COMPANY CONFIDENTIAL. DO NOT DISCLOSE.
Facilitating innovative (and fast!) product development
Analog Devices accelerates engineering cycles with robust statistical workflows and a commitment to Six Sigma
_____________________________________________________________________________________
Challenge: Balance the pressure to bring products to market quickly with the need for comprehensive, scientifically robust research.
Solution: Implement tools to streamline analytics-based workflows. Develop a library of custom scripts that standardize analyses, reduce error and save engineers time on repetitive tasks.
Result: By automating analyses with replicable scripts in JMP®, engineers reduced a computational method that would have previously required 1,200 person-hours to only require 8 hours. This efficiency – and others like it – help Analog Devices achieve both speed to market and scientific robustness.
_____________________________________________________________________________________
Offering an array of analog, mixed signal and DSP integrated circuits, Analog Devices serves the instrumentation, automation, communications, healthcare and automotive sectors, among others, with a diverse portfolio of technology. But like all companies in its industry, Analog Devices must constantly adapt its offerings to keep pace with ever-changing standards and customer needs. This particular challenge puts significant pressure on the company’s product development pipeline. It’s not enough for Analog Devices’ designers to come up with innovative products; engineers have to figure out how to scale new technologies and bring them to market quickly.
Six Sigma unites engineering and business goals
As product engineer at Analog Devices, it’s Stefan Nikles’ job to transform prototypes into manufacturable, marketable products. “Getting the factories to make the parts for us, doing all the testing and characterization, and developing a lot of the safety and reliability aspects of those parts – all this falls on our shoulders,” he explains.
With so many process elements to consider, Nikles, who works primarily on microelectromechanical systems (MEMS), describes the need to advance business priorities while also engaging in very precise engineering. In the industry, he says, there is often pressure to “get to a point where you say, well, this is good enough and move on – rather than burrowing down on a particular problem.” Speed is of paramount importance in the product development process, but without a measured approach, it can also be a blind spot. That’s why Nikles and his colleagues rely on careful data analytics to maintain scientific rigor in the course of rapid-fire product development.
“It's really important for us to be able to focus on what needs to get done as opposed to what would be nice to have but doesn't necessarily need to get done,” he says. With large datasets collected in the course of research, product engineers depend on statistical methods to help them tease out the important information quickly without drowning in a pool of data.
For Nikles, Analog Devices’ Six Sigma training program offered solutions to this common problem. So he signed up. “I wanted to get to a point where I could converse a bit more intelligently with my peers, and I knew [Six Sigma] would help me get better at analysis,” he remembers. Fortunately, the program not only boosted his knowledge of industry best practices, it also proved to him the benefits of one essential tool: JMP® software.
Scripting opens the door to new efficiencies
Prior to joining Analog Devices, Nikles relied primarily on Excel for statistical applications. “I resisted [JMP] at first,” he recalls, confident that his ability to build macros in Excel with VBA coding would allow him to accomplish anything that could be done with JMP. “But that wasn’t true,” he says. “Eventually I conceded.” Equipped with a firm grasp of Six Sigma methodology, Nikles set out to learn how to not only utilize a suite of applications in JMP for quality, reliability and experimentation, but also to automate aspects of his analytical process with custom scripts. Now, Nikles says he uses the JMP Scripting language (JSL) on a daily basis: “I can even write scripts [in JSL] that have nothing to do with JMP...It’s a very helpful tool.”
Scripting now makes possible a number of analyses that Nikles had previously been unable to run. “For instance, the kind of drift analysis that I want to do, I’ve yet to really find a straightforward way to do it, but it’s not a problem because I just write a JSL script that does it for me,” he says. Moreover, JMP helps with daily tasks like migrating, collating and organizing disparate data sets. In his work on MEMS, automated testers create files with hundreds of columns of variables and hundreds of rows of data. “It's a huge chore to go through and format the data and get all of the column names lined up and fixed,” Nikles relates. JMP streamlines this process by converting cumbersome STDF-formatted data directly into a JMP table. “From that point on, it’s all in JMP,” Nikles says of the process. “Once the data are collected, it's relatively easy to do the analysis.”
In one case, Nikles’ team had to manually change hexadecimal register values in a long script prior to testing. “It was impossible. Just not useful at all,” he recalls. Not only was it a tedious process, but one with little room for error. The data were so critical that the team couldn’t afford any mistakes in the code. “I knew a lot of people were going to be doing this same activity and using it for their own studies... so why not just use JMP to generate this script?” With future users in mind, he developed the script with various customizable options, turning a lengthy task into an instantaneous one. “And I still get compliments on it,” he says.
Reducing computational processes from 1,200 hours to 8
Though the efficiencies and improvements made possible by Nikles’ statistical efforts may not always have an immediate, visible effect, they have cumulatively had a real impact on Analog Devices’ bottom line. Time-to-market is key and Nikles and his colleagues have found an expedient way to minimize “extra” activities that could extend timelines or introduce costly quality issues and variation.
Since the further rollout of JMP at Analog Devices, Nikles estimates that computational processes that would have previously required 1,200 person-hours per year now only require 8 hours thanks to custom JMP scripts that enable automatic replications. “One person-year is about 2,000 hours, so a savings of 1,200 hours is about nine months’ worth of salary,” Nikles explains. In an environment where time is the number one resource, this dramatic reduction not only justifies his use of JMP but promotes it to others.
Building a community of JMP users at Analog Devices
A scripting pioneer of sorts within his organization, Nikles has built up a library of scripts for shared use. The automation of common processes like drift analyses and column merges are proof of the undeniable practicality of learning to script. “Those scripts save a lot of time,” Nikles says; these time savings are only amplified as more engineers at Analog Devices learn to use existing scripts to speed their own workflows. Not only does scripting increase efficiency, sharing scripts ensures consistency and limits the possibility of user error. Spending time up front writing a script facilitates structured and rapid subsequent analyses. And, Nikles explains, it’s a “method of recording what operations I used to create a certain data analysis,” allowing him to track his steps and think through the ideal approach.
By offering a systematic, reliable, customizable tool for streamlining product development, JMP has attracted a growing fanbase at Analog Devices. To support the company’s internal JMP user community, Nikles created a forum that meets biweekly to share new techniques, troubleshoot and brainstorm solutions to scripting challenges. “Script-writing has increased since I started the forum,” he says; the group has been especially critical for supporting new users and building internal JMP and scripting capacity, which only increases the profile of these approaches at Analog Devices. With productive engineers and markedly reduced product development timelines, quality or time concessions are increasingly a thing of the past for this technology leader.
