Rajesh Vashist (RV): Dynamic performance—the capability to maintain high performance under harsh and changing environmental and system conditions—is critical for many applications such as 5G. Quartz oscillators, however, cannot handle dynamic conditions like vibration, shock, and rapid temperature changes. SiTime solves these problems through our systems solutions approach, and we believe that SiTime products are the timing solutions of choice in high performance systems such as 5G. We believe our customers and partners such as Intel have validated our leadership in 5G. In 2018, we announced our collaboration with Intel on MEMS timing for 5G.
Reliability is also a problem for quartz devices. The reliability of quartz oscillators is between 100K and 30 million hours MTBF. SiTime typically delivers over 1.5 billion hours MTBF. The reliability of our MEMS timing solutions is very important for high performance systems such as core telecom infrastructure and automotive that will remain in the field for long periods of time.
Another area of concern for quartz devices is their supply chain which is geographically concentrated. Kyocera, a quartz vendor, also supplies a large percentage of the world’s ceramic packages. If there is a problem or natural disaster at their manufacturing location, such as the 2011 Tsunami in Japan, it could disrupt the entire quartz supply chain. SiTime uses the semiconductor supply chain and in most cases, uses standard plastic packages that are available from multiple suppliers.
RH: What are the drawbacks of MEMS oscillators? (Price? Yield?)
RV: There are no technical limitations or drawbacks. The biggest challenge is that customers need to know that such devices are available. Today we have 10,000 customers, but there are up to 150,000 electronics customers that could use our products.
Since quartz timing has existed for 90 years, quartz companies have developed many more types of products than currently exist with MEMS timing. For example, quartz companies offer standalone resonators, which we are in the process of developing. SiTime does not yet offer mobile TCXOs. However, over time, SiTime MEMS timing products will cover the entire market.
On price and yield—we are very competitive. Our MEMS yields are in the 90% range.
RH: You have now sold more than 1 billion MEMS timing solutions. Which conclusions can you draw from the customer responses, and which conclusions can you draw from field returns?
RV: In the past, many OEM/ODM designers learned to overcome the shortcomings of quartz with various work-arounds. When the timing solution is working well enough, the oscillator/resonator choice is not often top-of-mind. However, as new standards such as 5G come into play, a MEMS timing solution has massive advantages over quartz. There are similar significant advantages for MEMS timing in other areas such as IoT and ADAS. Once customers see our solutions and benefits, they are eager to design SiTime into their applications.
Our MTBF is well in excess of 1.5 billion hours. Our quality is 0.6 DPPM, which is 10 to 100 times better than most Tier-1 quartz companies. Customers have come to expect that SiTime provides much higher levels of quality and reliability than quartz—and once they shift to MEMS, they rarely go back to quartz.
RH: Recently, SiTime announced its Emerald Platform. Can you please tell me more about that platform? What is new?
RV: Emerald products are the first MEMS-based OCXO. They are designed to solve problems associated with quarts OCXOs, which are very sensitive to high temperature, thermal ramps, and vibration. Designers often need to make 3 to 5 board iterations to place the quartz OCXO correctly, and if the design still shows sensitivity, they must use a plastic or metal cover for the quartz OCXO to isolate it from the ambient environment. Our Emerald Platform™ eliminates this problem. Our robust OCXOs can be placed anywhere on the board. Customers get the placement right the first time, can introduce products faster and accelerates time to revenue.
In addition to solving board design issues, the Emerald Platform sets new performance benchmarks in dynamic stability (in system, in the presence of vibration, shock, temperature ramps) that is up to 20 times better than quartz—a must-have for new 5G infrastructure deployment. Emerald products are programmable and always available in any frequency from 1 to 220 MHz, plus a range of frequency stabilities, operating temperatures, output types, and packages including a solution that is 75% smaller than similar quartz OCXOs.
For customers, we’ve eliminated a long list of headaches for our customers, allowing them to sleep better at night. For SiTime, the Emerald Platform allows us to expand into the higher end of the $1.5 billion telecom and networking timing market.
RH: Emerald is designed for the 5G infrastructure. In the press release you stated that timing was potentially the single biggest point of failure in 5G systems. Can you please clarify this?
RV: 5G technology will put pressure on the timing solution far more than in the past for a number of reasons. 5G radios use millimeter-wave technology where coverage is several hundreds of meters rather than the kilometer distance of 4G. Therefore, the 5G network will be much denser. A much denser network bringing equipment closer to the customer and out into harsher, less controlled environments, such as deployment on traffic lights, street lamps, buildings, rooftops, and parking garages. These locations have a lot of environmental stressors such as vibration, shock, high temperature, and rapid temperature changes. These stressors cause quartz devices to fail and we believe that this could be the single-biggest point of failure in 5G. Emerald products are 20 times more immune to such environmental stressors and will ensure that the network, and the mission critical services that run on the network, run well.
Keeping in mind the denser deployment profile of 5G, higher bandwidth, and deployment of mission critical services, the standards bodies are making the radio-to-radio time accuracy spec 10 times tighter. In 4G, it was 1.5 microseconds. In 5G, it is 130 nanoseconds (ns). To meet this tight spec, we believe that Emerald OCXOs will be required.