Is Going Soft A Good Thing? How About Full-Auto?

Is Going Soft A Good Thing? How About Full-Auto?

Compressed design cycles and increasingly complex devices are making software and automation techniques an integral part of an engineer's toolkit.

Download this article in .PDF format
This file type includes high resolution graphics and schematics when applicable.

A common theme in every sector of the RF industry is the rapid acceleration of product cycles and technological advances. This trend could stem from the inclusion of RF technology in many widely adopted consumer products. Or it could result from a few technology breakthroughs that have expedited the manufacturing process. Either way, design cycles for custom devices are forced to match this velocity and deliver increasingly complex and multi-domain designs. Fortunately, software tools and automation techniques are helping engineers keep up with these rapid and highly integrated deliverables (Fig. 1).

1. Gold plating thin-film filter components and the use of metallic materials, such as brass, increase the performance and reliability of RF/microwave components, which are susceptible to corrosion conditions and thermal stress. (Courtesy of Mercury Systems)

In past years, the interaction between custom RF/microwave multi-function device manufacturers and their customers involved a predefined specification and an outline drawing. The specifications were relatively straightforward. In addition, they had several months to several years to deliver a final product. These custom multifunction modules (CMMs) lacked the digital, analog, and power components present in modern-day electronics.

With the increased focus on high levels of integration and multi-domain functionality, a more hands-on approach was needed. According to Anthony Pospishil, general manager for Mercury Systems’ West Caldwell Advanced Micro Center, “When we work with a customer to develop their architecture, we break down their requirements into block diagrams and then into components and elements. Next, the system is simulated from end to end according to the performance necessary to meet their requirements. We develop a specification using statistical models to see what is really producible and feed that back to the customer. The specification doesn't always match their needs. And sometimes, requirements and performance criteria shift and there is a back and forth with the specifications. “

A much more detailed approach to early design stage has been enabled by software that can model high-level to low-level capability. Previous spreadsheet-based methods functioned on cascade analysis, which was done with simple mathematical equations and the generic simulation of component operations. Yet this approach failed to account for many linear and nonlinear design aspects, as complex mathematical models would be needed to predict the performance of a more complex system. “Software is essential to the development of custom integrated assemblies,” said Douglas Okamoto, senior electrical engineer for Microsemi. “It is important to understand the interactions and bottlenecks within the integrated assembly to optimize performance parameters. This need has always existed. But over the years, we see an evolution of software that enables designers to work more quickly and efficiently.”

Today’s system-design tools can now include advanced behavioral models, which target portions of the device that are not yet fully defined as well as inter-stage interactions. These aspects are mixed with more detailed parametric models of previous designs. Statistical models that more accurately resemble real systems also can be integrated into this early process, allowing potential failure modes and design challenges to be discovered earlier.

Today’s advanced software design suites are taking a large leap by including linear and nonlinear modeling and simulation. “Designers must understand the capabilities and limitations of software CAD tools,” noted Okamoto. “High accuracy is often at odds with development time, so it is imperative to know how large of a problem the software can handle to yield results in an acceptable amount of time. Although CAD tools and computing power have both improved over the years, designers must still verify the accuracy of simulations.”

Electromagnetic (EM) simulators are often used to delve into the interactions among the various components in a CMM. These solvers can be used to predict coupling, interference, and even electrothermal effects for a complete CMM--including the enclosures. Certain components of a CMM--mainly filters, mixers, and oscillators--are heavily influenced by parasitics and enclosure electrodynamics. Additionally, the interference of analog, digital, power, and RF signals should be simulated and considered. After all, these mixed-domain interactions can cause failure modes that are very difficult to identify later in the design process.

Many simulation tools can be used to create parasitic extractions and coupling models, which enable the appropriate shielding, filtering, or isolation to be applied. Some simulation suites are even capable of including EM, thermal, and structural analysis in one package. Rigorous standards, such as military and aerospace, can then be met without many design and manufacturing iterations.

For complex CMM designs, another innovation has quickly become common practice: the use of a controlled design process involving representatives of each design aspect as the project progresses. These design reviews often involve parties from RF, analog, digital, power, test, manufacturing, and quality control. This approach ensures that each disciplines’ needs are met early in the process. As design cycles have shrunk, such a capability has of course become more critical. Redesigns to account for oversights or inadequate preparation are becoming too costly, and company internal design guidelines are heavily stressed.

Design for automated manufacturing is another critical time-reducing step in today’s design process. It also requires additional considerations in the early design stages. Thankfully, automated manufacturing steps, such as wire bonding, pick-and-place, and quality-control testing, are growing increasingly reliable. They are helping to produce small to high volumes of CMMs can be consistently produced to meet high-end standards with less wastage (Fig. 2).

2. Automated pick-and-place machines help to ensure consistent and reliable assembly modules in a fast and cost-efficient assembly process. (Courtesy of Mercury Systems)

As the machine manufacture and assembly of the latest electronics becomes more common and electronics grow more complex, automated manufacturing processes will be able to exceed the handmade manufacturing of even the most refined technicians. This has not been the case yet with many sensitive components, such as filters. After all, many of the components that traditionally compose these devices require custom and hands-on manipulation for tuning. Yet future components, which are designed specifically for automation, may soon replace the older hand-style components for the benefits of cost, reliability, repeatability, and quality control.

Design for test is a significant consideration in this process, as the models’ precision can be refined to avoid costly troubleshooting. If test points aren’t specifically designed for in advance, there may not be adequate access to the necessary signals in a component or integrated circuit (IC). Additionally, designing for test can mitigate interference from the test setup and probes in advance, thereby enabling more accurate descriptions of the component functions of the CMM.

“We can get a lot more electrical testing done now with automated testing,” said Mercury Systems' Pospishil. “For example, we have complex modules with 10 inputs and outputs. To test a module like that, we will use automated testing with a controller and thermal platform with software written around the test requirements.”

Automated testing can provide tremendous amounts of data, which can later be used to fuel future models, validate current ones, or test extremely complex multiple-input multiple-output (MIMO) components/devices. Those latter devices would otherwise be inaccessible or require shortcuts (Fig. 3). “It is at the point where we press a button and the automated system does all of the testing for us,” Pospishil said. “In some cases, these tests could take all day. Some products now are so complex that the testing takes up to 16 hours. This can happen now, as there is better test equipment that is smarter, faster, and reduces test time significantly.”

3. Testing automation can enable characterization and analysis of complex module performance in ways that may be too costly or time intensive otherwise.

New software tools, design-process techniques, testing techniques, and process automation are enabling the rapidly shrinking design cycles of the most advanced multifunction devices. These new approaches deliver significant cost and time savings. At the same time, however, they may create distance between a design engineer and the circuits/components with which they are working. As these software tools and models are increasingly relied upon, they are replacing the extensive physical testing and validation that used to be the precedent.

Changes like these raise the question of how far these simulation tools and models can be trusted. Yet another argument is that the new wealth of data enabled by these methods is enabling skilled designers to have even greater understanding of the underlying mechanics of today’s devices. These technologies and tools can more readily transform that understanding to actionable designs. 

Download this article in .PDF format
This file type includes high resolution graphics and schematics when applicable.

 

Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish