Industry view: Strategy Analytics on the RF market
7/17/2012 5:03 PM EDT
KL: What are the challenges on the analog side?
high frequency analog bits, you have to worry about the drive current
and noise performance. You need low noise amplifiers, for example. As
they make the transistors smaller and smaller, you lose drive
capability, you're running at lower voltages, you have higher noise you
have more problems with margins. It's not that it can't be done, it’s
just really challenging and RF doesn't really scale the same way that
KL: What major trends do you see?
seeing more and more digital content in the transceiver. We’re not yet
at the point where we can do an analog to digital conversion at the
antenna but we’re getting close. On the receive side, it's feasible to
do almost everything in digital but on the transmit side you have power
amplifiers where you have to match the output of the transceiver to a 50
ohm high-power environment.
You also need to protect the
receive side from the transmit side. Typically, for the receiver, you
need to knock any interfering signals from the transmit side down to
-130 to -140 dB. That means you need filters with very specific RF
frequencies corresponding to the different bands and very steep skirts.
This gets to the challenge of the rest of what's in the handset—not the
chipset but the RF front end. In a typical phone now you have eight or
more bands, and eight bands is going to go to 16 with LTE. You’ve got a
lot of different frequencies and you need to maintain coexistence and
prevent interference, what Steve Mollenkopf of Qualcomm calls self
KL: Now, when you’re doing this much
filtering the first thing I think of is loss. No matter how good your
filters are, you're losing some your signal.
you have insertion loss for the filters, which means you have to crank
the power up and there's another trade-off—if you put all these bands
into the phone with all the filters and switches your battery life will
take a hit. You can make a very simple handset with a single band that
does nothing but voice calls or you can put together a modern smartphone
that does everything but bake quiche and then you're going to need a
KL: And the losses generate heat, and then you’ve got a thermal management problem.
operators and the OEM handset makers don't always have full
appreciation of the limits. They’ll go to the chipset and power amp guys
and say, ‘We want X, why can’t you deliver this?’ and it’s physically
impossible. I don't have any statistics on this but certainly we hear
all the time anecdotes from operators that users complain that their
handsets just get too hot.
KL: What’s on the horizon for competing technologies?
far as doing D/A on the transmit side, I don't think they've figured
out how to do well. There’s a lot of research going on in so-called
digital PAs but those are really anywhere near feasible or practical
yet—if they can be done, they can’t be done cost effectively. On the
receive side, we’re not there yet but the A/D conversion can be done
fairly close to the antenna, usually after the low noise amplifier. If
you do put a converter right at the antenna, with today's technology it
just draws too much current. You could do it in a base station if you
wanted to but it's not really practical today for handsets.
KL: What keeps your customers up at night?
about where they should spend their R&D dollars to stay
competitive. Some of things we’re looking at specifically in handsets
are multimode, multiband power amplifiers. Those are power amplifiers
that can handle more than one air interface at once so instead of having
a separate power amp for edge and wideband CDMA, you can now have a
power amp that can do both.
When you're talking about a handset
that has eight cellular bands, until recently that required eight
separate power amplifiers which takes up an awful lot of space. If you
collapse that into let's say two—one for the high bands and one for the
low bands—then that's pretty attractive to the handset makers.
KL: How much activity is going on there?
December 2011, we estimated that shipments of multimode PAs were around
$140 million in 2011, which is really significant considering that the
year before they were maybe only $25 million. It’s a hot area right now
and envelope tracking will be a part of it. I’m working on a report now
on envelope tracking. Most people in the cell phone industry agree that
this is coming sooner or later and it's a way to improve efficiency of
the power amplifiers. It's something that will probably offered by the
chipset makers but it has to work with PAs, so it's got to be
cooperative. The chipset guys generally call the shots but the PA guys
will have to develop and market power amplifiers that work with whatever
the chipset guys do.