The National Astronomical Observatory of Japan has constructed a prototype 12-m antenna of the Atacama Compact Array to evaluate its performance at the ALMA Test Facility in the NRAO VLA observatory in New Mexico, the United States. The antenna has a CFRP tube backup structure (BUS) with CFRP boards to support 205 machined Aluminum
surface panels. Their accuracies were measured to be 5.9 m rms on average. A chemical treatment technique of the surface panels has successfully applied to scatter the solar radiation, which resulted in a subreflector temperature increase of about 25 degrees relative to ambient temperature during direct solar observations. Holography measurements and panel adjustments led to a final surface accuracy of 20 m rms, (weighted by 12dB edge taper), after three rounds of the panel adjustments. Based on a long term temperature monitoring of the BUS and thermal deformation FEM calculation, the BUS thermal deformation was estimated to be less than 3.1 m rms. We have employed gear drive mechanism both for a fast position switching capability and for smooth drive at low velocities. Servo errors measured with angle encoders were found to be less than 0.1 arcseconds rms at rotational velocities below 0.1 degrees s-1 and to increase to 0.7 arcseconds rms at the maximum speed of the 'on-the-fly' scan as a single dish, 0.5 deg s-1 induced by the irregularity of individual gear tooth profiles. Simultaneous measurements of the
antenna motion with the angle encoders and seismic accelerometers mounted at the primary reflector mirror edges and at the subreflector showed the same amplitude and phase of oscillation, indicating that they are rigid, suggesting that it is possible to estimate where the antenna is actually pointing from the encoder readout. Continuous tracking measurements of Polaris during day and night have revealed a large pointing drift due to thermal distortion of the yoke structure. We have applied retrospective thermal corrections to tracking data for two hours, with a preliminary thermal deformation model of the yoke, and have found the tracking accuracy improved to be 0.1 - 0.3 arcseconds rms for a 15-munites period. The whole sky absolute pointing error under no wind and during night was measured to be 1.17 arcseconds rms. We need to make both an elaborated modeling of thermal deformation of the structure and systematic searches for
significant correlation among pointing errors and metrology sensor outputs to achieve the stable tracking performance requested by ALMA.