Evolution of the LBT Telemetry System

The Large Binocular Telescope (LBT) Telescope Control Syst em (TCS) records about 10GB of telemetry data per night. Additionall y, the vibration monitoring system records about 9GB of telemetry data per night. Thr ough 2013, we have amassed over 6TB of Hierarchical Data Format (HDF5) files and lmost 9TB in a MySQL database of TCS and vibration data. The LBT telemetry s ystem, in its third major revision since 2004, provides the mechanism to captur e and store this data. The telemetry system has evolved from a simple HDF file system with MySQL stream definitions within the TCS, to a separate system using a MySQL database system for the definitions and data, and finally to no database use at a ll, using HDF5 files. 1. Initial System Design and Implementation The LBT telemetry system is designed to provide the following: system data for commissioning, failure diagnosis and system tuning; monitoring of key systems for failure mitigation; and access to historical data. The telemetry collection system is requir ed to support sample streams as large as 16kB, at a rate up to 40Hz (Edgin 2012 ; Edgin & Cushing 2010). Currently, telemetry streams are collected at rates up to 1kH z. From inception, analysis tools were not part of the LBT telemetry system. Third party roducts were presumed for accessing and analyzing telemetry data. 1.1. HDF Data Files with MySQL Stream Definitions The first generation telemetry system for the LBT was implemented as a subsys tem of the TCS. The data were implemented as sequences of time-tagged values from telescope subsystems to HDF (NCSA 2003) files (De La Pe ña & Axelrod 2006). As a TCS subsystem, it had a GUI similar to the other TCS GUIs. Telemetry stream type s and contents were defined in a MySQL relational database via the GUI. The database also maintained stream status and a catalog of the archived data files. From th e GUI the user could start /stop streams from the pre-defined collection, set acquisition parameters, monitor the state of a stream, monitor the state of the telemetry system, and generate graphics via MATLAB’s GUI builder, GUIDE. Only a couple of the TCS subsystems used this system, beginning in February 2004. A few post-processing tools using MATLAB were provided for sp ecific data streams, but nothing generic. This was implemented as a shared library. Because it was a subsystem of the TCS, the implementation in the TCS subsystems (the telemetry producers) was very minimal. However, this was also the major drawback of the design only TCS clients could write data, and it was limited to co l-