Thursday, March 28, 2024
Login

Software-Designed Instrument Revolution

Parent Category: 2014 HFE

By Bill Driver and Vimal M. Fernandez

The ability to test today’s emerging products before bringing them to market is not scaling proportionally to product complexity. Each subsequent generation of technology includes more pieces of sophisticated technology and the cost of using traditional instrumentation to test these devices is only increasing. 

One way to minimize hardware costs and reduce test time has always been to use virtual instruments and modular I/O. But a new approach—software-designed instrumentation—is giving engineers the ability to achieve test time reductions that are orders of magnitude beyond what was previously possible. 

Introduction to Software-Designed Instrumentation

For years, test engineers have used software such as LabVIEW—instead of the fixed software built into traditional box instruments—to customize measurement systems and reduce cost. While this approach provides flexibility and takes advantage of the latest PC and CPU technologies, the CPU is still a bottleneck in many demanding test applications. 

CPUs inherently limit parallelism, and typical software stacks result in latencies that reduce test system performance in cases where measurements need to be adjusted dynamically based on values or device under test (DUT) state. To dramatically reduce test times, you need to combine custom digital logic with multicore CPU technology to give your test system a balance of low latency and high throughput. 

While off-the-shelf instrumentation hardware traditionally has fixed capability, NI is leading the way with more open, flexible measurement devices based on FPGA technology. FPGAs are high-density digital chips that you can customize to directly incorporate custom signal processing and control algorithms into measurement hardware. The result is off-the-shelf hardware that has the best of both worlds—fixed, high-quality measurement technology, the latest digital bus integration, and user-customizable logic that is highly parallel, provides low latency, and is tied directly to I/O for inline processing and tight control loops. 

Extend Your Knowledge to Hardware Customization

FPGAs continue to gain design wins and market share from application-specific standard products (ASSPs) and application-specific integrated circuits (ASICs) because they keep up with Moore’s Law better than other devices and dramatically lower development costs, resulting in smaller test system size and lower power consumption. Very capable FPGAs are entering the market and defining the hardware capabilities of many devices, but the IP they contain is vendor defined and the FPGA’s power may not be accessible to you. This is largely because the specialized hardware description language (HDL) knowledge needed to program these devices requires a steep learning curve and is generally restricted to digital design experts.

System design software such as LabVIEW make the latest FPGA technology accessible to a range of engineers and scientists. Using graphical programming, you can implement logic to define the behavior of an instrument in hardware and reprogram the instrument when requirements change. The graphical dataflow nature of LabVIEW is well suited for implementing and visualizing the type of parallel operations that can be implemented in digital hardware. 

Software-Designed Instrumentation vs. Traditional Instruments

User-programmable FPGAs in your measurement system hardware provide three key benefits ranging from decreased test time to CPU load reduction. The following sections describe various usage scenarios in more detail.

Decrease Test Time and Increase Confidence with In-Hardware Measurements

Today’s instruments can perform a limited number of measurements in parallel, but software-designed instruments are limited only by the available FPGA logic. You can process dozens of measurements or data channels with true hardware parallelism, removing the need to choose between measurements of interest. With software-designed instruments, functionality such as real-time spectral masking is achieved with dramatically higher performance and at a fraction of the cost compared to traditional box instruments.

Focus on Data of Interest with Custom-Defined Triggers

Traditional instrument options for low-latency trigger behavior are fixed according to the hardware being used, but with software-designed instruments you can incorporate custom triggering functionality into your device to quickly zero in on situations of interest. Flexible hardware-based triggering means that you can implement custom spectral masks or other complex conditions as criteria for either capturing important measurement data or activating additional instrumentation equipment.

Reduce CPU Processing Loads with Real-Time Onboard Signal Processing

Processing large amounts of data can tax even the most capable commercial CPUs, resulting in systems with multiple processors or extended test times. With software-designed instruments you can preprocess data in the hardware, potentially reducing the CPU load significantly. Computations such as fast Fourier transforms (FFTs), filtering, digital downconversion, and channelization are implemented in hardware, reducing the amount of data passed to and processed by the CPU.

Next-Generation Software-Designed Instruments

The trend of more instruments enabling user programmability of the FPGA will change how long-term test strategies are approached for future programs. The benefits not only reduce capital costs due to eliminating redundant functionality, but also minimize the maintenance cost associated with code modifications. This reduction can also be maximized when software is properly architected in a manner that abstracts the core code blocks from those that will be known to need modification throughout the product life cycle. 

Custom Designed Test Instrument Personalities – Spectrum Monitoring

For spectrum monitoring and electronic warfare applications, users often need to look at large portions of spectrum and capture a broad range of signals: from transient, elusive signals to continuous wideband interferers. Real-time spectrum analysis is a prime example of an application that requires significant real-time – and often custom – signal processing. Approaching this application with software-designed instrumentation allows simultaneous processing of large amounts of data in hardware, and adds custom features like real-time frequency mask triggering. 

A Real Time Spectrum Analyzer (RTSA) is an evolution of the modern FFT-based spectrum analyzer that computes spectrum measurements in real time. Swept-tuned spectrum analyzers (SA) calculate power levels for a single frequency point at a time, which makes it easy to see how an SA could miss signals that change faster than its sweep rate. VSAs, on the other hand, acquire the full real-time bandwidth in each acquisition. However, VSAs traditionally require that the data be transferred to the processor before the next acquisition can begin. A VSA could miss signals that occur during this transfer time. Triggering could help in both of these situations, but that requires detailed knowledge of the signal power and frequency – which isn’t necessarily obvious with dynamic interferers. With RTSAs, the user can acquire the full real-time bandwidth and concurrently process the acquisition by performing overlapping FFTs in the hardware. This results in a gapless conversion to the frequency domain that can be exhibited using a number of common RTSA displays, such as a persistence plot. 

1412 HFE software 01

Figure 1 • A comparison between the appearance of a gapless RTSA persistence plot and a VSA spectrum plot. 

FPGAs are the backbone of a user-programmable RTSA due to their inherent ability to deterministically process data at high rates. For example, using the Xilinx Kintex 7 FPGA on the high performance NI PXIe-5668R VSA, one can capture more than 750 MHz real-time bandwidth of 16-bit IQ data at 1 GS/s – an effective data rate of 4 GB/s. In order to process the data in real time – a neighboring FPGA co-processor is used to process the data.  In this case, the 4 GB/s of data must be bit-packed into 12-bit IQ data to ensure the maximum data rate is less than the 3 GB/s, which can be continuously transferred through the backplane of a PXI chassis. 

The IQ data has now been acquired by the VSA, processed, bit-packed, and then streamed over the PXI bus to a neighboring NI PXIe-7976R FPGA module. The FPGA module then converts the received raw IQ data to the frequency domain via a 50% overlapped windowed FFT. At 3 GB/s, the NI PXIe-7976R performs an astonishing 2 million overlapped and windowed FFTs per second that can capture signals as brief as 1.5 µs. 

The spectrum can then be displayed using common RTSA displays, such as a persistence plot. The persistence plot overlays 10s of thousands of FFT traces onto a 2 dimensional display, which is represented as an intensity plot as seen in Figure 3.  Further customization of this display is achieved through three additional controls: (1) the persistence, or how quickly signals fade from the display, (2) the emphasis, or the ability to highlight less frequent events, and (3) the saturation, or the ability to compress the color mapping for better contrast. Additionally, trace statistics (max/min hold and averaging) and real-time frequency mask triggering are added to the persistence plot. 

The real-time frequency mask triggering allows the user to draw a power level boundary on the display and then capture the signal that penetrates this boundary. The captured signal can then be sent to the host computer for analysis or streamed to a redundant array of inexpensive disks (RAID) for later use. In addition to creating a persistence plot, the FPGA concurrently calculates time domain, frequency domain, and waterfall plots. These plots retain generic VSA features like the ability to change the reference level, center frequency, RF attenuation, or span. 

1412 HFE software 02

Figure 2 • More than 750 MHz of real-time bandwidth acquired by the NI PXIe-5668R VSA streamed at 3 GB/s to a neighboring NI PXIe-7976R FPGA Module for co-processing to create an RTSA. The 26.5 GHz NI PXIe-5668R VSA contains a user-programmable FPGA.

Performing customizable high-speed gap-free signal processing is only possible with the integration of the user-programmable FPGA in RF instrumentation. These features are created in hardware on the FPGA with the LabVIEW design environment, where the user decides what tradeoffs to make. RTSA is only one of many custom FPGA personalities that can be created, and the user-programmed FPGA produces a new class of instrument, one where the user is actually in control.

Future-Proofing

Vendor-defined instruments and fixed-capability off-the-shelf instruments will remain available for years to come, but increasingly complex devices and time-to-market pressure will lead to the rise of software-based instrumentation systems. The continuation of device complexity and time pressure means that software-designed instruments will play an increasingly important role in test instrumentation—starting right now. 

Software-designed instrumentation provides the highest level of flexibility, performance, and future-proofing currently possible with off-the-shelf hardware. As your system requirements change, software-designed instruments will preserve your software investment across different pieces of modular I/O and also ensure your existing I/O can be modified according to the application at hand.

About the Authors:

Bill Driver has over 13 years of experience in test and measurement, including sales and product management roles at Tektronix and LeCroy (now Teledyne LeCroy).  He is currently a senior product manager for National Instruments in Austin TX, and focuses on automated test applications. He graduated with a BSEE from Virginia Polytechnic Institute and State University as well as a MBA from the University of New Hampshire.   

Vimal M. Fernandez has more than three years of experience in test and measurement. He started his career at National Instruments as an Applications Engineer, then transitioned into R&D as an RF Test Engineer. Vimal is currently a Product Marketing Engineer with a focus on RF and Microwave Test.  He graduated with a BSEE from McGill University in Montreal, Quebec.  

 

Search

CLICKABLES