Sample Rate

DPD measurement provides two modes for configuring the measurement sample rate, namely Reference Waveform and User.

Reference Waveform mode uses the sampling rate of reference waveform for computation as well as application of DPD lookup table or polynomial. Measurement skips software resampling throughout signal processing chain.

User mode allows you to configure custom sample rate. Measurement enables software resampling before computing and applying DPD lookup table or polynomial.

The minimum recommended sample rate, expressed in samples per second, is given by:

where,

B is the bandwidth of the signal in Hz,

and K is the highest order of PA non-linearity

Synchronization Method

For computing the gain and phase characteristics of the DUT, DPD measurement requires reliable time-synchronization between the acquired and reference waveforms.

DPD measurement provides two synchronization methods, Direct and Alias Protected.

Direct

This method synchronizes the acquired and reference waveforms with the assumption that the sample rate is high enough for preventing aliasing in intermediate nonlinear operations during the estimation of sub-sample delay between the two waveforms.

Alias Protected

This method synchronizes the acquired and reference waveforms while also ensuring that intermediate nonlinear operations during the estimation of sub-sample delay between the two waveforms are not affected by aliasing. This synchronization method should be used for the following test configurations:

  • signal is either single-carrier or contiguous multi-carrier, and measurement sample rate is less than approximately three times the occupied-bandwidth.
  • signal is non-contiguous multi-carrier and the component carriers are sparsely allocated in frequency.

This method uses the offset and bandwidth information of all component carriers comprising the signal. If you set the DPD Auto Carrier Detection Enabled property to True, the measurement auto detects the carrier offsets and bandwidths. If you set the DPD Auto Carrier Detection Enabled property to False, you must configure the carrier offsets and bandwidths by setting the DPD Number of Carriers, DPD Carrier Offset, and DPD Carrier Bandwidth properties.

The following diagram illustrates the offsets and bandwidths of the component carriers in a multi-carrier signal with a sample rate of FS(S/s).

Target Gain Type

The average gain of the predistorter and the PA, considered together as a linearized DUT, can be controlled by tweaking the gain of the predistorter. The following figure shows a simplified representation of the signal processing chain where PA is preceded by a predistorter.

DPD measurement supports three target gain types, namely Average Gain, Linear Region Gain and Peak Input Power Gain.

Average Gain

The predistorter is computed such that the average gain of the Predistorter + PA System is equal to the average gain of the PA.

Linear Region Gain

The predistorter is computed such that the average gain of the Predistorter + PA System is equal to the linear region gain of the PA.

Peak Input Power Gain

The predistorter is computed such that the average gain of the Predistorter + PA System is equal to the gain provided by the PA to all the samples in the reference waveform with power levels in the vicinity of the peak power level in the waveform.

The figure below shows the expected gain vs input power plot after DPD for various target gain types.

Threshold and Curve Fit

NI recommends that the Log LUT-based DPD use the appropriate threshold settings to reject low-power, noise-corrupted samples. Polynomial curve fit order should typically be between 3 and 10. The recommended polynomial curve fit type is Least Absolute Residual. Use appropriate threshold settings to reject low power noise corrupted samples. The default threshold is set to -20 dB relative to the peak power of the acquired waveform.