Spectrometer Systems Overview

Most radiation detectors require pulse processing electronics to extract and interpret the energy or time information involved with radiation interactions. Some techniques only need to detect whether a pulse is present or not (logic pulses), while others like the spectroscopy methods necessitate a more complex analysis (linear pulses):

 

  • Logic pulses carry information only by their presence or absence. A good example is in nuclear medical imaging, where positron emission tomography (PET) systems detect pairs of gamma rays emitted indirectly by a tracer that was previously introduced into the body of the patient. The system detects and associates pairs of digital pulses to determine their origins. Three-dimensional images of tracer concentration within the body can then be constructed.
  • Linear pulses provide valuable information through their amplitude and shape. This information is used in a wide range of applications such as the magnetic resonance spectroscopy or in baggage and cargo scanning. For the cargo scanning, the Z-spectroscopy technique  measures the full energy spectrum of the received X-rays pulses by analyzing the pulses characteristics, which provides information about the average atomic number (Z) of the material being scanned.

 

Although spectrometer systems widely differ by their intent, their radioactive energy source (visible, x-ray, gamma, etc.) or the type of material used as detector, they all require similar pulse processing electronics. For a long time, most of the pulse processing was performed by analog components. But recently (in the last decade or so), the development of fast analog to digital converters (ADCs) with a high dynamic range, coupled with digital signal processors (DSPs) and field programmable gate arrays (FPGAs) have changed the game.

 

Classical approach

The figure below presents a simplified overview of the general pulse processing used in spectroscopy systems. Right after the detector, there is a preamplifier which acts as an interface between the detector and the following pulse processing electronics.

 Classical pulse processing overview

 Figure 1: Classical pulse processing overview

 

In traditional systems, the pulse processing is composed of many functions implemented through the use of analog and digital circuits. Pulse shaping (filtering) and amplification, pulse pile-up detection and rejection were usually done by analog elements, while amplitude analysis and energy spectrum generation (histogram) were performed by digital circuits as ADCs and memories.

 

The new way to process pulses

As technologies advance, each year the market introduces faster and higher resolution ADCs, reaching the point where there is no longer a need to process the signal with analog elements. The previous figure can now be redrawn as follows:

  Today’s pulse processing overview

Figure 2: Today’s pulse processing overview

 

In the above solutions, the ADC comes right after the preamplifier, allowing a quasi-complete digital processing chain.

In Figure 2 A), most of the processing is performed by a PC. This is possible with the development of high-speed digitizer cards (ADCs) which have deep memory for data storage. The samples are transferred from the digitizer card to the PC memory before being analyzed. This kind of architecture offers a great flexibility for the processing algorithms and greatly simplifies the development. Unfortunately, with today’s systems requiring higher sampling rate and resolution, the PC rapidly becomes the bottleneck, especially for multichannel systems.

In Figure 2 B), the pulse processing is performed by an FPGA. Its highly parallel architecture results in a higher throughput capacity and a lower processing latency compared to PC based or DSP processor based implementations. The continued growth of user friendly tools helping designers to develop FPGA digital signal processing algorithms makes the FPGA a more accessible solution for designers, which previously was reserved for FPGA experts only. One good example is System Generator from Xilinx. This high-level tool offers FPGA programming in Matlab® Simulink® environment.

 

But why digitally process the pulses?

There are many advantages to digitizing the electrical signal right after the preamplifier and performing digital pulse processing compared to an analog implementation. This section highlights the main ones.

  • Higher ADC resolution: This might sounds weird since analog signals have an infinite resolution while the digital samples are limited by the ADC width. But in fact, non-ideal analog components add noise to the electrical signal as it passes through, and then degrade it. On the other side, digital processing is predictable and immune to environmental conditions like temperature. Combined with a high flexibility, digital processing techniques often result in better noise suppression and higher resolution.
  • Higher throughput: For a while, the ADCs sampling frequency was limiting the pulse throughput of digital solutions. But for the systems of today, with very high speed ADCs, the pulse overlap (pile-up) appears now to be one of the main limitations. Typically, the time of arrival of the pulses is not uniform. For high counting rate system, two or more pulses can overlap with high probability. Analog techniques to detect pile-up and to separate the pulses are less efficient than digital implementations, resulting in a higher rejection rate and ultimately a lower throughput than digital solutions.
  • Time of arrival: In some systems, the time of arrival of pulses is critical since they need to be associated with pulse occurrences on other detectors. This is easier to manage in a digital implementation where all delays are well known.
  • Reduced size: High density low-power FPGAs reduce size and improve portability of spectrometry systems compared to analog electronic components. This is especially true for systems with tens, hundreds or even thousands of channels, since the same FPGA might be used to handle different channels.
  • Flexibility: FPGAs have a high level of flexibility compared to analog components. They provide the ability to dynamically adjust different settings, such as the parameters of a shaping filter to optimize the performance, or to set some initial values during the calibration phase. They can even be completely reconfigured to modify their operation without any hardware changes. This is very useful in laboratory research.

 

There are many other advantages, such as the ease to update the design to improve both it and the system stability over time.

 

Next generation digital spectrometers

Digital spectrometer systems will continue to benefit from the advancement of technology. As the ADC’s performance continues to increase in terms of sampling frequency and resolution, higher counting rates system with better resolution will progressively appear on the market. The increase of the FPGA density will also allows designers to implement more complex pulse processing and handle more channels per FPGA, reducing the size and the cost of their solutions.

At the present moment, most systems combine a PC with FPGAs in order to control the system, as well as to perform the last processing steps and display the results. The next generation of digital spectrometers would probably be fully embedded, using single SoC with embedded programmable logic and processors such as the Xilinx Zynq FPGA, which includes a lot of programmable logic and a dual core ARM CPU.