
CEO interview: Building a European sensor unicorn
Nanusens has big plans in the consumer business. It has designed a micromachined MEMS accelerometer that can be built on a low cost, standard CMOS process, and crowdfunded the development of the first silicon for the sensor.
The sensor is aimed at providing controls via tapping for earbuds and has been built at SMIC in China, although the technology can be used on any CMOS process. This is a key advantage that will drive large volumes quickly, says founder and CEO Dr Josep Montanyà.
“My bet is three to five years to reach $1bn in sales,” said Josep, “We are going into a market that is dominated by reference designs – there are five ear bud reference designs, and in mobile phones there are two. So either you have significant market share or you have nothing, We have something that is unique is size and performance, especially when you go to combination devices, its a 30 percent cost advantage and half the size of the smallest in the market and ten times smaller than the others.”
The company crowdfunded the development of the first silicon and is now looking for more investment, hence the bullish outlook. But this is not his first rodeo, as the saying goes.
He started Baolab Microsystems in Barcelona, Spain in 2003 to develop a low voltage RF switch from university research.
“I got initial funding of €300,000 and set up Baoloab and had samples after two years. We showed these to a series of companies in the US and realised my professor was wrong and there wasn’t the demand. The issue was cost, size, yield and integration with CMOS. So in 2005 the focus changed to focus on MEMS, on a specific manufacturing process but then I realised we need to move the MEMS into CMOS. There was a lot of research on that but no one had managed to take it into volume manufacturing. In the end it was clear the best way was to use Vapour HF to etch the silicon oxide but the challenge was how to use those metal layers that were never intended to be parts of MEMS. We were still trying to develop a RF switch but the challenge was the contact.
“Then in 2010 we realised we had a MEMS process we could use for a non-contact design and moved quickly to a proof of concept of a compass. We were close to releasing a compass to market in 2012 but our lead investor had been in the company too long so the investment fund had to return money to investors after ten years, so they wanted to sell BaoLabs over the next two years. There were offers for investment for acquisition, but they rejected everything, plus they were thinking in terms of the fund and in the end the company shut down in 2014.
“A few months later I set up Nanusens. The number one lesson I learned was not to allow a single investor to control more than 50 percent of the company to avoid the situation where you have a great technology but the company needs to shut down.”
“We initially did engineering services and raised seed funding in 2016 with three investors from Barcelona and the Netherlands. So in 2016 I could start again but the specs for a compass had increased so that wasn’t feasible so we focussed on an accelerometer for the first product.”
“My experience is its not easy to raise funds for semiconductors. Semiconductor startups are tough, especially in the beginning. Now we have silicon working I think it will be different and now I am having interesting conversations with funds,” he said.
The crowdfunding, and the associated tax break, called EIS, also led to setting up the company in the UK.
“Most of the funding today apart from the initial investment was business angels and that’s why in 2018 we set up the crowdfunding and we have done four rounds with Crowdcube. With have 1400 investors but most are syndicated – we had some investments at £10,000, so we have maybe 100 direct investors. Most are from the UK, hence the move,” he said.
“In 2018 I wanted to try the crowdfunding funding – you can do that from Spain, but it happened that we were not getting to the target and I looked at other companies that were successful and the common factor was EIS. So I set up the legal structure in the UK and when you apply for EIS the funding was complete in a few days and we started overfunding.”
“With the larger investors we go direct and this gives us connections with potential customers in lots of different industries,” he said. “With IoT for example there’s applications in building, retail, lots of things.”
“We don’t discount using crowdfunding going forwards but the larger rounds are more challenging. We raised £1m then £2m on Crowdcube, now we are looking for $6m. It could be doable but we want to explore the corporate investors. The difference is we have the silicon and most of the business angels can’t help with that – corporate VC in semiconductor can assess the due diligence.”
The company set up in Torbay in the southwest of the UK.
“Then we had to subsidiarise the Spanish company and after two years started to set up the real structure. Initially we had the HQ in London as the legal structure, and then we started looking at options and there was a lot of support from Torbay. They had a small clean room and electron microscope, and half my time I live in Paignton and we are in process of staffing up and looking for people.”
Despite focussing on the MEMS accelerometer sensor, the original low voltage RF switch has also become a viable product.
“Every time we tape out a silicon run we add something and last year we got a big surprise with our RF switch for consumer and it was working very well for the reliability, beyond one billion cycles without an issue so we decided to promote the switch in parallel,” he said. “This is a capacitive switch rather than ohmic and talking to customers for antenna tuners we realised we had to change the name to the digital tunable capacitor (DTC). We have both devices in silicon with the mechanical part tested, now we are adding the electronics and will sample the RF DTC by end of June and later for the accelerometer. We plan to start volume production next year so we need Series A to fund that.”
“We use a 180nm process with 4 to 6 layers so we use all the metal layers at SMIC, we have very good service from them and we plan to use them for volume production. Most if not all our customers will be Chinese but we are in CMOS so as part of the plan we will need second sourcing as all the customers want that. That’s one of the beauties of our approach to go to other foundries and even other nodes. We have also tested at GlobalFoundries and LFoundry.”
“For lower nodes with larger metal layers it would be possible to place the MEMS on top and that’s an interesting future for what we have and we are designed to go to lower nodes.”
“Next year we should start delivering samples for more products. It has taken time for us to solve some important challenges on yield and the inner structures, but now to move to the next products a lot of what we have is the same. We are also in talking with semiconductor companies to embed the sensor into their chips for motion detection in an MCU, there’s lots of possibilities there – that’s IP.
“We can go to many markets but our focus is in consumer where size is most important as well as power consumption and cost. There we make the biggest difference. For automotive and other markets we see that more through IP,” he said. “Initially our focus is on product and secondly on IP. This gives us a lot of revenue short term and reinforces the IP business.”
Related articles
- European equity crowdfunding platforms to merge
- CMOS NEMS for tunable RF switching
- Baolab buries MEMS into CMOS production line
- Baolab plans to create low cost, smart, reconfigurable IMU
Other articles on eeNews Europe
- NI teams for 6G terahertz research
- World’s first 700MHz 3GPP network for public safety
- NTT buys Fujitsu unit in OpenRAN and 6G photonics push
- GraphCore joins €2.6m project for real world AI
- Siemens, Google team to scale cloud AI in manufacturing
