Voice controlled applications require more advanced microphones
4 mins read
Voice control of built environments, and of all devices used in and between them, is one of those concepts whose fate will be decided by whether people actually want it or not: do we really want to get to the stage of telling a light switch to turn on or off, rather than pressing it?
The technologies behind the concept are essentially already here, although they need to be developed to meet some performance and power criteria. These developments are being made to meet the requirements of other related markets, an example being the progress being made with microphones.
Traditionally, every mobile device contained an ECM – an Electret Condenser Microphone. According to analyst iSuppli, the ECM market was worth approximately $500million in 2011. It will be worth more or less the same for every year until 2015, the end of the projection. But, by this time, the market for mobile microphones will have topped $2billion. The remaining 75% will have been taken by the explosive growth in MEMS microphones, which have become popular for a number of reasons.
Most obviously, they are smaller than ECMs and, as smartphones require multiple microphones to be used, rather than the one required for a 'talk only' mobile phone, size is important. MEMS are also attributed as having other advantages in terms of performance consistency, operating temperature and orientation.
Mark Hesketh, general manager for MEMS technology at Wolfson Microelectronics, conceded that the acoustic signal chain was only as strong as its weakest link – and that could just as easily be a poor ECM or a poor MEMS microphone. However, he noted: "The performance of the MEMS microphone in terms of the signal to noise ratio and dynamic ranges has improved and that performance has been achieved with significantly smaller devices. Having a high performance, small package in your signal chain means that MEMS is helping us have the very best acoustic signal chain possible."
So how have MEMS devices moved ahead in the microphone race? At the heart of the microphone is the transducer. "We have a developed a transducer device that has a very linear performance and very low noise floor," explained Hesketh. "We not only design the MEMS transducer, but Wolfson also owns the process for their manufacture, so we can also tune the material design to get the very best performance from our transducers. We use silicon and variants – others use other materials – but we believe there are advantages in our material set, coupled with our design, that delivers the performance."
The other aspect is the associated asic that controls both input to transducer and the output of the device. The output can be either analogue or digital if converted to a pdm [pulse density modulation] stream.
There are three main building blocks in the asic. Usually at the front end is a low noise amplifier, followed by an a/d converter if it is a digital microphone. The other core block is a charge pump that generates the high voltage via for the transducer.
In a typical handset or tablet application, the outputs (if there are multiple microphones) will be channelled to an audio SoC. The latest of these from Wolfson is the WM5110 – an audio hub with 'a new level of integrated audio processing' of mixed signals. But the whole audio chain has a 'microphone chain' as a component part. Hesketh said: "When you have just the transducer and an asic, it is very important that the asics are high performance. Wolfson has a pedigree in high performance, low power mixed signal cmos for audio, so we have leveraged that into the microphones. All of the design expertise that we use to deliver our SoC chips is also used to deliver the microphone asics. So, when Wolfson bought in the MEMS technology, it was a compatible match because it sat right alongside the existing portfolio of mixed signal work. It really allowed us to mature the technology in the microphone space."
This ability to develop both transducer and electronics in tandem is now going a step further. The company has taken its system in a package microphones and put them on a single chip. "MEMS was brought into Wolfson and we put it beside our mixed-signal heritage. That gives us our MEMS cmos chip and then our packaging. One of the advantages of owning the design and then the processing technology – one of the key bits of IP that came with the acquisition – was the ability to manufacture our MEMS transducer in cmos foundries using low temperature processing.
"What that does is it to open up the ability for us to coprocess the cmos and the MEMS parts on the same wafer and thereby achieve the single die or single chip microphone. No longer do we have two different parts wired together and glued in a box; we now have a single piece of silicon which has the transducer and the asic. We can do that in such a way that we can manufacture the cmos and then post process the MEMS without affecting the cmos element – it becomes a very efficient way to do the manufacturing."
One advantage of this approach is that there is no longer an interface between transducer and asic to worry about as there is in multichip devices. "The reason that we believe in the longer term that the integrated solution will have advantages is first of all from a size point of view. People are continuing to push in the x and y direction to reduce size," said Hesketh. "However, one of the areas that we also see as important is the z-axis, because most traditional MEMS microphones are about 1mm high. We believe that, with work we have done with the integrated microphone, we can significantly reduce that height. As things get thinner and things need to be placed right next to the skin of the product, having a very thin microphone will offer more opportunities in both placement and performance. Size will become more important as people add more and more microphones."
And there is no reason why the technology cannot be pushed further as it lends itself to other sensor applications. Pressure, altimeters and ultrasonic transducers are examples that use the same sort of membrane technology. Hesketh commented: "In terms of the business, then we are always looking for opportunities in complementary areas for our technologies. The 5110 is a device that has the ability to take inputs from many devices and different sensors and as we move forward we will look to expand our technology and capability."