REPORT ON SOME EXPERIMENTAL LONG-PERIOD SEISMOGRAPHS By RALP~ GILMAN

The unexpected usefulness in research of the Press-Ewing seismograph system (Press, Ewing and Lehner, 1958) has stimulated further efforts to obtain even higher seismograph sensitivity in the period range beyond 100 see. A major difficulty has been in maintaining pendulum stability at long periods. This paper describes a stable long-period seismometer and gives results from several se, ismograph systems incorporating long-period pendulums and galvanometers. A PERIOD LENGTHENING DEVICE IN ORDER to improve pendulum behavior, auxiliary restoring forces are used in addition to those due to springs and gravity. This method, here applied to an ordinary horizontal pendulum, takes the form of a pair of moving coil transducers of the type commonly used as velocity sensors in seismometers. As shown in figure 1, the transducers are arranged so that the coils are attached to opposite sides of the pendulum mass and the permanent magnets are fixed to the instrument frame. A constant current is passed through the coils which are connected in series (figure 2), and polarized to produce restoring forces on the mass opposite in direction to the restoring forces due to gravity. This results in a lengthened period. Since the coil current is constant and the magnetic field strength is uniform along the axis of the pole piece, the resulting force is very nearly proportional to the penetration of the coil in the gap. In this way an approximately linear relationship between force and displacement is achieved. The total current is regulated by a variable resistor and provides for period adjustment. The amount of current passing through each coil is regulated by a balancing potentiometer and affords a convenient centering system. Coil current of 200 u A is sufficient to lengthen the pendulum period from 20 sec. to 100 see. Constant coil current is assured by using mercury batteries. An important advantage of this system arises from the possibility of using remote controls so that centering, period adjustments and tests may be made without disturbing the instrument. Although the mass and boom were made from high grade electrolytic copper, the presence of magnetic impurities in these parts contributed to instability at periods of about 100 sec. In addition to instability, the effect manifests itself by producing a dependence of free period on amplitude. To correct this condition, the pole pieces of the auxiliary transducers were tapered outwardly toward the direction of greater boom deflection. Nonlinear restoring forces are thus provided, opposite in sense to those caused by magnetic contamination. The test for proper compensation is independence of free period on amplitude. Excellent seismometer behavior is thus achieved, operating periods being limited by ground stability. Test oscillations of 180 see. have been produced in the vault of the Seismological Laboratory at Pasadena. Ground tilts from earth Manuscript received for publication March 17, 1960. 553 5 5 4 BULLETIN OF THE SEISMOLOGICAL SOCIETY OF AMERICA Fig. 1. Photograph showing arrangement of components attached to horizontal pendulum. M--pendulum mass; Tl--auxiliary transducer; T2--veloeity transducer; T~--displacement transducer. Fig. 2. Diagram of operating controls for auxiliary transducers. Rl--centering potentiometer, 100 K~; R2--period regulator, 30 K~; A--microammeter, 0-500; Bl--mercury cell, 4.0 V. tides, local thermal effects, and artificial disturbances limit continuous operation to periods below 100 see. Such difficulties are associated with high tilt sensitivity in long-period horizontal pendulums. When these methods are used with a vertical pendulum, it is hoped tha t the absence of difficulties from tilt will result in even greater success. SOME TRANSDUCER AND RECORDING SYSTEMS The following transducer and recording systems have been used with the longperiod pendulum. In all cases seismometer and galvanometer damping was adjusted to the critical value. EXPERIMENTAL LONG-PERIOD SEISMOGRAPHS 555 1. Pendulum period 100 sec. Displacement transducer driving an ink-writlng galvanometer of 0.25 sec. period. 2. Pendulum period 100 see. Velocity transducer driving a Lehner galvanometer of period 480 see. 3. Pendulum period 60 see. Displacement transducer driving an ink-writing galvanometer of 0.25 sec. period. 4. Pendulum period 60 sec. Displacement transducer driving Lehner galvanometer of period 110 sec. Filter with time constant of 50 see. interposed between transducer and galvanometer. PERIOD-SECONDS Fig. 3. Ampl i tude response curves. 00 5. Pendulum period 60 sec. Displacement transducer driving Lehner galvanometer of period 480 sec. Filter with time constant of 200 see. interposed between transducer and galvanometer. Amplitude response curves are shown in figure 3. Curves were calculated for systems using the displacement transducer. A calibration method similar to that of Murphy and others (1954), was used to calibrate the To = 100, T~ = 480 system. Numerous comparisons of waves of nearly sinusoidal character recorded by the various arrangements have verified the near correctness of the curves. Calculated phase response curves for systems using long-period galvanometers are shown in figure 4. RESULTS FROM VARIOUS TRANSDUCER AND RECORDING SYSTEMS 1. To = 100 sec., T~ = 0.25 sec. Displacement transducer.--In the early stages of this work it became apparent that a system for remote monitoring of pendulum 556 BULLETIN OF THE SEISMOLOGICAL 8OCIETY OF AM]!IRIC_& position would be required. A displacement transducer which provides a voltage output proportional to mass position was installed and coupled with a recorder using an ink-writing galvanometer of period 0.25 sec., and paper speed 0.75 inch per hour. Pendulum period was adjusted to 100 sec. Figure 5 reproduces a trace taken from this recorder. Ground tilt due to earth tides is recorded together with two large earthquakes (14 Sept. 1959; 14:00:39; 281⁄2 S, 177 W; M = 71⁄4:15 Sept. 1959; 05:29:42; 281⁄2 S, 177 W; M = 6~). Although this recorder was installed to check pendulum centering, its low peak magnification of 75 provided a convenient means for estimating the magnitudes of large earthquakes which otherwise would be off scale.