electrical-engineering-principles
Understanding Harmonics and Spurious Signals in Signal Generator Outputs
Table of Contents
Introduction
Signal generators form the bedrock of modern electronics testing, providing the precise electrical stimuli needed to validate everything from simple audio amplifiers to complex wireless communication systems. The ideal signal generator output is a pure, mathematically perfect waveform. In reality, the output of any physical signal generator is burdened by deterministic imperfections that limit its usefulness in demanding applications. These imperfections manifest primarily as harmonic distortion and spurious signals.
Understanding the difference between these two classes of artifacts, their root causes, and their effects on system-level measurements is essential for any engineer or technician working with RF, microwave, or high-speed digital systems. Harmonic distortion consists of signals at integer multiples of the fundamental frequency, arising from non-linearities in the signal path. Spurious signals are discrete, unwanted frequency components that are not harmonically related to the fundamental, often caused by internal clocking schemes, power supply noise, or mixed-signal interference. Mitigating these artifacts allows engineers to maximize the Spurious-Free Dynamic Range (SFDR) and Total Harmonic Distortion (THD) of their test setup, leading to more accurate characterizations of the Device Under Test (DUT).
This article explores the internal architectures responsible for these artifacts, details the mathematical and physical underpinnings of harmonics and spurs, examines their practical impact on real-world applications, and provides actionable strategies for minimizing their presence in critical measurements.
The Root Causes: Signal Generator Architectures and Imperfections
The specific type and severity of unwanted signals are directly tied to the fundamental architecture of the signal generator. Modern generators typically fall into three main categories: Direct Digital Synthesis (DDS), Phase-Locked Loop (PLL) based, and traditional analog generators. Each introduces a unique set of imperfections.
Direct Digital Synthesis (DDS) and Quantization Artifacts
DDS generators create waveforms by digitally constructing the signal sine wave point-by-point from a reference clock. The digital numbers are fed into a Digital-to-Analog Converter (DAC). The purity of a DDS output is limited by several discrete effects. Phase truncation occurs because the generator stores enough memory for a finite phase increment, discarding less significant bits. This truncation creates a periodic phase error that modulates the output, generating specific spurious tones. Amplitude quantization arises from the finite resolution of the DAC and the sine lookup table, which adds a noise-like floor but also discrete spurs depending on the waveform periodicity. Most significantly, the non-linear integral and differential non-linearity of the DAC itself introduces harmonic distortion. The DAC’s inherent non-linearity is often the dominant source of harmonic distortion in DDS-based instruments. When selecting a DDS generator, understanding the correlation between the master clock frequency, the output frequency, and the resultant spur frequencies (often predictable) is critical for frequency planning.
Phase-Locked Loop (PLL) Architectures and Reference Spurs
PLL-based generators generate high-frequency signals by multiplying a lower-frequency reference. The phase detector and charge pump within the PLL are sources of periodic disturbance. When the phase detector compares the divided output to the reference, mismatches in its current sourcing or sinking create voltage ripple on the control line of the Voltage-Controlled Oscillator (VCO). This ripple frequency-modulates the VCO output, producing spurious tones at offsets equal to the phase detector frequency. These are known as reference spurs. In fractional-N PLLs, the modulus control creates additional deterministic jitter that folds into the passband as fractional spurs. While the loop filter is designed to suppress these artifacts, practical filter limitations mean a residual level of spurs always exists. Adjusting the loop filter bandwidth and selecting a phase detector frequency that places spurs outside the operational bandwidth of the system are standard techniques for managing PLL spurs.
Analog Signal Generators and Output Amplifier Non-Linearity
Traditional analog generators rely on high-stability LC or Wien bridge oscillators. While these can achieve exceptionally low phase noise and spurious free performance at low frequencies, they are highly susceptible to thermal drift and mechanical vibration. The final output stage, an amplifier designed to deliver a wide range of power levels (e.g., -120 dBm to +20 dBm), is a major source of harmonic distortion. Amplifiers have a linear operating region. When the demanded output power approaches the amplifier’s saturation point (P1dB), the transfer function becomes highly non-linear, generating substantial harmonic power. This is why operating a generator with a built-in attenuator at a lower power level (amplitude back-off) significantly improves harmonic performance. Understanding the trade-off between maximum output power and linearity is fundamental to making clean measurements.
Harmonic Distortion: The Predictable Foes
Harmonics are the most well-understood type of deterministic distortion. They appear as discrete tones at precise multiples of the fundamental frequency (2*f0, 3*f0, 4*f0, etc.) and are directly attributable to non-linearities in the signal generation path.
The Mathematics of Harmonics and Non-Linear Transfer Functions
A perfectly linear system produces an output that is a straight multiple of the input. Any deviation from this straight line introduces distortion. If the transfer function is represented as a power series, *y(t) = a1*x(t) + a2*x²(t) + a3*x³(t) + ...*, applying a pure sine wave *x(t) = A*cos(ωt)* results in output components at frequencies 2ω, 3ω, etc. The even-order terms (a2, a4) generate even harmonics, while odd-order terms (a3, a5) generate odd harmonics. Odd-order distortion is particularly troublesome because the third harmonic often falls within the same octave as the fundamental, making it impossible to filter out with a simple low-pass filter.
This mathematical relationship highlights a crucial distinction: the relative level of harmonics is highly dependent on the output amplitude. A 10 dB reduction in output power typically yields a 20 dB improvement in second harmonic distortion (a2 term) and a 30 dB improvement in third harmonic distortion (a3 term). This predictable behavior allows engineers to use amplitude back-off as a powerful tool for achieving cleaner signals.
Total Harmonic Distortion (THD) and THD+N
THD is the standard metric for quantifying harmonic content. It is defined as the ratio of the sum of the powers of all harmonics to the power of the fundamental. It is usually expressed as a percentage or in dBc (decibels relative to the carrier). THD+N includes the noise floor in the measurement, providing a more comprehensive view of signal purity. For high-precision audio or power integrity testing, THD levels below -80 dB (0.01%) or -100 dB (0.001%) are often mandatory. Generator datasheets will specify THD at a given amplitude and frequency. When interpreting a datasheet, pay close attention to the test conditions (frequency and amplitude), as THD degrades rapidly near the generator’s maximum rated output power.
Practical Identification on a Spectrum Analyzer
Identifying harmonics is straightforward. Place the signal generator’s output on a spectrum analyzer set to a wide-span sweep. The signal at the fundamental frequency (f0) will be the highest. Harmonics appear as discrete peaks at 2*f0, 3*f0, and 4*f0. A quick check is to change the signal generator’s frequency. If all the discrete peaks remain at exact integer multiples of the new fundamental frequency, they are harmonics. If a peak does not move when the fundamental is changed, it is likely a spurious signal unrelated to the fundamental. This simple diagnostic test is highly effective in distinguishing between the two.
Spurious Signals: The Unpredictable Interlopers
Unlike harmonics, spurious signals do not follow a simple integer multiple pattern relative to the fundamental. They are caused by a wide variety of internal mechanisms and are often the limiting factor in high-sensitivity receiver testing and wide-bandwidth systems.
Primary Sources of Spurious Content
DDS Truncation Spurs: As mentioned earlier, DDS architectures generate specific spurs based on the phase truncation and DAC quantization. The exact frequency of these spurs depends on the ratio of the master clock to the output frequency. These spurs can be predicted and often avoided by careful frequency selection.
PLL Reference and Fractional Spurs: These appear as discrete tones at a specific offset from the carrier. For integer-N PLLs, reference spurs appear at offsets of ±f_pd (phase detector frequency). For fractional-N PLLs, spurs can appear at fractions of the reference frequency. These spurs are particularly damaging in communications because they can fall within the channel bandwidth and directly degrade the Error Vector Magnitude (EVM).
Power Supply and Digital Feedthrough: In modern generators, high-speed digital logic (FPGAs, DSPs) coexists with sensitive analog output stages. Insufficient isolation allows digital clock signals and switching noise to couple into the analog output. This often manifests as spurs at the frequencies of internal clocks or the switching frequency of power converters. Careful PCB layout, shielding, and the use of low-noise power supplies are essential to minimize this.
Sub-Harmonics and Intermodulation: Some high-frequency multipliers used in millimeter-wave generators produce sub-harmonics (e.g., f0/2). Additionally, when a generator produces a modulated signal (like a two-tone test), the non-linearities in the amplifier create intermodulation products (IMD) which are spurious signals. The third-order intermodulation products (IM3) are especially critical for receiver testing.
Spurious-Free Dynamic Range (SFDR)
SFDR is arguably the most important specification for a signal generator used in frequency-domain measurements. SFDR is defined as the ratio of the RMS amplitude of the fundamental signal to the RMS amplitude of the next largest spurious or harmonic component within a defined bandwidth. This component could be a harmonic, a DDS spur, a reference spur, or a power-line-related artifact. SFDR is expressed in dBc (relative to the carrier). A high SFDR (e.g., >80 dBc) means the generator can provide a clean signal that will not mask the DUT's own spurious responses. For testing high-performance ADCs or narrow-band filters, the generator’s SFDR must significantly exceed the expected dynamic range of the DUT.
Distinguishing Spurs from Harmonics and Phase Noise
On a spectrum analyzer, spurs appear as sharp, discrete peaks. The key differentiator from harmonics is their behavior relative to the fundamental frequency. While harmonics track the fundamental proportionally, spurs often maintain a fixed frequency offset from the fundamental (in the case of PLL spurs) or remain at a static absolute frequency regardless of the generator's setting (in the case of power supply noise). Phase noise, by contrast, appears as a widening of the carrier signal itself, looking like a "skirt" around the fundamental. Understanding these distinctions guides the troubleshooting process. Identifying a spur as a 60 Hz power supply artifact suggests a different mitigation strategy (e.g., better grounding) than identifying it as a PLL reference spur (e.g., adjusting the loop filter).
System-Level Implications: Why Clean Signals Matter
The presence of harmonics and spurious signals is not just an abstract specification. It has direct and measurable consequences on the accuracy and validity of tests performed across the entire electronics industry.
ADC and DAC Testing
Characterizing the dynamic performance of an ADC is one of the most demanding applications for a signal generator. The standard test involves applying a pure sine wave to the ADC and analyzing the output FFT. Any harmonic or spur present on the input signal will be indistinguishable from the ADC's own internal distortion. If the test source has a THD of -80 dBc, the engineer cannot accurately measure an ADC with a SINAD (Signal-to-Noise and Distortion) performance better than approximately 13 ENOB (Effective Number of Bits). The test source must be significantly cleaner than the device being tested. This often necessitates using external band-pass filters to strip away harmonics from the generator output before it reaches the ADC.
Wireless Communications Transceiver Testing
In cellular and wireless connectivity testing (Wi-Fi, Bluetooth), signal generators are used to simulate base station or interfering signals. If the generator has high out-of-band spurs, it will desensitize the receiver under test. According to standardization bodies like 3GPP and IEEE, a transmitter must not exceed a specific spectral mask to prevent interference to adjacent channels. If a signal generator used as a test stimulus has spurs exceeding this mask, it becomes impossible to test the receiver's true adjacent channel selectivity (ACS) or blocking performance. The test equipment must be more ideal than the standard requires. Spurs located in the receive band of a device will directly corrupt measurements of receiver sensitivity (TIS / Rx Power).
Mixer and Filter Characterization
When characterizing a mixer intermodulation products, the test relies on two signal generators. Any spurs or harmonics on either generator will produce extra mixing products, making it impossible to isolate the mixer's true IMD performance. Similarly, characterizing a band-pass filter's shape factor or stop-band rejection requires a signal source that is clean across a wide frequency range. A generator with high harmonic output can fool the engineer into thinking the filter has poor high-side rejection when the measurement is actually seeing the generator's second harmonic folding through. Pre-filtering the test signal is often mandatory for accurate filter and mixer characterization.
Mitigation Strategies for Optimal Signal Purity
While no source is perfect, several established techniques allow engineers to achieve signal purity that far surpasses the raw specifications of the signal generator.
Operational Strategies: Amplitude Back-Off and Attenuation
The simplest and most effective way to reduce harmonic distortion is to use the generator at a lower output power. If the test requires a +10 dBm signal but the generator is capable of +20 dBm, set the generator to +10 dBm directly. Do not set the generator to +20 dBm and use an external 10 dB attenuator unless the generator's internal attenuator is optimized for low harmonic generation. Operating closer to the generator’s maximum output pushes the output amplifier further into its non-linear region, drastically increasing harmonic levels. As a rule of thumb, backing off from the maximum output power by 5-10 dB provides a significant improvement in THD.
External Filtering: The Engineer’s Best Friend
External filters are the most powerful tool for achieving an ultra-clean signal. A standard signal generator may have -30 dBc harmonic distortion. A simple low-pass filter placed at the generator's output can suppress the 2nd and 3rd harmonics by more than 40 dB, resulting in a signal with less than -70 dBc distortion. The choice of filter depends on the application:
- Low-Pass Filters: Ideal for suppressing harmonics when the fundamental is well below the filter cutoff.
- Band-Pass Filters: Excellent for rejecting both harmonic and broadband spurious noise when testing narrowband devices.
- Notch Filters: Used to remove a specific problematic spur (e.g., a clock feedthrough artifact) without affecting the fundamental.
Step attenuators and programmable filter banks are integrated into many high-end signal generators, but external cavity or ceramic filters offer the highest performance for critical applications.
Frequency Planning and Clock Selection
In DDS and PLL systems, the engineer often has some choice over the operating frequency or the internal clock settings. Understanding the spurious profile of the generator allows for intelligent frequency planning. For example, if a DDS generator has a known bad spur at a specific output frequency due to phase truncation, the engineer can shift the test frequency by 5-10 kHz, avoiding the spur while still validating the DUT performance. In PLLs, choosing a reference crystal frequency that places harmonics and spurs outside the operational band is a standard design practice.
Advanced Calibration and Digital Pre-Distortion (DPD)
High-end instrumentation often employs some form of digital pre-distortion or internal calibration. The instrument uses a high-speed feedback path and a digitizer to measure its own output. It then calculates its non-linear transfer function and pre-distorts the digital waveform before converting it to analog. This technique can effectively cancel out low-order harmonics and even some deterministic spurs. While this requires significant internal hardware (FPGAs, high-speed ADCs, and DSP cores) it represents the state-of-the-art in signal purity. DPD can reduce harmonic distortion by 20-30 dB over the raw analog performance, pushing THD below -90 dBc at moderate output levels.
Conclusion
The difference between a mediocre measurement and a definitive, publication-quality result often comes down to the purity of the stimulus signal. Harmonics and spurious signals are unavoidable realities of physical signal generation, rooted in the fundamental architectures of DDS, PLL, and analog circuits. However, they are not arbitrary obstacles. They are deterministic artifacts that can be understood, predicted, and effectively mitigated.
By mastering the concepts of non-linear transfer functions, phase truncation, reference spurs, and amplitude back-off, the test engineer gains the ability to extract maximum performance from their equipment. The specification sheet for a signal generator tells a story; SFDR, THD, and phase noise are its critical chapters. Applying external filtering, intelligent frequency planning, and careful operational techniques ensures the signal generator becomes a transparent tool rather than a limiting factor in the measurement chain. Ultimately, a deep understanding of these signal imperfections empowers engineers to confidently validate their designs, ensuring system-level performance and reliability.