Speed limits on American highways are set as much by law as by engineering. So, too, for research data collected at the Laboratory¹s three experimental halls, where the laws of physics and the limits of technology hold sway. But this summer those boundaries will be stretched with a pedal-to-the-metal approach that will accelerate Hall B into the data-collection fast lanes.

Equipment and software upgrades should boost rates of data acquisition by almost 50 percent, from 11 megabytes to as many as 16 megabytes per second. Given that a byte is the equivalent of a single character (such as a letter, a numeral or a punctuation mark), Hall B will be the envy of even the most prolific writer, creating some 3.2 million "words" every second. At such rates, Hall B could theoretically produce a trillion bytes of raw information per day. A week's worth of collected data could thus approach the total number of characters contained within the 17 million books currently catalogued in the Library of Congress.

According to Graham Heyes, JLab head of data acquisition, the increase in collection rates is good news for researchers tracking the outcomes of experiments, as well as for Laboratory staff overseeing and running the halls.

"Most of these experiments are based on statistics. Precision is important, so a large number of events is very important," he says. "If we can substantially increase the rate of data acquisition and transmission, we substantially decrease the amount of time required to do the experiment. Taking the same amount of data in three days that once was taken in a week saves everybody time and money."

Cutting Through the Clutter

Ingenuity has played a major role in the upgrade's planning and development. Although there are physical constraints on how quickly data can be collected and routed, Lab designers have designed workarounds combinations of computer programs, hardware configurations and the pending installation of high-capacity fiber optics that permit more efficient data gathering and transmission.

In particular, designers contended with the telecommunications links between Hall B and the Lab¹s Computer Center. It has taken time and thought, says Dieter Cords, head of Hall B¹s On-line Group for CLAS experiments, but the problem has been resolved.

"It's more complicated than just buying equipment," Cords explains. "You have to take a number of steps to remove bottlenecks. As with any emerging technology, it's a matter of price and convenience."

Storage currently poses little problem; culled information is eventually stockpiled in the Lab's Computer Center, in a device known as the "Silo". Thus far, roughly 40 terabytes of data from all three halls have been amassed on 1,000 of the Silo's 6,000 tapes. Even with Hall B's increased rate of data acquisition, it will be some time before the Silo's capacity is taxed.

Such information inundation is made possible by the detector at the heart of Hall B. Called CLAS, for CEBAF Large Acceptance Spectrometer, the detector's components include time-of-flight counters, energy-measuring calorimeters and particle-tracking drift chambers. CLAS records, on average, 2,200 particle interactions per second on 40,000 data channels.

However fast they prove to be, not all particle interactions are created equal. Particle-interaction recordings may be contaminated by extraneous events. This "noise" can be caused by such factors as background radiation in the hall, electronics associated with detector subsystems, even electrical spikes generated by broken wires.

"These systems don't have to be foolproof, but they do have to be highly calibrated," Cords says. "Many physicists can be involved in just one experiment. Every user wants to ensure that all the data we take is analyzable."

To the rescue has come corrective software developed by physicists working on the CLAS team. Over the past years they have produced data analysis and reconstruction programs for Hall B that interpret experimental results, adjusting for CLAS system changes and weeding out those events that are incomplete or noisy.

"You have to keep running very hard to stay still and even faster to get ahead," says data acquisition head Heyes. "But for the foreseeable future Hall B is where we want it to be."

maintained by webmaster@jlab.org