IntroductionThis document provides a brief description of the Combined Array for Research in Millimeter-wave Astronomy (CARMA), suitable for planning observations and submitting an observing proposal. CARMA is a 15 antenna aperture synthesis telescope which operates at wavelengths of 3 mm (85-116 GHz) and 1 mm (215-270 GHz). The array is located at Cedar Flat (elevation 2200 m; latitude 37.3°; longitude -118.1°) in the Inyo Mountains of eastern California. The CARMA operations base is 25 minutes' drive away at the Owens Valley Radio Observatory near Big Pine, California. CARMA is operated by the CARMA Association, consisting of the Owens Valley Radio Observatory of the California Institute of Technology, the Radio Astronomy Laboratory of the University of California, Berkeley, the Laboratory for Astronomical Imaging of the University of Illinois, and the Laboratory for Millimeter Astronomy of the University of Maryland, with support from the National Science Foundation.Telescopes and Array ConfigurationsNine of the CARMA antennas are 6.1 meters in diameter; the six are 10.4 meters in diameter. The measured surface accuracies of the antennas are typically 40 microns rms (lambda/75 at 3 mm wavelength) or better. The halfpower beamwidth is 100"/60" (6/10 m antennas) at 115 GHz, and the typical radio pointing accuracy is 5" rms.The antennas may be located at various stations distributed over an area 2 km in diameter, with spacings from 7 m to 2 km. Normally the antennas are deployed in one of 5 standard configurations (A, B, C, D, E), providing angular resolutions of roughly 0.3", 0.8", 2", 5", or 10" at 100 GHz. For the 2008 January-June observing period, the B, C, and D antenna configurations will be offered, with baselines 82-813, 26-370, and 11-148 m respectively. ReceiversThe antennas are equipt with 3 mm (85-116 GHz) and 1 mm (215-270 GHz) SIS receivers. One linear polarization is received. It is not possible to use both receivers simultaneously.Both sidebands of the first local oscillator are received, and are separated in the correlator by phase-switching the local oscillators. The intermediate frequency (IF) band is 1-5 GHz. At 3mm, the useful portion of the IF band is limited. On the 10-m antennas, the useful portion is 1-4.5 GHz, with less sensitivity near 1 GHz. On the 6-m antennas, the useful portion of the IF band is 1.1-2.4 GHz. At 1mm, the best sensitivity is achieved within the band 1-4.5 GHz. The Correlator SpectrometerThe digital correlator presently provides up to 366 complex channels per sideband for each of the 105 interferometer baselines, within a maximum bandwidth of 1.5 GHz per baseline per sideband. In cross-correlation mode, the upper and lower sidebands of the first mixer are separated and the lower sideband appears in the lower half of the channels while the upper sideband appears in the upper half of the channels.Signals entering the correlator are split into 3 separate bands. By setting the frequencies of the second local oscillators these bands may be positioned independently within the I.F. bandwidth. The bands may be in any order, and may be overlapped. It is possible to observe several widely separated spectral lines simultaneously, or with different velocity resolutions. However, as noted above, the useful IF bandwidth at 3mm is presently limited by the receivers to 1.1-2.4 GHz when all antennas are to be included. The following figure shows this situation. The second LOs are positioned at x, y, and z GHz. |
|
|
|
For example, two lines separated by less than 2.4 - 1.1 = 1.3 GHz (such as HCN at 88.632 GHz and HCO+ at 89.189 GHz) may be simultaneously observed in one sideband. Alternatively, two more widely separated lines could be observed with one in each sideband. In this case the separation would have to be more than 2 ± 1.1 = 2.2 GHz and less than 2 ± 2.4 = 4.8 GHz in order for the lines to appear in the passbands of all receivers. Depending on the project, it may be preferable to observe lines separately rather than to observe them simultaneously if they would appear in lower-sensitivity parts of the IF (i.e. below about 1.5 GHz). Each band may be configured in one of five modes: |
| Nom.Bandwidth | Actual Bandwidth | Channels | Chan.width | dV[3mm] | Vtot[3mm] |
| (MHz) | (MHz) | (per sideband) | (MHz) | (km/s) | (km/s) |
| 500 | 468.75 | 15 | 31.25 | 94 | 1410 |
| 62 | 61.52 | 63 | 0.977 | 3.4 | 214 |
| 31 | 30.76 | 63 | 0.488 | 1.7 | 107 |
| 8 | 7.69 | 63 | 0.122 | 0.42 | 26.5 |
| 2 | 1.9226 | 63 | 0.0305 | 0.10 | 6.3 |
An analog filter attenuates the edge channels of the 62-MHz band. An overlap of about 6 channels is recommended if multiple 62-MHz bands are to be used to cover a single line.
Field of View; Sensitivity to extended Structures; MosaicingThe antenna half-power beamwidth is 100" x (115 GHz/freq) for the 6 m antennas and 60" x (115 GHz/freq) for the 10 m antennas. Sources anywhere within the primary beam can be imaged, but the sensitivity decreases and pointing errors become more critical beyond the half-power points. The sensitivity to large scale structure is determined by the spacing of the sampled points in the aperture u-v plane. Typically the array has little sensitivity to structures which are larger than about 10 times the synthesized beamwidth.In order to image objects larger than about 1 arcminute in size, it is necessary to make observations at multiple pointing centers (mosaic) with the interferometer. The mosaic observations should cover the object to be imaged, and be spaced by no more than lambda/(2 x the 10m antenna diameter), approximately one half the beamwidth, in order to sample the short u-v spacings. The mosaiced interferometer observations may be combined with single dish data. The single dish observations should ideally be obtained with an antenna which is at least twice the diameter of the 6 m antennas, as these antennas will typically have the shortest baselines. It is important that the single dish map be sampled at the Nyquist rate - that is, at no more than half-beamwidth spacings. Since the visibility function of an extended object is strongly peaked at the origin, the short u-v spacings contain most of the information about the large scale structure, and sparser sampling further out in the u-v plane mostly results in loss in sensitivity, rather than distortions in the image. Map SensitivityThe CARMA Sensitivity Calculator applet is the most convenient way of computing the expected rms noise for a CARMA observation. As inputs, you will give the array configuration, the observing frequency, the source declination, mosaic pattern, and the spectral resolution; the applet returns the synthesized beamwidth and the expected noise in mJy/beam and in Kelvin for observations of a single track. At 3mm a sensitivity of better than 0.5 mJy/beam rms (5 K in 1 km/s for a 1" synthesized beam) can be obtained in good weather at 95 GHz with an 8 hour track (70% time on source). The 1mm receivers are still being characterized, and the system performance will be further evaluated during the fall. The Sensitivity Calculator reports the current expectation.For some projects, it will be necessary to request more time than can be allocated for a single track. In these cases, proposers should specify the total number of hours requested for the source. In general, such sources will be observed above an elevation limit of 30°, with this limit decreasing to 15° gradually for projects below 30° declination. In E configuration the elevation limits are higher to minimize shadowing. If other elevation limits are required they must be requested and justified in the proposal. Sensitivity-limited projects can often be observed in smaller pieces without regard to hour angle. An increase in the time allocated will be made to compensate for any reduction in sensitivity from that expected from longer tracks. CalibrationThe instrumental gain phase is calibrated by observing a nearby quasar every 15 - 30 minutes. A few minutes' observation of a 1 Jy quasar with ~1 GHz bandwidth provides a good calibration. Any systematic errors are minimized by selecting a quasar close to the source, but since the array geometry is determined to an accuracy of about lambda/10 it is usually better to select a stronger quasar within 30-50 degrees, than a weaker quasar which is closer to the source. A suitable calibrator can typically be found within 20 degrees of a source.The flux density scale (Jy/K) is determined from planet observations. Since the quasars are usually time variable it is necessary to calibrate the quasar flux from observations of planets at short baselines where the planets are not resolved. A history of calibrator fluxes is maintained within the MIRIAD reduction software package. The passband calibration is done in two parts. The correlator IF passband is measured with high signal-to-noise using injected noise sources. The RF part of the passband must be measured from observations of an unresolved source, usually 3C84 or 3C273, or, at short baselines, the planets. Preparing Observing FilesCARMA is operated by running observing scripts which have been prepared and checked in advance. The PI will be notified following the selection of an observing proposal, and must then submit a script which contains the complete observing sequence of source and calibration observations as well as tuning and correlator settings.Further InformationInformation for proposersThe current call for proposals Observing preparation tools, including the CARMA Sensitivity Calculator applet |