Taipei, Wednesday, September 24, 2014 11:15 (GMT+8)
Reed switches and MEMS: A conversation with Coto Technology
Michael McManus, DIGITIMES, Taipei [Friday 21 June 2013]

One of the oldest companies in the electronics industry, Coto Technology has been designing and developing small signal switching solutions for over 90 years. These days, the 93-year old company is a major player in the automatic testing equipment (ATE) industry where it provides reed relays for testing devices. While reed technology predates the digital age, Coto has also made moves in one of latest growth areas in the semiconductor industry, MEMS technology. Earlier this year, Coto made a splash when it announced the availability of what it claims is the smallest MEMS-based reed switch available on the market today.

During the Globalpress Electronics Summit 2013 in Santa Cruz, Digitimes had the opportunity to sit down with Stephen Day, VP of technology, and Bill Gotschewski, VP of sales and marketing at Coto Technology to find out more about Coto Technology and their ability to develop a MEMS reed switch with a footprint of less than 2.5 mm2.

Q: Before we discuss your MEMS switch, can we touch on reed switches. Reed technology (switches and relays) has a bit of elegance to it because of its combination of simplicity and historic staying power. Can you tell us a bit about that history?

A: The reed switch was invented in 1936 by a researcher at Bell Labs named WB Elwood. Elwood basically took a piece of glass tubing and put two soft nickel/iron magnetic blades inside and fused them to the tube. He then had nitrogen blown into to the tube to provide a clean, inert atmosphere and the whole thing was hermetically sealed. Precious metals, usually ruthenium or rhodium, are now used on the contacts to make them last longer. They used to use gold but gold is too sticky.

The way reed switches work is there is a tiny gap between the two blades and if you bring the switch close to a magnetic source, such as a magnet or coil (to make a relay), the two blades are induced to north and south poles and attract, coming together to complete a circuit.

Despite their simplicity, for their size reed switches can switch at high power. The movement of the blades is also so far inside their limit of elasticity that they can close literally billions of times. So reed switches have enormous lifetimes. Moreover, reed relays are enormously reliable because they are sealed hermetically, compared with electromechanical relays that are affected by the outside atmosphere. And they are not prone to damage from electrostatic discharge, unlike some solid state switches.

We've made switches tested to five billion mechanical cycles without failure. Now, if you start to flow current through the switch, the amount of watt power will affect the lifespan to some extent. For a 5V 10mA load, the life cycle is about a billion cycles but that would drop to 100 million cycles for a 5V 100mA load.

Q: Historically reed relays were used in telecom, but not so much anymore. What are the main applications for reed switches?

A: We should first explain the difference between a reed switch and reed relay. A reed switch is a standalone device that can be operated by a magnet, a current-carrying coil, or a combination of both. A reed relay combines a reed switch and a coil into one component.

Reed switches are used in enormous numbers as sensors in areas such as alarm systems and medical devices, among other applications. One of our principle applications has been to wrap a coil around the switch and make it a relay for use in automatic test equipment (ATE) solutions or anywhere you need to switch a large current with a small current. They are like a power amplifier in a way.

These days in the ATE industry, each tester has 10,000-20,000 relays inside and the system may go down if just one relay fails. So the number one objective is reliability. You have to be switching at 500 million to one billion cycles, which requires enormously high reliability on each individual piece - or an overall reliability rate of 99.999%. We have really focused on super-high reliability, and over the past 30 years we have dominated in the ATE space. This is the area where we have hung our hat, testing anything from Apple iPhones to the next Intel processor in a range from high precision to high frequency.

Q: The ATE industry is still using glass solutions?

A: The glass solution has lasted from 1940 until now but the technology is hitting a wall. Over the years, the industry has wanted to get more throughput by including more channels and higher densities in the testers. For example, if Foxconn wants to test more Apple iPhones in a 15-minute period it will look for smaller and smaller solutions.

Unfortunately, we think there are fundamental physical limitations being reached where you can't make a reed switch any smaller. The way reed switches are made, a lot of heat is needed to fuse the glass. If you make the switch too short, the heat travels by thermal induction down the blade of the switch and it destroys the precious metal coating.

In 1940 reed switches were 50mm long, now they are down to about 5mm and that is about the practical limit. If you include the length of the wire, realistically the device ends up being about 7mm long.

So if there were 100% reed switches in a system 15 years ago, it is more like 30% today. MOSFETs have kicked in as a replacement, as have electromechanical solutions. But if you ask an engineer what would be the preferred solution, the answer would absolutely be a reed.

Q: Is this what has led you to developing a MEMS solution?

A: Based on our industry perspective, we understood that there would be continued strong demand for a magnetically operated reed switch that is much smaller than existing types, that can handle similar electrical switching power, and that can be attached to a circuit board by surface mounting. But it still needed to retain the benefits of reed technology. MEMS was an ideal fit.

So about six years ago we met with a company called HT Micro, a MEMS and microfabrication specialist located in Albuquerque, New Mexico. Management at HT Micro basically all worked at Sandia National Laboratories previously, doing military impact switches and nuclear device detonation switches. Thank god for all of us that was not a big growth market, so they were interested in joining forces with us to develop more mainstream products. That is how we got started. We have since set up a joint venture called RedRock to develop the technology.

HT Micro has its own fab, which is very important for being able to control production. These are not manufacturing processes that are amenable to conventional semiconductor foundries.

Q: Can you talk about the MEMS reed switch you recently announced?

A: What we have done is develop a new type of reed switch based on high aspect microfabrication. The switch maintains the desirable properties of conventional reed switches - high current carrying capability, hermetically sealed contacts, high resistance to electrostatic discharge (ESD) and zero power operation, in a package about one-tenth the size of the smallest available reed switches.

Instead of using blades, our MEMS reed switch has a metal cantilever that bridges two isolated metal blocks that act as magnetic field amplifiers. There is a small gap between the cantilever and one of the blocks and when magnetic flux from an external magnet builds up in the gap, it pulls the cantilever into electrical contact with the block. Much like traditional reed switches, the contacts are coated with Ruthenium.

Q: You say your switch is the smallest MEMS reed switch in the market. How have you been able to achieve that?

A: We use what is called high aspect ratio microfabrication (HARM) instead of planar MEMS. From our experience most switch users are much more concerned about footprint of the switch (PCB real estate) than they are about height. In traditional planar MEMS, the blade is electroplated on top of a base substrate, and then a layer under most of the blade is etched away, freeing up the blade so it can bend. But making thin, wide blades the planar MEMS way by using conventional electroplating is difficult and if you try to maximize the cross sectional area of the blades by plating them wider, it increases the footprint.

Using HARM, the blades are grown by electroplating, but they are grown edge-on, and vertically relative to the switch substrate. That way, we can make them as high as we want without increasing the footprint of the switch.

Another thing about HARM, is that it produces switch structures that generate closure force that is much greater than that shown by previous MEMS-based magnetic switches. This enables hot switching up to several hundred milliwatts. The high retract forces in the switch when it opens also prevents the switch from sticking shut during hot switching or after long closure periods.

This is important because while some customers are looking for a switch to perform hundreds of millions of cycles, others need the switch to sit for almost two years and then be used once. This is very important, for example, in applications used in the medical industry.

Q: Your products are not priced to target the mass market, such as for smartphones. What are some other possible applications? Is the target market the ATE industry?

A: The MEMS reed switch can be used anywhere you need higher power in a small space because the switch dissipates the power very efficiently. In areas such as robotics and sensor applications, the MEMS reed switches are ideal as actuators. Other spaces where the device would be ideal is where low power or no power activation is required. For example battery sensitive applications like hearing aids. A lot of 70 year old guys don't want to always be replacing the battery in their hearing aids.

In the ATE industry, our focus will be on a MEMS reed relay, which is being developed in parallel to our MEMS reed switch. This product will come in the future.

Stephen Day, VP of technology, Coto Technology

Stephen Day, VP of technology, Coto Technology