The Automated Test Blog

I am hearing increasingly from customers and other vendors in test and measurement about the need for “Open Analysis”. The need is driven by the increasingly diverse set of applications and thus measurement requirements driven by, among other things, macro trends outlined in the book The Long Tail. As communication standards continue to proliferate, and the pace of change increases, users need tools that can adapt just as rapidly to their changing measurement needs. Instrumentation vendors such as Tektronix, for example, are increasingly offering options for users to plug in third party analysis tools, or export their data to analysis environments for custom processing. Tek’s OpenChoice is “a collection of software libraries, utilities, samples, industry-standard protocols and interfaces”. An example of OpenChoice software is SignalExpress Tektronix Edition, which is used to automate Tektronix instruments and bring data into and open, PC-based environment for further analysis. Tektronix AEs, third parties, or customers themselves can add in their own custom analysis to meet their specific application needs.

I’ve recently been using the term Instrumentation 2.0 to refer to the changes happening in the architecture and use of test instrumentation. Rick Nelson, Editor of Test and Measurement World, wrote about this trend in his recent editorial and blog, I’d like to clarify what this concept means in terms of architecture and usage, for the two are related, but can be independent.

The internal architecture of test instruments is changing. Rapidly improving capability of PC processors, bus technology, and FPGAs have changed the architecture of both standalone instruments as well as modular devices such as PXI. Standalone oscilloscopes, for example, used to use proprietary technology from the front-end conditioning, through digitization, memory, storage, and display. Instrumentation companies were once the pioneers of early display technology, like the DVBST CRT in the Tektronix 4014. Signal processing on these devices has often been done using custom Digital Signal Processors, or DSPs. The investment in commercial PC and consumer electronics technology has created a compelling alternative to these proprietary architectures, however. Increasingly, standalone instruments use commercial operating systems, displays, memory, and storage. Indeed, most high-performance modern instruments contain a standard PC motherboard, transfer data over a commercial bus such as PCI, and run the Windows operating system. Increasingly, I believe the digitization and signal processing capability inside standalone instruments will also migrate to commercial technologies. Commercial A/Ds, driven by applications such as direct conversion for cellular base stations, are now available up to several GS/s. And studies have show that for many applications general-purpose processors and FPGAs can outperform dedicated DSPs. Modern instruments, will, in fact, start to look a lot more like a PC, at least architecturally speaking. A typical standalone instrument will have a acquisition block connected over a commercial high speed bus, such as PCI Express, to a general purpose processor. This processor will both run Windows and drive the display, as well as perform the signal processing necessary for real time measurements. In the highest performance instruments, FPGAs will be used to augment the analysis capability of the processor.

The change is architecture, combined with evolving needs of design and test engineers, has led to a change in usage. Increasingly, users require flexibility in their measurement systems and desire the ability to define capability specific to their application. This trend towards ever increasing customization and the ability through technology to capture this demand, is well articulated by Chris Anderson in his blog and book, both titled The Long Tail. He focuses on the trend in media markets such as music and film, but examples of long tail demand abound in test and measurement as well. Wireless technologies, for example, continue to evolve at a fast pace, and devices often include more than one wireless link. The model of a “one-box” tester specific to a single wireless protocol, therefore, has become a dinosaur. Users need a common platform where they can deploy measurements for new standards as they are deployed, without replacing their hardware. The relationship between these two trends is that the ability of users to define their own measurement capability is enabled by the architectural changes noted above. Instruments built on a PC architecture that use commercial technolgy components can enable users to write their own algorithms in software for processing the acquired data. Even better if they can switch out the acquisition front-end when the requirements necessitate it.

Of course, modular, or PC-based, instrumentation has been at the forefront of this trend. By their nature, these devices build on PC busses and processors and separate the acquisition from the measurement analysis. And the rapid evolution of converters, buses, and PC processors, has increased the capability of this approach over recent years, so that in many cases it is equal to or greater than the alternative standalone instrument. Software tools such as LabVIEW have been used for many years to create customized analysis for measurement applications. Newer developments, such as LabVIEW FPGA actually target LabVIEW programs to FPGA components in a test system to create very powerful embedded processing that is still defined by the user to their needs.