Customer Solutions

Measuring Amplifier Spectral Purity/Phase Noise Using NI LabVIEW and PXI

Author(s):

John Payne, Niksar Australia Pty Ltd.

Industry:

Telecommunications

Product:

LabVIEW, PXI/CompactPCI

The Challenge:

Measuring the single-sided noise sidebands of a high-powered HF amplifier system with a frequency resolution of 1 Hz and resolving spectral levels down to -110 dBc.

The Solution:

Using National Instruments LabVIEW and PXI to generate a reference signal to drive the amplifier, capture the amplifier output signal, and compute the required spectral information for quick and efficient measurements.

At Niksar Australia, we specialize in RF test and measurement systems. We recently required a system that could efficiently and quickly measure the sidebands of a high-powered HF amplifier system (spectral purity and phase noise) with a frequency resolution of 1 Hz and resolve spectral levels down to -110 dBc.

Inexpensive, High-Performance System with LabVIEW

We knew this would require a system with high dynamic range and low effective noise sidebands close to the carrier signal. Using the NI LabVIEW environment, we engineered a system that met these requirements. For the system processor, we used the NI PXI-8350 rack-mount controller with an NI MXI-4 interface. Our system also included a 3.0 GHz Pentium 4 processor with hyperthreading technology and 2.5 GB of RAM and an NI PXI-5122 high-resolution digitizer with 256 MB of memory.

To make sure the amplifier was suitable for applications requiring sensitive amplitude, phase, and signal frequency characteristic measurements, we measured the spectral noise that contributed to an amplifier unit under test (UUT) in regions close to the main carrier signal frequency. In the past, we used expensive noise-measuring test equipment for this purpose. However, with the NI PXI-5421 arbitrary waveform generator and the PXI-5122 high-performance digitizer, we achieved the necessary performance level.

To measure signal characteristics to a 1 Hz resolution required capturing a minimum of 1 second of signal data. When operating at high frequencies, we had to avoid aliasing errors and artifacts by setting a sampling regime that gives a sampling rate at least twice the upper frequency of interest, and generally higher than that for a margin. In transforming from the time domain to the frequency domain, we typically use a weighting window to reduce the discontinuity effects between the start and end of the series. This introduced unwanted frequency domain artifacts and reduced the frequency resolution.

To avoid the artifacts of time series windowing, we had to carefully select the carrier frequency, sampling rate, and number of carrier cycles to obviate the need for applying time series weighting. We required strict adherence to capturing an integer number of carrier cycles as well as to maintaining a sampling regime where there is an integer number of samples per carrier cycle.

To reduce the instrument noise-level contribution, we had to maintain coherent signal generation and sampling to minimize instrument-generated noise and eliminate differential clocking drifts.

Signal Processing

After capturing signals, we had to manage the amount of data to minimize the use of processing resources (memory and CPU time) and to derive the required frequency spectrum with the desired high resolution. We achieved this by initially processing the captured data block by block to reduce the amount of data being transformed. We then efficiently conducted final data processing and normalization to obtain the desired frequency spectrum for testing against the specification limits. We readily obtained an analysis bandwidth of 7,500 Hz with respect to the carrier.

We then applied spectral averaging to arrive at a smoothed (average) spectrum. During the averaging process, for each data acquisition, we measured and plotted the relative gain and phase stability of the instrumentation units (arbitrary waveform generator and digitizer).

Instrument Performance

Based on the individual specifications for our two PXI modules alone, we could not achieve the required performance. Based on the noise level, spurious-free dynamic range (SFDR), and total harmonic distortion (THD) specifications for these instruments, we found that ~75 dBc would be the best performance achievable. However, by applying sound signal processing practices available within LabVIEW to the signals produced and sampled by the NI PXI modules, we could achieve considerably better performance when examining spectral purity and phase noise characteristics of the UUT.

For more information, contact:

John Payne

Niksar Australia Pty Ltd.

135/45 Gilby RoadMount Waverly

VIC 3149, Australia

Tel: 613 9558 9924

Fax: 613 9558 9927

E-mail: jpp@niksar.com.au