Phase Noise Results: Choosing the Best Signal Generator

An RF or microwave signal generator’s phase noise performance is a critical factor in determining how well it fits an application. For example, phase noise performance is essential in testing high-performance systems such as Doppler radars or software-defined radios (SDRs). Excellent phase noise specifications are also important when using a signal generator for oscillator substitution or analog-to-digital converter (ADC) testing.

When evaluating an RF or microwave signal generator for use in such applications, several performance factors, such as spurious, harmonics, broadband noise, AM noise, and phase noise, are worth further examination. Looking specifically at phase noise performance, it is affected by the instrument’s internal architecture and the features and capabilities layered on top of that architecture. The most common architectures are single-loop and double-loop, which will be explored later. Available features include digital modulation capabilities, pulse capabilities, and multi-unit synchronization, and their presence can affect phase noise performance.

Phase noise performance includes tradeoffs such as cost, switching speed, and optimization at various frequency offsets from the carrier signal in both the design and evaluation of a signal generator. To address various requirements, some signal generators offer two or more levels of phase noise performance (e.g., standard and optional capabilities). Others allow optimization of phase noise performance at wide or narrow offsets. Still others may allow the user to selectively degrade phase noise performance and observe the effects on the device-under-test (DUT).

This blog discusses the phase noise fundamentals and then examines architectural choices and the effects of various functionality alternatives.

The foundation: stability and noise

Any discussion of phase noise is mainly concerned with a signal’s frequency stability. An oscillator’s long-term stability may be characterized by hours, days, months, or even years. Short-term stability refers to frequency changes that occur over a few seconds or less. These short-cycle variations significantly affect systems that rely on extreme processing to extract more information from a signal. For that reason, this discussion will focus on short-term stability.

Short-term stability can be described in many ways, but single-sideband (SSB) phase noise is the most common. The US National Institute of Standards and Technology (NIST) defines SSB phase noise as the ratio of two power quantities: the power density at a specific frequency offset from the carrier and the total power of the carrier signal. This is most commonly measured in a 1-Hz bandwidth at a frequency “f” away from the carrier, and the units are dBc/Hz or “decibels below carrier frequency power over a 1-Hz bandwidth.”

The phase noise level is deterministically related to the carrier frequency, increasing by 6 dB for every doubling in frequency. When characterizing the performance of components integrated into advanced radar and communication systems, measurements of phase noise for a 1 GHz carrier may extend from roughly -40 dBc/Hz at close offsets (1 Hz or less) down to -150 dBc/Hz at far offsets (10 kHz or more). These measurements will be about 18 dB higher with a carrier frequency of 8 GHz. At such low levels, the measurement noise floor is affected by two microscopic electronic effects: thermal noise from passive devices, which is broad and flat (white noise), and flicker noise from active devices, which has a 1/f shape (pink noise) that emerges from the thermal noise at lower offsets. Both contributors are unavoidable because they are present all along the signal chain: in the measuring instrument, the device that produces the signal-under-test (SUT), and even the cables that connect the two.

Another sometimes overlooked noise source is any amplifier in the signal chain. While the primary purpose is to increase the power level of a weak carrier signal, the amplifier adds its noise and boosts any input noise. The net effect is that amplifier, thermal, and flicker noise combine to give any phase noise plot a characteristic shape and, more significantly, reduce the theoretical lower limit of any phase noise measurement (Figure 1).

Figure 1. The three main contributors to noise create a theoretical lower limit for phase noise measurements

These effects all show up in the phase noise characteristics of a high-performance signal generator. For example, the underlying noise sources can be traced to the major sections of the instrument block diagram (Figure 2). For offsets below 1 kHz, the noise is dominated by the reference oscillator’s performance, which is multiplied up to the carrier frequency. The other major contributors are the synthesizer at offsets of 1 kHz to roughly 100 kHz, the yttrium-iron-garnet (YIG) oscillator from 100 kHz to 2 MHz, and the output amplifier at offsets above 2 MHz. When well understood, these effects can be minimized and optimized within a system design to ensure maximum performance.

Looking at the relationship between phase noise and frequency, phase noise increases deterministically as frequency increases. This is especially true when frequency-multiplying techniques are used inside or outside the signal generator. Small changes in frequency or band often affect the actual relationship.

Figure 2. Contributions to a signal generator’s phase noise performance can be traced to the major sections of its internal architecture.

Signal generator architectures

Two types of architectures are common: single-loop and multiple-loop phase-locked loops (PLLs). The single-loop approach is less complex, which makes it simpler to design and optimize. It also tends to be less expensive, but the lower cost comes with a tradeoff: single-loop synthesizers tend to provide moderate levels of phase noise performance (but outstanding adjacent-channel power ratio or ACPR).

Multiple-loop designs are more complex and, therefore, typically more expensive. The additional elements may include a fine loop, an offset or step loop, and a summing loop, contributing to lower spurious levels and significant improvements in phase noise performance (Figure 3). If certain loop-adjustment controls are user-accessible, a multi-loop synthesizer provides greater flexibility in optimizing phase noise performance to suit a specific application.

Figure 3. This triple-loop architecture significantly improves phase noise performance, as implemented in the Keysight PSG and MXG signal generators.

To illustrate the differences, Figure 4 shows phase noise plots from three Keysight X-Series signal generators: the EXG, the standard MXG, and the MXG with the “enhanced low phase noise” option (UNY). There are clear differences in performance between the single-loop EXG and the multi-loop MXG, with or without the low-phase noise option.

Figure 4. While the single-loop EXG is suitable for many situations, the multiple-loop MXG offers a substantial improvement for high-performance applications.

Digging one layer deeper within any architecture, the oscillator type can also affect phase noise performance. For example, signal generators that use a voltage-controlled oscillator (VCO) will generally provide worse phase noise performance than one that uses a YIG oscillator in the synthesizer section (Figure 5). This has a tradeoff: YIG-based signal generators generally provide slower switching speeds than VCO-based designs.

Figure 5. Inside a signal generator, the architecture and oscillator type combination affect the overall amount of phase noise and the distribution of phase noise versus offset and frequency.

One more architectural element is worth mentioning: reference sources, either internal or external. Within the signal generator architecture, the frequency reference’s noise performance significantly influences phase noise. Under the heading “designer’s choice,” most RF and microwave signal generators include a high-quality internal 10-MHz reference and offer a higher-performance 10-MHz reference as an option. In the “user’s choice realm,” most signal generators provide an external input for a known, high-performance 10-MHz reference.

Taking this one step further, the Keysight PSG signal generators can be configured with an input that accepts a 1-GHz external frequency reference (option H1S). In this mode, the PSG bypasses the internal reference assembly, negating its additive phase noise when using a 10-MHz reference (internal or external).

Signal generator capabilities

There are different capability tier levels for RF and microwave signal generators. At the basic level, the first choice is between continuous-wave (CW) or analog signal generators (analog-modulation capable) and vector signal generators (VSGs; capable of analog and vector modulation). For those who need a “golden source” in a lab setting, a high-performance CW or analog signal generator is often the default choice. However, suppose you need more functionality. In that case, a VSG includes vector, or digital, modulation (e.g., I and Q modulation inputs) and sometimes offers pulse modulation, which helps simulate Doppler radar signals.

A VSG may include a baseband arbitrary waveform generation (AWG) capability and deep internal waveform memory. Some, such as the MXG vector signal generator (N5182B), also support real-time simulation of complex real-world signals. This is typically done in combination with signal-creation software such as Pathwave Signal Generation software to support a variety of applications:

• Cellular communications, including Verizon Pre-5G, LTE/LTE-A-FDD/TDD
• GNSS with up to 32 GPS and GLONASS line-of-site satellites
• DVB-T/H with a continuous PN23 or up to two hours of video playback
• Custom modulation such as AWGN or 1024QAM
• AWGN and phase-noise impairments

The various digital capabilities within the signal generator may affect phase noise performance. The specifics of the possible interactions vary and can be complex. As a result, it’s essential to consult the detailed specifications for any signal generator and compare the performance levels to your complete set of application requirements.

Keysight Signal Generators

Keysight’s signal generators provide excellent performance regarding phase noise, output power, ACPR, error vector magnitude (EVM), and bandwidth. The Keysight AP5001A and AP5002A analog signal generators, EXG N5171B analog signal generators, and N5172B vector signal generators are cost-effective single-loop synthesizers. The new AP500xA portable signal generators offer excellent noise performance (Figure 6) in a compact size.

Figure 6. AP500xA Phase Noise Measurement at 10 dBm with an internal reference

The Keysight N5181B analog signal generators and N5182B vector signal generators utilize the triple-loop PLL architecture shown in Figure 3 to provide excellent spurious and phase noise performance.

In the MXG, one key to improved phase noise is the frequency plan, which is optimized for the triple-loop topology. The frequency plan addresses several attributes: the choice of oscillator and reference frequencies in the synthesizer sum and offset loops and the associated frequency conversion (mixers and multipliers) and filtering.

The triple-loop approach allows optimized frequency spacing that ensures effective filtering of nonlinear artifacts (e.g., images) by pushing them outside the bandwidth of the synthesizer circuits. In the MXG, the plan arranges the frequency references and conversions such that the largest nonlinearities are far from the desired frequencies, and modest filtering can heavily attenuate the remaining spurious signals. Moving the large nonlinearities also allows internal signal levels to be set higher, resulting in relatively lower broadband noise and improved dynamic range.

With these capabilities, the MXG supports the development of components and receivers that meet the challenges of increased interference, data throughput, and signal quality in applications such as commercial wireless, military communications, and radar.

For example, today’s aerospace/defense environment requires enhanced radar performance to detect weak signals at long distances. To provide the pure and precise signals needed to test these designs, the MXG delivers phase noise performance as good as –146 dBc/Hz at 1 GHz and 20 kHz offset with option UNY, enhanced low phase noise. Option UNY is recommended for signal-generation applications such as LO substitution or blocking signals that require spectrally pure signals or outstanding modulation accuracy. For developers of radar components such as mixers and analog-to-digital converters, the MXG also features spurious performance of –96 dBc at 1 GHz.

Conclusion

Phase noise performance is often the critical factor in determining a signal generator’s suitability for a demanding application. Getting the best possible phase noise performance depends on various factors: internal architecture, type of oscillator, internal and external frequency references, and the effects of additional built-in capabilities. The tradeoffs around these attributes include switching speed, optimization for close-in or far-out offsets, and cost.