ORTEC DSPEC 50, Руководство пользователя по оборудованию

Страница: 17 / 150

932502G / 0618 1. INTRODUCTION
The protection-time settings with the low-frequency rejector (LFR) filter
off
range between:
!
Minimum PT (highest throughput) = (1 × rise time) + (1 × flattop)
!
Maximum PT =
(2 × rise time) + (1 × flattop)
With LFR on , the protection-time settings range between:
!
Minimum PT = (3 × rise time) + (2 × flattop)
!
Maximum PT = (6 × rise time) + (3 × flattop)
1.5. The Low Frequency Rejector (LFR) Filter
In designing an MCB that can be used in conjunction with mechanical coolers, ORTEC has
developed a new digital filter, the Low Frequency Rejector (LFR) filter, capable of correcting
the pulse output signal for changes in the baseline caused by cooler-induced microphonics. In
many ways digital filters are easier to understand than their analog counterparts. Figure 2 shows
the voltage step output produced at the preamplifier by the collection of charge produced by
absorption of a gamma-ray and the resulting trapezoidal weighting function in a digital spectro-
meter. The difficulty in the measurement is to precisely determine the height of the step pulse
because the baseline contains noise. A fairly obvious estimate of the step signal is obtained by
averaging the digitized samples of the signal before and after the step. M samples immediately
after the event are first ignored, to allow for a maximum rise time of M times the sample inter-
val. N samples of the baseline prior to the step pulse are averaged and then subtracted from the
average of N samples taken after the step pulse. This simple procedure produces a trapezoidal
weighting function with a rise time of N sample intervals and a flattop of M sample intervals.
The maximum value of the trapezoid output, occurring at the end of the flattop, is the best esti-
mate of the step height and therefore the gamma-ray energy. With a proper selection of M and N,
this filter is very nearly the optimum filter for a system with noise arising only from the detector
leakage (parallel noise) and the FET current (series noise).
The trapezoidal filter is essentially independent of dc offsets, since the averaging and subtracting
removes the dc component of the signal. Unfortunately, it is just as sensitive as analog filters to
slowly varying signals such as those produced in microphonic noise. Figure 2 shows the output
of the trapezoidal filter is equal to the slope of the baseline signal multiplied by the full width at
half maximum (FWHM) of the trapezoid. If a step pulse were to be measured on such a base-
line, the filter output value would be too great by an error equal to the difference between the
average values A1 and A2. Since the microphonic noise component in a signal is approximately
a sine wave, as illustrated in Fig. 3, the error induced can be positive, negative, or zero. This
error signal adds to the width of the spectral lines, appearing as degraded resolution performance
from the detector, and can in many cases be a dominant noise source, especially at lower
energies.
7