Baseline Restorer

The Baseline Restorer

To ensure good energy resolution and peak position stability at high counting rates, the higher-performance spectroscopy amplifiers are entirely dc-coupled (except for the CR differentiator network located close to the amplifier input). As a consequence, the dc offsets of the earliest stages of the amplifier are magnified by the amplifier gain to cause a large and unstable dc offset at the amplifier output. A baseline restorer is required to remove this dc offset, and to ensure that the amplifier output pulse rides on a baseline that is securely tied to ground potential.

Figure 20 illustrates the basic principle of a baseline restorer. In the case of the simpler, time-invariant baseline restorers, switch S1 is always closed. The time-invariant baseline restorer behaves just like a CR differentiator. The baseline between pulses is returned to ground potential by resistor R_BLR. In order not to degrade the signal-to-noise ratio of the pulse-shaping amplifier, the C_BLR R_BLR time constant must be at least 50 times the shaping time constant employed in the amplifier.

The simple, time-invariant baseline restorer does not adequately maintain the baseline at ground potential at high counting rates. Since the time-invariant baseline restorer is really a CR differentiator, the average signal area above ground must equal the average signal area below ground at the baseline restorer output. At low counting rates, the spacing between pulses is extremely long compared with the pulse width. Consequently, the baseline between pulses remains very close to ground potential. As the counting rate increases, the baseline must shift down, so that the area of the signal remaining above ground potential is equal to the area between ground potential and the shifted baseline. The amount of baseline shift increases as the counting rate increases. Diode networks are typically incorporated to reduce this shift, but such solutions are unable to make the shift negligible.

The gated baseline restorer virtually eliminates the baseline shift caused by changing counting rates. In Fig. 20, switch S1 is opened for the duration of the amplifier pulse, and closed otherwise. Therefore, the CR differentiator function is active only on the baseline between pulses. The effect of the signal pulse is essentially eliminated. The gated baseline restorer perceives that it is operating at zero counting rate, and maintains the baseline firmly at ground potential, independent of the actual counting rate.

The stability of baseline restoration at very high counting rates with the gated baseline restorer depends on the ability of the gating control circuits to distinguish between the pulses and the baseline. In the simpler circuits, this is accomplished with a discriminator whose threshold is manually adjusted to sit just above the noise that surrounds the baseline. The more sophisticated amplifiers include automatic noise discriminators and more complicated pulse detection methods to perform this task more effectively. Figure 21 is an example of the results obtained on a high-performance baseline restorer. Peak shift and resolution broadening are both negligible over a very wide range of counting rates. At some upper limit on counting rate, there is inadequate baseline between pulses for the baseline restorer to control. Above that counting rate, the baseline will shift strongly with increasing counting rate. If counting rates must be processed above this limit, then a shorter amplifier shaping time constant must be selected.

Figure 20. A Simplified Diagram of a Baseline Restorer.

Figure 21. (a) Resolution, and (b) Peak Position Stability as a Function of Counting Rate with a High-Performance, Gated Baseline Restorer. Measured on the 1.33-MeV gamma-ray line from a ⁶⁰Co radioactive source, using a 10% efficiency GAMMA-X PLUS detector.