The Automated Test Blog

Embedded.com recently published an article by an engineer at On Semiconductor about their use of PXI and LabVIEW for the characterization benches.  In the article, Ray Morgan describes the benefits of this system over previous approaches, saying, among other things that “the PXI platform set a new standard for semiconductor design validation, breaking many of the paradigms and constraints of previous testing methodologies”.

I found the article particularly inersting as it articulates some common challenges I’ve been seeing in semiconductor characterization that I believe PXI and LabVIEW are very well-suited to solve.  We are, in fact, seeing a major uptick in the use of these platforms in applications such as these.  Another technolgy that I believe will be important in these applications is the user-programmable FPGA.  I discussed this trend in a previous post on Protocol Aware ATE.

I am, by nature, and optimist.  So when confronted with the very difficult business conditions we now face, I can’t help but look for the opportunities.  After all, opportunities do often lie in the the most unexpected places.  In this case, I have been researching the phenomena that a lot of very good innovation comes in the face of great adversity.  I recently wrote a column for Electronic Design on this subject.  The gist of the article is that while R&D spending clearly decreases in a recession, good ideas do tend to rise to the top, and they face less competition, in these tough times.  In fact, there are quite a few examples of companies and products that have succeeded not just despite a recession, but because of it.  I also discuss some strategies that you and your company can employ in this time to maximize the opportunity.

Two weeks ago, Martin Rowe from Test and Measurement World posted to story Will LXI Really Grow 24 Percent on his blog.  I was glad to see Martin dig into the number to uncover some of the creative spin in the way it was positioned.  Martin is dead-on – this number can be misleading for at least three reasons:

1) A large number of LXI instruments already had Ethernet connectivity and then added LXI compliance;

2) Many LXI instruments also have other buses that can be used to connect them to a PC, such as GPIB and USB; and

3) Not all standalong instruments are connected to a PC at all.

Not that these issues are a problem for users – it makes sense to have standards around implementation of Ethernet instruments and it makes sense to give users a choice of connectivity options.  The only problem is that these numbers are positioned this way to influence public opinion; in this case, to imply that the use of LXI is greater then it actually is.  I don’t blame the LXI group for trying, but I do commend Martin Rowe for questioning the report.

As I stated in my last blog post, I’m planning to discuss one of three industry trends per blog entry over the next few weeks. My 3rd and final trend is:

Expansion of Wireless and Protocol-Aware Test
In addition to emerging technological advances, software-defined instrumentation has proved ideal for rapid-growth areas such as wireless and protocol-aware test. For example, consumer electronics devices including cell phones and automotive in-dash entertainment systems often integrate multiple communication protocols and standards such as GSM, GPS, and WLAN. Test engineers using traditional instruments rely on vendors to develop dedicated, stand-alone instruments to test each standard. With software-defined instruments, engineers and researchers can test multiple standards using common modular hardware components and implement emerging and custom wireless protocols and algorithms in their test systems regardless of the maturity of a new wireless standard.

For example, Dr. Umberto Spagnolini at the Polytechnic Institute of Milan is using LabVIEW to prototype algorithms for emerging standards such as WiMAX. Researchers such as Dr. Spagnolini can directly control system parameters, including channel coding, power, and modulation scheme, while adding fading and multipath interference to determine system immunity as a prototype of emerging WiMAX algorithms.

In the semiconductor industry, the demand for increasingly functionality and integration continues unabated.  As a result, semiconductor companies are heavily investing in complex systems on a chip (SoCs) and systems in a package (SiPs) technologies. It is often difficult to fully verify these devices using traditional ATE, which has led to an increased demand for so-called protocol aware test or the ability to test devices by emulating the real-world signals connected to them.

These increasing requirements for semiconductor test and the need to reduce total test costs have led industry organizations such as the Semiconductor Test Consortium (STC) and the newly founded Collaborative Alliance for Semiconductor Test (CAST) to investigate standards around open test architectures that support the integration of modular, software-defined instrumentation such as PXI into traditional semiconductor ATE. By using software-defined, FPGA-based instrumentation in these semiconductor test systems, engineers can achieve real-time responses with the standard pin electronics found in traditional ATE, lowering the total cost of test through better use-case coverage and improving the user’s ability to debug failures.

As I stated in my last blog post, I’m planning to discuss one of three industry trends per blog entry over the next few weeks. My 2nd trend is:

Trend#2: Increased Adoption of Parallel Processing Technologies

Multicore technology has become a standard feature in automated test systems and a necessity for today’s electronic devices that are processing unprecedented amounts of data. Software-defined instrumentation takes advantage of the latest multicore processors and high-speed bus technologies to generate, capture, analyze and process the gigabytes of data required to properly design and test electronic devices. Multicore architectures can present a challenge when used with traditional text-based programming environments that are not inherently parallel and require low-level programming techniques. However, test engineers quickly can realize the benefits of multicore technology through inherently higher level programming environments such as LabVIEW, which automatically distributes multithreaded applications across multiple computing cores for maximum performance and throughput.

Many test engineers I talk to are already experiencing the challenge of programming multicore in that, for the first time, they are not seeing an increase in test system performance when updating the PC in the system. In fact, due to the potentially slower clock rates of many multicore processors, their systems may actually run slower!

On the other hand, Alejandro Torres, senior manufacturing test engineer at Sanmina-SCI, provided an example of the potential business benefits attained by using programming tools tuned for multicore technology when he stated, “By leveraging the multicore technology in LabVIEW and the latest NI multicore PXI embedded controller, we were able to increase our test throughput by one additional workday per week. Best of all, we achieved this throughput increase by simply upgrading from a previous-generation PXI single-core embedded controller to the latest NI PXI multicore embedded controller with only minimal changes to our code.”

Another area of growth for software-defined instrumentation is the increase in system-level design tools for FPGAs. Many modular instruments now come equipped with FPGAs, including several released in the past year that offer the high-performance Xilinx Virtex-5 FPGA. These FPGA-based instruments provide test engineers with the ability to implement more complex digital signal processing at faster rates than ever before. Because software programs such as LabVIEW give test engineers the ability to program FPGAs without requiring knowledge of VHDL, the performance benefits of FPGAs are no longer limited to a subset of hardware engineers with extensive knowledge in digital design.

Next week, I’ll post on the third trend, the Expansion of Wireless and Protocol-Aware Test.

Yes, its 2009 and time again to make some predictions about the technologies and trends that I think will shape our industry this year.  Of course, making predictions for the rest of this year right now is a pretty risky proposition. But, one thing I know to be true is that in a tough economy, you have to be able to do more with fewer resources. Test and measurement often comes under particular scrutiny in an economic down cycle, and test engineers will need to be prepared to optimize our approach to verification and production test, or even look at alternatives to our existing test engineering strategies

These demands have led to three major trends that I believe will significantly influence the Test and Measurement industry over the next year. Instead of blogging them all here today, I will share one per entry over the next few weeks.

Trend#1: Growth of Software-Defined Instrumentation

The adoption of software-defined instrumentation is the most significant trend in test and measurement for 2009. Software-defined instruments, also known as virtual instruments, consist of modular hardware and user-defined software that give engineers the ability to combine standard and user-defined measurements with custom data processing using common hardware components. This flexibility has become critical as electronic devices such as next-generation navigation systems and smart phones integrate diverse capabilities and rapidly adopt new communication standards. Using software-defined instruments, engineers rapidly can reconfigure their test equipment by modifying software algorithms to meet changing test requirements.

In addition, engineers are using software-defined instrumentation to achieve new levels of measurement performance and lower test costs by applying the latest technological advancements such as multicore processors and field-programmable gate arrays (FPGAs) in their test systems to meet the demands of new application areas such as wireless and protocol-aware test.

Because of the flexibility and cost-effectiveness of this approach, thousands of companies are adopting software-defined instrumentation and industry standards that build on this approach continue to grow, even in the difficult world economy.  For example, according to the PXI Systems Alliance, more than 100,000 PXI systems will be deployed by the end of 2009, and the number of deployed PXI systems is expected to double in the next decade.

Jessy Cavazos, test and measurement industry manager at Frost & Sullivan, recently confirmed that PXI is influencing this trend when she stated, “The open, modular architecture of software-defined instruments such as those in PXI have proven beneficial to a wide range of industries, and, as a result, PXI revenue in measurement and automation is expected to grow at 17.6 percent CAGR through 2014. The performance delivered by the PXI platform has successfully addressed areas such as RF applications in radar testing, mobile phone testing and other wireless applications that were previously impossible to address with other instrumentation.”

Next week, I’ll post on the second trend, the increased adotoption of parallel technologies.

There is no doubt that we now are facing an economic headwind. Customers I’ve visited around the world are trying to understand how this headwind will affect their business and what they can do to put themselves in the best position to weather this storm, and perhaps even come out of this down cycle stronger than they are now. More and more test engineers are looking at how to optimize their approach to test more with less.

Testing more with less means that you may have to reevaluate your approach to test, and indeed, it is in disruptive conditions such as we now face that new ideas are the most prone to take hold. I believe that modular, software defined test systems provide the greatest opportunity for testing more with less – regardless of the dimension where optimization is needed. For example, optimizing the test speed of a production system is often the best path to decreasing test costs. For other applications, reconfiguring a single tester to test multiple devices yields the best results. And in very complex testers, the capital cost may the focus for cost reduction. For each of these situations, and in many others, the software-defined approach has proven time and time again to deliver dramatic improvements.

Let me share a few examples. In a production system, time is, quite literally, money. If you can reduce test time by ½, then you may be able to put half the number of testers at the end of a line. Wireless devices, in particular, are often expensive and time consuming to test. Many cellular phones are still tested with a inefficient method called “call-processing”. In this approach, a device called a “one-box tester” is used to actually simulate a phone call to the device under test. Bringing a phone up into a call is slow and is unnecessary to test if the device is correctly assembled. This would be like testing a television by watching a movie on it, instead of simply sending it a test pattern and verifying the result. More and more cellular handset manufacturers are moving to a software-defined test system that instead tests the physical layer signal of the device and uses signal processing in software to perform the necessary tests to verify each type of wireless standard. This technique is often 2-5 times faster than call-processing, which results in huge savings for the manufacturer.

Another example of achieving cost savings with a modular, software-defined approach is to use the flexibility of software to reconfigure a system to test many different types of devices. Another challenge in wireless test is that many devices have multiple wireless standards. My beloved iPhone, for example, now has 5 radios! Often, these different standards have required different instruments with their own vendor-defined measurement routines. A software-defined system can be reconfigured to test each standard with the same hardware. And when standards inevitably change and evolve, a software-defined system is in a much better position to be able to react to these changes.

A final example is in semiconductor ATE. Many semiconductor devices are tested on so-called “Big iron” testers. These testers have the sophisticated digital infrastructure and pin-electronics to test high performance semiconductors such as processors and SOCs. For simpler devices with low target prices, such as a MEMs sensor or an RFID, however, these testers may be overkill. Because they have the infrastructure to support high performance, it is difficult to scale them to these simpler requirements. A modular, open system such as PXI, though, has a very low entry cost and can be configured with only the minimum required capability, which results in lower capital expenditure.

So, in these turbulent economic times, it will be up to test engineers to innovate using the latest technology to meet the challenges of testing more with less. And those companies and individuals that do this the best will be able to come out of this difficult time stronger than before and ready to take advantage of new opportunities as conditions improve.

As recently discussed in a post by Rick Nelson of Test and Measurement World magazine, the Semiconductor Test Consortium (STC) has begun work on defining a Portable Test Instrument Module (PTIM) – a standard plug-in module for performing ancillary measurements on existing semiconductor ATE.  My colleague, Luke Schreier, delivered a presentation at the last STC global meeting that was very well-received which proposed PXI as a suitable specification to build from.   The business case is very compelling – traditional ATE architectures are built to accommodate the densest and highest speed test pins possible – 1 kilowatt of power per board is not uncommon.  This is necessary for the high speed digital electronics needed to test the latest processors, for instance.  When you need to add some audio or RF measurements into the system, however, the infrastructure can be overkill.  Moreover, to fully integrate an instrument into a tester requires expertise of the ATE vendor, so to make these measurements, the vendor may be required to invest significantly in development of measuement functionality already available on the open market, just in other form factors.

It is interesting to me that the semiconductor test industry is recognizing some of the features we designed into the original PXI specification.  In fact, one of the slides we used to use in the early days of PXI showed a set of rack and stack instruments on one side and a “big iron” semiconductor tester on the other, with PXI right in the middle.  The point was that PXI borrowedconcepts from both of these markets – the measurement quality from box instruments, and the card modular form factor and integrated timing and synchronization from semiconductor ATE.  It looks like after 10 years, its finally come full circle.

I was recently invited to give a talk at the VLSI Test Symposium titled “Migration of PXI Instruments into Semiconductor Test“. The session focused on emerging trends in semiconductor ATE and on work that is currently going on in both vendors consortia, to migrate PXI into semiconductor test applications. The presentation covered the key challenges currently facing engineers that are validating and testing increasingly complex devices such as SoCs and SiPs. As I previously blogged, Protocol Aware ATE is a new technique for testing these complex devices at a system level. In the presentation, I also covered existing work to build semiconductor ATE based on PXI, including examples of augmenting existing ATE, creating testers with a PXI measurement core, and work by the Semiconductor Test Consortium on a Portable Test Instrument Module, or PTIM. The PTIM initiative is designed to provide a way to add ancillary measurement capability to existing ATE platforms. The STC has been evaluating various options and has the desire to standardize PTIM on an exiting industry standard. PXI has been proposed as the PTIM platform and is currently being discussed at the Global STC Conference this week in San Diego. This proposal will enable ATE customers and vendors to leverage the large commercial investment in PXI and extend ATE capability by using the 1500 existing PXI modules currently available.

Last week, I was in China for a conference. The economy continues to boom there; its amazing to see how much has changed since my last trip just two years ago. The skyline in Shanghai has grown immeasurably, the piracy and knock off brands have been pushed ‘underground’ (though, its still pretty easy to get a fake Rolex), and the evidence of China’s growing middle class is apparent – its almost as easy to get a real Rolex now at the many upscale shops around town. But one of the most pronounced changes is the growing pollution in Shanghai and Beijing. I only flew through Beijing, but I couldn’t even see the terminal from our plane as we taxied in. As one of my colleagues noted, you could look right at the sun, which was only a soft glow behind the yellow smog. Shanghai wasn’t quite as bad, but as we drove into the city, we all noticed our sinuses clogging and a distinct itch in the back of our throats. Another colleague attempted to take a jog in the city, but gave up after a few blocks.

So that raises the question that everyone has been asking: How is Beijing going to host the world’s premier athletic competition in a mere few months? The Chinese have instituted a decade long plan to clean up Beijing’s air, but opinions vary on how well it has worked so far (and my experience last week would suggest that it hasn’t). There are also more drastic contingencies planned, such as shutting down factories, restricting automobiles, and even seeding clouds to force rain, in an effort to “cleanse the air”. Some athlete’s aren’t so confident – one of the world’s premier marathon runners today announced that he may pull out of the Olympic due to health concerns related to the pollution.

So what’s the long term solution to the pollution in China and elsewhere around the world? Probably not restricting traffic or seeding clouds. The problem will be ultimately solved through consumer demand and engineering innovation, or green engineering . Consumer demand is what is causing the huge uptick in green products and corporate sustainability plans. Sure, there is a lot of hype out there (now called “Greenwashing“), but the net effect of all the focus on environmental sensitivity is a positive one. Even more sustainable are the discoveries and innovations in the scientific and engineering community. China’s pollution, for example, is primarily the result of coal power plants (a new plant opens in China every week to 10 days) and automobiles. Both of these industries are seeing huge investments in research to fund cleaner, more sustainable alternatives. The global investments in renewable energy reached $100B in 2006 and its projected to grow to more than $750B in the next 10 years. Research abounds in more efficient solar, wind, and even wave energy production. There is also a large investment in researching and productizing alternative fuel and zero emissions vehicles.  We don’t yet know which of these innovations will prove most successful, but innovation, combined with a capitalistic system which rewards it, will ultimately triumph.  Its only a matter of time before these inventions are not just a novelty, but a economically sustainable business delivering zero emissions but also better performance and lower cost of ownership than their fossil fuel predecessors. It might not be in time for the Beijing Olympics, but hopefully it will be in time to clean the air for the next generation of China’s urban population.