MENU

Will Apple drive analog ICs?

Will Apple drive analog ICs?

Feature articles |
By eeNews Europe



But Apple’s extraordinary ability to get semiconductor suppliers to develop new devices for them — in fact, its ability to swallow entire companies without even belching — suggests Apple could influence the analog market for years to come. So it’s worth asking what Apple is currently doing in analog, what improvements would help them and how many of those improvements are manufacturing related.

We know the North First Street facility includes chip manufacturing equipment from Applied Materials, Hitachi, Novellus, and ASML. The fab will produce roughly 7,000 eight-inch wafers a month at geometries from 0.6-micron down to 90nm. Note: The sweet spot for analog manufacturing is still in the 0.35- to 0.18-micron range.

We also know that Apple has been buying an estimated $2-$2.5 billion a year in analog parts for its phones and tablets. The bulk of these parts include custom-made power management ICs (PMICs), less-custom audio codecs and a variety of sensors including motion sensors and touch-sensitive screens. If we were placing bets on which of the three part types Apple will tweak in the new facility, power management looks like the low-hanging fruit. No matter how cleverly crafted, the analog parts can turn into multi-sourced commodities. But shrinking power management functions remains a challenge.

With its Haswell-generation processors, for instance, Intel attempted to integrate power management functions onto the CPU. Its goal was to shrink the amount of space the PC motherboard devotes to power and cooling, and thus enable new miniaturized PC form factors. But machines that move dozens of amperes around are not easily shrunken, and Intel reverted back to more traditional Vcore regulator architectures.

In the case of mobile devices, the PMICs are custom-configured for each phone or tablet, and — depending on the feature set of the phone, can be quite complicated. There are as many 26 or 28 separate devices on one chip, including two or three 300mA switch-mode regulators, 22 or 24 low-drop out regulators (LDOs), a lithium-Ion battery charge monitor controller, and several LED backlight drivers. Apple uses Dialog Semiconductor as the supplier for these parts.

It takes a lot of hard work, rather than any special tricks, to do this integration: You want the LDOs and other voltage controllers to sequence devices on-and-off (or to clock them down) in response to commands from the cell phone applications and baseband processors.

The BiCMOS or BCD processes used to implant power transistors on a CMOS substrates are now well-known — even among the Asia-based foundries that needed to unlearn memory manufacturing to serve analog clients. The power transistor implants buffer the power sources (batteries or AC adapters) from the CMOS control logic fabricated in 0.18- or 0.13-micron CMOS. We bet money that the fabrication facility Apple acquires from Maxim includes a transistor implant mechanism.


Analysts are speculating that the new fab will support advanced sensor prototyping. Apple spends about $750 million on motion sensors each year. They include footstep counters that support location-based services, tracking your movements with navigation aids and ad servers.

Interestingly, sleep researchers are the University of California at Berkeley claim that, by itself, an Apple iPhone has enough built-in sensor sensitivity and resolution to monitor your sleep from under your mattress. In this way, wearable bands from companies like Jawbone, Fitbit and a variety of smartwatches could be redundant.

While Apple has likely worked its sensor vendors pretty hard, especially on pricing, I’m not aware of any recent breakthroughs in the underlying technology.

There is on-going research in MEMS sensor manufacturing, intended to reduce dependency on bulk chemical etching. In the manufacturing of MEMS, you are typically digging a trench or a tunnel or a hole in silicon — a cavity for micro-miniature moving parts to move around in — and then sealing the hole against moisture and other environmental contaminants.

MEMs companies such as InvenSense and mCube have come up with interesting ways of doing this. To date, Apple has used parts from Bosch, InvenSense and STMicroelectronics. In any case, the goal here is to reduce costs by eliminating manufacturing steps. It does not fundamentally change what the sensor does. And, with the proliferation of MEMS specialty fabs, this a problem Apple probably does not need to solve with its own R&D facility.

More interesting problems with RF MEMS may call for a prototype factory. Increasing the number of insertion cycles for miniature antenna switches might represent a challenge for Apple’s engineers, but it’s a 20-year old problem with initial R&D funded by DARPA in 1996. If anyone can get RF MEMS off the ground, analyst Tony Massimini of Semico seemed to imply in one report, it could be Apple. And this facility may be the place to do it.

We suspect there may be other sensor challenges the company is looking at. For example, Apple might be interested in research on the high-resolution gas and chemical sensors using spectrometry or other wavelength sensing semiconductors.

There are plenty of established and emerging sensors to explore. Among other possibilities are pulse oximeters, already used in the Apple Watch, skin hydration sensors and EKG probes. One company is developing immiscible semiconductors that can read the chemical composition of liquids on a molecular level, and another has an infrared spectrometer provides molecular-level analysis of foods, medicines, and fuels.

While the success of these devices will likely depend on specially-tuned semiconductor manufacturing, it is hard to tell how much Apple will participate in their development. Applications interfaces like iHealth make it easy for sensor developers to attach their devices to the iPhone. But there remains a question as to what extent the company participates in the development of these attachments. This is a very big question, as Apple could be the driver for a lot of analog product development.

About the author:

Stephan Ohr is Consultant, Semiconductor Industry Analyst.

If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :    eeNews on Google News

Share:

Linked Articles
10s