The following article describes the “PIRANHA” data acquisition/analysis system design and build by Lockheed Martin Missile and Space (LMMS) at Sunnyvale, California, in 1997. The material is drawn from a technical paper which reviews the philosophy of large data based acquisition and analysis systems, entitled “New Digital Data Acquisition/Analysis System Technology for Large-Scale, Structural-Dynamic Testing Facilities” by Strether Smith, Steve Katz, Bill Hollowell and Eric Olson of Lockheed Martin Missiles and Space's Advanced Technology Center and Al Brower, Bob Franz and Scott Snyder of DSPCon.

The Piranha System The complete paper can be downloaded in Portable Document Format (PDF) for more in-depth review. (Visit www.adobe.com to download the viewer: Acrobat Reader.)

Strether Smith is a TTi instructor, whose Digital Data Acquisition, Digital Signal Processing, Data Acquisition and Analysis and Vibration and Shock Test Control Techniques courses are always very popular.

New Tools and “Off-the-shelf” Technologies

An important goal of any system development task is to provide an adequate system at a minimum cost. Components and tools of the highest level, consistent with the flexibility required, are usually the most “efficient”. Use of an available hardware subsystem is more cost effective than building one from lesser components. In software, using a mature development product with a well-evolved “toolbox” is far cheaper than developing the functions yourself. To this end, the Commercial-off-the-Shelf (COTS) market offers an ever-growing family of hardware and software products that reduce the need for a developer to construct his/her own capabilities. In fact, many of the components of a modern, large-scale, data acquisition system can be bought at your local computer store.

The most important new tools and technologies that help the system developer produce “better, faster, cheaper” systems are:

The Sigma-Delta ADC

The Sigma-Delta () acquisition concept is the most significant advance in digital, audio-frequency data acquisition technology in the past 20 years. These systems, originally developed for the commercial audio industry and found in all compact-disk recorders and players, offer superior accuracy and low signal distortion at a fraction of the price of conventional ADC systems. The concept of, though complex in implementation, is relatively simple in concept. The primary “secret” is that a sigma-delta system “over-samples” the data at a speed many times greater (typically 256) than the desired sample rate. The converter quantizes the signal with a simple, 1-bit comparator, and digitally filters and decimates the sampled signal to yield the desired resolution and sample rate.

This approach provides four fundamental advantages when compared to conventional ADC methodologies. First, it eliminates the need for expensive, analog anti-aliasing filters. While alias protection is still required, a simple one-pole RC network is adequate because of the 's high sample ratio (sample rate/desired bandwidth). Second, the low-pass filtering performed in the digital calculations yields characteristics that are far superior to analog filters. Aliasing errors are effectively eliminated and, when properly implemented, these devices provide 92 dB of useful signal dynamic range. Third, the filter characteristic provides essentially perfect pass-band characteristics with magnitude errors less that 0.1% and constant delay of all frequency components. Last, the devices are inexpensive and are implemented in an ADC-per-channel architecture, eliminating the problems associated with multiplexed acquisition systems.

High-Performance “Personal Computers”

Second only in importance to the converter for the systems under discussion is the (r)evolution of high-performance “personal computers”. These machines, currently led by machines based on the Intel Pentium family, are proving to be excellent number crunchers challenging, and in the system described below, exceeding, the performance of their conventional workstation brethren. They do this at a much-lower cost while offering the capability of running “main-stream” software programs for auxiliary tasks.

When combined with high-performance, “hot-box” systems based on VME or VXI architectures, “personal computers” offer a very efficient combination of ease-of-use, performance, and low cost to the large-scale system developer.

Data Stream Management Systems

Performance of VME and VXI based data-acquisition sub-systems is largely due to digital signal processing (DSP) devices. These are powerful, independent, computing devices that are specialized to perform arithmetic calculations and manage streams of data. For the systems under discussion, their most important function is to “hose” data streams between devices such as A/D converters and disks. The primary destination of the data stream is a high-bandwidth recording device such as a SCSI disk.

The streams can also be routed to other DSPs for further run-time processing and then to display systems via a network as shown above.

High-Speed networks

High-Level Programming Environments

From the cost standpoint, modern software tools may be the most important development. New graphical, object-oriented languages are proving to produce robust, high-performance systems at a fraction of the cost of the conventional (C or Fortran) programming approach.

PIRANHA

The Piranha system with 320 channels and a bandwidth of 45 kHz was developed for use in spacecraft mechanical tests, supporting large-scale sine/random vibration, high-level acoustic, and pyro-shock experiments in LMMS's new Commercial Spacecraft facility in Sunnyvale California. The system is based on Intel “PC” computers running National Instruments LabVIEW software tools.

The success of the Piranha has been demonstrated in the first few tests, where the cost savings are justifying its construction. This project has proved that high-performance systems can be developed using available “commercial off-the-shelf” components, and that such systems satisfy the needs of almost any experiment in the structural-dynamic environmental simulation areas.

Basic System Capabilities:

• Number of channels: 320 (Five 64-channel modules operating independently or in concert)
• Bandwidth: 45KHz
• Data Storage: Complete time history.
• Recording Duration: 10 minutes: All channels at full bandwidth
• Run Time Calculation/Display: Time History, Sine Response, 1/Nth Octave, Power Spectral Density (Display on any Workstation)
• Signal Conditioning: “Internal Electronic” and Voltage (Single-ended or Differential). Fully automated setup with run-time saturation detection and logging
• Data Post Processing:
  - Time History
  - Sine Response (Tracking Filtered Complex or Magnitude)
  - 1/Nth-Octave: Sound Pressure Level (SPL) - Power Spectral Density (PSD)
  - “Narrow-band” Power Spectral Density (PSD)
  - Shock Response Spectra
  - Transfer Function, Coherence

• Data Processing/Archive Duration: All processing and data backup for a 320-channel test is completed within 1/2 hour of test completion.

System Architecture

The PIRANHA system is made up of

• Four PC “Workstations”. One is designated as “master” and three are designated as “monitors” during data acquisition. At other times any of the workstations may be used for test definition and/or data reduction. One of the stations is equipped with a large disk-storage system and is designated the “disk farm”. In other respects, the workstations are identical.
  During data acquisition, the “Master Workstation” serves as the controller for the acquisition system. It acts as a “traffic cop”, queuing commands to the acquisition modules and continually verifies the status and general health of the system. The monitor stations can access and display data from any channel that is being acquired.
• Five 64-channel “Data Stations” that are configured as nodes on a high speed network. These modules acquire and store the data, provide data eavesdropping, and perform run-time calculations. They can operate independently as 64-channel systems, or in concert as a fully-integrated system. Additional data stations can be added to increase the channel count without affecting individual-channel performance. Three wires, that carry the network, calibration, and acquisition-clock signals, are the only connections between the data stations. They are connected to the workstations by the 100 Mb/second network.
• Peripheral hardware, such as the system master clock and printers.

Information and Data Flow

The following figure cartoons the information flow through the system.

The first step is test setup/definition that is performed by the Test Manager system. This software system includes a multiple-tier data base and a variety of utilities that allow the Test-Definition Database to be loaded manually or via externally-generated (Excel) files. All testing parameters, and much of the post-processing operations, are defined here.

At “run time" the parameters from the Test-Definition Database are downloaded to the Data Stations and diagnostics are performed. The operator controls the data-acquisition operations, such as acquisition start and stop, at the Master Workstation. The Data Stations acquire and store all of the data and “eavesdrop” blocks of data which are passed directly to network-accessible data buffers (time-history for all channels) or to DSP processors (1/3 octave, PSD ... for selected channels). The master and monitor stations can access any of the eavesdropped data sets. They display the DSP processor results from the preselected channels, or individual-channel data displayed as time histories, or locally-calculated spectra.

At the conclusion of the test, the data is transferred from the raw data files on the Data Stations to “CATS files” on the “Data Server” Workstation. From there, they are analyzed in 1/N-octave, narrow-band, sine, or shock-response spectra format by any of the workstations. Using a 200 MHz “Pentium Pro”, data extraction, analysis, and plotting of a 320-channel test is completed in less than 30 minutes.

Toll-free tel. 866-TTi-4Edu (866-884-4338)
E-mail:Training@ttiedu.com

© Technology Training, Inc. 2006

Return to Resources

060919 BJA