posted by Bryon Moyer
The MEMS Industry Group sponsored a webinar recently with a focus on switches. Literally; mechanical switches. Just really tiny ones made of a beam that can be actuated by an electrical signal.
OK, so I guess it’s not completely mechanical, it’s electromechanical, but the suggestion is that you could configure complete circuits with these.
The presenter was Maarten De Boer of CMU, and he painted a picture of what could happen with the continued evolution of micro- and nanoswitches. The “pros” of such an approach are:
- Lower power (only needed to actuate; don’t necessarily need holding power)
- Better “off” characteristics; there’s no leakage (also contributes to lower power)
- You can carry RF signals
The “cons” are:
- They’re slower to respond (if you had circuits made out of them, the suggestion was to parallelize as much as possible to avoid ripple delays in serial circuits)
- They’re larger – at present (the suggestion being that this could evolve… I don’t know about competing with sub-10-nm sizes… Yes, I know the whole transistor isn’t sub-10-nm, but still…)
- This only works for digital – there’s no amplification, so you clearly wouldn’t replace analog transistors with switches
Reliability is still a work in progress; work is underway to determine failure times and modes.
To be clear, this wasn’t a suggestion that SoC designers around the world should stop their work and re-evaluate whether to replace billions of transistors with billions of switches (what could possibly go wrong??). But it was an interesting look at what could be possible as the relatively large switches we have today scale down into the nano realm. You never know…
You can view this and other past webinars a the MIG website.
posted by Bryon Moyer
Not long ago we looked at Peratech’s QTC technology. You might remember it as a functional ink that’s highly sensitive to pressure. Our focus at the time was how the technology works; designs seemed to be in process at that point.
Shortly after, they announced a touch screen solution. Because of the ink’s sensitivity, they can actually put the touch sensor behind the screen, reducing the cost of the screen itself and getting the electronics out of the way on the edges. They can also make the screens arbitrarily large. But more importantly, the touch layer is no longer in the light path, meaning that it doesn’t absorb any light, meaning that power can be reduced for the same effective light output.
They recommend it in particular for e-paper and OLED screens, although it will work with anything except LCD (which doesn’t like to be pressed). You need to deflect by about a micron for it to register, but it can also measure the amount of deflection, meaning you get a pressure/z-axis component as well as the usual x and y components of the press.
You can find out more in their release.
posted by Bryon Moyer
Quite some time ago, we reported on WiSpry, a MEMS company that was using its technology to switch capacitors so that the antenna tuning can be optimized and changed in real time as conditions and needs change.
Much more recently, a new solution was announced based on collaboration between Taoglas, who makes antenna assemblies, and Peregrine, who produces an array of digitally-switchable capacitors (amongst other things). They’ve combined the two into a module that can fit into phones and other devices like automobile telematics and patient monitoring devices that have to be small and yet communicate afar. You might think this sounds just like what WiSpry is doing, but, while they’re attacking the same basic problem, their solutions are very different.
Peregrine’s capacitors aren’t actuated by MEMS elements; they’re switched electronically using Peregrine’s UltraCMOS process, which relies on silicon-on-sapphire technology to provide good RF performance. So they’re purely electrical where WiSpry (and also Cavendish Kinetics) is electromechanical.
So which one is better? I asked what the benefit of the electrical version is, and I can oversimplify the answer as being, “We can actually produce ours reliably.” (They didn’t articulate that in a snarky fashion, to be clear… Yeah, I’m sexing it up to keep your attention…) Which suggests, of course, that MEMS makers can’t.
So I asked both WiSpry and Cavendish Kinetics about this; I can’t imagine either one of them saying, “Oh yeah, our production sucks!” even if it were true (and, for the record, I’m not saying it is). But it’s only right to let them respond, so I checked in. Cavendish Kinetics’ Marketing and Biz Dev EVP Larry Morrell said that they have real customer designs in the works, but that they haven’t reached production status yet.
But significantly, he said, “Based on our collective management experience (and the management team has done all this before), we are on a normal yield learning curve for a CMOS process. So we are tracking to our plan and the yields are improving monthly. Our current yield levels are well above minimum requirements to be able to predict fab output to support customers.” Carefully worded; it suggests to me that yields aren’t great today (a threshold of predicting output simply means stable, not high) – but if they can support customers without going out of business, that’s all that matters to customers. They expect production this year and capacity in the 10s of millions per month by the end of the year. [Update note: more clarification on Cavendish Kinetics yields can be found here.]
I did not receive a reply from WiSpry by “print” time.
You can find out more about the Peregrine/Taoglas offering in their release.