Semiconductor testing grapples with cost and complexity

Developing semiconductor test instrument solutions is getting prohibitively complex and expensive. “Having to put in 15-25 thousand man hours is commonplace these days. Very few companies can afford that,” says Paul van Ulsen, CEO of Salland Engineering. The hefty price tag is very much Van Ulsen’s problem, since developing automatic test equipment (ATE) and test and measurement setups is Salland’s bread and butter. The Zwolle-based company also offers test engineering services and low to mid-volume test services with a large engineering support content.

To continue Salland’s explosive growth trajectory – 40 percent per year over the past three years – Van Ulsen therefore had to change things up. Taking a cue from chip design, Salland has started to offer test IP: customers can license technology solutions for use in their test equipment, instruments or modules. “This is more efficient than offering custom solutions: it enables us to sell our work more than once,” Van Ulsen explains.

It’s an innovative approach, not seen before in the semiconductor equipment sector. But Van Ulsen feels the time is right for it. “Traditionally, test engineering has been an internal, fenced-off activity. Companies simply like to develop their own solutions. For a long time, that model worked perfectly well. But now complexity has passed a critical threshold. It’s getting too expensive to keep all necessary expertise in-house, except in high-volume businesses.”

“When your customers can no longer afford your services, it’s obvious you need to take action. But you can’t compromise on other aspects. The product still needs to be delivered yesterday and it has to perform flawlessly. We can meet those demands by offering IP blocks. It’s more efficient.”

It’s early days, but already Salland’s updated business model seems to be resonating in the market. “We’ve developed one of the world’s most advanced digital I/O cores, which we offer as a PXI module featuring twice as many channels at twice the performance as the next best product currently available. The same IP will be licensed to a Japanese customer, who will use it to build a test instrument.”

If you buy a car, you’ll almost certainly find Bosch components under the hood. Get a new racing bicycle and chances are that you’ll end up with Shimano gears and brakes. And if Van Ulsen’s dream comes true, there will be a bit of Salland in every piece of test equipment. A physical bit that is, because Van Ulsen estimates that Salland already makes some type of engineering contribution to 90 percent of the world’s major-brand test equipment. “And that fills us with pride,” he says.

End-to-end

Salland’s expanded portfolio will certainly be a topic of conversation during the next Future of Test Symposium, 27 and 28 October in Zwolle. The company first hosted the event five years ago, when it celebrated its 25th anniversary. “We can’t expect our customers and partners to travel all the way to Zwolle for a speech and a glass of champagne. A technical symposium seemed like a good way to convince them to come here.”

The symposium was an instant hit and Salland decided to make it a yearly event. Having skipped the past two years due to the pandemic, this year’s third edition coincides with Salland’s 30th anniversary. In between the two jubilees, Salland has expanded its test and lab floor by 1,000 square meters, quadrupling the total. “That, too, is something we’d like to celebrate,” says Van Ulsen.

“There’s no event in the world in which the entire test supply chain is in attendance, including the five largest ATE manufacturers and major chipmaking end customers.” Van Ulsen is expecting 100-150 representatives from 40-50 companies and research organizations. “It’s good to come together and exchange ideas. Especially in Europe, we need to join hands to enable end-to-end solutions. The days in which a single company can do it all on its own are well behind us. I think our industry is very aware of that, and that’s why people are eager to attend.”

All-electrical

Even offering technology as IP blocks and collaboration won’t be enough to keep test-related costs in check, Van Ulsen stresses. “In the future, we need to test more practically and do away with the huge currents and speeds. At Salland, our philosophy is that designs should be validated in the lab. In manufacturing, validation should be all about the manufacturing process. That can be done by comparing relatively straightforward electrical measurements with validated simulations.”

Showing the way, Salland spearheaded the Meteoriet consortium, in which three companies and two research institutes collaborated to develop all-electrical MEMS testing. Traditionally, MEMS testing involves a physical stimulus, making it hard to standardize testing methodology across different types of chips. The Meteoriet partnership, however, has shown that movement inside MEMS chips can be induced electrically and that the movement also has an electrical signature. Taken together, this enables testing MEMS chips solely by electrical means.

“This is the way forward for testing all semiconductor devices, including emerging markets such as photonic integrated circuits and quantum chips. In photonics, for example, the need to in and outcouple light stands in the way of a traditional testing strategy. Either you need to test fully electrically or include a light source for testing purposes in designs. The latter option may be an interesting business opportunity for Salland, too. We might hire design-for-test engineers to support customers or offer design-for-test IP.”

Recruiting

One last topic that will be popular among the Future of Test’s participants will be market developments. It’s been a hectic few years for semiconductor companies, including Salland. The chip shortage caused business to explode, yet the same shortage stood in the way of fully capitalizing on the opportunities that presented themselves. “We’ve already had to delay revenue into next year. We’re constantly having to take into account the availability of components and even do redesigns to be able to deliver.”

“I expect this situation will persevere for at least two more years. The market for PCs and smartphones is cooling, but companies are still investing heavily in technologies such as the Internet of Things, 5G and automotive. These are the segments Salland derives most of its business from. I expect Salland will grow 25-35 percent per year in the next few years. If we can find the engineers – we’re already actively recruiting.”

“The ambition of Europe, the US and some Asian countries to reduce technological dependence on other regions will spur growth as well. Apart from the overall capacity increase boosting business, companies tend to prefer locally sourced components and services. As semiconductor activities expand in Europe, the local test ecosystem stands to gain.”

This article was written in close collaboration with Salland Engineering.