Snap decision

1 December 2008

Installing RPC chambers, which work with the trigger

The pressure is on right now for the muon Level 1 Trigger system. In the absence of beam, and with muons being the only charged cosmic particles able to penetrate underground and right through the detector, the muon trigger finds itself acting as a reference point by which all the sub-detectors are attempting to calibrate themselves.

The LHC is designed to engineer bunch crossings of protons every 25 nanoseconds, meaning that the L1 trigger has to be exceptionally fast, making snap decisions about whether or not data is worth recording, before the next bunch crossing occurs. “When you’re taking the decision, the clock goes on,” notes Francesca Pastore, who is working on commissioning the muon barrel trigger, “so the detector we’re using, the RPC, has a time resolution of about two nanoseconds.”

The actual trigger decisions are taken by electronics, to cut out the time that software-based decision-making takes. “By the time the particles have reached the detector, another interaction could be taking place,” warns Francesca’s colleague, Mimmo Della Volpe, “so the risk is that you see a particle and attribute it to the wrong interaction, meaning the physics will be screwed up.”

Because the detectors have such huge information ‘pipelines’, there is a tiny period of time during which all the data streaming off the machine is still present, before it is either stored or rejected.

The L1 trigger labels which signals have come from which collision events and decides which are worth keeping for analysis, all within the 25 nanosecond time-window before the next bunch crossing occurs. But the central trigger, which ultimately tells the detector read-out software which time-windows to record information from, requires a complex integration of many different bits of electronics.

“Our trigger is made of three different shells,” explains Francesca. “We require coincidence of hits in time and in space” she says of the conditions for triggering. But the different lengths of cable between different parts of the detector and the trigger logic can result in hits being registered out of synchronisation. “What we’re doing now is using the huge amount of cosmic tracks to understand, and subtract, the delays between all the different bits of hardware,” says Francesca.

“It’s heavy work because cosmics are quite different,” says Mimmo. “Right now, we’re setting up the trigger in a way which is not really what we want in the end.” For a start, cosmics don’t offer the opportunity to calibrate for the geometry of particle tracks which will be forced to curve under ATLAS’s magnetic field. But the lack of this information is no bad thing at this stage – it allows the latencies in the electronics to be studied more easily, and their effects to be disentangled.

More of a hindrance is the fact that cosmics pass through the detector from top to bottom, rather from the interaction point at the centre of the detector, outwards. In other words, the pattern of hits recorded in the top half of the detector as a cosmic muon crosses it is opposite to how it will look for beam. But, since this is only an effect of the geometry of the detector, these differences can be calculated for using a simulation.

By doing this, it should be possible to flip from one configuration – cosmics – to another – beam – pretty easily. “Of course this must be tested,” says Francesca, “but when we have real data we can check it.”

There are 64 independent trigger sectors in all, each with their own calibration constants to be found. During the original commissioning phase, ahead of September 10th, each was calibrated individually, but not correlated as part of the whole. “Now we’re under pressure to have all the timing of the sectors aligned,” says Mimmo. “In less than 10 days we’ve collected 150 million muons,” he adds, reporting that, “After some timing adjustment, we’ve been able to improve the rate of tracks recorded by 100 per cent for the SCT and 70 per cent for the pixels.”

Work on this ‘global alignment’ began at the end of September, and is ongoing. The plan now is to have a combined running week at the end of November, in order to provide the ID teams with plenty of tracks for studying the behaviour of their sub-detectors, with alignment studies expected to continue well into the winter shutdown.


Ceri Perkins

ATLAS e-News