A B S T R A C TThe micro-channel plate intensified CCD photon counting detector developed at University College London has been upgraded to allow time-resolved spectroscopic optical data to be acquired on periodical sources such as pulsars. First observing trials have been carried out, acquiring spectroscopic data on the Crab pulsar. The detector was phase locked to the pulsar period and a temporal resolution of 41.4 ms employed. The phase locking allowed the coaddition of time slices over a large number of pulsar periods to build up quantifiable spectroscopic data when observing in a flux-limited regime.Key words: instrumentation: detectors ± methods: observational ± techniques: miscellaneous.
I N T R O D U C T I O NIn most imaging applications using a CCD camera, whether a fullframe imager as normally employed in astronomy or a frame transfer device as used in video applications, the acquisition of an image can be split into two phases; the integration period and the readout period. For the integration period, the clocks connected to the imaging area of the CCD are held at constant levels to allow a photo-electron charge to be accumulated, thus building up an image on the chip. In the readout period this image is then accessed by activating the pixel clocks so that the image is transferred row by row to the readout register where the image is read out one pixel at a time. The highest rate at which frames of data can be transferred, which limits the temporal resolution available, is governed by the readout period. This is typically in the range of milliseconds/frame, for small format CCDs that employ frame transfer, to many seconds/frame for large-format full-frame imagers.If, now, the integration period is removed and the CCD clocks are allowed to operate continuously with each vertical transfer being followed by a row readout (i.e. the CCD is operating in a Freerun mode) then, for two-dimensional images, all features will be smeared out vertically as shown in Fig. 1. However, for onedimensional images which subtend a single CCD row (such as a stellar spectrum), each row will become a time slice through the input with the integration time being equal to that required to shift and read out a row. In this mode of operation there is no vertical spatial dimension to an image and the same result would be obtained with a simple linear CCD.Hence, in principle, high time-resolution spectroscopy on stellar sources can be achieved by continuous clocking. However, a problem exists. Except on very bright sources, the flux acquired per time slice will be very low and readout noise on the CCD (even at the lowest levels) will dominate. As an example, on a 10th magnitude star, the expected signal with an 80 per cent efficient CCD using a 4.2-m telescope would be ,0.06 e 2 /A Ê in a 40 ms time slice (based upon measured efficiency figures for the William Herschel Telescope ± see website www.ing.iac.es/,sjst/ isis). This leads to the conclusion that this operating mode, called Freerun mode', can only realistically be employed w...