First Year Performance Paper - Section 2.2
2.2 Production and testing
IceCube DOMs are assembled at three different facilities worldwide. Each site produces complete DOMs from the incoming materials and tests them. The total production rate from all three assembly sites is in excess of 50 DOMs per week.
Because it is impossible to replace a bad unit or access it for repairs, careful testing is essential. The DOMs are tested in three stages. Before final assembly the separate subsystem components are tested. Fully assembled DOMs must then pass a series of performance tests in special dark freezer laboratories where up to 120 DOMs can be tested simultaneously. The tests mimic the temperatures, the data acquisition procedures and the optical signals that the DOMs will experience in the field. The DOMs are tested at four different temperature settings: room temperature, -45 °C and -20°C, which correspond to the temperatures of the highest and lowest DOM in the string, and -55°C, which corresponds to the surface temperature in winter of the IceTop DOMs. The freezing and warming cycle is repeated twice. The entire test cycle takes 3 weeks. DOMs that pass all the testing are shipped to the South Pole, where they are retested inside their shipping crates before deployment in the ice.
In addition to certifying DOMs as good for deployment, the testing measures critical performance characteristics for all DOMs in a controlled laboratory setting . In particular, short laser pulses create single PE events which reveal a time resolution between 2 and 2.5 ns, with 8% of hits in a tail extending to 60 ns . The late pulsing is a known PMT effect , and is due to electrons that were elastically scattered on the dynode, and have longer flight time in the PMT. This delay is less than expected for late arriving photons due to ice scattering (see Fig. 19).