved. It is however quite easy to measure the relative phases of different spectral line components in a single source, provided they fall within the bandpass of the correlator. These phase differences give information about the angular separations between the different components.

The correlator

The Dutch Autocorrelation Spectrometer (DAS) at the JCMT was used to form the complex cross-correlation function of the two IF signals. There are 2048 lags which produce 1024 complex frequency channels (although only about 800 of these contained good data because of the bandpass of the filters). The lags could be distributed in many different ways amongst up to 8 sub-bands, each with a bandwidth of about 125 MHz. Some examples of the configurations available are: 800 channels covering 125 MHz with a resolution of 0.156 MHz; 100 channels in each of 8 sub-bands with 1.25 MHz resolution - total bandwidth covered 920 MHz (there is some overlap between the sub-bands); 400 channels covering 500 MHz with a resolution of 1.25 MHz. and the remaining 400 channels covering 125 MHz at 0.313 MHz resolution.

The system temperature and sensitivity

The system temperature was the geometric mean of the CSO and JCMT system temperatures. These were determined by the standard method by measuring the power from the sky and an ambient load, so that the contribution due to the atmosphere is taken into account. Single-sideband for operation at 345 GHz are presently in the range 500 to 1000 K. The value is lower at 230 GHz, but increases going to low elevations (high zenith angles) and near to the main atmospheric absorption lines. The conversion factor from effective antenna temperature to units of flux is expected to be about 50 Jy/k, assuming an efficiency of 50%. The true value on a given observing run is determined by observing quasars as flux calibrator sources.

The interferometer starts to resolve sources of size > 0.5". A thermal source 0.5" x 0.5" must have a brightness temperature of at least 12K to produce a flux of 260 mJy at 345GHz.

The integration time of 10 seconds implicit in the above is the shortest which was normally used. For most purposes the records would be added together (after correcting for the effects of the changing delays) to give longer integrations. There is however a limit to this which is set by the phase drifts due to uncertainties in the baseline, thermal effects and the atmosphere. This is quite often no longer that 100 seconds. To continue integrating coherently for longer than that it is necessary to have a source in the beam giving a signal to noise ratio of greater than 1 which can be used as a phase reference.

The effects of the atmosphere

The interferometer is more sensitive to the weather than the individual antennae. The Earth's atmosphere introduces random phase fluctuations that decorrelate the signals. The rms phase fluctuation depends on the weather conditions, the observing frequency and the zenith angle. There is a strong diurnal effect because the heating of the ground by the sun drives much of the turbulence. In good weather, observations may be possible between ~5pm and 7am, but often the start has to be delayed by and hour or two while the atmosphere settles. Observing beyond 7am is limited because direct sunlight must not be allowed to fall on the CSO's surface. A typical night rms phase fluctuation value is 30 degrees for observations at 345GHz. Sources can be observed down to zenith angles of ~65 degrees, allowing up to 8 hours of observations on a source of moderate declination. In practice the observations of the main source must be interleaved with those of a quasar for phase, flux and passband calibration.

Data reduction

The output of the DAS was processed using software written specifically for the CSO-JCMT Interferometer. This performs the fine delay correction, the passband correction and has facilities for adding and smoothing spectra. The final output is usually either a complex spectrum or an average flux and phase for the specified part of the passband. The output can be written out in the form of a UVFITS file for further processing with other software packages.

For further information contact:

John Carlstrom (jc@astro.caltech.edu) or
Richard Hills (richard@mrao.cam.ac.uk)

or see

  "The CSO-JCMT Submillimeter Interferometer," Carlstrom, J.E., Phillips, T.G., Hills, R.E., Lay, OO.P., Force, B., Hall C.G., and Schinckel,  A.E., 1994, in Astronomy with Millimeter and    Submillimeter Wave Interferometry, A.S.P. Conf. Series

"IF Fiber Optics and Digital Correlating Spectrometer on the JCMT-CSO Interferometer," Lay, O.P.,     Hills, R.E. and Carlstrom, J.E.  1994, in Astronomy with Millimeter and Submillimeter Wave  Interferometry, A.S.P. Conf. Ser.

"Protostellar Accretion Disks Resolved with the JCMT-CSO Interferometer," Lay, O.P., Carlstrom, J.E.,    Hills, R.E. and Phillips, T.G., 1994,  Ap. J. (Letters),  434, L75.

"NGC 1333 IRAS 4:  Further Multiplicity Revealed with the CSO-JCMT Interferometer,"  Lay,  O.P., Carlstrom, J.E., and Hills, R.E., 1995, Ap. J (Letters), 452, L73.

"Resolving Circumstellar Disks with the CSO-JCMT Interferometer", Carlstrom, J. E., Lay, O. P., Hills, R. E., and Phillips, T. G., 1995,RevMexAA (Serie de Conferencias), 1, 355

"Astronomical Interferometry at Submillimetre Wavelengths", O. P. Lay,  Ph. D. Thesis, 1994, University of Cambridge.
------------------------------------------------------------

back to CSO reference page

------------------------------------------------------------