Handset Power Budget

So far in this chapter we have described a number of hardware items that are being
added to the handset. These hardware items each consume significant amounts of
power. The objective with 3G handset design is to keep the overall power budget equal
to, similar to, or, preferably, lower than 2G handset power budgets.
Consider a typical power budget for a GSM phone using the StarCore
(Motorola/Lucent) core processor. The example in Table 4.10 is for a traditional superhet
with baseband, IF, and RF stages.
In practice, if the baseband can be supported on a 0.9-V supply, the call state power
drain can be reduced to less than 40 mW, rather than the 110 mW stated in the table.
Also, as network density increases, the RF PAcan be run at lower power levels (though
sometimes with some decrease in efficiency).

Assuming the overall power budget can be reduced to 400 mW (RF/IF and baseband
receive and transmit on a typical duty cycle), a 700 milliamp hour battery at 3.6
W delivers 2.4 Watt-hours of energy and will support 6.25 hours of use.
Table 4.11 shows how the power budget increases as you add in an image sensor,
MPEG-4 encoder/decoder, and LCD. (A transmissive backlit LCD will consume rather
more than 200 milliWatts; a passive reflective display rather less.)
As a rule of thumb, you can say that adding multimedia/rich media functionality to a
handset easily doubles the power consumption—even before you take into account the
additional RF power budget needed to send and receive all the additional image bandwidth
created by the device. We have reduced 6.5 hours of use to 2.4 hours if the same
capacity battery is used. If we opt for a larger-capacity battery, it may be bigger and
weigh more or need to use a more exotic battery chemistry. Either way it will cost more.
In later chapters we also describe how it is the job of handset and network software
to increase session complexity and session persistency. Session complexity involves
supporting multiple users each with multiple code streams. Because the session has
continuous activity, the duty cycle will be 100 percent rather than the 35 percent more
typical with existing voice exchanges, though the amplitude (bit rate) of the exchange
will be continuously changing as the session progresses.
Both the RF and processor power budgets will need to comprehend this continuous
duty cycle. Batteries will also have to be capable of supplying significant peak loads
(instantaneous bandwidth) and a high peak to average peak-to-mean ratio. The peakto-
mean ratio will be significantly greater than present GSM handsets. This is a problem
for RF PAdesigners, since it is difficult to get an RF PAto run efficiently when it is
lightly loaded. In addition, as session persistency increases, the overall capacity of the
battery will need to increase.