Method for operating a signal filter and radiation detection system

Abstract

In an embodiment a method for operating a radiation detection system having at least one radiation detector and at least one signal filter includes supplying an input signal to the at least one signal filter by the at least one radiation detector, the input signal having step-shaped signal rises, each step-shaped signal rise having a rise time, determining the rise time of a respective step-shaped signal rise, specifying a waiting time for the respective step-shaped signal rise in each case such that the waiting time is greater than or equal to the rise time of the respective step-shaped signal rise and producing an output signal of the at least one signal filter, data point pairs of the input signal being processed in which a time interval of data points from each other is equal to the waiting time for the respective step-shaped signal rise, wherein at least 80% of rise times of the step-shaped signal rises lie between 10 ns and 800 ns inclusive, and wherein the at least one radiation detector includes a silicon drift detector having a radiation entry window of at least 5 mm.sup.2.

Claims

1. A method for operating a radiation detection system comprising at least one radiation detector and at least one signal filter, the method comprising: supplying an input signal to the at least one signal filter by the at least one radiation detector, the input signal having step-shaped signal rises, each step-shaped signal rise having a rise time; determining the rise time of a respective step-shaped signal rise; specifying a waiting time for the respective step-shaped signal rise in each case such that the waiting time is greater than or equal to the rise time of the respective step-shaped signal rise; and producing an output signal of the at least one signal filter, data point pairs of the input signal being processed in which a time interval of data points from each other is equal to the waiting time for the respective step-shaped signal rise, wherein at least 80% of rise times of the step-shaped signal rises lie between 10 ns and 800 ns inclusive, and wherein the at least one radiation detector comprises a silicon drift detector having a radiation entry window of at least 5 mm.sup.2.

2. The method according to claim 1, further comprising generating a signal band in the output signal for each signal rise, and wherein a height and/or an integral of the signal band is proportional to a step height of an associated step-shaped signal rise.

3. The method according to claim 1, further comprising buffering the input signal so that the data points of the input signal relating to the respective step-shaped signal rise are stored at least temporarily.

4. The method according to claim 3, wherein supplying the input signal to the at least one signal filter by the at least one radiation detector, buffering the input signal, and producing the output signal of the at least one signal filter are performed continuously and simultaneously, wherein determining the rise time of the respective step-shaped signal rise and specifying the waiting time for the respective step-shaped signal rise are each triggered by one of signal rises so that the waiting time remains the same until a new waiting time is determined in a further determination of the rise time on a basis of a subsequent step-shaped signal rise, and wherein a change from the waiting time to the new waiting time takes place abruptly.

5. The method according to claim 1, wherein waiting times assigned to signal rises are partly below 0.8 times and partly above 1.2 times of an average waiting time.

6. The method according to claim 5, further comprising buffering the input signal to the at least one signal filter by the at least one radiation detector so that the data points of the input signal relating to the respective step-shaped signal rise are stored at least temporarily, wherein buffering the input signal to the at least one signal filter by the at least one radiation detector comprises buffering to a maximum waiting time so that enough data is buffered in order to be able to calculate a filter output with the maximum waiting time, so that the data points are discarded after the maximum waiting time has elapsed, and wherein the maximum waiting time is fixed for a particular input signal and the maximum waiting time exceeds the average waiting time by at least a factor 1.5 and by at most a factor 3.

7. The method according to claim 1, further comprising rejecting a further step-shaped signal rise, which follows a preceding step-shaped signal rise within a sum of the waiting time and a pulse forming time, in a signal evaluation of the input signal.

8. The method according to claim 1, further comprising generating the output signal of the at least one signal filter from the data point pairs by a trapezoidal filter, a Gaussian filter, or a Cusp filter.

9. The method according to claim 1, wherein the input signal to the at least one signal filter by the at least one radiation detector comprises a pre-amplified signal and a digitized signal of the at least one radiation detector.

10. A radiation detection system comprising: at least one radiation detector; a charge-sensitive amplifier; and evaluation electronics comprising a preamplifier, an analog-digital converter (ADC), and at least one signal filter configured to perform digital signal processing, wherein the evaluation electronics follows the charge-sensitive amplifier, wherein the evaluation electronics includes, seen from the charge-sensitive amplifier, the preamplifier, the analog-digital converter (ADC), and the at least one signal filter, wherein the at least one radiation detector comprises at least one radiation-sensitive region and at least two electrodes connected to the at least one radiation-sensitive region, wherein the at least one radiation detector is configured to supply an input signal to the at least one signal filter, wherein the at least one signal filter is configured to provide an output signal of the evaluation electronics, wherein the input signal comprises step-shaped signal rises, each step-shaped signal rise having a rise time, wherein at least 80% of rise times of step-shaped signal rises lie between 10 ns and 800 ns inclusive, and wherein the at least one radiation detector comprises a silicon drift detector having a radiation entry window of at least 5 mm.sup.2.

11. The radiation detection system according to claim 10, wherein the evaluation electronics is configured to generate X-ray spectra, and wherein the at least one radiation detector is configured to detect X-rays.

12. The radiation detection system according to claim 10, wherein the radiation detection system is an Energy Dispersive X-ray spectroscopy (EDX) system.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) In the following, a method and a radiation detection system are explained in more detail with reference to the drawing on the basis of exemplary embodiments. Same reference signs indicate the same elements in the individual figures. However, there are no references to scale shown, rather individual elements may be exaggeratedly large for a better understanding.

(2) In the figures:

(3) FIG. 1 shows a schematic perspective sectional view of an exemplary embodiment of a radiation detection system;

(4) FIGS. 2 and 3 show schematic representations of an input signal for a method;

(5) FIGS. 4 and 5 show schematic block diagram representations of signal filters;

(6) FIGS. 6 to 8 show schematic representations of input signals and output signals for the methods;

(7) FIG. 9 shows schematic representation of rise time distributions for radiation detection systems; and

(8) FIG. 10 shows a schematic representation of time curves of waiting times and output signals for a method and for a modification of a method.

DETAILED DESCRIPTION OF ILLUSTRATIVE EMBODIMENTS

(9) FIG. 1 shows an exemplary embodiment of a radiation detection system 1. The radiation detection system 1 is especially configured for energy dispersive X-ray spectroscopy (EDX). The energy dispersive spectroscopy of X-rays has far-reaching applications in research and industry, especially in material analysis. The non-destructive analysis of the atomic composition of a sample using EDX is particularly widespread, for example in industrial quality control, in the geosciences or in waste recycling.

(10) The basis for this is the excitation of samples, mostly with primary X-rays or electrons, and the energy dispersive measurement of the emitted fluorescence radiation through the transition of electrons excited in the sample back to the ground state. By comparing the measured fluorescence spectrum of the sample with the known energies of characteristic X-ray lines of the periodic table of chemical elements, the elements present in the sample can be identified and, after suitable calibration, the concentration of the respective element can also be determined from the intensity of the respective characteristic lines.

(11) In practice, X-ray fluorescence analyses are mostly performed in laboratory benchtop devices, with portable handheld devices or via EDX modules in electron microscopes such as SEMs or TEMs. The requirements for the detector are a compact design, a high energy resolution for the exact identification of X-ray lines, large active areas and a low background noise. Therefore, semiconductor radiation detectors and in particular silicon drift detectors, SDDs for short, are mostly used.

(12) A modern form of semiconductor radiation detectors is the silicon drift detector, see FIG. 1. With such a radiation detector 10, a large, usually round n− semiconductor volume is depleted as a radiation-sensitive region 11 via the principle of sideways depletion from the direction of a p.sup.+ rear contact 13 towards a radiation entry window 17 and metallic drift rings 12a by applying a reverse voltage. The anode 12 is an n+ contact 19a, which is preferably implanted in the detector center. The thus depleted area of the semiconductor serves as an active volume 11 for the absorption of X-rays X.

(13) The type of doping illustrated in FIG. 1, the materials used and the exact structure, for example, with regard to ring voltage generation via voltage dividers, are only examples. Depending on the design, the structural and material set up of the radiation detector 10 may differ significantly from that shown in FIG. 1. Only the basic principle of reading charges at an electrode by generating a suitable potential field shall be illustrated. The method described here can therefore also be used for other detector types.

(14) In this volume 11 electron-hole pairs resulting from radiation interactions are separated by the applied voltage. The number of electron-hole pairs is proportional to the energy of the absorbed photon. To read out these signal charges, lower and lower voltages V1 . . . Vn are applied to the drift rings 12a with decreasing radius, so that the potential gradient for electrons in the active volume 11 always points to the anode 12, 19a. Electrons drift to the anode 12, 19a and are read out there. Holes flow off via the rear contact 13.

(15) On the rings 12a there are preferably p.sup.+-doped areas 19b. Between the rings 12a, areas of an insulation layer 18, for example, of SiO.sub.2, can be placed. The radiation entry window 17 can be formed by a p.sup.+-doped layer and/or by a window material which is transparent to the radiation to be detected and which protects the radiation detector 10. A further electrode 16 is preferably attached to one of the outermost rings 12a as earth connection GND. The anode 12 is electrically followed by a charge-sensitive amplifier 14, which is connected to an evaluation electronics 15 preferably by means of a preamplifier 151, an ADC 152 and a digital signal processing 30 with signal filters 2. The evaluation electronics 15, for example, is an electrical circuit on which analog components and ICs are connected and/or which is integrated in a computer. An energy spectrum 40 is preferably output by the evaluation electronics 15.

(16) Signal electrons read out by the radiation detector 10 are thus preferably converted into a voltage signal in the detector module by an application-specific integrated circuit, ASIC for short. In particular, this ASIC is a very low-noise, charge-sensitive CMOS amplifier 14 which collects electrons from the radiation detector 10 on a feedback capacitance and generates a voltage signal proportional to the time integral of the input current by connecting operational amplifiers as integrators at the output of the detector module. Other ASICs or individual transistors are also possible as charge-sensitive amplifier 14.

(17) If an electron cloud arrives at the anode 12, the charge on the feedback capacitance and thus also the output signal of the charge-sensitive amplifier 14 increases sharply. Typical signal rise times range from several tens of nanoseconds to several hundred nanoseconds. The rise time depends on the drift time of the electron cloud and thus in particular on the location of the photon absorption in the detector and on the detector size. The height of the pulse is proportional to the number of signal charges and thus to the energy of the absorbed photon.

(18) According to the example in FIG. 1, the radiation detector 10 contains the charge-sensitive amplifier 14, which converts signal electrons into voltage signals. On the other hand, the preamplifier 151 for filtering, conditioning and/or amplification is located in the evaluation electronics 15. The radiation detector 10 and the evaluation electronics 15 can be separate components, symbolized in FIG. 1 by dash dot lines. However, it is also possible that the charge sensitive amplifier 14 and the preamplifier 151 are both integrated into the detector module. The explicit separation of the preamplifier 151 and the charge-sensitive amplifier 14 and/or the division into the detector module and the evaluation electronics 15 shown is therefore only optional.

(19) FIG. 2 shows an example of a voltage pulse at the output of the detector module due to a radiation interaction in the radiation detector 10. A time interval of approximately 1 ρs is shown. The signal shown essentially corresponds to the input signal 31 at the signal filters 2, whereby a preamplifier and an ADC are preferably connected in between. The input signal 31 has exactly one step-shaped signal rise 32 in the time range shown. Before and after the signal rise 32, the input signal 31 is approximately constant, with the input signal 31 rising very slowly because of the detector leakage current before and after signal rise 32.

(20) The object of the evaluation electronics 15 is to determine the signal levels of these voltage pulses 32 at the output of the detector module. Since the height of such a voltage pulse 32 is proportional to the number of charges released by the radiation interaction and thus is also proportional to the energy of the absorbed photon, the qualitative form of the energy spectrum 40 can be inferred from the relative frequency of the pulse heights. This analysis is performed using digital signal processing 30 after the signal has passed through the 151 preamplifier system and an ADC 152.

(21) The voltage pulses 32 from radiation interactions are superimposed by the detector leakage current. This leakage current is also read out by the potential curve in detector 10 and integrated into the charge-sensitive amplifier; because that the voltage signal shows a slight increase caused by the detector leakage current even without voltage pulses. Due to the finite capacitance in the charge-sensitive amplifier, this may be discharged at a certain threshold voltage, also referred to as a reset, resulting in a sawtooth-like output signal at the detector module, see FIG. 3.

(22) The radiation detection system, which can be regarded as a spectrometer, thus consists essentially of the energy dispersive X-ray detector 10, which is mostly a semiconductor radiation detector such as the silicon drift detector, and of the evaluation electronics 15, preferably comprising or consisting of preamplifier 151, ADC 152 and digital signal processing 30.

(23) FIGS. 4 and 5 illustrate schematic block diagram representations of examples of methods in signal filters 2 described here. The input signal 31 is routed in parallel to an optional data buffer 36, to a fast filter 37 and, if necessary, to a unit 39 for determining the rise time. The determined waiting time T is transmitted to an energy filter 22, which is also optionally connected to the data buffer 36. Furthermore, a signal evaluation/communication 35 is preferably available, which can receive data from the energy filter 22 as well as from the fast filter 37. The task of the signal evaluation/communication 35 is, among other things, the extraction of values from the energy filter 22 at suitable times and the calculation of the energy spectrum 40.

(24) In the evaluation electronics 15, further analog signal processing often takes place, for example, amplification and filtering, before the preamplifier signal is digitized in an analog-to-digital converter 152. For example, an ADC of type AD9649 of the manufacturer Analog Devices can be used. The digital further processing of the signal for the measurement of the X-ray spectrum is preferably carried out in a programmable digital computing unit of the evaluation electronics, which is usually an FPGA, whereby FPGA stands for Field Programmable Gate Array, or (at the customer) stands for programmable logic gate arrangement. For example, a Spartan-7 FPGA from Xilinx is used.

(25) Via an external communication interface, for example, via USB, Ethernet or SPI, commands can be sent to the evaluation electronics, for example, commands such as ‘Start measurements’, ‘Configure filter’, and so on, and data can be read out. For example, a spectrum 40 can be transmitted to a terminal device for further evaluation.

(26) Such a structure is sketched, for example, in the publication U.S. Pat. No. 5,870,051 A, see in particular FIGS. 1 and 3. Contrary to what is shown in this publication in FIG. 2, the levels FIPPI and DSP can be implemented in the same FPGA and a further D/A conversion can be omitted. The disclosure content of this publication, in particular with regard to FIGS. 2 and 3, is incorporated by reference.

(27) In the programmable digital computing unit 30 of the evaluation electronics 15, the energy spectrum 40 is calculated from the ADC data. Several digital filter outputs are usually calculated in parallel from the ADC data. The most common filters are sliding mean value filters, often with trapezoidal step responses, but so-called Cusp filters are also possible, in which the ADC data is weighted differently.

(28) Preferably at least two digital signal filters 2 are calculated:

(29) A first filter with a very small time constant, which has a high temporal resolution, is used to detect signals, for example, to determine the number and timing of the voltage pulses. Common names for this filter 37 are Fast Filter, Pulse Detection Filter or Peak Detect Filter.

(30) A second filter with a longer time constant is used for efficient noise suppression and energy determination. Common names for such a second filter 22 are Energy Filter or Slow Filter.

(31) In addition, further filters with average time constants can be used, which are used in particular to correct the detector leakage current (also known as baseline correction). A functional block diagram of the programmable digital processing unit is shown in FIG. 1 of document U.S. Pat. No. 7,763,859 B2 or in FIG. 6 of document U.S. Pat. No. 6,609,075 B1. The disclosure content of these documents, in particular with regard to the figures mentioned, is incorporated by reference.

(32) To detect step-shaped signals, the fast filter output is preferably tested against a limit value, and if this value is exceeded, the energy value is taken from the energy filter at a suitable time shortly thereafter, for example, by searching for the maximum in the energy filter output in a narrow time range after the limit value has been exceeded. This is shown, for example, in connection with FIG. 5C or 8A in document U.S. Pat. No. 5,873,054 A. The disclosure content of this document, in particular with regard to these figures, is incorporated by reference.

(33) A further task of the fast filter is the rejection of sum events, so-called pile-ups, which occur when two signals follow each other quickly. In this case, the energy filter cannot be evaluated because its height does not correspond to the correct signal height, but is composed of the sum of several individual events. If the fast filter output exceeds the limit value several times in a defined period of time, the signals are rejected, also known as pile-up rejection. This is also shown in document U.S. Pat. No. 5,873,054 A.

(34) The longer the energy filter is in time, the greater the time interval between two signals must be in order to be able to evaluate both. The minimum time interval for correct evaluation is the sum of the peaking time and the gap time of the energy filter. These two terms are explained below. As an example, the step-shaped response of a trapezoidal filter is shown in FIG. 5A of document U.S. Pat. No. 7,807,973 B2. The disclosure content of this document, in particular regarding this figure, is incorporated by reference.

(35) To determine the energy spectrum, especially in energy dispersive X-ray spectrometers, step-shaped signals must be evaluated in the detector signal, whereby the step height of the signals contains information about the energy of the detected radiation. A filter, also known as an energy filter, with a trapezoidal step response, which is calculated from the difference between two moving averages, is often used to determine the energy.

(36) At each time step, the mean value is calculated from a certain number of the most recent values of an input signal 31, preferably from ADC values 152. The duration of the averaged ADC values is called Peaking Time, and in the example in FIG. 6 this is two time steps. FIG. 6 schematically shows the input signal 31, ADC in the upper half, and the output signal 33 in the lower half.

(37) For the first moving average, the two ADC values A are averaged. The second moving average value is calculated from the same number of values from the adjacent, next-older ADC values B. To calculate the filter output, the older moving average value B is now subtracted from the newer moving average value A, so that a new value C is available in the filter output. An ideal step signal of the ADC values thus becomes a signal band 34 of triangular form in the filter output 33, where the maximum of the filter output 33 equals the step height of the ADC values, that is, one.

(38) However, signal pulses in the detector output signal of a radiation detector 10, for example, a silicon drift detector, SDD for short, are not ideal steps, but have rise times between a few 10 ns and several 100 ns due to drift times in the detector, electrical capacitances and finite amplifier slopes, for example. If one calculates the filter output 33 for a pulse with a rise time of two time steps according to the scheme described above, the time curve of the output signal 33 shown in FIG. 7 is obtained.

(39) Due to the finite rise time R of the signal rise 32, the signal band 34 widens in the filter output signal 33. In addition, the full step height of the ADC signal 31 is no longer reached in the filter output 33. At present, the height of the signal band 34 is about 0.6 instead of 1 in the input signal 31. A correct determination of the step height 32 is therefore not possible with this energy filter. For the correct evaluation of the step height 32, the energy filter 2 therefore uses a time interval, that is, the waiting time T or gap time, between the two moving mean values, see FIG. 8.

(40) As a double arrow, FIG. 8 also illustrates the time interval between the data points A and B, which in this case are moving averages with an averaging duration of two steps. This distance is equal to the waiting time T (two time steps in this example) plus the peaking time (two time steps), that is, is equal to four time steps.

(41) ADC values in this time interval are not used in this time step to calculate the energy filter output 33 or are weighted for this time step with the factor zero. The time interval T in this example is two time steps and is called Gap Time. If the gap time is equal to or greater than the rise time R of the step signal, the filter output 33 reaches the correct step height. In this example, T=two time steps=R. The reason for the correct determination of the step height is that at one point in time, one moving average value is completely before the step signal 32 and one moving average value is completely after the step signal 32. This is always fulfilled as long as T≥R applies.

(42) In the example shown here, all ADC values within a moving average are equally weighted. Thus, the latest moving average is calculated by forming the sum of the two ADC values A and dividing by the number of values. It is also possible to weight the ADC values differently, for example, to take the most recent value into account twice. This results in step responses that deviate from the simple trapeze. However, this does not change the further procedure of the evaluation and the necessity of a waiting time T. A moving average value without individual weighting, that is, a trapezoidal filter, is preferred; alternatively, moving average values with weighting, like Cusp filters, can be used.

(43) One problem now lies in the fact that in particular a silicon drift detector 10 does not have a constant rise time R due to its principle of construction. The reason for this is that an absorption of X-ray photons takes place both close to the anode, that is, at the center of detector 10, and far from the anode, that is, at the edge of detector 10. Absorption remote from the anode leads to a longer drift time in the detector 10 and, thus, to a longer rise time. For a uniformly irradiated detector 10, this results in a distribution of rise times that is wider the larger the detector area is. This is illustrated in FIG. 9 for two silicon drift detectors 10a, 10b, where the rise time R is plotted against the relative frequency W for each.

(44) In the signal evaluation, as far as possible all impulses should be recorded correctly, which is why the waiting time T is longer than the longest expected rise time in the application of conventional methods. For example, as shown in FIG. 10, the maximum rise time for the SDD detector 10a with a radiation entrance area of 20 mm.sup.2 is 70 ns, and for the SDD detector 10b with a radiation entrance area of 80 mm.sup.2 it is 220 ns.

(45) The disadvantage of this procedure with a fixed waiting time, however, is that the gap time is selected longer than necessary for a large proportion of the pulses and that the evaluation takes more time than necessary, since the gap time of the longest expected rise time is also used for signals with relatively short rise times. This extends the duration of the signal bands in the filter output signal 33* and increases the probability that a second pulse occurs in the detector signal during the evaluation time and that both signals must be rejected, since the pulse heights of the individual pulses can no longer be determined correctly, corresponding to a pile-up rejection, see above.

(46) Contrary to that, the gap time is selected adaptively in the procedure described here. The improvement in reducing dead time is in particular due to the fact that the gap time is set individually for each pulse, so that this specific gap time is equal to or greater than the rise time of the corresponding pulse. Instead of the conventional estimation for the worst case, that is, that the gap time is always fixed statically to the maximum rise time to be expected, the gap time of the energy filter 2, 22 is dynamically adapted to the rise time R of the relevant signal rise 32 for each pulse and, thus, for each signal rise 32 in the method described here.

(47) The advantage of this method is in particular a faster evaluation, since the gap time is in many cases smaller than previously selected. This allows the throughput of signals to be increased, especially for large detectors with wide rise time distribution.

(48) In this example, the method follows the following algorithm: First, the input signal 31, for example, the ADC data, is optionally buffered for so long that the filter output can be realized with the maximum waiting time. In particular, this means buffering data for at least twice the peaking time plus waiting time. Secondly, the rise time R of the detected pulse, that is, of the respective signal rise 32, is determined. Third, the energy filter output, that is, the associated signal band 34 of the output signal 33, is calculated with the minimum gap time T for this pulse.

(49) A simulated output signal 33 for a method described here with a dynamic waiting time T and a filter output signal 33* for a method with a static waiting time are illustrated in FIG. 10. The terms dynamic waiting time and adaptive waiting time are used synonymously.

(50) Curve 37 denotes the fast filter which is operated without gap time and which is used for pulse detection. Curve 33* refers to a static energy filter with a fixed gap time of 350 ns according to a modification of a method. Curve 33 illustrates an energy signal filter 2 with an adaptive gap time according to the method described here. Finally, the curve T shows the temporal course of the gap time. All signals shown have a nominal signal height of 30 mV, which is identical for the two curves 33, 33*.

(51) The first two signals, see area I in FIG. 10, have a relatively short rise time. Since the filter output 33 is adapted to the rise time R in the method described here, the adaptive filter 2 uses shorter waiting times. The waiting time of the static filter 33*, on the other hand, remains at 350 ns, which would be necessary for the slowest possible signal of the detector 10. Both static and dynamic waiting times lead to the correct filter height of 30 mV, whereby the dynamic method takes less time to evaluate.

(52) The voltage pulse on the right in FIG. 10, see area II, is a rather slowly rising signal, therefore a rather long waiting time is chosen in the adaptive method and the filter durations of the adaptive and the static method are only slightly different. Both the static and the dynamic determination of the waiting time lead to the correct filter height of 30 mV.

(53) In the middle area III of FIG. 10 it is shown that there is a fast sequence of three rapidly increasing signals 32, 32*, 32**. The static procedure uses for all three signals 32, 32*, 32** the statically set waiting time of 350 ns. Since the corresponding signal band 34 in the filter output signal 33* is thus longer than the time interval between the signals 32, 32*, 32**, overlapping effects occur in the filter output 33*. The height of the filter output 33* exceeds the actual signal height of 30 mV. It is therefore not possible to resolve these three signals 32, 32*, 32** with a static waiting time of 350 ns.

(54) However, the dynamic method recognizes that the maximum waiting time of 350 ns is not necessary for these rapidly following pulses 32, 32*, 32**, since the rise time of these pulses is significantly less than 350 ns. Instead, smaller waiting times are used, namely for each of the three pulses 32, 32*, 32** a waiting time slightly above the actual rise time is used. This shortens the signal bands 34 in the output signal 33. In this case, the energy filter 22 can resolve the pulse sequence 32, 32*, 32** in time and the correct height of 30 mV results for all three signals 32, 32*, 32**. This leads to an improved pulse throughput in the adaptive method described here.

(55) At times when the dynamic filter 2 changes the waiting time, discontinuities occur. However, these discontinuities are not relevant for the pulse evaluation, since they occur at times when the energy filter 2 is not evaluated. By buffering data, the times at which the waiting time changes can be set and therefore these changes can also take place at times at which the energy filter is not evaluated and the discontinuities are not relevant.

(56) There are several possible methods for determining the rise time. For example, a comparison can be made between a slope of the signal and a threshold value. The rise time can also be determined by a duration of the fast filter above a threshold value.

(57) It is possible to change the signal form during analog signal conditioning in the evaluation electronics 15 in order to simplify digitization in the ADC 152. For example, a high-pass filter in the preamplifier 151 can be used to filter the low-frequency leakage current rise from the signal, or an analog subtraction of the estimated leakage current can be performed. This leads to other types of signals arising from the step-shaped signal in the preamplifier output in the ADC signal, which can then be digitally recalculated to the original preamplifier signal. The signal is then digitally processed in the same way as in the case without analog signal conditioning.

(58) The invention described here is not limited by the description given in the exemplary embodiments. Rather, the invention includes any new feature and any combination of features, which in particular includes any combination of features in the patent claims, even if that feature or combination itself is not explicitly mentioned in the patent claims or exemplary embodiments.