Next: 11 Initial Checkout Up: 10 Data Encoding Previous: 10 Data Encoding

10.1 Data Analysis

Typically the Master will do a block read to dataspace until SS=2, which will produce a variable-length data block, depending on the input data. The length of this data block is available to the analysis program.

The analysis program starts at the end of the block. The last word in the block must be the channel header for the last channel that had data. Using the channel word count, it is possible to backskip over channels to find a particular channel. Normally, all the channels would be analyzed.

Working backwards from the channel header in bytes, the program will skip between 1 and 3 bytes containing zero. The preceding two bytes will be the cluster header for the last cluster in that channel. Using the cluster width, it is possible to chain backwards through the data block.

In the case of a pedestal run on a 256-bin CCD, where all bins have data, special handling is required. The data will consist of one cluster which is 256 bins wide. The cluster width in the cluster header will be zero, since the 8-bit counter will have overflowed. The bin address will also be zero since the cluster started at the first (zeroth) bin. The 256 data bytes will take 64 words exactly, so the cluster header will start on a word boundary, and be followed by 2 zero bytes to fill to the next word. The word before the channel header will therefore consist of 4 zeroes, which can only occur in this particular case and thus may be identified by the analysis program.

Next: 11 Initial Checkout Up: 10 Data Encoding Previous: 10 Data Encoding