Pixel Progress

27 May 2008

Pixel Commissioning

In the pit during the commissioning of the electrical system

The silicon pixel detector is only about the size of a human being, which is downright dinky in comparison with a detector the size of a cathedral. Its rectangular patches of silicon, with sides hardly thicker than your hair, are responsible for distinguishing the dense tracks that will emanate from the collision point just five centimeters away at the closest. The commissioning group is making sure that it works.

At the centre of ATLAS, the pixel detector was last to get connected. The electrical systems started getting hooked up on February 5th after nearly a year sitting in the cavern. High-voltage electronics guide particle signals across the silicon pixels while the chips that read the data run on a lower voltage. Simultaneously connecting and testing, the mechanical and test crews worked side-by-side in the pit, making connections and operating them using laptops on the network. This work finished February 26th.

The results showed that all the electrical connections that had worked during the last test, in the summer of last year, were still functioning. Unfortunately, 7 new modules – units which contain an average of 46,000 pixels each – were found to lack high voltage, out of 1,744. They are impossible to repair as they are deep inside the near-complete ATLAS detector.

Once the electrical system was functional, crews moved onto the optical system. The work started on March 17th, this time with a mechanical team in the pit making connections and the test team in the control room. Jean-Francois Arguin, a post-doc from Lawrence Berkely Laboratory recalls the brief moment of panic during the first attempt to send information through the system. “There was no reply from the detector, nothing was working.”

It turned out to be a simple matter of crossed wires. To save space, multiple detector channels are read through a single optical fibre that splits on its way to the rest of the data acquisition system. The split ends were connected to the wrong places on the “crates” - electronic devices that turn the optical signals back into electrical signals.

“We spent, basically, four days of Hell, where we corrected everything,” says Jean-Francois. While fire and brimstone would pose major safety hazards in the ATLAS pit, unplugging and reconnecting over 500 fragile optical fibers is its own sort of torture. The fibres are now correctly connected.

On the other hand, the testing of the “optoboards” went surprisingly well. These devices change the data from electrical to optical format for speedy transmission through the many metres of cable. During initial installation, they had trouble coordinating communication on the two ends of the optical cables – the transmitter and receiver weren’t quite agreeing on what constituted a “zero” or “one”. The problems were remedied before installation and did not resurface during commissioning.

The only area of concern in the optical system is the lasers that send information back to the detector. Twenty-five of the 2,000 channels were dead. Unlike the optoboards on the detector, however, the lasers on the crates are accessible. Most of them have already been replaced, but the reason for the failures is yet unknown.

Testing of the cooling system started on April 25th and came to an abrupt halt on May 1st when a problem arose in the compressors of the cooling plant, accompanied by what Jean-Francois describes as “a burning smell”.

The cooling system is critical to the detector, which produces the same amount of heat energy as 70 100-Watt incandescent bulbs. For low risk operation, the detector temperature stays below 40˚C. Above that, the silicon pixels may begin to peel themselves away from the electronics that register particle signals, rendering it useless.

To test the 88-loop cooling system, the commissioners divided the detector into four quadrants. They tested each loop individually and then ran a quadrant all together. “Overall, we tested 77 cooling loops out of 88,” says Claudia Gemme, run coordinator. “For the ones that are missing, we didn’t have the chance to switch on the detector at all.”

The 44 loops in the top half of the detector worked rather well. “As far as we know, we have three circuits with leaks, two are small, and we think we can operate them,” says Claudia. “One is big and the locations of the leaks were not completely clear, so we decided not to operate it.” Without this loop, 12 modules will not run.

The commissioning team encountered more trouble in the bottom half of the cooling loops, 11 of which haven’t been tested at all. “There are a few cooling loops that we have to learn how we can operate, if we can operate,” Claudia explains. The crew wanted to thermally image the detector for more information. As the testing of the cooling system is on hold until the cooling plant is repaired, thermal imaging will not be possible. "We have the feeling that we don’t have huge problems in operating the detector if the cooling is working properly," says Claudia, "but we had very few hours to test.”

In the meantime, they are still testing electronic parts of the detector at about a quarter of running energy, making sure to distribute the areas in operation to avoid overheating. Until cooling is back and commissioned, it is impossible to say exactly what proportion of the pixels will take data.





Katie McAlpine

ATLAS e-News