This article presents a two-step deconvolution routine to compensate for the measurement distortions of transient electrical fields in the picosecond domain in an electromagnetic pulse exposure facility. Because of the low-pass nature of connection cables and the limited bandwidth of a transient digitizer (Tektronix SCD 5000), the measured signal is a distorted output of D-dot (dD/dt, where D is the electric displacement) sensors. An empirical transfer function of the cable-digitizer system was evaluated using a reference impulse generated by a picosecond pulse generator. The reference impulse was injected into the connection cable at the D-dot sensor end and measured with the SCD 5000 while the cable was kept in the same position as for making D-dot measurements to ensure an in-position compensation. Two types of asymptotic conical dipole (ACD) D-dot sensors were utilized as the electrical field sensing devices: an axial model ACD-1(A) and a radial model ACD-1(R). Due to its right-angle structures, the ACD-1(R) gives a different output from that of ACD-1(A) to the same pulse, especially to the fast leading edge. The right-angle bends in ACD-1(R) cause reflections of a sensed signal in the sensor. To correct the errors due to the reflections, the impulse response of the ACD-1(R) D-dot sensor was proposed as a summation of the δ function and the parameters were determined with a pulse measurement using the ACD-1(A) D-dot sensor and an optimization procedure with the Levenberg–Marguardt algorithm. With this routine the measurement accuracy for the electrical pulse in picosecond domain has been improved significantly.
Read full abstract