Software-defined Radio (SDR) Configuration Pitfalls and Reference Signal Coherence Experiments

Software defined radios (SDRs) are devices which apply digital signal processing to implement many functions traditionally performed by radio frequency (RF) circuits. The availability of relatively inexpensive SDRs, and their ability to facilitate the recording and processing of signals, makes them a valuable tool for research. The ETTUS N210 is a low-cost, but highly capable, SDR that is routinely used by the scientific and engineering communities. As part of an experimental setup, the SDR internal clock is ideally disciplined with a high-quality 10 MHz reference source, which is often derived from global positioning system (GPS) signals. However, when using an ETTUS N210, there are several options for discipling clocks in the event that an external reference signal is not available. During a recent experiment using the BasicRX daughter cards, an outage occurred where the external 10 MHz reference was not available for a period of time. During this time, the SDR had defaulted without warning to using its internal clock source for discipling, despite the software being set to use an external source. An inspection of the data acquired during the outage of the external reference revealed that signals collected during these times had a linear phase drift, while generally retaining favorable coherence properties. The ability to maintain phase coherence between multiple receive channels is particularly important for phased-array applications. Following these observations, the software was modified to interrogate the SDR prior to data collection, in order to verify the detection and use of external reference signals as prescribed. In addition, a series of tests were performed to understand the difference in the coherence properties of signals acquired using external and internal sources. While it was anticipated that confirming external clock state and use would be trivial, this was not the case in practice. After navigating through several software configuration obstacles, invaluable lessons were discovered which would be beneficial to discuss with a broader technical community. After resolving the software configuration issues, a series of laboratory tests were conducted with a controlled input sinusoid fed into two separate N210 receiver channels in different SDRs. To evaluate the influence of the different reference sources, the phase difference between the input signals as sampled by the two ADCs (analog-to-digital converters) on the separate SDRs, was analyzed as a function of time, reference signal source, and SNR (signal-to-noise ratio). It was found that with the external reference source, the difference between the phase of the signals remained largely constant with a time variance that was dependent on input SNR of the RF input test signal. For the internal reference signal, we found that the phase difference changed linearly as a function of time in a manner that was surprisingly well behaved and predictable.