Simplified electronic signal processing in the small nuclear physics laboratory

Simplified electronic signal processing in the small nuclear physics laboratory

ARTICLE IN PRESS Nuclear Instruments and Methods in Physics Research A 551 (2005) 487–492 www.elsevier.com/locate/nima Simplified electronic signal p...

185KB Sizes 0 Downloads 17 Views

ARTICLE IN PRESS

Nuclear Instruments and Methods in Physics Research A 551 (2005) 487–492 www.elsevier.com/locate/nima

Simplified electronic signal processing in the small nuclear physics laboratory P.A. DeYoung, G.F. Peaslee Department of Physics and Engineering Hope College, Holland, MI 49422-9000, USA Received 5 January 2005; received in revised form 26 May 2005; accepted 6 June 2005 Available online 14 July 2005

Abstract Small nuclear physics laboratories of all kinds traditionally have processed the signals from radiation detectors with a variety of discrete NIM- or CAMAC-based electronic modules. The logic signals associated with signal processing are often passed through gate generators, coincidence modules, fan-in/fan-out modules, delay units, counters, and other assorted logic modules. These multi-component systems generate gates for acquisition systems, gates for specific linear electronics modules (ADCs and TDCs), or measure count rates and dead times. This can involve a significant number of individual modules each of which can be quite costly and each of which provides only limited functions. We describe here an upgrade to our acquisition system where all the needed logic functions are performed in just a single unit: a Universal Logic Module based on a Field Programmable Gate Array (FPGA) from JTEC Corporation. This module also contains flash memory that holds three separate configurations allowing for rapid changes from one electronics configuration to a different one. Both CAMAC and VME versions of the unit are available. The system described here is just one example of the huge variety of functionality that can be programmed into this single module. It can accommodate very complicated circuits and is easily reprogrammed. In the small nuclear physics laboratory the Universal Logic Module can save cost when upgrading systems and reduce the number of instances where one has an insufficient number of channels of a particular function. r 2005 Elsevier B.V. All rights reserved. PACS: 07.05.Hd; 07.50.Q+ Keywords: Field programmable gate array; Signal processing; Data acquisition

1. Introduction

Corresponding author.

E-mail address: [email protected] (P.A. DeYoung).

Many research activities at Hope College involve the detection of various types of radiation. This is especially true for the work associated with the recently installed Hope College Ion Beam Analysis

0168-9002/$ - see front matter r 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2005.06.055

ARTICLE IN PRESS 488

P.A. DeYoung, G.F. Peaslee / Nuclear Instruments and Methods in Physics Research A 551 (2005) 487–492

Laboratory where routine analyses are done with Proton-Induced X-ray Emission (PIXE) and Rutherford Back Scattering (RBS). Those familiar with such work will realize that in addition to the standard need to amplify and shape the signals from the various detectors, ADC gates and computer triggers are also needed to initiate the processing of individual events. The creation, manipulation, and counting of the necessary logic signals often require a large number of wires and discrete modules to accomplish the generation of the various gates, overlaps, coincidence functions, and trigger generation. With a single universal logic module, based on a Xilinx Inc. Field Programmable Gate Array (FPGA) [1] one can incorporate most, if not all, the logic functions associated with a typical experimental setup. The device we use is capable of storing three separate experimental electronics configurations. This allows for rapid (less than a minute) reconfiguration of the electronics setup when switching between different experimental techniques. The particular device we use is based on CAMAC, but similar devices are available for VME-based systems as well. The use of FPGAs for logic processing in physics experiments is widespread and even a simple literature search will reveal hundreds of references to FPGAs (some examples are given in Refs. [3–8]). However, generally FPGAs have only been used in conjunction with complex experimental devices and/or when there is sufficient expertise at the laboratory to incorporate the FPGA with other custom electronics. In addition to finding references to the use of FPGAs in nuclear and high-energy efforts, there are also reports of their use in other fields such as astronomy [9,10], medical imaging [11], and biology [12]. Here, we describe the use of an FPGA in a small laboratory setting with simple detector systems and a single data acquisition system. We assert that the ease of use and potential cost-savings of an FPGA could make it valuable for many other small laboratories.

2. Device description Specifically, the device on which we have based our acquisition system is the ULM52C from JTEC

Instruments [2]. The heart of this CAMAC module (as for all the universal logic modules produced by JTEC) is a Xilinx Inc. FPGA. In this model the FPGA is a Spartan XC2S200 with 200,000 system gates. FPGAs are clocked devices and this particular module is clocked at 100 MHz. The device has six LEDs that are useful for status and diagnostic purposes. The module has 52 ECL input/outputs. These pins can be configured as input or output, in groups of four, by inserting the appropriate logic translator chips and resistors inside the module. While the heart of the universal logic module is the FPGA, there are a number of other features of the module that make it more user friendly and more versatile. The modules do include general memory that can be accessed by either CAMAC commands or by the FPGA software. The devices include circuitry to interface with the CAMAC backplane and to respond to CAMAC commands. There are some reserved commands but the FPGA can also be configured to respond to user-defined CAMAC commands as well. One of the most outstanding features is the flash memory included in the device, which allows various FPGA programs (electronics configurations) to be loaded into the flash memory and quickly copied into the FPGA. There is room for three separate configurations in the flash memory of the ULM52C. This facilitates changing from one experimental setup (e.g. PIXE measurements) to another (e.g. RBS measurements) with only a few CAMAC commands. Although the ULM52C cannot be so equipped, some universal logic modules can be ordered with on-board digital signal processors for involved calculations. The device was easy to install because it is a standard CAMAC module. Any laboratory familiar with CAMAC addressing should be able to use the loading/booting software provided by the manufacturer to install and use the ULM52C. Technical assistance from the manufacturer was also exceptional. The device does come with a preinstalled diagnostic program in flash memory which can be loaded into the FPGA, so the unit can be checked with CAMAC commands without any additional programming.

ARTICLE IN PRESS P.A. DeYoung, G.F. Peaslee / Nuclear Instruments and Methods in Physics Research A 551 (2005) 487–492

3. Software description Software is required to develop the FPGA programs and to load them into the unit. A simple Visual Basic program to load the device was provided by the ULM52C manufacturer [2]. Selecting the software to program the device is more complicated. There are several packages available from Xilinx Inc. [13] with different combinations of features and prices. There is free software, ISE WebPACKTM , that includes a schematic (graphical circuit) editor, an HDL (line-oriented coding) editor, synthesis, implementation, and simulation tools. Unfortunately, this package does not include the utility that allows one to schematically design a logic circuit with custom logic elements. This feature is very powerful and very useful but not strictly necessary. A more complete package, ISE BaseX, is available ($695) that includes this very useful component. Of course, there are a number of other commercial and shareware efforts but we have relied only on tools from Xilinx Inc. The authors are willing to share their designs upon request. The overall process associated with programming the FPGA, once one understands the design goals and requirements, begins by starting a new design. This top level design can either be linemode coding (HDL) or a graphical circuit as shown in Fig. 1. HDL elements can be included as blocks on a graphical design. Once the design is complete and the inputs and outputs are completely connected and specified, the design is implemented. This process translates the design into information that can be downloaded to the chip to complete its configuration. This step also checks syntax, checks for proper wiring, removes unused logic, optimizes logic, places the individual logic elements on the FPGA, and optimizes the timing. The last step of the design process is to use a simulation tool to verify the operation of the design. One simulates the expected inputs (widths and relative timings) and the simulation tool displays the output waveforms or intermediate internal signals. The ability to check the internal signals is crucial since one cannot check the operation with an oscilloscope as is traditionally done when using discrete modules. When one is

489

satisfied with the design it is loaded into the flash memory of the ULM52C via CAMAC commands (automated with the supplied Basic program). The ULM52C is then rebooted with a few CAMAC commands and the process is complete. Successfully making a new signal processing system with this CAMAC module and associated software is still dependent on the user’s ability to conceptualize a proper design that will meet the experimental design requirements. The software assists with implementation but does not create the design. Similarly, the user must understand how basic and advanced elements (be they specific NIM or CAMAC modules or virtual logic symbols in the software layout) can be combined to accomplish the conceptual design. Just as a C++ compiler cannot be used to invert a matrix without understanding the mathematics behind the process, one cannot expect this hardware and software to accomplish things the user does not know how to do. Fortunately, it is true that the software package does not require one to understand digital design details such as the number of loads that can be driven by a particular TTL chip or power requirements. Nor does it require one to understand the details of optimum wiring layout of the individual gates within the FPGA itself. In addition, the software does make the design process easier because of the large number of elements available. Unlike using hardware modules that often do not do exactly the needed function or which do not have the needed number of inputs or outputs, with the combined FPGA-software approach if one needs precisely three inputs and with one inverted input, it is possible to have exactly that element in the design. In order to be practical for other small laboratories, the effort needed to learn to use the software and to create a successful electronics design must be reasonable. In our opinion, the difficulty of learning the programming software is somewhere between modern wordprocessors (easier) and high-level programming languages such as IDL (harder). One should note that as the use of these devices becomes more common at larger laboratories, one can call on colleagues there (as we did) for advice and help. The supplier of the particular module described here was also very willing to help and review our early designs. Of course, the time to produce a working

ARTICLE IN PRESS 490

P.A. DeYoung, G.F. Peaslee / Nuclear Instruments and Methods in Physics Research A 551 (2005) 487–492 VCC

OBUF

OBUF

Place holder for future additional in hibits

OBUF

OPAD

OBUF

START CLOCK

IPAD

P164

IBUF

AND3

IPAD

P165

P179

OBUF

P176

OBUF

P148

IBUF

AND3

Digital Integrator(5) AND2

OBUF

Left Green LED

Right Yellow LED

OPAD

Right Red LED Right Green LED ADC Gate (11)

Q live scaler (12)

P147

OPAD

Q scaler (13)

P142 COREGen Module CLK Q[19:0]

IPAD

P185

OPAD

GATE

OPAD OBUF

Clock

OPAD

P178

Left Yellow LED

IBUF

IPAD

P162

OPAD

OBUF

ADCGATE

Busy (4)

OPAD

P175

Left Red LED

P174

cnst=190 4usgate

ADC Trigger(3)

OPAD

P180

Timer (10)

BUFG

CLOCK_DS

Q15

Q[19:0]

OBUF

OPAD

P150

Fig. 1. FPGA schematic design used when making Rutherford scattering measurements. It generates the required ADC gate and performs other logic functions including inhibiting gate generation while busy and inhibiting the integrator pulses for dead time measurements. The circuitry in the upper right simply sets front panel LEDs so the user can verify that the proper program has been loaded.

design is a strong function of the complexity of the design and one’s experience. The complex trigger for the MoNA detector [14] at the National Superconducting Cyclotron easily required over 1000 man-hours of time. Of course like most large programming efforts, these hours were applied over a long period as the design complexity was increased in incremental fashion. The design discussed below was developed at Hope College in matter of hours (by someone with experience).

4. Local use Signal processing and data acquisition at Hope College are done with CAMAC-based digitizers

and scalers. These are supplemented with some NIM electronics for amplifiers and constantfraction discriminators. Now that the ULM52C, a CAMAC module, is used no other logic modules (e.g. gate-and-delay units, multiplicity-logic units, or overlap gates) are needed. All of the logic functions have been programmed into the FPGA found in the ULM52C. Since most of our digitizers and scalers are based on the NIM logic standard, it was necessary to employ NIM-ECL converters for the input and output to the ULM52C because it is based on the ECL logic standard. If one were planning a completely new system, rather than just upgrading an existing system, the number of these converters used could be reduced.

ARTICLE IN PRESS P.A. DeYoung, G.F. Peaslee / Nuclear Instruments and Methods in Physics Research A 551 (2005) 487–492

5. Current designs A simple example program is shown in Fig. 1. Although very simple, this configuration replaces a number of discrete modules (e.g. a quad gate and delay generator and a downscaler in this case). This design uses only a small fraction of the available FPGA space but illustrates many of the features of the FPGA approach. For example, the particular gate (two input, three input, etc.) needed for the function is the element used, customized modules (ADCGATE and CLOCK_DS) that efficiently perform a very specific function are employed, and a schematic diagram is automatically produced. (More complex designs simply have more design elements, more wires, and can have multiple layers, or subroutines.) To date we have developed three FPGA programs for use at Hope College. Two are nearly identical and are designed to handle a single detector when doing either RBS or PIXE. The program shown in Fig. 1, generates a 4:0 ms gate for the ADC and performs a number of logic functions. These logic functions include the generation of a precise timer based on downscaling of the onboard clock (100 MHz) by a factor of 65 536, gating of the digital integrator output in order to measure the system deadtime, and routing signals to the LEDs that serve as status indicators. We use three of the LEDs to indicate the particular FPGA program which is running (upper right corner of Fig. 1). The remaining program provides the same functions but also interfaces with the automatic target position scanning ability of our microprobe station. The control system for this target position mechanism has the ability to automatically raster the target so that the beam sweeps over an area of the target. The FPGA program insures that data are not recorded while the target is in motion and also provides a signal to a scaler to coordinate the position of the detector with the data stream.

6. Conclusion In our opinion, the FPGA based ULM52C module is a tremendously powerful device that

491

greatly enhances the electronics capabilities of the small nuclear physics laboratory. While the initial cost of the module, software, and associated NIMECL converters of approximately $8000 is not insignificant, the cost of the modules that can be replaced by this system is potentially much greater given that it can replace a very large number of discrete devices. It is true that there is a learning curve involved in programming the FPGA correctly, but the authors did not find this time excessive. In fact, this type of design and testing task would be appropriate for most electronicssavvy students in a smaller laboratory. The time to develop new reliable designs is equivalent to that of hardware setups with the advantage that designs can be stored and reloaded. This module is certainly appropriate for laboratories that are expanding or updating. One significant benefit with this module is that electronics setup will not be limited by an insufficient number of some specific logic module. Additionally, this approach is much more flexible and allows for specific logic elements to be greatly customized. Finally, it provides for quick changes from one standard setup to another with a high degree of reliability.

Acknowledgements This work was funded through support from the National Science Foundation under grants MRI0319523 and RUI:PHY-0354920.

References [1] See hwww.xilinx.comi for company information. Specific information about the Spartan-II FPGA can be found using the pull-down menu at the bottom of the products and services page under the site listed above. [2] JTEC Instruments, 32 Thompson Rd., Rochester, NY 14623. Phone (585)334-7215. See hwww.jtec-instruments.comi for on-line information. [3] Volker Lindenstruth, Ivan Kisel, Nucl. Instr. and Meth. A 535 (2004) 48. [4] D. Bazin, J.A. Caggiano, B.M. Sherrill, J. Yurkon, A. Zeller, Nucl. Instr. and Meth. B 204 (2003) 629. [5] B.B. Back, (PHOBOS Collaboration), et al., Nucl. Instr. and Meth. A 499 (2003) 603. [6] R. Grzywacz, Nucl. Instr. and Meth. B 204 (2003) 649.

ARTICLE IN PRESS 492

P.A. DeYoung, G.F. Peaslee / Nuclear Instruments and Methods in Physics Research A 551 (2005) 487–492

[7] M. Ebine, M. Katagiri, A. Birumachi, M. Matsubayashi, K. Sakasai, S. Sato, Nucl. Instr. and Meth. A 529 (2004) 429. [8] F.V. Vevchanovski, B. Bebauer, E.I. Litvinenko, A.S. Nikiforov, V.I. Prikhodfo, Ch. Schulz, Th. Wilpert, Nucl. Instr. and Meth. A 529 (2004) 413. [9] K. Nurdan, T. Conka-Nurdan, H.J. Besch, B. Freisleben, N.A. Pavel, A.H. Walenta, Nucl. Instr. and Meth. A 510 (2003) 122.

[10] S.E. Boggs, W. Coburn, D.M. Smith, J.D. Bowen, P. Jean, J.M. Kregenow, R.P. Lin, P. von Ballmoos, New Astron. Rev. 48 (2004) 251. [11] J. Varela, Nucl. Instr. and Meth. A 527 (2004) 21. [12] T. Sugie, T. Ito, T. Ebisuzaki, Comp. Phys. Comm. 162 (2004) 37. [13] Select ISE Design Tools under Design Resources from the Products and Services page at hwww.xilinx.comi [14] T. Baumann, et al., Nucl. Instr. and Meth. A 543 (2005) 517.