Do Dutch high tech companies invest enough in software development? Bits&Chips poses the question to Guustaaf Savenije and Ton Peijnenburg of VDL ETG, the largest tier-1 high tech supplier and hardware company in the Netherlands. “Software is just one of the disciplines – no more, no less. Better participate in the multidisciplinary world!”
“We’ll never get rich from software alone.” Guustaaf Savenije, the CEO of VDL ETG, immediately tempers the expectations when asked about his company’s software strategy. The tone is set.
Savenije counts ASML and Thermo Fisher among his customers. If they no longer build hardware, he says, then they no longer have to develop software. That’s also where the priorities of the Netherlands’ largest high tech supplier lie.
VDL ETG does build high-end system software. It develops and makes ASML’s wafer handler and is responsible for the machine’s future generations, including the software. “We deliver complete, functionally tested modules. If there’s a servo loop in it, we also provide the software for that. But it starts with physics and mechanics. I’m telling you, point blank, because that’s our core. Software is part of it, but just like electronics, it’s one of the supporting disciplines. Thermo Fisher distinguishes itself with image processing software, but without their electron accelerators and detectors, they can’t process images,” Savenije illustrates.
He prefers not to talk about software, but about product strategy. “From there, we look at physics, mechanics and software. Software has no special position in our organization. We do discuss this at a high level with other VDL divisions. But then it’s mostly about how we organize our factories and manage production processes. We mainly look at what software’s available on the market.”
Savenije may occasionally like to provoke, but his right-hand man, Ton Peijenburg, provides some nuance. Peijnenburg, who as deputy director is jointly responsible for technology and development at VDL ETG, makes it clear that there are two aspects involved. His engineers deal with embedded software in products on the one hand and with software in factory systems on the other. “In both, VDL needs to follow current trends, such as data science, artificial intelligence and machine learning, and make use of them.”
Don’t you want to be at the forefront of software to achieve the precision you require in your machine modules?
Peijnenburg: “We’ll apply algorithms for compensation, learning control or machine learning when those technologies have been tested, for example, by a university. We’re certainly not a pioneer there. We pick it up, just as we pick up a different coating technique. Software is also a way for us to implement functionality and improve machine performance, just like a higher-quality aluminum or a faster milling machine.”
Savenije: “Contamination! We support PhD research there. That’s differentiating for us, that’s our profession. Just like new manufacturing techniques. We’re experimenting with industrial 3D metal printing. We want to lead the way when that becomes available. In software, we just have to use state of the art.”
You can’t differentiate yourself with software?
Savenije about the logistics and automation of his manufacturing processes: “If we don’t do software, it will definitely hurt. Then our costs will go up. But with software alone, we won’t distinguish ourselves. We can, however, with knowledge about cleanliness, how to get the last nanometer out and with which materials to make the best electron lenses, with assembly and functional qualification.”
Are you a fast follower?
Savenije: “We’ll never win by investing purely in software. But we need to have a modern logistics system for our factories. We need to do tracking and tracing, as well as save and mine the data.”
Peijnenburg on product software: “With increasing complexity and precision, aligning and making agreements about requirements and specifications has become very complicated. It’s not just about accuracy, but also about standards with thousands of requirements, for example, the regulations that the American FDA places on medical modules. The items we supply have ten to a hundred thousand requirements. The A4 sheet with specs has grown over the past decades into a veritable stack of books – hard to oversee. The digitization of the workflow helps there. We’re taking steps to ensure that this is better aligned.”
Database techniques and software tools are indispensable for this, says Peijnenburg. “If you organize a review, you can’t expect 40 people to have read a hundred pages. There, interactive visualization with diagrams and graphs helps.”
The requirements also change per machine generation, so traceability is required, as are baselines to record the configuration history. Peijnenburg: “In the software, baselining is a standard technique. Even the robot football team that I coach does that every week. For the competition, we press the ‘build software’ button and it works. With product data management and product lifecycle management, we also need to record our development history. And the measurement data of the products that leave our factory – to be able to trace the root cause of a problem later.”
“We’ll have to get used to saving all the information. The worst thing we can do now is to delete data. Product data from a final test? Save! So we need storage. What’s the cheapest? The cloud! For many, that’s about the equivalent of throwing the data out on the street. We have to deal with such matters.”
Savenije: “The mindset needs to be: record everything. That’s no joke. Do you get aluminum from Japan and was there a tornado during casting? Capture that! Make a label ‘weather conditions during casting’. Who knows, in ten years this could lead to the root cause of a problem. You can never do too much there, but you can easily do too little. For me, that’s not software. Software does help me to process huge amounts of data.”
“Frankly, software makes a lot possible, both for us and the entire region. But to be honest: if developments in the field of data management were to stop, we wouldn’t be bothered by that for the next ten years. Because by that time, we’re still busy implementing the current knowledge in that area. Of course, you need a few visionaries who say it can always be done better and more efficiently.”
To what extent does your software development serve your business?
Savenije underlines that software at VDL ETG is simply a servant to the business. “We don’t sell software. In that sense, it doesn’t differentiate us. But it can cause great dissatisfaction if you don’t pay attention to it.”
The big digitization threat he sees is mainly with his customers. “Companies that supply the systems that show the x-ray and MRI images to doctors in a hospital have a much more intimate relationship with these customers. So, if the suppliers of medical scanners aren’t careful, those who make the higher software layers will soon determine which instruments are or aren’t good for their customers. So, scanner manufacturers are also forced to supply those top layers.”
For the same reason, ASML’s holistic approach to lithography also contains a strategic component, says Savenije. “It’s not just about better performance. It makes their link even stronger and reduces the risk of someone saying: a stepper from another supplier is also good enough here.”
How should we maintain our position in the field of precision technology?
“We have to, we really have to.” This clearly touches on a subject that’s important to Savenije. He finds high tech vital for the Netherlands. “We’re currently able to develop more complex systems than the frontrunners from the past – Germany and Japan.”
Peijnenburg: “If you look at software for managing data, the US is at the forefront of innovation. But for the most advanced machines, you have to be in the Netherlands. Our strength is in our culture. We have no problem with discussing the content of a technical problem with each other. We do this based on equality here, regardless of rank or position. We mobilize the entire workforce, while everyone in Asia looks at their boss. America is also much more hierarchical than they’d like to admit. This way, we achieve a faster and better result with multiple disciplines. We have a culture that embraces systemic and multidisciplinary thinking like no other. Combined with the fact that we know how to work with diamond cutters, how to work ultra clean, how to handle aluminum and so on.”
According to Peijnenburg, the high demands of companies such as ASML, Océ, Philips and Thermo Fisher also play a role. “They demand the utmost. Their applications challenge their employees and suppliers. The mindset is there, and the technologies are available. And when they’re not there, we’ll make them or get them.”
While Savenije and Peijnenburg serve this positive story, they’re stuck with a slight irritation about the attitude of some of the embedded-software engineers. With over thirty years of development experience at ASML, Assembléon, Besi, Philips and VDL Group, Savenije believes that the software community is still too isolated. “I like to exaggerate a bit, but software people tend to only ask for requirements and leave it at that. They should think along with the electrical and the mechanical engineer. Software is just one of the disciplines – no more, no less. So better participate in that multidisciplinary team!”
Peijnenburg calls it “the gap between software and the rest.” He also encounters it at Eindhoven University of Technology’s High Tech Systems Center, where he works one day a week. “Software and the rest” don’t easily talk to each other when the question comes up who’ll solve the problem. Peijnenburg: “Will software get started with calibration and metrology software? Or does mechanics have to make a more precise re-design? Discussions like that. To be able to integrate even further and better, we need to focus on this in the near future.”
Savenije: “IT people also have a duty to visit developers from other disciplines and try to understand their problems. They have to enter the hardware rooms. Occasionally screw up something, break something, write a specification themselves and review that with hardware colleagues. Create and test simulations and prototypes together. That’ll make it nice and balanced.”
How do we close the gap?
Both Savenije and Peijnenburg believe the key is in systems engineering. “We feel uncomfortable in neighboring domains because we’re afraid to say stupid things. The best software engineer is a lost physicist, mechanic or electrical engineer. The latter also need to know more about software, that’s also good. It has to go both ways,” says Savenije.
It’s all about an open atmosphere. “Everybody needs to have a sincere interest in the profession of one another. The mechanic has to have the mindset to occasionally look over the shoulder of the software developer and ask what he’s actually typing. That’s all. On occasion, the software guy has to go into the factory and ask: what’s that? A five-axis machine? How do you actually program it? That’s how problems get solved.”
Peijnenburg: “To maintain our competitiveness, we need to work together even more intensively. Team up on data analysis, compensation and metrology, for example, to make systems work even better. Part of the problem isn’t in the nanometer, nor more control over the physics or the understanding of mechanical-physical phenomena. It’s in the complexity of a system. What do you have to calibrate in a module with 36 motors and 3200 sensors? These kinds of metrology calibration techniques require automation – and therefore much more software.”