Multichannel Combining

The devices just described allow us to multiplex different operators using different RF
channels onto shared RF hardware (the antenna). In addition, we need to separate RF
channels within the base station transceiver. If a single channel is placed through a
power amplifier, the amount of linearity required depends on the modulation used.
For example, constant envelope GSM can tolerate a nonlinearity of >-20 dBc. Nonconstant
envelope modulation typically needs a linearity of <-45 dBc.
As sufficient linearity over multichannel system bandwidths have been difficult
(read: technical and financial) to achieve, a separate PA has been used for each channel
and combining is performed after amplification. Typically two choices of combining are
considered—wideband hybrid combining and narrowband cavity filter combining.
The advantage of a wideband combiner is that it needs no retuning for channel
additions. The disadvantage is that it has high loss. The disadvantage of a narrowband
cavity filter is that it requires retuning with changes in band planning. The advantage
is that it has relatively low loss.
If sufficient linearity is available, such that intermodulation products (even at base
station powers) are dramatically low (<-75 dBc), then combining may take place at low
power, as shown in Figure 13.17 before the multicarrier amplifier. Given sufficient linearity,
even multistandard combining may be used—for example, CDMA and AMPS
or PCS1900, GSM and CDMA1900 in geographic proximity.
Cost and size constraints are making vendors move relatively quickly to linear
amplifier solutions and baseband selectivity. The counter argument is to find a way of
improving the Q of the RF filters used and relaxing baseband processor overhead (for
example, the DSP overhead involved in PApredistortion and linearization). This is one
of the rationales for superconductivity filters.