Preliminary Technical
Data
Rev. PrA | Page 43 of 82
These peaks increase the overall dynamic range needed for the OFDM signal through a signal chain, leading to an increase in the
Peak to Average Ratio (PAR) of the signal.
Modern communication PAs used to amplify such OFDM waveforms are only linear for a certain power range. Most of the input
signal (average power) will be within this linear range. However, the signal may have peaks that exceed the PA’s linear operation
range. To avoid saturation of the output signal due to these peaks, we could potentially attenuate the desired signal. This method
ensures that the range required by the signal is within the PA’s linear range. However, this is undesirable as it would reduce the
average power at the expense of maintaining a given PAR, making the system less efficient. An alternative to attenuation is to use
CFR – where instead of attenuating the whole signal, we attenuate portions of the signal that are above the PA’s linear range. This
results in a constant output power while reducing the PAR and thus ensuring that the signal remains within the PA’s linear range.
This is summarized in
Figure 52. Effect of Signal Attenuation vs. CFR
It is important to note that CFR leads to higher in-band and out-of-band noise levels. This effect results in Error Vector Magnitude
(EVM) degradation while also increasing the noise power spectral density, resulting in increase in Adjacent Channel Leakage
Ratio (ACLR). It is important to optimize the CFR algorithm to make sure that the CFR block’s impact is within the customer’s
system level specifications (derived from 3GPP specifications or other regulatory standards).
CFR ALGORITHM OVERVIEW
Several CFR algorithms and implementations exist in the literature – only a few have been commercially viable.
Clipping and
filtering
and
pulse cancellation
are two of the more popular techniques. Even though optimal cancellation of peaks is technically
achievable (target PAR requested is met exactly), the latency requirements imposed by modern communication standards makes
the design of a real-time CFR engine quite challenging. Spectral regrowth of corrected peaks by interpolating stages in the
datapath, such as interpolating filters and Digital to Analog Converters (DACs), is also a concern. An ideal CFR block has very low
latency and zero missed peaks.
The ADRV9029 implements CFR using a variation of the
pulse cancellation
technique by subtracting a pre-computed pulse from
the detected peaks to bring the signal within the PA’s linear range. The CFR block consists of three copies of CFR engines, each of
which uses a detection threshold to detect the peaks and a correction threshold to which the detected peaks are attenuated. The
pre-computed pulses may be stored within the device at the start-up or be updated during run-time. These spectrally shaped
correction pulses are subtracted from the data stream to bring the signal within the PA’s linear range. The correction pulse needs
to be spectrally shaped to manage the noise leakage into adjacent bands.
shows the architecture implemented in each of
the CFR engines.