Check out our NI Connect 2022 coverage.
Testing is not only critical for figuring out whether semiconductors and other devices will work out in the field, but it is also increasingly key for companies such as GM and Qualcomm to optimize performance, said NI CEO Eric Starkloff.
“Testing can not only be backward-looking to assess pass/fail, but also part of a forward-looking process to optimize performance," said Starkloff in a keynote at the company's annual NI Connect event last month.
Founded over 40 years ago, NI offers automated test and measurement hardware connected by software tools, including TestStand, SystemLink, and LabView. The systems help bring virtually every technology out there to market faster, cheaper, and with better performance, from robotics and rockets to airplanes and autonomous cars. The company’s technologies also aided in the development of 3G, 4G LTE, and 5G networks.
“Software and automation are at the core of what we do at NI," said Starkloff. But in recent years, it shifted to strengthen its software position with new technologies such as data analytics and artificial intelligence.
The company continues to enhance its product offerings to propel innovation in areas such as wireless, aerospace and defense, semiconductors, and autos. It is also releasing new tools designed to help engineers save precious time, reduce costs, and wring insights out of data to help them innovate and, ultimately, get to market faster.
Last month saw the rollout of its latest analytics software tool called DataStudio. It creates a bridge between data from product definition, chip design, and verification to validation in the lab and production testing in the fab.
Starkloff and other executives said companies have to go about test and measurement differently. Instead of leaving it to the end of the development process, they must put testing at the heart of everything they do.
"Test data is crucial," he said. "It is the only data that tells you how your product performs in the real world."
Dialing in Batteries
Starkloff said it is working with GM on a software-connected battery-test solution that uses SystemLink to connect data from testing at different phases of the battery development cycle and provide new insights.
GM has become aggressive in its efforts to sell more electric vehicles (EVs). It is promising billions of dollars to expand its lineup in the years ahead and to design batteries that charge faster and have higher energy-storage capacities.
But none of the innovations mean anything if the battery lacks vehicle-grade reliability, safety, and security.
“It is all about the battery,” said Steve Tarnowsky, head of global battery cell engineering at GM, during the keynote. He added building batteries at scale are key to a future where EVs are affordable for the masses.
At the heart of GM’s strategy is Ultium, a modular set of building blocks spanning the powertrain, electric motor, power electronics, and batteries that will sit at the heart of all models in GM’s electric vehicle lineup. Ultium will allow GM to plug-and-play parts for different EVs, including heavy-duty trucks as well as high-end cars, without having to swap out wire harnesses or communications systems for each model.
Inside Ultium batteries, GM said that it bundles together up to 24 battery cells in larger modules that protect the cells from harsh vibrations, severe temperatures, electric interference, and many other rigors of the road. Up to 12 of the modules are wired together to build out the battery pack that gets placed in a heavy-duty enclosure lining the base of the vehicle, potentially carrying hundreds of cells that can weigh up to thousands of pounds.
Tarnowsky explained that “the battery has an outsize impact on the cost and performance of a vehicle. So, getting it right means we must ensure that everything performs at its best from the cell level to the pack level.”
Testing Plus Monitoring
Ongoing testing is critical because if any one of the cells in the battery is damaged, the rest of the system becomes more likely to fail. For GM, which hopes to sell at least one million electric cars every year by mid-decade, each carrying a battery with up to hundreds of cells, it needs to safeguard everything from the cells to the pack that feeds energy to the electric motor and other systems that impact the EV's range and safety.
But designing batteries and testing them creates colossal amounts of data, including from its production facilities to testing labs to companies it works with in the supply chain. To handle the deluge of data, GM is leveraging NI’s SystemLink software platform, which GM said helps to reduce risk—no “blind spots” in the battery development, as Tarnowsky put it—and spot areas of improvement in its batteries more efficiently.
He said NI’s SystemLink helps it securely store test data from its proprietary batteries and strictly control access to it. The co-developed test solution also maximizes automation to allow it to scale up as needed.
The system is also based on 80% commercial off-the-shelf technology, noted Tarnowsky, allowing GM's engineers to tap into open-source software and different databases and programming languages to analyze all the data.
GM said it is also tracking data from electric vehicles now on the road from customers willing to share it. The data will be used to gauge the performance of its batteries and figure out when they need maintenance.
Less Time to Test
Software and hardware to test chips is another key pillar of NI's business. It is working closely with industry giants like Analog Devices, NVIDIA, Qualcomm, and even technology giants investing in custom chip design.
Like batteries, designing a modern chip creates vast amounts of data that can gain data insights to help them innovate and improve performance. And chip firms are constantly trying to take better advantage of it.
Gaurav Verma, senior VP of engineering at Qualcomm, said the current pace of chip development is faster than ever. A decade ago, it would require a minimum of 24 months to move from prototype to customer sample. Shipping a million units would take up to eight months on top of that. But market pressures have pushed it to the point where the whole process must wrap up in six months.
“One of the biggest challenges is the sheer complexity that has changed from 3G to 5G,” he said during the keynote, citing the 50X increase in throughput and major challenges with the millimeter waves used in 5G.
NI said it is working closely with Qualcomm to standardize the hardware and software in all of its test labs to reduce the time to market for its radio-frequency ICs and millimeter-wave modules for 5G smartphones.
Verma said it has also adapted the software at the heart of NI’s PXI-based semiconductor test system (STS) hardware, giving it the ability to reuse test programs from the lab to its production plants. That is providing deeper insights into missteps in the development cycle and improving collaboration between engineers at different stages of the process. “We can no longer approach testing the same way," he pointed out.
Lifecycle Analytics
Qualcomm also uses OptimalPlus, NI’s AI-based lifecycle analytics software that is used to collect, analyze, and gain insights out of the company's manufacturing data and give it more visibility into its supply chains.
It also runs data from its fabs through OptimalPlus to tease out tiny differences between batches of chips, manage suppliers, and benchmark its foundries against each other to bolster quality, yield, and cost.
In addition, OptimalPlus is being used to create algorithms that identify and removing potentially faulty chips before being shipped by its foundries. The technology has improved the outgoing quality of its products.
Qualcomm is not only feeding data to its foundry partners to improve how its chips are manufactured. It is also storing it for later to leverage with artificial intelligence and machine learning to identify possible areas for improvement in its chips.
"The reality is that this is not far away in the future," said Verma.
Check out our NI Connect 2022 coverage.
