There are a few things you could do here to make this easier. For one, rather than make a batch file+script for each instrument's software, you could just make a single wrapper which would take the instrument as a parameter, do the logging and then look up the correct software to call (from a hash, for example, or a database if you have already invested in that infrastructure).

You could also write plain batch file wrappers (or again, a single wrapper) which dump the user/study/instrument/start-time to a single log and then build a perl script to parse through that file. To get end times, I'd just have the wrapper batch file make a log entry before and after the instrument code runs. You could even ftp these logs to a central server where your perl script could do the parsing, meaning you'd only have to maintain one instance of perl, if that's an issue.