News from the pit

10 March 2008

It can’t have escaped the attention of any ATLAS collaborator that last week was M6 week. As the final “Milestone” before the run-up to actual data-taking, more emphasis this time was on the higher-level trigger systems, and the logistics of how the control room will operate.

“The basic idea is that we stress test the system. We try to put as many pieces as we have into one system. Then we can iron out any inconsistencies,” explained Run Coordinator, Thorsten Wengler. “Usually people are very good at getting their own systems to run smoothly, but you always get interference when you start combining them.”

“Perhaps more importantly, we try and run the thing at the speed that we will have to run when we start getting real collisions,” he added.

While the first half of the week saw the muon and muon trigger systems, inner detectors, calorimeter and calorimeter trigger systems come under the spotlight, Thursday and part of Friday were devoted entirely to testing the software and operation of the higher level trigger system.

Speaking last Monday, Thorsten explained that in previous integration weeks, software testing has tended to be pushed to the end: “If something’s wrong fundamentally with some of the electronics systems, it can take months to re-build, so in the past we’ve put the emphasis there. But now the time has come to thoroughly check all the software algorithms and infrastructure, to make sure that events are arriving at the right place, and are selected in the way we intend.”

From Friday lunchtime, the system was put into integrated running mode, combining all the available parts of the sub-detectors and software, until Monday morning. During this testing, the combined system was run using cosmic tracks and artificial random triggers.

In addition to trying to get the whole system to run smoothly, one of the big aims of the M6 weekend exercise was to figure out the most efficient way for the control room to operate. “Do the muon system people ever need to walk across the room and speak to the inner detector people, or do they both just talk to the shift leader?” asks Thorsten. “Of course, we can think about this before, but the only way to really know is to actually do it.”

Deciding the most useful things to display on the seven large projection screens, and how to manage extra bodies in the control room were other important goals. On top of the shift crew of 15, there are often other key experts around. “They then sit all over the place; they sit on the floor and balance their laptops on the walls!” smiles Thorsten. “I’m pleased that people are so keen, and I absolutely do not want to interfere with that. But on the other hand, at a certain point it becomes impossible to operate this experiment if you can’t make yourself heard over a two metre distance. Sixty people, even talking quietly, make a lot of noise.”

To address this issue, a suite of satellite control rooms have been added, some of them available for the first time during M6. Located about 30 metres away from the main building, each room allows a number of experts from each area or subsystem to work and support those in the control room.

“There’s a tendency – if you put a bunch of physicists in one room – that they’ll all cluster around one screen, and discuss, and try and work things out. We’re not very well trained in this sort of regimented style of working where everyone sits and has a job, and communicates in a formal way,” Thorsten concedes. “But we have to give in to necessity to a certain extent; otherwise we could never operate a thing like this.”

Having said that, he is prepared to take an organic approach to finding the optimum operational mode for the control room: “I think a certain amount of chaos is inevitable, especially as we come to terms with the running phase. But we do have to establish an operation model, and define the basic rules, or we won't converge ... You can’t take a book and look up how to do this.”



Colin Barras


Ceri Perkins