Note The Phase Noise measurement requires license activation. If you install RFmx SpecAn but do not activate the Phase Noise measurement license, you will have 30 days from the first time you use the measurement to evaluate it.

Oscillators are ubiquitous to radio frequency (RF) communication systems. Oscillators are used for frequency mixing during up conversion and down conversion stages in signal generation and analysis, and as reference clocks. The signal produced by oscillators is corrupted by, among other impairments, the phase noise.

Phase noise is the random phase fluctuations in a signal produced by the oscillator. Phase noise increases the inter-carrier interference in RF communication systems that rely on multi-carrier technologies such as Orthogonal Frequency Division Multiplexing (OFDM) [1], Orthogonal Frequency Division Multiple Access (OFDMA) and Single Carrier Frequency Division Multiple Access (SC-FDMA) [2], and thus becomes a figure of merit when evaluating oscillators.

An ideal sinusoidal signal, generated by an oscillator, at a frequency fc, amplitude A and initial phase Φ0, sampled at frequency Fs can be represented as:

Phase noise, L[k], is conventionally expressed as a single side band PSD of Φ[n][4]. An example of a phase noise plot, 10 logL[k] vs k, is illustrated in the following image.

The frequency axis k is usually expressed in a logarithmic scale and hence the phase noise plot is also called as the log plot. The units of phase noise, 10logL[k], is dBc/Hz.

Range Definition

RFmxSpecAn PhaseNoise measurement allows you to configure phase noise offset frequency range by setting the Range Definition property to Auto or Manual.

Auto

In this mode, you can specify the complete phase noise offset frequency range using the Start Frequency (Hz) and Stop Frequency (Hz) properties. The measurement divides this range into multiple decades (sub-ranges ). A separate acquisition is done for each sub-range. This division into multiple sub-ranges is done for two reasons:

  • To get similar number of trace points for each decade when the frequency axis of plot is viewed in a logarithmic scale. A single acquisition for complete measurement range leads to fewer number of trace points for lower offset frequencies and larger number of points for far offset frequencies.
  • To optimize the measurement for speed if the measurement range happens to be wide.

The resolution bandwidth, RBWBin Width(Hz), is specified as a percentage of the start frequency of each sub-range by configuring the RBW(%) property. This property affects the measurement interval and the number of trace points for each sub-range,i, as shown in the following equations.

In this mode, the measurement uses pre-computed number of averages for each sub-range, as shown in the following table.

The averaging counts for different sub-ranges are heuristically selected to strike a balance between measurement speed and quality.

Manual

In this mode, you can configure each offset frequency sub-range by setting the Range Start Frequency (Hz), Range Stop Frequency (Hz), Range RBW Percentage (%) and Range Averaging Count properties for each sub-range.

Averaging Multiplier

The averaging count for each sub-range can be increased by setting the Averaging Multiplier property. This multiplies the averaging count for each sub-range by the factor specified. This setting applies to both the Auto and the Manual range definitions. When the range definition is Auto, the pre-computed averaging count for each decade range is increased by this factor. When the range definition is Manual, the averaging count specified by you for each range is increased by this factor.

Smoothing

In order to extract the underlying trend of phase noise across frequencies, trace smoothing can be performed on a measured log plot trace. The following image illustrates measured and smoothed log plot traces.

The three supported smoothing types are Linear, Logarithmic, and Median that perform linear moving average, logarithmic moving average, and moving median filtering respectively as expressed in the following equations.

The parameter w controls the amount of smoothing, and is derived from the Smoothing Percentage (%) property.

Spot Noise

Spot noise is the phase noise, in dBc/Hz, at a specified offset frequency.

Integrated Noise

Phase noise generates unwanted modulation products, also called as residual noise. Residual noise can be measured by integrating the phase noise log plot trace over a specified frequency range.

In addition to integrated phase noise, PhaseNoise measurement computes Residual PM, Residual FM and Jitter.

Integrated Phase Noise (dBc)

Residual PM (rad)

Residual PM is the contribution of phase noise to phase modulation of the carrier at frequency fc evaluated over a frequency range.

Residual FM (Hz)

Residual FM is the contribution of phase noise to frequency modulation of the carrier at frequency fc evaluated over a frequency range.

Jitter (s)

Jitter is the time domain fluctuation of a signal due to phase noise evaluated over a frequency range.

where,

Spur Removal

The phase noise log plot trace may contain spurs due to presence of non-random interfering signals. The presence of such interference skews phase noise measurement results and therefore should be eliminated from the measured log plot trace. The following image below shows two traces, one with spurs and the other with spurs removed.

The spur removal algorithm detects the spur region on a measured log plot trace and replaces this region with a smoother trace.

Spur detection is based on peak excursion specified by the user and is performed on the difference between the measured and smooth trace, with the peak threshold set to 0 dBm. Refer to the Marker topic for more information on peak excursion and peak threshold.

Phase Noise Cancellation

Phase noise cancellation is used to reduce the contribution of the signal analyzer’s phase noise to the phase noise measurement of the device under test (DUT). A reference source, whose phase noise is at least 10 dB better than that of signal analyzer, is used to obtain a good approximation of signal analyzer’s phase noise, LRef[k]. The obtained trace is subtracted from the phase noise trace of device under test, LDUT[k], to get the post cancellation trace, LPost Cancellation[k].

The following image shows the effect of phase noise cancellation.

The cancellation follows the following relationship.

where,

LRef[k],LDUT[k], and LPost Cancellation[k] are expressed in dBc/Hz.

T is the threshold, in dB, which specifies the minimum difference between LDUT[k] and LRef[k] that must exist before cancellation is performed.

Note
  1. Sufficient averaging should be performed for obtaining reasonable estimates of LRef[k] and LDUT[k].
  2. Same hardware and measurement settings should be used to obtain estimates of LRef[k] and LDUT[k].

Order of Operations

The following block diagram depicts the order of operations used by the RFmxSpecAn PhaseNoise measurement.

Recommended Settings

The following are the recommended settings for the phase noise measurement.

Reference Level

Reference level should be set equal to the peak power of the input signal to utilize the full-scale of the digitizer more efficiently. Setting such a reference level brings the power level of the IF signal at the input of digitizer closer to the digitizer’s full scale.

Supported Analyzers: PXIe-5665, PXIe-5668, PXIe-5644/5645/5646, PXIe-5840

RF Attenuation

The input signal for phase noise measurement is usually a continuous wave signal. This allows for a higher signal power at the input of first mixer in the signal analyzer to get a better dynamic range. Refer to the dynamic range chart of the device mentioned in the device specification sheet. The mixer level follows the relationship shown in the following equation.

For a fixed reference level, the RF Attenuation can be decreased to get a higher mixer level.

Supported Analyzers:PXIe-5665, PXIe-5668

Note
  1. The RFmxSpecAn PhaseNoise measurement sets the IF Output Power Offset as 6 dB by default.
  2. The measured log plot trace may contain discontinuities at the edges of sub-ranges under low Carrier-to-Noise Ratio (CNR) conditions. This behavior can be easily seen when no carrier is present at the RF input of signal analyzer. In this scenario, the acquired complex signal is almost entirely Additive White Gaussian Noise, whose phase is distributed uniformly between [-π,π) radians. Unwrapping this phase is highly prone to errors, leading to discontinuities in measured log plot trace.

References

[1] Wu, Songping, and Yeheskel Bar-Ness. "OFDM systems in the presence of phase noise: consequences and solutions." IEEE Transactions on Communications 52.11 (2004): 1988-1996.

[2] Sridharan, Gokul, and Teng Joon Lim. "Performance analysis of SC-FDMA in the presence of receiver phase noise." IEEE Transactions on Communications 60.12 (2012): 3876-3885.

[3] Cohen, Leon. "The generalization of the Wiener-Khinchin theorem." Acoustics, Speech and Signal Processing, 1998. Proceedings of the 1998 IEEE International Conference on. Vol. 3. IEEE, 1998.

[4] Vig, John R. "IEEE standard definitions of physical quantities for fundamental frequency and time metrology–random instabilities (IEEE standard 1139-1999)." IEEE, New York 1 (1999).