Defining a Repeatable Power Rail Ripple Measurement

A report that says a rail has twenty millivolts of ripple is incomplete without the measurement conditions. Different bandwidths, probe connections, acquisition durations, and operating states can produce different peak-to-peak values on the same board. Define the measurement before using its number to accept or reject a power supply.

Separate the questions being asked

Decide whether the objective is switching ripple, broadband noise, startup behavior, or response to a load event. These may require different acquisition settings. A bandwidth limit useful for a specified ripple test can conceal a fast transient that matters in another investigation.

Rohde & Schwarz discusses this tradeoff in its power integrity measurement guidance: probe attenuation and system noise can obscure small rail variations, while limited bandwidth removes higher-frequency content. The measurement system must be chosen for the question rather than for an attractive trace.

Specify the connection and instrument state

• Identify the rail and exact local reference point on a board photograph.
• Record the probe or coaxial connection and the instrument input configuration.
• State bandwidth, coupling, vertical scale, offset, and acquisition mode.
• Define the observation interval and whether results are single-shot, averaged, or accumulated.
• Record input voltage, load activity, firmware state, and ambient conditions.

Use a connection appropriate to the rail and instrument ratings. A short local reference reduces ambiguity from a large pickup loop. If the connection loads the rail, include that effect in the review rather than assuming a direct cable is electrically invisible.

Establish the measurement floor

Characterize the setup’s residual noise using the instrument manufacturer’s appropriate procedure and comparable settings. Do not subtract two peak-to-peak numbers blindly; their statistical behavior may differ. Instead, determine whether the system can resolve the variation of interest with useful confidence.

For example, compare idle and active processor states using the same connection and acquisition interval. If the active capture contains occasional load steps, report those separately from steady switching ripple. A longer observation may reveal rare events that a short screenshot misses.

Save representative raw captures and explain how the reported value was calculated. When comparing board revisions, preserve the method as well as the operating conditions. If a setting changes, rerun an overlap measurement to understand its effect. The resulting record lets another engineer reproduce the rail assessment and prevents measurement choices from being mistaken for improvements in the power circuit.

Include the location of local bypass capacitors when photographing the connection; measuring across a distant ground path can answer a different electrical question.