Reading an Oscillator Frequency Stability Budget Without Counting Terms Twice

An oscillator’s advertised frequency stability may already include several sources of error, while another product’s headline number may cover only temperature. Comparing those numbers without reading the conditions can produce the wrong timing budget. Identify the terms included in each limit before adding margins or choosing a part.

Read the scope of the specification

The SiTime SiT1572 data sheet explicitly describes what its total frequency stability includes and lists aging separately. This is an example of why the note attached to a ppm figure is as important as the figure itself.

Record nominal frequency, temperature range, supply range, output load, and any specified assembly conditions. Check whether the ordered option actually carries the limit being used in the design.

Build a nonoverlapping error budget

List included terms and separate terms in different rows. Do not add initial tolerance again if the total specification already includes it. Likewise, do not omit aging merely because temperature stability is already included.

Determine whether the application requires a worst case bound or a justified statistical estimate. Independent typical values cannot be combined and presented as a guaranteed maximum without an appropriate model and evidence.

Translate ppm into system behavior

A frequency error of 20 ppm corresponds to 20 microseconds of accumulated timing difference per second against an ideal reference, if that error remains constant. Over one day, the arithmetic gives about 1.73 seconds.

Real temperature and aging behavior can vary with time, so this constant error example is a scale estimate rather than a prediction of every daily clock error. For an interface, compare the local and remote reference requirements as specified by that interface.

Verify the relevant conditions

Measure frequency using a reference with suitable accuracy and a measurement interval appropriate to the oscillator. Include the intended board load and supply. A room temperature reading immediately after adjustment does not verify the full environmental specification.

If calibration removes initial error, document what remains after calibration and when recalibration is required. Temperature change and aging are not automatically removed by one stored correction value.

Save the exact part option, specification notes, budget terms, calibration assumptions, and measured conditions. A sound selection explains how the complete frequency error fits the application over its intended service interval. Comparing only the smallest ppm number can hide excluded conditions, while indiscriminately adding every printed tolerance can count the same contribution more than once and reject an otherwise suitable oscillator.