The talk will address various aspects of timing systems. Timing system (TS) is at the heart of the accelerator, it dictates and coordinates the operation of almost all devices in the whole accelerator chain, and helps devices to fulfill machine-physics and equipment requirements. We’ll systematically address the TS concepts and the roles of the TS, both functional and in terms of TS clients. This will give us a base for further discussion. We’ll explore what constitutes TS integration: a) domain knowledge, b) hardware development, and c) accelerator-specific real-time application, which orchestrates the operation of the machine. We’ll dive into the concept of TS transport layer and outline differences among different realizations. An overview of a selection of past and existing TS related projects will be given. We’ll also touch machine protection system (MPS), since it is from the technological and system-integration point of view similar to the TS.
In the presentation I would like to explain we are totally indifferent regarding transfer layer choice (being WR or MRF or something else) - what we perceive as the core of work is the connection to all timing system clients and machine specific customization to provide correct interplay of the entire machine.
Presentation explains that understanding timing system is much more than 'simply' being expert in certain communication technology. It's all about understanding what services the rest of the machine requires from the timing system to successfully fulfill the mission of the complete machine. What we see most of the time is that people quickly jump in technology playing (because FPGAs are cheap and everybody "knows it"), but it later turns out the system does not nicely fit with the rest of the machine. Once additional timing systems are introduced (and other ugly patches...), the system design is already deep down the drain, but everybody sticks at it since they already invested so much effort...