ORTEC Logo

Home | Applications | Contact Us
Products | Service | Training
  Digital Signal Processing (DSP)

In the previous few pages the functions incorporated in linear pulse-shaping amplifiers have been described in terms of analog signal processing components. Alternatively, most of these functions can be implemented by means of Digital Signal Processing (DSP). Basically, the DSP method converts the continuous analog signal at the output of the preamplifier to a stream of digital numbers representing the history of the preamplifier output voltage. The technique is implemented by using a flash ADC to repeatedly sample and digitize the preamplifier signal. The constant interval between samples is typically small so that the digital numbers represent the pulse profiles with reasonable accuracy. For every analog pulse processing function in the continuous time domain, one can construct an equivalent function in the discrete time domain of the digital representation. Thus, the equivalent signal processing can be implemented in a computer. Because software computation would be too slow to keep up with the data rates, the processing is done in a hardware circuit known as a DSP (Digital Signal Processor).

Figure 25(A) shows the block diagram of a typical DSP MCA, which is a complete digital signal processing system for gamma-ray spectrometry. The digital signal processing in this system incorporates the low- and high-pass filters, automatic pole-zero adjustment, the baseline restorer, fine gain adjustment, a spectrum stabilizer, and means for measuring and histogramming the amplitudes of the digital pulses. This latter function replaces the multichannel analyzer normally used with analog signal processing.

Figure 25(B) illustrates the typical digital filter response in the DSPEC. The flat top is employed to eliminate the degradation of energy resolution normally caused by the variations in charge collection time in HPGe detectors (ballistic deficit). For very wide pulse widths, the flat top becomes negligible, and the pulse shape approaches a cusp. The cusp is the ideal filter for achieving the optimum signal-to-noise ratio at the noise-corner time constant. A reasonable approximation to the cusp can be readily implemented in digital signal processing, whereas it is virtually impossible to achieve using analog signal processing. The cusp shape can be easily changed to a trapezoid, which yields optimum energy resolution for shaping-time constants that are small compared to the noise-corner time constant (for higher counting rates).

The benefits of digital signal processing are:
  • greater flexibility in realizing the optimum pulse-shaping filter over the entire range of shaping time constants,
  • improved temperature stability,
  • ballistic deficit correction at short shaping time constants and optimum energy resolution at long shaping time constants, and
  • computer automated optimization of the pulse-shaping filter to suit the detector and data acquisition conditions.

Figure 25(A). DSPEC Block Diagram.

Figure 25(B).  Digital Filter Response.