Abstract-We describe the current status of the SCUBA-2 submm camera for astronomy. We describe the first measurements made while operating an entire sub-array, rather than small subsections as has previously been the case. These results demonstrate that the procedure required to set up the multiplexer for full sub-array operation can be carried out entirely automatically, and that the pixels are sufficiently uniform for the whole subarray to be usefully operated simultaneously.SCUBA-2 is the first wide-field, ultra-sensitive camera for sub-millimetre astronomy. With over 10 000 pixels, and operating in two frequency bands centred at 450 and 850 μm (670 and 350 GHz), it will enable research which addresses many of the fundamental questions in modern-day astronomy.There has been a huge revolution in sub-millimetre astronomy (wavelengths of a few hundred μm to a few mm) in the past decade. The SCUBA [1] instrument on the James Clerk Maxwell Telescope in Hawaii was largely responsible for this. With over 100 pixels, and (for its time) state-of-the-art germanium semiconductor detectors, it enabled the sky to be mapped much more quickly and with improved fidelity when compared to the single pixel instrument which preceded it. However, with less than 1% of the sub-millimetre sky having been studied in any detail, there is a need for instruments with greatly increased mapping capability.This requires a large increase in number of pixels over SCUBA. In addition, the overall noise performance of SCUBA was limited by the detectors. Ideally, detectors will achieve background limited performance (BLIP), in which photon noise from background radiation dominates.Achieving the twin goals of improved noise performance and substantially increased pixel count has required a change in detector technology. Semiconductor detectors are not background noise limited for the best ground based telescopes, and have reached fundamental noise limits. In addition, extending the SCUBA design to large pixel counts is not feasible for two reasons. Firstly, it is not possible to use a multiplexed readout without an unacceptable noise penalty. Without multiplexing, an instrument such as SCUBA-2 would require an unworkable size of readout electronics, as well as number of wires to the detectors. Secondly, the germanium thermistor chips must be individually glued and wired to each pixel.SCUBA-2 uses superconducting detectors (transition edge sensors). These have several advantages over semiconductor bolometers. As well as being able to achieve better signal to noise, all the fabrication stages are carried out on the array as a whole using standard micromachining techniques rather Fig. 1. The response to a bias modulation measured simultaneously by all pixels in one sub-array. For each pixel, the x-axis shows time, and the y-axis shows detector current. than separately on individual pixels. Finally, it is possible to constructed a multiplexed readout using superconducting electronics with only a small degradation in noise properties compared to indiv...