For years NXP has had an integrated platform solution for mobile phones. Now it wants a single chip mobile phone. What makes that possible is 45nm process technology. NXP's single chip aspiration is a success, many years ago, in the TV business.

"In the old days if you opened up a TV, there would be zillions of components", Rene Penning de Vries, chief technology officer at NXP, tells Electronics Weekly, "in the early 90s some of my colleagues set out to make a single chip TV. No one thought it was possible. Then three or four years later, we had engineering samples and two years later we went into production and, two years later, we had 60 per cent of the market."

A single chip mobile phone will have to support all the generally used telephony standards, past and current, and all the generally used data communications standards, past and current.

"We think our engineering talent is so fantastic, both in our company, and in the industry, that with time, with ambition, and with perseverance, it's going to happen," says de Vries.

Will it require collaboration? "Of course," replies de Vries, "we will need to massively collaborate."

NXP intends to be first semiconductor vendor to come out with an integrated chip supporting all the cellular standards up to LTE (Long Term Evolution) and all the legacy standards, plus WiMAX, in the second half of next year. What makes it cost-effective is 45nm CMOS process technology.

"In terms of RF CMOS on 45nm, TSMC doesn't make it publicly known, but we are number one," said de Vries, "I don't know of any company that is pushing the envelope so much in terms of integration of RF functionality."

NXP divides the RF section of the chip into three sections: a low data rate section for NFC, Bluetooth, Zigbee, Wibree and UWB; a reconfigurable mid-data rate section for cellular; and a reconfigurable high data rate section for WiFi, WiMAX and LTE.

The 2008, 45nm chip will have 2G, GSM, GPRS, EDGE, 3G, 3.5G, TD-SCDMA, HSDPA, HSUPA, LTE, Bluetooth, FM radio and WiMAX.

An internal NXP debate is on which other data connectivity standards, WiFi, Zigbee, UWB etc, to add in as well.

De Vries reckons the chip would need only two radios. "I think, conceptually that could be done. The broadband would be the LTE and the WiMAX, and all the other guys are more narrow band."

When it is pointed out that Qualcomm was TSMC's lead partner on 45nm development and is already producing 45nm chips, de Vries replies: "Our friend in San Diego is mainly digital in cellular. It's what you put on that plain vanilla CMOS that determines where you are in the marketplace. If you add special functionality, such as RF-CMOS functionality, then suddenly it becomes a more complex, but also a more value-adding enterprise. Our ambition is to integrate various functions so we have a single chip solution."

Have you solved the problem of interference? "At the moment it is solved by physical separation," replies de Vries.

In a current phone, the chips supporting different standards are located as far as possible from each other on the PCB.

What happens when everything is on one chip?  "That is the issue everyone faces if they want to integrate various RF functions on a single chip," replies de Vries, "sometimes it is not so severe, when you're not having two functions at the same time. But Bluetooth is almost omnipresent.  So, for sure, you need to focus on Bluetooth and other standards. That is a big problem. That is a technical problem. It has to do with design, with the layout of the components, and with the architecture of the chip."

How far is NXP from finding a solution? "What we are facing at the moment is the cohabitation of Bluetooth and FM radio", says de Vries, "we are pretty strong in FM radio, and we have FM radio module in every phone we sell in this market.

Now how people listen to FM radio is through earphones with a wire you have to connect to your phone. That's a pain in the neck. You don't want it. You'd rather have a Bluetooth connection between the headphone, and a small earplug, and you can then dispense with the cable. If you do that then FM radio and Bluetooth have to operate simultaneously."

How long does he think it will be before NXP can do it? "In the second half of next year we will be able to do it on 45nm RF CMOS from TSMC", responds de Vries.

Is NXP thinking of doing the front end? "If you take apart one of the most advanced phone which has all the cellular functions, it may have Bluetooth, FM radio, GPS, you will see in the layout of that phone there are as many antennas as there are pipes. That is a headache. That is a puzzle to be solved."

Are you addressing that problem at NXP? "I wouldn't want to make any comment," says de Vries, "it's something we have in the back of our mind, but it would not be appropriate to be specific. It's certainly not high on the agenda of our productisation plans, but we have some ideas."

Is there a possibility of a collaboration on that? "Yes that's a likely way forward," adds de Vries, "the front end is not about components, it's about modules and there are a few big module-makers in the industry. More than likely, the key to SDR is the combination between the module makers and the more CMOS-oriented baseband companies."

A possible partner on the front end is BitWave of Boston which has an alpha product using software to control analogue RF channels configurable from 700MHz to 4.2GHz for protocols with bandwidths from 25kHz to 20MHz. Beta product is due for Q2 2008.

Another early addition to the single chip capability will be GPS which is why NXP Semiconductor bought GPS fabless semiconductor specialist GloNav last December.

NXP's internal approach to GPS was software-based, and NXP decided they needed a hardware-based solution to cut down on the MIPS, and on the time to first fix. GloNav already had an RF CMOS single chip GPS chip. This year NXP intends to incorporate the GloNav solution on the Nexperia platform, and next year to incorporate it into the cellular baseband.

How does he see the battle between LTE and WiMAX for 4G? "It's not either or. There will be reasons to go for WiMAX, and reasons to go for LTE. But it would be great for the industry, and the world, to have one standard."

But he's not optimistic. The deciding factors are, reckons de Vries: "Politics; the ability to re-use the baseband infrastructure because of the massive investments in baseband infrastructure; and spectrum efficiency. But I doubt if the deciding element will be spectrum efficiency."

Have the lawsuits over IP in the telecoms world, mostly involving Qualcomm, had an impact in Europe?

"It's a reflection of the fact that IP ownership in our industry is getting more and more important," says de Vries. The semiconductor industry has been very relaxed on IP ownership, and that is changing massively."