Addressing the challenges of tighter design margins in today’s electronic innovations
An Interview with Duane Lowenstein and Scott Leithem
Tighter design margins are a significant challenge in modern electronics manufacturing. As devices become smaller, faster, and more energy-efficient, the tolerances for errors shrink dramatically. This problem will only increase with today’s market pressures and required pace of innovation. To explore what engineers can do to strike the right balance of costs, yields, and time-to-market, I sat down with test process expert Duane Lowenstein, Keysight Solutions Fellow, and Scott Leithem, Keysight Solutions Applications Engineer, to discuss what they recommend to Keysight customers who are at the forefront of innovation and facing these very challenges.
Sandra: Hi Duane and Scott. Thanks for joining me today. Can you start by telling us a bit about your role at Keysight and your work with our customers?
Duane: Thanks for having me, Sandra. I have been with the company now over 25 years in various roles, including as new product introduction manager, where I helped one internal organization achieve a 30% reduction in manufacturing production ramp time while increasing overall production capacity; another where I reduced customer returns by 65% and improved overall profitability by 35%.
I currently work with Keysight customers to deliver Manufacturing Process Consulting (MPC), Test process Consulting (TPA), Instrument Migration Planning Services (IMPS), Cost of Test solutions (COT), and other services to help them reduce cost of test and test time and to increase yields.
Scott: Happy to be here as well, Sandra. I am an application engineer focused on planning and designing services for Keysight solutions, with an additional focus on aerospace and defense applications. I have been working for Keysight for 12 years, with a focus on calibration, understanding and calculating measurement uncertainties, ensuring the accuracy of measurements, and developing our competitive differentiators.
Sandra: Wow, that’s impressive and I’m grateful for your time and wealth of knowledge. So, tell us what you each see as the primary challenges associated with tighter design margins?
Duane: As you mentioned in your introduction, tighter design margins are a challenge of modern electronic manufacturing. The challenge isn’t new, but it is getting more severe as devices become smaller, faster, and more energy-efficient. It has pushed designers and their simulation tools to the edge, and it has kept test and test strategies one step behind the growing measurement needs.
Scott: For example, as the world becomes increasingly reliant on wireless connectivity, the need for faster and more reliable communication networks has never been greater. Researchers expect sub-THz (sub-terahertz) frequencies to play a significant role in future wireless communication networks such as 6G, thanks to their ability to provide the tens of GHz of bandwidth necessary for the terabit-per-second data rates required for some emerging applications.
Testing devices at sub-THz frequencies presents many challenges, from degraded signal-to-noise ratio to frequency response and mismatch that vary significantly over the band of interest. This highlights the need for testing solutions that can accurately characterize devices with minimal uncertainty.
The result has frustrated design engineers and delayed time to market when product designs go to validation and are failing testing due to false passes and failures during the design phase.
Sandra: It seems that wanting smaller, faster, cheaper has always been the case. Why is it such a problem now?
Duane: You’re right in that change has been the number one constant in electronic innovation. Although the objectives you mention have always been the same, the obstacles have been changing. We have overcome many of them with improvements in component quality, process controls, and implementing better DFx (design for manufacturing, quality, test assembly, etc.). But today, we are facing the biggest obstacle of all, margin stack up.
Sandra: Margin stack-up. Can you explain what that is?
Duane: The simplest way of visualizing margin stack up is to think of Matryoshka dolls, also known as Russian nesting dolls. These are the curved, colorfully painted wooden shapes of dolls that, in decreasing sizes, are placed one inside another. It is not unusual to see them having up to 10 – 12 copies of the same doll, but different sizes.
So, what do dolls have to do with margin stack up? Each of the dolls have intricate curved shapes that must be copied for each doll but, precisely smaller than the next bigger one. If the dolls are copied with too tight of a tolerance, the dolls will either not be able to be stacked or if stacked they will be hard to get apart and will ruin the painted decor on the wood. On the other hand, if the tolerance is too big, they will easily be stacked but when moved, they will rattle and ruin the painted decor on the wood because of the excess space.
In the manufacturing of electronics, it is all about understanding the continual stack up of tolerances from chip to component, to micro-circuit, to circuit card, to subassembly, to final product. If the tolerances are too tight, the product will possibly fail, if they are too loose, they may not meet the desired customer specifications.
Sandra: That’s a really helpful visual. In your experience, is margin stack up difficult to address?
Duane: Yes, the difficulty lies in how product design and development is typically done today. The ability to design faster, smaller, lighter, less power, less noise, etc., products with simulation tools is getting more powerful, and faster. Yet the design, build, programming, and qualification of a test station to address the more complex measurements and tolerances require time. This would typically not be a problem except that, in reality, the design of test only starts at the breadboarding of the product, well into design, making test the bottleneck. At that point, making changes to take what is theoretically possible and mass produce it in a repeatable, predictable manner become more difficult.
Scott: An example comes up in the satellite industry, where smaller startups and traditional companies alike see the potential for new applications leveraging low earth orbit (LEO) networks and larger constellations of satellites with cheaper, faster launches. It is a merging of the wireless industry with the space industry in terms of the production requirements (high volumes, faster test times, more room for error), versus the traditional satellite companies who have spent millions of dollars on a single satellite for weather or GPS applications.
In that case, the satellite has to work because they basically have put all of their money on a single satellite and launch, and it is difficult to impossible to fix major issues once the satellite is in space.
On the other hand, the low earth orbit satellite constellations require 100’s to 1000’s of satellites to be produced and launched. The requirements for everything to work perfectly are somewhat relaxed, as it is cheaper to launch a new working satellite than to fix a broken one and the cost to do so is much reduced from when single satellites were produced and launched.
Tighter tolerances are very important as this is still about launching satellites into space and controlling precisely where they are going to go and how they are going to communicate within their networks. So, you have to worry about the rates of no trouble founds (false failures) and escapes (false passes) when taking a LEO satellite constellation business into production and manage those tight tolerances well.
Sandra: What are then your recommendations to address the challenges of tighter design margins?
Duane: There are several areas that I advise our customers, and the more comprehensive their approach to incorporate these, the better the results:
-
Enhance Collaboration Between Design and Manufacturing: Bridging the gap between design and production is crucial. Early collaboration ensures realistic design expectations and helps align manufacturing capabilities with design goals. Designers and manufacturers must jointly address tolerance stack-ups and measurement uncertainties to minimize errors.
-
Adopt Advanced Test Strategies: Testing is the bridge between design and manufacturing. Robust test methodologies help predict, identify, and mitigate margin-related issues. My recommendations include:
- Design for Testability (DFT): Embed features that simplify testing during the design phase, making it easier to detect and isolate defects.
- Pre-screening Parts: Tighten pre-environmental stress screening (ESS) test specifications to catch marginal components early, reducing post-ESS failures.
- Use of Guard Bands: Establish tighter internal test limits to account for uncertainties in environmental conditions, equipment drift, and variability.
-
Prioritize Measurement Uncertainty Management: Understanding and controlling measurement uncertainty is critical. This involves evaluating factors like:
- Instrument accuracy and drift.
- Environmental effects such as temperature and humidity.
- Calibration methods and intervals.
Tools like the Test Uncertainty Ratio (TUR) can help normalize system accuracy against test limits. A higher TUR, such as 4:1, ensures more reliable results, reducing false passes and failures. Monte Carlo simulations are also useful for understanding how variations impact yield and quality.
- Implement Environmental Stress Screening (ESS): Environmental testing, such as temperature cycling or vibration, reveals weaknesses in materials or designs. However, pre- and post-ESS test strategies must account for material variability over temperature ranges to improve yields and reduce delays.
- Utilize Statistical Methods for Margin Analysis: Statistical approaches, such as Monte Carlo simulations and Guide to the Expression of Uncertainty in Measurement (GUM), enable teams to analyze and quantify the impacts of design and test uncertainties. This data-driven approach helps refine test limits and ensure optimal performance.
Scott: In addition, regular instrument calibration is key: Routine calibration of test systems and instruments minimizes uncertainties and ensures consistent performance. Establishing a metrology-driven approach to calibrate instruments under varying conditions can significantly reduce errors.
We see increasing requirements for standardized practices around regular calibration of test equipment, specifically in the form of accreditation. This is driven by the required accuracy and consistency of results that drive certain applications: for example, safety and reliability is paramount in the automotive industry. Accurate testing in aerospace defense applications (as we’ve spoken about before) is critical to ensure the safety of people around the world. And finally, as connected devices become more commonplace in the medical industry, the leaders in that industry are asking important questions about reliability, repeatability and accurate measurements during the testing processes. It is an interesting difference from the Internet of Things (IoT) industry in general, as the IoT industry is mostly about low cost and low power, and trying to get as many devices produced as quickly as possible.
The fact is, for these applications and their supply chains, having accredited certification is a requirement. It is important to look beyond the required accreditation stamp on the calibration report. We need to focus on the reason that these applications require accredited calibration (consequences of an uncomprehensive calibration) and to check the scope of accreditation for the calibration provider to ensure that the scope covers that of the measurements used to perform the calibration.
It is also critical to check that the verified parameters actually match the ones that matter for their application and that the provider is using equipment with a sufficiently low measurement uncertainty to be able to properly verify the specifications.
The way I look at it is this: accreditation gets us to the table, but whether the service provider’s accreditation is actually relevant to the customer and whether the calibration is comprehensive enough to provide confidence in measurement accuracy is what should be considered when selecting a provider. Keysight strives to ensure that the scope of accreditation always covers the specifications in each instrument where we offer accredited calibration and tests every specification, every time.
Sandra: This is incredibly useful information. Do your customers find it surprising that regular instrument calibration features so prominently in the strategy to address tighter design margins?
Duane: Yes. Many customers believe that if they purchase high-end instruments from a leading-edge vendor like Keysight they wouldn’t, or shouldn’t, have to worry about ongoing calibration. But quite the opposite is true. The more precision your instrument delivers, the more you have to make sure it stays true to specifications. And it is critical that the calibration be performed by a reputable vendor, with the right equipment, the right processes, and who has been certified by a third party to perform the specific calibration required for a customers’ application.
Sandra: Thank you, Duane and Scott, for your excellent insights. There are clearly many strategies that must complement the design stage in order to bring leading-edge designs successfully to market.
I invite our readers to find out more about what makes for a good calibration by visiting our calibration services page and reviewing resources such as How to Select a Calibration Vendor.
And I encourage you to check out KeysightCare to make sure your instruments are regularly scheduled for the most thorough calibration that is the best fit for your application, to ensure your high-precision instruments continue to work like-new, every time. KeysightCare Enhanced includes a calibration service of choice based on the equipment’s recommended calibration interval to an accredited calibration. KeysightCare Enhanced offers full protection for your innovation investment including prioritized technical support, repair, and calibration coverage. Are you ready to discuss your calibration options?
About Duane Lowenstein:
Currently Duane engages with Keysight’s customers to deliver Manufacturing Process Consulting (MPC), Test Process Consulting (TPA), R&D Work Flow Analysis (WPA) Asset Management Programs, Instrument Migration Planning Services (IMPS), Cost of Test solutions (COT) and, Design and Test simulation Software (DaT). His experience has spanned many industries including computer, wireless, automotive, aerospace and defense, as well as commercial electronics. His focus on measurable results has led customers to dramatic reductions in the cost of test, test time, reduction in work- in-progress, shorter development cycles and increasing yields.
Duane has published over 18 papers covering a wide variety of test and measurement subjects focused on the electronic industry. These papers include operational, financial, managerial, product development, production, support and life cycle management. He has also lectured around the world on many of these topics.
In previous roles at Hewlett Packard/Agilent, Duane was responsible for new product introductions for the Medical Products Group (MPG). There he achieved a 30% reduction in manufacturing production ramp time and increased overall production capacity. The outsourcing strategy that MPG (now part of Phillips) currently uses was part of the strategy defined and implemented by Duane. During his time in the Stanford Park Division the focus of his tenure was the increasing on-time delivery by 30% and reduced customer returns by 65% and improving overall profitability by 35%.
Before joining HP Duane was an officer in the US Navy where he was one of the lead engineers in the design of the USS Seawolf class submarine. He also was a key player in the Naval Shipyard Six Sigma and Lean initiatives.
Duane’s has a Bachelor of Science in Engineering from Rutgers University and a MBA from The Pennsylvania State University. Duane is an adjunct professor at Merrimack College and a retired Commander from the US Navy.
About Scott Leithem:
Scott is an application engineer focused on planning and designing services for Keysight solutions, with an additional focus on aerospace and defense applications. He has been working for Keysight for 12 years, where he has spent the last 8 years focused on calibration, understanding and calculating measurement uncertainties, and ensuring the accuracy of measurements. Before that, Scott worked as a support engineer and product manager, specializing in signal sources and signal analyzers. He has a Master of Science in electrical engineering from the University of Illinois.