(ed.). Gravitational waves (IOP, 2001)(422s).

122 LISA: A proposed joint ESA–NASA gravitational-wave missionmeasurement accuracy clearly is extremely high.Each optical assembly has two lasers for redundancy, and a switchingmechanism for switching between them. Each laser is capable of 2 W outputover a mean lifetime of several years, but is operated at1Wtoextend the lifetimeby a substantial factor. A schematic drawing of the optical bench for one opticalassembly was shown earlier in figure 10.5. Light from the active laser is takenby a single mode fibre onto the optical bench, and then formed into a beam ofroughly 5 mm diameter. A few per cent of the beam is split off for other purposes,and then a beamsplitter sends almost all of the power out to the telescope, whereit is transmitted to the matching optical assembly on the other end of the arm. Thereturn beam from the laser in the distant optical assembly is sent by the telescopeto the test mass, where it is reflected from the face of the test mass, and then beatagainst a small amount of power from the local laser.The beat signals detected in the six optical assemblies are the main data usedto make the gravitational-wave measurements. In the present baseline scheme,one laser in spacecraft SC-1 is considered the master laser and locked to a stablereference cavity on its optical bench. The other laser in SC-1 is phase lockedto the first with a small offset frequency of perhaps a couple of kilohertz. Thetwo laser beams are transmitted along the adjacent arms of the triangle to SC-2and SC-3, where they are reflected from the test masses in the receiving opticalassemblies. The local lasers are then phase locked to the received beams, alsowith small frequency offsets, and their beams are sent back to SC-1. Finally, thesecond lasers in SC-2 and SC-3 are offset locked to the first lasers, their beams aresent over the third arm of the triangle, and the beats between them are recordedon both ends of the side, after reflection from the test masses.The frequencies of the beat signals can range from well under 1 MHz toroughly 15 MHz, depending on the exact orbits of the spacecraft, or rather theorbits of the freely floating test masses in them. In one possible measurementscheme, the signals are beat down to a convenient frequency range for precisephase measurements, such as perhaps 10 kHz, using outputs from frequencysynthesizers. The drive frequencies for the synthesizers come from ultra-stablecrystal oscillators (USOs) on each spacecraft. In a second measurement scheme,the signals are first sampled at a high rate such as 40 MHz, and the results areanalysed in the software. In this case, the sampling times are controlled by theUSOs.If only the signals from the two arms adjacent to SC-1 are considered,the antenna is like a single groundbased detector, with changes in the armlength difference being the quantity of interest. However, it does not appearpractical to keep the difference in arm length nearly constant, since this wouldrequire applying quite large forces to the test masses to overcome the changinggravitational forces on them due to the Sun and to the planets. If such large forceswere applied, it would be very difficult to avoid noise in those forces withinthe frequency band of interest from giving undesired changes that could not bedistinguished from real signals. To avoid this problem, the test masses are left