First published as feature article in the Microwave Journal March 2006 issue.

Synthetic Instruments: A New Horizon

Use of synthetic instruments and synthetic measurement systems to configure modular components to emulate the complete functions of traditional box instruments and test systems

From: Vol. 49, No. 3 | March 2006 | Page 22
by David P. Menzer, Auriga Measurement Systems LLC

Fig. 1 An SI generic network analyzer block diagram.

Modular test components have been with us for many years in many forms. Going back to the early ’70s I can remember HP BWO oscillator/sweeper mainframes with their plug-in modules to address the various extremely high frequencies above 1 GHz. While this modularity in test equipment exists today, a significant new aspect is now sweeping across the landscape — the ability to configure modular components to emulate the complete functions of traditional box instruments and in many cases that of complete test systems. The purpose of this article is to further explore what we mean when we talk of these synthetic instruments, where we might want to apply them and what some of the significant trade-offs are.

Before we can begin to explore what we can do, why we would do it and what the advantages are, we should start with some definitions of terms. Appropriate to the way we work today, an appropriate definition from Wikipedia reads like this: “It describes a functional mode or personality component of a synthetic measurement system that performs a specific synthesis or analysis function on a device under test (DUT) using specific software running on generic, non-specific physical hardware. Typically the generic hardware is a dual cascade of three subsystems: digital processing and control, A/D or D/A conversion (codec) and signal conditioning. One cascade is for stimulus, one for response. Sandwiched between them is the device under test (DUT) that is being measured.” How we achieve this synthetic instrument (SI) or synthetic measurement system (SMS) will be further explored here.

Today we can utilize components from three standards-based approaches, as well as non-standards-based approaches to achieve an SI. What is critical to a successful SI implementation is not the standard the hardware is based on, but the support available from the manufacturer and the design of the software architecture the system is implemented in.

Most of us are familiar with VXI,™ as this standard has been available for over 20 years. It has only been in the last few years, however, that significant numbers of microwave components have become available in this format. VXI has the advantage that a large installed base of components are available to draw on to build an SMS.

More recently a lot of interest has been focused on PXI,™ as it offers to be a lower cost, more convenient standard to work in and offers built-in convenience of working inside a PC. While initially focused on more traditional data acquisition tasks, vendors are increasingly offering higher frequency building blocks that users can now begin to use to assemble a viable SI in the microwave regime.

Most recently, with ratification in late September of last year, a new microwave oriented standard has become available — LXI.™ With a large group of the most prominent microwave instrumentation companies behind it, and building off of 1G Ethernet, LXI has the potential over time to replace much of the GPIB instruments we have all become familiar with. LXI also incorporates a number of triggering capabilities, including IEEE-1588 Precision Trigger Protocol (~200 ns jitter) and inter-unit communication features with performance characteristics exceeding those of the other two standards mentioned here.

Since an SI or SMS is not restricted by definition to any specific hardware standard, we may also construct a system using a mixture of the previously mentioned three standards, along with standard instrumentation or other hardware that does not conform to any specific “system” specification. In cases like this, the system integrator will of necessity take on the responsibility for any software required to construct a functional SI or SMS.

We will more fully explore the implications of the software task in building an SI or SMS, but to date be aware that none of the hardware systems standards, VXI, PXI or LXI, address software much above the driver level, and certainly not sufficiently high enough to insure a successful SMS implementation. This is still an open area of discussion and the responsibility of the system integrator.

Fig. 2 An SI vector network analyzer block diagram.

What Can We Implement with Synthetic Instruments

If we take a brief look at some of the elemental functions available in VXI, PXI or LXI, the question arises — what can we accomplish of significant value? Clearly the hardware vendors believe that the building blocks that are available will hold our attention longer than the intellectual satisfaction of making a simple power or voltage measurement. Let us now take a closer look at what can be accomplished.

Today, we can purchase a number of items: mass interconnects, switching and scanning units, power supplies, digital I/Os, counter/timers, A/D, D/A, signal conditioning and signal generation. Announced products include wideband up and down frequency converters as well as high performance wideband arbitrary waveform and signal generators.

To demonstrate the utility of the SI or SMS approach we can illustrate how several popular, powerful standard instruments can be implemented using this modular approach. Using a single RF source, a down converter, signal conditioning and an A/D module, we have all of the hardware we need to implement a generic spectrum analyzer, as shown in Figure 1.

To then build on this approach, let us look at one of the most fundamental, yet high performance instruments we commonly use, a vector network analyzer. We can construct this from a combination of signal generators, down converters, signal conditioning and A/D modules, with the application of a suitable amount of software. A representative block diagram is shown in Figure 2.

Fig. 3 An SI implementation of a pulsed VNA.

Using this same block diagram, but some additional signal conditioning at RF and IF, we can achieve a noise figure analyzer. As a final demonstration of the utility of this approach, if we add pulse generation and timing modules, we can convert our CW network analyzer to a highly capable pulsed network analyzer, as shown in Figure 3, or a pulsed power meter.

Trade-offs of SI vs. Traditional Instrumentation Approaches

Given the above examples of the flexibility and re-usability of the hardware modules needed to implement a broad spectrum of instruments, why would this not be our primary approach? As a wise man was heard to say, there is no free lunch, and that is true here as well. While we can achieve a number of different instrument/measurement capabilities in a compact format in SI, it comes at the cost of in general a much more significant software task for the end user or system integrator and generally a much longer time/measurement. As an example of the latter characteristic, which admittedly is easier to quantify, the above SI VNA is capable of generating a complete set of the four S-parameters at a given frequency, fully error corrected, in approximately 200 μs. That is in contrast to the capability of today’s standard, one-box VNAs of making the same measurement in ~200 μs. To quantify the former characteristic of a much greater software task, we can get a measure of that, in that at present there is only one commercially available SI VNA, and a large number of traditional VNAs from a number of suppliers.

Fig. 4 SI capabilities vs. traditional instrument approaches.

We have now seen an example of how, with minimal hardware changes, we can develop several standard instruments. We need to explore when SI or SMS are appropriate and what trade-offs are involved. To help understand the trade-offs, let’s examine why PXI (see Figure 4), VXI and now LXI instruments have been developed. VXI is the oldest standard, going back to the early 1990s and is based on earlier work on the European standard VME bus architecture. PXI was first standardized in 1997 with the intent to leverage off of the growing popularity of using PCs in the ATE environment. All three standards aim to provide a degree of modularity not available in traditional test instrumentation. Up to the last several years that was essentially the goal of PXI and VXI, and although there were several attempts at SI early on, they struggled due to the state of computer science and hardware. LXI is the first hardware and software standard developed with SI and SMS fully in mind. High speed communications, hardware and software triggering and software driver architectures are all incorporated into the standard, as well as a high degree of self and system discovery.

Table 1

In broad general terms, what we gain from SI or SMS is a much more modular system architecture and the ability to rapidly reconfigure the hardware to meet the task at hand. Due to this modular approach, we can, to a much higher degree, acquire “just enough test.” The test engineer configuring the test system can specify at a much lower level the required performance. As an example, if our application only requires 60 dB of dynamic range, we can choose a 12-bit digitizer as a system resource. As an example, essentially all commercial spectrum and network analyzers contain 14- to 18-bit digitizers, and so have more “test” than our supposed application. Additionally, if our requirements change or better hardware becomes available, we can upgrade just this 12-bit digitizer by only replacing that item. With proper system and software design, our application code will continue to run with the new digitizer with no additional changes. This flexibility allows our test system to more closely track the state of the art, as it is expected that manufacturers will be able to introduce modular components more quickly to the market as compared to the development time required for fully configured traditional “box” instruments.

What do we give up with SI? Unfortunately, as with most things, there is a price to pay. In the case of SI and SMS, the price is generally slower measurements and an increased software development task. As we showed earlier, the advantage of dedicated hardware, execution from firmware and a tight coupling between the hardware and firmware as found in traditional instruments results in higher measurement speeds. Also, since in many cases the calibration routines, measurement science and the integration between the various modular pieces is now the responsibility of the system integrator, much more software will have to be developed. Vendors are certainly working hard to minimize this impact, but the manufacturer of that digitizer cannot assume responsibility for all of the potential “instrument” applications. While they will surely develop applications that they may be familiar with, the balance will fall to the end customer to complete.

Fig. 5 SI VNA software interface panel.

SI- and SMS-based systems as configured today do not have a convenient hardware-based control or display panel as H-P’s MMS solutions offered. Here, we look to use the host PC display and controls as a convenient interface. Figure 5 shows just such a display panel developed to allow convenient operator interface to an SI-based VNA.

One of the side benefits we have noticed in our lab is that since so many of today’s instruments basically expect a mouse to be available to interact with the instrument GUI to access many functions not accessible from the traditional pushbutton panels, we often have had up to four rodents wandering about the workbench. Significant time has been involved in locating the correct rodent to control the specific instrument we are looking at. With an SI solution, the PC is the only device that requires a mouse; just think of the cost savings there in the elimination of up to three mice from every test system.

The acquisition cost of an SI or SMS system is an interesting question. On the basis of a single function, we will generally pay more for the synthetic solution at this time. As we add functions, SI becomes much more attractive in general. Cost comparisons are very dangerous in the abstract, so these comments, and Table 1, should be taken with a grain of salt. But the generality today is that without the manufacturing volume, and in the particular the case of VXI, the need for a mainframe adds significant cost unless it can be amortized across several functions. This situation may change in the future as the balance of manufacturing volumes changes between traditional instruments and the various SI approaches.

Device Modeling as a Demonstration of the Flexibility of SI

Fig. 6 Typical DC/RF and load-pull system block diagram.

Our experience with the capabilities of both traditional and SI approaches can be illustrated by a real world situation. Developing state-of-the-art device models for the PHEMT market entails some unique test requirements. To develop these models we require a VNA that is capable of short pulse operation, sometimes as short as several hundred nanoseconds. A typical test configuration is shown in Figure 6 with the short pulse VNA, SA, DC pulse modulation capability. Finally, to frost this cake, we layer on the requirement of large-signal load-pull capability to verify the resulting model.

The full range of measurements required here cannot be achieved by current commercial instrumentation. In particular, low duty cycle short pulse operation, to control thermal and trapping effects, leads to unacceptably long test times. A typical measurement suite may require approximately 200 bias point measurements to characterize a device. These measurements on commercial instrumentation would take weeks. By parsing the test system and implementing as an SMS system, we have driven this measurement time to approximately eight hours.

What Are the Markets for SI/SMS

So what markets are the SI and SMS approaches focused on? Clearly aerospace and defense are addressed by the stability and longevity of VXI and LXI. We can grant LXI these characteristics as it is based on well-established standards in both the physical characteristics, 19” rack mount and in its communications as they are based on Ethernet, which has been around now for 40 years. PXI, being based and somewhat more dependent on computer hardware, runs the risk of being obsoleted by that market place. Manufacturing is an area in question. Typically, manufacturing places a premium on speed of test. With the exception of areas where SI performance can exceed traditional solutions, SI is not expected to offer speed advantages due to the required software, communications and triggering overheads involved with SI vs. traditional solutions.

Fig. 7 A standard C-size VXI-based SI VNA/SA.

SI and SMS solutions should be more compact than competing traditional solutions, PXI and VXI have already demonstrated significant space savings and Figure 7 shows a standard C-size VXI-based SI VNA/SA demonstrating the reduction in space possible. This VXI mainframe, which is approximately 14 inches high, contains essentially all of the instrumentation to implement an SI VNA, SA, NFA and pulse power meter, which combined would require 25 inches of rack space, a 44 percent reduction.

Conclusion

The most important feature of SI and SMS will be the ability moving forward to update the capabilities of a test system without wholesale replacement of components. Modification of a front-end down converter can extend the frequency range of all instrument functions, while new, faster and higher bit A/D and D/A functions become available to extend our dynamic range. As more test systems become SI-based, we can expect both different groupings of functions as well as more granularity of function. Both of these developments will affect how we upgrade in the future to meet new needs.

A brief summary of the main advantages of SI as a function of market space includes:

• Aerospace and Defense
• Re-configurability supports long productive lifespan
• Solutions potentially more compact than traditional solutions
• Research and Development
• Peaked performance around specific performance characteristics
• Manufacturing
• Just enough test
• Potentially more compact
• Rapid re-configurability

We discussed some of the cost, capability, and space and performance comparisons of SI vs. traditional instruments. Each implementation must clearly consider all aspects of the various technologies before a final solution is chosen. Fortunately, the well-designed specifications for all of the approaches we have considered here will allow us to co-mingle and intermix solutions, with minimal impact on overall system performance.

David Menzer received his BS and MS degrees from Rensselaer Polytechnic Institute in 1975 and 1976, respectively. He is currently president and chief operating officer of Auriga Measurement Systems LLC. Prior to forming Auriga, he was director of business development for ACCO-USA. There, he oversaw the development and execution of sales programs, corporate and product marketing, and identified and developed new business opportunities. In support of international sales, he cultivated and supported three sales representative organizations in Europe and Asia. During his tenure, two major products were introduced, and over 10 major proposals were written, resulting in three contracts and five proposals in negotiation. A number of brochures and data sheets were also published under his direction.