How to Evaluate Phase Noise Specification in OCXO Data Sheets

https://www.sapphireosc.com/products/low-phase-noise-ocxo-frequency.html

An OCXO data sheet can be dense with specifications, but the phase noise table is often where high-performance designs are won or lost. Unlike frequency stability (a single number) or aging (a simple drift rate), phase noise is a curve—and understanding how to read that curve prevents costly mismatches between your oscillator and your application.

Here is a systematic framework for evaluating phase noise specifications, distinguishing marketing claims from guaranteed performance, and matching oscillator characteristics to real-world system requirements.

The One-Sentence Summary

Phase noise evaluation reduces to three questions. What offset frequencies matter to my system? Does the specification represent typical or guaranteed performance? And was the measurement conducted under relevant conditions?

Step 1: Understand the Phase Noise Specification Format

Every phase noise specification in an OCXO data sheet follows a consistent format: a value in dBc/Hz (decibels relative to the carrier per hertz of bandwidth) at a specific frequency offset from the carrier.

Consider a typical specification from a high-performance OCXO data sheet. At 10 Hz offset, you might see -128 dBc/Hz. At 100 Hz offset, -145 dBc/Hz. At 1 kHz offset, -155 dBc/Hz. At 10 kHz offset, -160 dBc/Hz. And at 100 kHz offset, -168 dBc/Hz.

What each entry means is straightforward. At 1 kHz away from the carrier frequency—for example, 10 MHz plus or minus 1 kHz—the noise power in a 1 Hz bandwidth is 155 dB lower than the carrier power. The more negative the number, the better. So -168 dBc/Hz at 100 kHz is superior to -155 dBc/Hz at 1 kHz.

What the units do NOT tell you is equally important. These single-point measurements do not reveal spurious signals, which are discrete noise spikes, or the noise floor's flatness. For critical applications, always request a full phase noise plot, not just tabulated values.

Step 2: Match Offset Ranges to Your Application

Different applications are sensitive to phase noise at different offset frequencies. A specification that looks excellent for one use case may be entirely irrelevant for another.

Close-in offsets from 1 Hz to 100 Hz are critical for radar, Doppler systems, FMCW LiDAR, and precision timing applications. This part of the noise curve determines your ability to detect slow-moving targets and directly affects short-term stability, often measured as Allan deviation.

Mid-range offsets from 100 Hz to 10 kHz matter most for digital communications, cellular base stations, and satellite links. Noise in this region impacts Error Vector Magnitude (EVM) and bit error rates in QAM and QPSK modulated signals.

The noise floor from 10 kHz to 1 MHz becomes dominant for high-speed data converters (ADCs and DACs) and clock distribution systems. Clock jitter at these offsets directly degrades signal-to-noise ratio in sampled systems.

Here is how application matching works in practice. A 5G small cell cares most about phase noise at 1 kHz to 100 kHz offsets because this range directly impacts modulation error rate on 256-QAM signals. A weather radar system cares most about 10 Hz to 100 Hz offsets because moving targets—wind shear and precipitation—produce Doppler shifts in this range. Choosing an OCXO optimized for the wrong offset range means paying for performance you do not need while lacking performance you do.

Step 3: Distinguish Typical vs. Guaranteed Specifications

This is the most critical—and most frequently overlooked—aspect of OCXO data sheet evaluation.

Most data sheet phase noise values are typical. This means they were measured on one device under ideal conditions: 25 degrees Celsius, a quiet power supply, and a vibration-free bench. A typical specification means "you will likely get performance close to this," not "every device will meet or exceed this value."

The real-world implication is significant. If a data sheet states -160 dBc/Hz at 1 kHz as typical, a production unit might measure -155 dBc/Hz. That is a 5 dB degradation that could push your system out of specification.

A guaranteed or maximum specification means every shipped unit meets that level. Guaranteed phase noise is typically 3 to 6 dB worse than typical values for the same offset. Some high-performance OCXOs, like the Rakon LNO 100 series, offer guaranteed low phase noise across temperature ranges, not just at room temperature.

How can you identify each in a data sheet? Look at the phrasing. "Typical," "representative," or "characteristic" means the value is not guaranteed and carries high risk. "Maximum," "guaranteed," or "limit" means every device meets the specification with low risk. If no qualifier is provided, assume typical and proceed with caution.

For production designs, request guaranteed phase noise limits in writing. For prototypes or low-volume applications, typical values are often sufficient.

Step 4: Verify Test Conditions

A phase noise specification is only as trustworthy as the conditions under which it was measured. Four questions must be answered.

First, what was the measurement temperature? Some data sheets specify phase noise only at 25 degrees Celsius. Others, like those from Rakon, provide guaranteed performance across operating temperature ranges such as -20 to +70 degrees Celsius. Phase noise typically degrades at temperature extremes—sometimes by 3 to 10 dB.

Second, was the device warmed up per specification? An OCXO can take three to five minutes to reach thermal equilibrium. Measurements taken before warm-up are meaningless. Look for explicit statements like "three minutes after power is on."

Third, what power supply was used? Power supply noise can directly inject into the oscillator output, artificially inflating or hiding phase noise contributions. High-quality data sheets specify the power supply characteristics used during measurement.

Fourth, was vibration isolated? OCXOs are sensitive to mechanical vibration, a property called g-sensitivity. A specification measured on a vibration-isolated optical bench may not represent field performance in a moving vehicle, aircraft, or outdoor installation.

Look for data sheets that reference recognized test standards. IEC 62884-3:2018 is the international standard for phase noise measurement of oscillators. IEC EN 60679-4-1 covers quality evaluation for OCXOs. MIL-PRF-55310 is the US military specification for oscillators. References to these standards provide higher confidence in the measurement methodology.

Step 5: Compare at the Same Carrier Frequency

Phase noise scales with carrier frequency. A specification at 10 MHz cannot be directly compared to a specification at 100 MHz because they represent different physical noise mechanisms.

The general scaling rule is that phase noise at a given offset typically increases—gets worse—by approximately twenty times the logarithm base ten of the frequency ratio. In practice, a 10x increase in carrier frequency adds 20 dB to the phase noise at a given offset.

Here is an example. An OCXO specified at -160 dBc/Hz at 10 kHz offset for a 10 MHz carrier would be approximately -140 dBc/Hz at 10 kHz offset for a 100 MHz carrier. That is a 20 dB degradation due solely to frequency scaling, not a difference in oscillator quality.

Always verify the carrier frequency in the phase noise table. When comparing two OCXOs from different manufacturers, ensure they are specified at the same carrier frequency or apply the scaling formula to normalize the comparison.

Step 6: Understand the Relationship Between Phase Noise and Jitter

For systems with high-speed ADCs, DACs, or clock data recovery circuits, integrated jitter—measured in femtoseconds or picoseconds—may be more relevant than spot phase noise values.

The relationship is mathematical. Phase noise across a bandwidth, such as 12 kHz to 20 MHz for JESD204B clocks, integrates to phase jitter. Lower integrated jitter means cleaner sampling clocks and higher signal-to-noise ratio in data converters.

What should you look for in a data sheet? Some OCXO manufacturers provide both phase noise plots and calculated jitter values over standard integration bandwidths. If these are not provided, the phase noise plot allows you to calculate your own jitter for your specific integration limits.

For applications using JESD204B or JESD204C interfaces, pay particular attention to the jitter integration range specified in the data sheet and ensure it matches your converter's requirements.