Precision measurement challenges in high-speed, high-frequency systems

Developing equipment for cutting-edge applications like high-throughput satellite systems, non-terrestrial mobile and next generation 6G research, and drone-detection radar requires precision measurements that typically involve complex test setups and long measurement times. A new dual-input spectrum and signal analyzer architecture now promises greater accuracy, easier setup, and faster results, previously deemed impossible.

Engin Kodal, Product Manager Spectrum Analysers and Signal Generators at Rohde & Schwarz

By Engin Kodal, Product Manager Spectrum Analysers and Signal Generators at Rohde & Schwarz

Eliminating Instrument Noise

Mobile and wireless communication systems are adopting wider modulation bandwidths and high-order modulation schemes to achieve higher data rates. However, as the higher-order modulation schemes introduce extra symbol states in the available vector space, the tolerable error vector magnitude (EVM) diminishes. Also, oscillator phase noise scales with frequency and therefore directly impacts EVM performance at higher frequencies. Designers must reduce system phase noise and improve amplifier linearity to prevent high occurrences of decoding errors. In practice, phase noise performance of at least -90 dBc/Hz is needed for Quadrature Phase Shift Keying (QPSK), which is challenging at Q-band and V-band frequency ranges increasingly used for next-generation satellite communication systems. Hence effective optimization of oscillators and synthesizers calls for test and measurement equipment that can accurately measure phase noise and modulation accuracy.

For high-end applications like this, conventional spectrum analyzers typically have excessive internal noise, including wideband noise as well as phase noise from the internal local oscillator. Removing this noise to permit accurate measurements can require expensive phase noise testers. Alternatively, cross correlation is an effective technique that allows noise to be subtracted from the measured signal and can greatly reduce the instrument’s contribution to EVM, as figure 1 shows. On the other hand, performing cross correlation with conventional instruments increases measurement complexity and test costs as two spectrum analyzers are required and additional signal processing is needed for time alignment and cross-correlation calculations.

Figure 1. Cross correlation significantly reduces instrument-related EVM.

Radar applications are also becoming increasingly challenging as demand grows for equipment capable of detecting targets such as drones. These typically have a small radar cross-section (RCS) and cause only slight Doppler frequency shift when flying at relatively low speed, so engineers must lower the radar receiver’s wideband noise floor and improve the local oscillator’s phase noise performance to distinguish the drone’s RCS reflection from background clutter. When testing with a spectrum analyzer during development, applying cross correlation can significantly enhance measurement sensitivity for checking system performance.

Engineers also need to locate small unwanted spurs and interferers in the system that could be misinterpreted as targets. Finding and eliminating these is critical to ensure the radar can achieve the required sensitivity, as the trusted level for detecting real targets would otherwise need to be increased. Without cross-correlation, decreasing the resolution bandwidth (RBW) can lower the noise floor of the signal analyzer although the sweep speed must also be reduced due to the prolonged settling time of the filters. This prolongs the measurement time needed to characterize the system and detect unwanted interferers. With cross correlation, the noise floor can be reduced down to the physical thermal noise limit without reducing the RBW and without adding a pre amplifier, which reduces the overall dynamic range. The improvement in noise floor is 5 x log(N), where N is the number of correlations.

Figure 2 compares measurements performed with RBW of 3 MHz using a Root Mean Square (RMS) detector and a cross-correlated detector, showing how cross-correlation lowers the noise floor. A modulated spurious signal is barely detectable with the RMS detector, yet is clearly revealed after cross correlation.

Figure 2. Cross correlation reveals spurs that are barely distinguishable against the instrument noise floor.

New Input Architecture

By adding an extra input channel, with preselection for both paths, the signal analyzer can suppress spurious signals at the image frequency and eliminate unwanted mixing products. With alignment, calibration and equalization applied to both channels, including the integrated splitter, cross-correlation calculations can be performed in the digital backend (figure 3) thereby saving external splitters and additional calibration steps. Rohde & Schwarz has taken this approach with the new FSWX signal and spectrum analyzer, making it the first such instrument to implement two independent input channels with cross correlation.

Figure 3. Dual RF input channels with cross-correlation digital backend.

As well as letting the instrument perform cross correlation internally, the dual-channel input offers further flexible measurement opportunities by routing two RF input ports directly to the receive unit. This can be a great benefit when testing WLAN and 5G MIMO cellular equipment, as well as characterising up converters, down converters, and transmit/receive modules (TRM) in satellite payloads and ground stations for non-terrestrial network (NTN) infrastructure.

A particularly useful application lies in testing phased array antennas for beam steering; a challenge also often encountered in radar applications. The instrument’s aligned and phase-coherent dual input channels let engineers measure the phase between different transmit paths quickly and determine the beam accuracy precisely (figure 4). Testing phased array antennas under modulated conditions, which involves comparing one element with others in the array, is also improved.

Figure 4. Wideband phased array / antenna alignment measurement.

There are other ways to perform measurements like these, such as by using a vector network analyzer (VNA), although the process can be slower as VNAs typically offer continuous wave (CW) swept measurements or narrowband modulated measurements. Oscilloscopes and PXI modules could also be considered. However, oscilloscopes tend to have insufficient dynamic range. Although PXI modules have a wider dynamic range despite similar LO noise, neither instrument typically applies preselection, which tends to lower dynamic range due to noise and spurs at the image frequency. This can be difficult and complex to overcome.

Compare and Characterise

A multi-input signal and spectrum analyzer like the FSWX can also help engineers overcome limitations of conventional test instruments when using digital RF memory (DRFM). While well-known applications include radar jamming, spoofing, and false target generation in electronic warfare, DRFM’s inherent high speed and fidelity also facilitate hardware-in-the-loop radar testing, channel emulation, and protocol testing. With two time-aligned input channels, users can capture input and output signals simultaneously, to analyze phase and amplitude in real-time and thus quickly characterise the signals.

Figure 5. Phase and time aligned dual RF input channels allow input and output comparison for EW systems in both frequency and time domain simultaneously.

In addition, by incorporating filter banks with independent high-pass and low-pass filters that cover a wide frequency spectrum, the FSWX can perform broadband IQ analysis more quickly and with less distortion than conventional spectrum analyzers that use swept narrowband YIG filters for preselection. While YIG filters are effective for detecting extremely small interference signals, filter banks can have bandwidth of up to several GHz as well as a flat frequency response. This lets the system accurately resolve IQ data for tests such as modulation analysis or EVM measurement. These filter banks can also facilitate spur searches. By also including optional YIG filters, the analyzer can ensure best sensitivity in tests that demand greater dynamic range.

Multi RF Input and Flexible Triggering

Because the FSWX can handle multiple inputs, different frequencies can be set for the local oscillators and various levels of gain or attenuation can be set at the different input. It is possible to measure on one frequency while triggering on another, which can help investigating the effects of short out-of-band bursts on a radar signal, for example. Moreover, when characterising components such as amplifiers or up-/down-converters, the ability to measure two signals simultaneously lets users directly compare IQ data captured at the input and output.

Conclusion

As applications that demand high-speed, high-accuracy wireless and radar systems become mainstream, engineers need faster and more straightforward ways to eliminate the effects of instrument noise when testing components and subsystems. Migrating multiple input channels and cross-correlation, until now only found in complex test systems, directly into test and measurement equipment helps simplify test setups and reduce measurement time.