CfA 1.2 m Millimeter-Wave Telescope:
Technical Information

Constructed and operated for more than a decade at Columbia University, the 1.2 m telescope was moved to the Center for Astrophysics in 1986. Regular observations from the roof of Building D began in January 1988. The relatively large beam, agile drive system, and extremely sensitive superconducting receiver make this telescope nearly ideal for conducting large-scale CO surveys.

Summarized below are the specifications of the telescope, with special attention to modifications since its installation at Harvard.


The antenna system consists of a 1.2 m parabolic primary and 17.8 cm hyperbolic secondary in a Cassegrain configuration with effective f/D=2.8. The antenna primary is a monolithic aluminum casting with f/D=0.375, numerically milled by Philco Ford to 40 microns surface accuracy (l/65 at 115 GHz). The telescope's focus, beam pattern, and main beamwidth were most recently measured and adjusted in the fall of 1994 using a transmitter in the intermediate field (1.4 km distant on the roof of Harvard's William James Hall). The beam pattern matches well the predictions of scalar diffraction theory. The beamwidth (FWHM) is 8.4+/-0.2 arcmin and the main beam efficiency 82%.

The antenna is housed in a 16 ft Ash dome with a 75 in slit. During normal observations, the slit is covered with a screen of Goretex, woven PTFE (polytetrafluoroethylene -- Teflon), selected for its near transparency to microwaves, its strength, and its resistance to aging. The screen keeps the wind out of the dome and makes possible regulation of the temperature inside. LO reflections from the Goretex screen were found to be the source of occasional standing waves in scan baselines; subsequent modification of the mounting plates at the bottom and top of the screen gave it a "V" shape, eliminating surfaces of constant phase for the reflected LO and solving the standing wave problem.

Mount and Drive

The telescope mount and drive systems are essentially unchanged from their configurations at Columbia. Because the telescope is small, direct-drive torque motors are used on both axes, with the advantage that the drive system has no gear trains. Although the motors provide only 11 ft lb of torque, the telescope can change orientation at 10 degrees per second. Both axes are monitored by 16 bit shaft encoders and tachometers read at 100 Hz by the telescope-control computer to calculate torque corrections for pointing.

The pointing of the telescope is fine tuned at the beginning of each season by using a coaligned optical telescope to observe a large number of stars covering a wide range of azimuths and elevations. A least-squares fit to the pointing errors is used to define 5 pointing parameters (offsets of the azimuth and elevation encoders, effective longitude and latitude, and the very small nonperpendicularity of the azimuth and elevation axes). Because the relatively large beam of the telescope makes continuum observations of planets inconvenient, pointing is checked weekly by radio continuum observations of the limb of the sun. Although during the observing season (fall, winter, and spring) the sun transits below the elevation of most CO observations, it is the only practical astronomical source for pointing checks. At elevations used for observations, the root mean square pointing errors of the telescope were less than about 1', about 1/9 beamwidth.


The heterodyne receiver, which uses a superconducting-insulator-superconducting (SIS) Josephson junction as the mixer, is the two-backshort design of Kerr (Pan et al. 1983). A scalar feed couples the microwave signal to the receiver, where it is mixed with a local oscillator (LO) signal to produce a 1.4 GHz intermediate frequency (IF) signal that is further amplified with a very low-noise high electron mobility field effect transistor (HEMT FET) amplifier, and passed to the IF section of the receiver. The IF section further amplifies the signal and heterodynes it down to 150 MHz, passing a bandwidth of 200 MHz to the spectrometer. A diagram of the signal path is given here.

The LO signal is generated by a Gunn diode oscillator whose frequency is controlled via a phase-lock loop system by a computer-controlled frequency synthesizer. The SIS mixer and the FET first stage amplifier are on the liquid helium-cooled cold stage of a vacuum dewar; the rest of the electronics are room temperature. Typical receiver noise temperatures at 115.3 GHz are 65-70 K single sideband (SSB). Although the performance improves somewhat to 55 K SSB if the helium dewar is pumped to 2.7 K, it is not standard observing procedure, because the sky noise at 115 GHz dominates at this level of receiver performance. On the best dry, cold days the total system temperatures are less than 350 K SSB, referred to above the atmosphere.


The telescope has two software-selectable filter banks of a modified NRAO design, each containing 256 channels. At 115 GHz, the 0.5 MHz per channel filter bank ( picture ) provides a velocity resolution of 1.3 km/s, and velocity coverage of 333 km/s, and the resolution and coverage of the 0.25 MHz per channel filter bank ( picture ) are 0.65 and 166 km/s, respectively. The spectrometers divide the 150 MHz final IF signal from the receiver into 16 bands of 4 or 8 MHz width, each centered on 8 MHz. The 16 bands are passed to an equal number of filter boards, each with 16 contiguous two-pole Butterworth filters of 0.25 or 0.5 MHz width. The outputs of the filters are passed to square law detectors. After amplification, the detected signals are accumulated in integrators. The sampling time is 48 ms, followed by a 5 ms hold for sequential read-out by an analog-to-digital converter, after which the integrators are cleared for the next cycle. The 256 values produced by the converter are stored in a buffer during the following cycle, allowing the computer a full 48 ms to read the data.

Computer System

Prior to January 1991, pointing, data taking, and calibration of the radio telescope were controlled by a Data General NOVA minicomputer ( picture ) running a custom telescope-control system. The control computer was fairly limited in speed and memory (having only 32 K byte of random access memory and 5 M byte of fixed disk storage), but it was fast enough to allow limited data reduction on-line. For futher processing, all scans were transferred via 1600 bpi 9-track magnetic tape to a Digital Equipment VAXstation II/GPX workstation.

In January 1991, the telescope-control functions were transferred to a Macintosh IIfx computer, running a translated and improved version of the telescope-control system written in C. Individual scans or more commonly concatenated files containing large numbers of scans can be obtained from the control computer directly over the Internet. Generally the data is analyzed as FITS-format "cubes" of Galactic longitude, latitude, and velocity. Such cubes can be built from the raw scan files either using custom Macintosh software or on Unix workstations with IDL or CLASS.

Calibration and Observing Techniques

The receiver noise temperature is calibrated at the start of every observing shift by measuring the difference in receiver response to ambient temperature and liquid nitrogen temperature loads. The loads are made of Eccosorb, a carbon-impregnated foam highly absorbent to microwaves and cone-shaped to prevent direct reflection of LO back to the feed.

CO line intensities are calibrated using the room-temperature chopper-wheel method and the two-layer atmosphere model of Kutner (1978). At the CO signal frequency the atmospheric opacity is appreciable, mostly due to molecular oxygen and water vapor, and corrections to the observed line intensities for signal attenuation must be applied. Kutner's two-layer model of the atmosphere parameterizes the elevation dependence of the correction factor in terms of only 3 parameters, each of which has a physical interpretation. Because oxygen has a much greater scale height than water vapor, the model assumes they can be considered separate layers, oxygen above water, with different characteristic temperatures and opacities. The temperature and opacity of oxygen in the upper atmosphere do not vary much seasonally and are assumed to be constant at 255 K and 0.378, respectively, at the signal frequency. The remaining parameters in the model, the temperature and opacity of water and the fraction of the received power from the sky, are determined through antenna tippings (measurements of the intensity of the sky signal as a function of elevation) at least once per six hour observing shift, and more frequently if the weather is changing. Typical zenith water opacities ranged from 0.10 to 0.15, with values as low as about 0.05 in the coldest, driest weather. A 1 second calibration is performed at the start of each scan to correct for short term variations of the receiver gain and atmospheric opacity.

The observing season for the 1.2 m telescope, like other millimeter-wave telescopes at temperate northern latitudes, generally runs from October to May, with the best conditions in November through March. Cold, dry days afford the best observations, because of the decreased atmospheric opacity due to water vapor and the colder sky in general. Overall, the weather permits operation of the telescope roughly half of the time between October and May.

To obtain flat spectral baselines close to the Galactic plane where emission typically covers a large range in velocity, spectra were acquired by position switching every 15 s between the source position (ON) and two emission-free reference positions (OFFs) selected by the telescope control program to straddle the ON in elevation. The fraction of the time spent on each OFF was adjusted so that the time-weighted average system temperature at the OFFs was equal to that at the ON, resulting in baselines that were very flat, and residual offsets that were typically less than 1 K. This offset was generally removed by simply fitting a straight line to the emission-free ends of the spectrum. The OFF positions used for our Galactic surveys are given here.

Away from the plane in those regions where only one or two relatively narrow CO lines are found, frequency-switching by 10-20 MHz at a rate of 1 Hz was often used instead of position switching. Since spectral lines remain within the range of the spectrometer in both phases of the switching cycle, data could be obtained twice as fast as with position switching, although higher order polynomials, typically 4th or 5th order, were required to remove the residual baseline. A telluric emission line from CO in the mesosphere, variable in both intensity and LSR velocity, is detected in frequency-switched spectra; because the LSR velocity of the line could be predicted exactly, blending with Galactic emission could be avoided by appropriate scheduling of the observations. In a few cases of large surveys (e.g., Taurus and Orion) a model of the telluric line was fit daily to spectra free of Galactic emission and used to remove the line from all spectra.

A typical CO spectrum toward the inner-Galaxy is shown here.

The 1.2 m Telescopes Millimeter-Wave Group Homepage