posted by Bryon Moyer
Continuing with the series of Sensors Expo conversations, I had a chance to discuss a couple of topics with Memsic (one of which we’ll talk about in a future entry). Today we’ll look at their magnetic sensor, which relies on AMR – anisotropic magneto-resistance.
Currents generate magnetic fields, but with AMR, the difference in direction by the field generated by current and some external field will impact the resistance of the material. Current is generally run at an angle in stripes of material, resulting in a so-called “barber pole” look. The good news about this technology is that it can read quickly – about 7 ms.
One of the challenges of any magnetometer for use in a compass is that you’re trying to sense the direction of a 200-mG field in an environment of magnetic fields that can be on the order of 4-16 G. So you have to “center” the measurement so that you can detect this small signal within the environment that threatens to overwhelm it. This centering, on the other hand, gives you more flexibility on where to place the magnet, since proximity to large anomalies is less detrimental.
In a new 2D mag sensor recently released, they’re targeting things like watches, so power has to be conserved. The problem is that the magnetized material in the barber pole can gradually lose its magnetization as dipoles lose their alignment. So you need to realign things occasionally – sort of like running a comb through it to straighten it all out.
Now, you could do that automatically, but it takes power. So instead, they have a set/reset function that sets the field and then reverses the field. This lets them re-center the zero point in addition to refreshing the magnet. But this has to be done manually (although I suppose it could be handled by the system integrator – perhaps hitting the “compass” button could first execute a refresh before measuring, which the user would never know).
The other thing they’re doing differently with this device is giving it a longer lifetime. Phones come and go, and historically, mag sensors have had commensurate lifetimes. But for non-phone system makers, it can be frustrating to evaluate a sensor and, when you’re finally ready to go into production or when you’re extending production, to find out that it’s no longer available. So this device will be kept around longer than typical fleeting phone lifetimes.
You can find out more in their announcement.
posted by Bryon Moyer
Chip design has always consisted of a series of loops. Do something, check the effects, fix things, check again, and hopefully converge on a solution. A big part of the focus of EDA tools developers has been to make each of these passes faster and reduce the number of passes.
One of the critical things that has to be checked at the end of each layout pass is that the layout meets the design rules. A couple years ago, the DRC-checking part started moving to real-time with Mentor’s Calibre tool, which dominates DRC checking. It started with Mentor’s own InRoute (for digital) and then RealTime, for integration into custom layout tools like Laker (was Springsoft, now Synopsys). This meant that DRC rules were checked immediately with each layout change, eliminating one aspect of the long loop.
Not comprehended in that, however, was the electrical aspects of layout – the Rs and Cs (mostly parasitic) that accrue as you lay your chip out. Cadence has just announced a change to that. They call it “electrically-aware design,” and it moves the extraction and parts of verification from the end of the loop to “real-time.” You can feed forward voltage/current points from circuit simulation and monitor as you do layout; you can establish constraints and track adherence; you can get warnings when something you’ve done in layout creates an electromagnetic issue. You push a polygon and the tools recalculate the parasitics and update the performance numbers immediately, alerting if necessary.
The big win here is that it allows designers to “converge by construction” instead of doing an entire layout and then finding all the issues. It also lets designers push the edge a bit more. If you’re tight on your schedule (who isn’t?), then you might over-design to get things to pass – you’re not then going to go back and “back things off” until they fail in order to optimize since that will take too long. But with the real-time view of the impact of layout, you can see if you’ve over-designed and then make immediate adjustments to achieve a better balance.
It’s a simple concept with interesting potential for custom and analog designers. (And if you’re wondering about real-time DRC, Cadence already has that in place as well.)
You can find more details in their release.
posted by Bryon Moyer
A couple of years ago I was stunned to find myself jazzed by batteries. I mean, how mundane can you get? A battery is a battery is a battery, right? And we keep hearing that there are needs for major breakthroughs to enable better electric cars and such, and yet those breakthroughs haven’t been obvious. Probably because they haven’t happened; development has been incremental, not a step function.
But something about Infinite Power Solutions’ story grabbed my curiosity and took me into the realm of thin-film batteries. Just to jog your memory, these guys made thin, flat, solid-state batteries – the rough size of a postage stamp. And they were pushing the limits of storage; it seemed like exciting stuff.
More recently, I was doing some battery-related stuff and wanted to contact them for comment. Ominously, they had a one-page website. It didn’t say they were dead, but… well, there are only two times when the website has one page: before launch and on the way out.
During Sensors Expo I had a conversation or two that confirmed IPS’s demise. Doors were closed. Not sure if anyone picked up the technology.
Now, this certainly wouldn’t be the first time some brash new technology company didn’t see its way through, and I don’t necessarily spend a lot of time covering folks that disappear. Prognosis is more fun than autopsy.
But I couldn’t help wondering, sort of like you wonder about those wonderful friends of yours that suddenly and without warning announce they’re getting a divorce, what happened? Things seemed so promising. Did the technology simply not pan out as expected?
I don’t have complete answers, but the things I heard had nothing to do with the fundamentals of their technology. It was about the side details. Like people being able to poke into it or bend it and create internal shorts. Or the fact that they had to be manually mounted on boards; they didn’t work well with pick-and-place automation.
If these are the things that made the difference, then, well, it’s a damn shame. Just like it is when any intriguing idea gets sidelined due to such “trivialities.”
It’s a tough call to make when you’re bringing out something fundamentally new and exciting. It’s not unusual for there to be these little nagging side issues, and the temptation is to, if not outright ignore them or pretend they’re not there (typical when the boss simply wants it to work and doesn’t want to hear about annoying details that might get in the way), then to assume that they’ll sort themselves out along the way.
And many times they do sort themselves, but it’s hard to know that in advance. In this case, if the tidbits I got were true, then they didn’t sort themselves, and they got in the way of adoption. Were they unresolvable? Did IPS simply not pay enough attention or focus on those issues? Who knows. It’s history at this point.
But it’s a reminder that it’s worth sweating the details. And if you think you have the luxury of ignoring a few annoying issues or deferring their solution, then I’ve always felt that, rather than pretending they’re not issues, everyone needs to agree that, yes, these could be issues, we don’t think they’ll be roadblocks for now, so we’ll move forward. But it also bears thinking through, what if we’re wrong? What’s Plan B?