Analog Devices ADRV9001 User Manual page 171

Preliminary Technical Data
Parameter
immediateLutSwitching
useSpecialFrame
resetLuts
Each of these parameters are described in more details as the following:
numberOfSamples
It specifies the number of samples used per DPD data capture with a limit of 4096. In general, the DPD performance could be improved
if more samples are used. For TETRA1, currently, the Linearization Channel (LCH) is not supported so the numberOfSamples should be
set to 4096. When LCH is supported, the numberOfSamples could be changed to a different value. More information will be provided in
the future.
For LTE, the Number of Sample should be set to 4096 without frequency hopping.
outlierRemovalEnable
This feature removes anomalous captured data before DPD computation. Currently it is not implemented and should be disabled.
outlierRemovalThreshold
Since outlier removal is not implemented, this threshold is ignored.
additionalPowerScale
This parameter is used to scale the higher power terms during the calculation of the auto-correlation matrix using transmit data d(n). It
is used to keep the nominal magnitude of each of the power terms about the same to avoid ill condition of the correlation matrix. The
scaling factor
could be defined as
the information through this parameter.
rxTxNormalizationLowerThreshold/rxTxNormalizationUpperThreshold
These are required parameters for the normalization of the magnitude and phase of the Rx and Tx data. These thresholds are used to
pick a linear region for normalizing the data. The region chosen should be below the compression point but above the noise. In the AM-
AM and AM-PM plots shown in Figure 142, a possible choice of the linear region is highlighted in red. In general,
rxTxNormalizationUpperThreshold should be set to 0.5 of the peak signal amplitude and rxTxNormalizationLowerThreshold should be
set to 0.3 of the peak signal amplitude. Once they are set, the threshold values should be fixed and not vary from capture to capture.
Therefore, by knowing the peak Tx signal in dBFS, rxTxNormalizationUpperThreshold should be set to "peak Tx dBFS – 6 dB" and
rxTxNormalizationLowerThreshold should be set to "peak Tx dBFS – 10.5 dB". Peak Tx is usually set at P1dB by adjusting the Tx
attenuation setting.
The user could enter those thresholds in dBFS through TES. If using API, the linear numbers should be used which can be calculated as
10 .
(threshold_dBFS/10)
immediateLutSwitching
When a new DPD solution is formed, the new solution is loaded into a spare LUT, which can then be switched with the active LUT.
There are two options regarding the LUT switching. When immediateLutSwitching is set to be "TRUE", the new LUT swaps out the
active LUT immediately when ready. When immediateLutSwitching is set to be "FALSE", after the updated LUT is available, LUT
swapping is triggered after the next data frame is completed. Note this only applies to TDD operations. FDD systems should always use
immediate LUT switching. Currently, it should always set to be "TRUE".
useSpecialFrame
In order to achieve optimal performance, DPD needs to capture the peaks of the signal. Some standards allow for transmission of special
data that can be used for DPD estimation. In other cases, the baseband processor may know in advance that a frame will contain good
data for DPD estimation. In these cases, the user may choose to control which frames are used for DPD. The flag indicates that the user
will control the frames that DPD can capture. This is currently not supported so it should be disabled.
Type Description
bool Determines whether the LUT
switches immediately or at
the end of Tx data frame.
bool DPD only runs on a user
indicated special frame.
Reset LUTs so that no pre-
bool
distortion is applied.
, where "std" stands for standard deviation. User could measure
Rev. PrA | Page 171 of 253
Min Max Default
TRUE
FALSE
FALSE
UG-1828
Note
FALSE not
currently
supported.
Currently not
supported.
User should reset
LUTs at the start
of DPD operation.
and then pass