SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : InvenSense (INVN) -- Ignore unavailable to you. Want to Upgrade?


To: Taylor Mill who wrote (174)11/4/2013 2:00:57 PM
From: brokenst0nes  Read Replies (2) | Respond to of 351
 
Agree does seem to be a mismatch with the high-end tier, however, I come back to what I wrote before, it really depends on what the input data is for the algorithm and therefore where it could run, certain cases will just need motion sensor data, while others could need data from non-motion sensors. It is also unclear to me whether the application processor(AP) would be on or off when certain algos are in play, seems some sensors would be paired with AP in all cases. Have a look at global.samsungtomorrow.com and think whether all those sensors would be active when the screen is off along with the AP.

Behrooz had the following to say on the last call

Well, actually, the sensor hub, as I talked about it before, it really highlights the complexity of fusion and calibration. You can either use a sensor hub with raw sensor with some of our competitors supply and try and do the calibration within the sensor hub or you do it within the sensor itself, within the sensor SoC, which is what we have. So I think if you look at the products that we ship today, the MPU-6500 family platform, we have basically the functions of calibration and fusion built in into the DMP with our software algorithms. And what you're going to see soon is as operating systems come out that have specs around doing more of a sensor processing closer to the sensor, you're going to see products from us that have that. So it's not quite the large sensor hub that some of these phones use in a very, very high-end, and there's really 1 or 2 phones, there are 2 products that I can say that have the sensor hub. The majority of them really are going to rely for cost and power reasons. They're going to rely on the more intelligent sensor SoC, which is why we're getting really significant traction in the customer base. In the last quarter, we have seen an accelerated design win based on the software algorithms, based on the fusion and calibration, and these are things that customers used to do or want to do in the sensor hub and now they're able to do it in our chips. So you'll see some product announcements, hopefully soon from us that will reflect that.

Splitting hairs a little, but Galaxy Note 3 and S4, and of course the latest iPhones make it more than 1-2 already.

Re the other sensors whether InvenSense is likely to expand, personally I will be disappointed if they don't expand into providing their own pressure sensor and magnetometer, even if they are implemented in multiple pieces of silicon.



To: Taylor Mill who wrote (174)11/9/2013 1:22:27 PM
From: GPS Info  Respond to of 351
 
Taylor,

On the other hand, no company is likely to supply everything.

Clearly, this seems to be the case. I’m very interested in discovering those companies that can support a set of consumer applications using MEMS sensors. A smart company would go after the largest markets with the fattest margins.

What do we have in smart phones now? As far as I can tell in the consumer market, there are MEMS gyros, accelerometers, magnetometers and some biometric sensors. I haven’t forgotten the microphone market, but this seems fairly narrow for now. We can only guess where the other half-dozen sensors will fit into the market in the distant future. In my view it would be foolish of InvenSense to spread themselves so thin as to lose focus on the current market.

Clayton Christensen suggests that in the early stages of technical developments, it is often necessary to tightly integrate different modules to maximize performance. The integration must be done by a relatively small group of experts, as opposed to doing it across an industry. Once the technology has matured enough, then different players can offer swappable modules. I suggest that this could take a decade for MEMS. I strongly believe that we are still in the early stages of development and that someone must provide the required performance and this will be done through a very tight integration using proprietary software. The big question for InvenSense will be if they can outperform other in-house solutions. I will make by bet and take my chances.

It's a big market --- I'm still confident that INVN will be a key player as they get this all sorted out.

I think INVN will have some success in the integration of MEMS sensors. They will need to hire some young, smart engineers to make it all happen. We will need to stay alert in the months and years ahead.

Regards