Ripple and Noise in DC Supplies: How to Measure and Reduce Them

2026-10-29
Ripple and Noise in DC Supplies: How to Measure and Reduce Them - Nenghua Power
Photo: Jinan Nenghua real products

A so-called DC output is rarely an ideal straight line; it carries a small AC component on top. This is usually split into two ideas: ripple is the periodic, fixed-frequency part, while noise is random, broadband spiking. It matters little for most loads, but for precision measurement, sensors, communications and amplification it translates directly into error.

Read the measurement conditions first

Ripple and noise figures are only comparable under the same method. Datasheet numbers usually carry conditions: measurement bandwidth (for example 20 MHz), load, whether an end capacitor is fitted on the output, and whether a dedicated ground spring rather than a long ground lead was used. Change the method and the reading can shift a lot, so confirm matched conditions when comparing products.

Common measurement mistakes

MistakeConsequence
Long ground lead on the probeForms a loop antenna, picks up ambient noise, inflates the reading
No bandwidth limitCounts high-frequency interference that is not from the supply
Poor probe contactContact jitter adds spurious spikes
Switching-frequency content read as a faultMisdiagnosis of a normal unit

Ways to reduce noise

Selection and acceptance

Define the load's tolerance and bandwidth first, then compare figures; at acceptance, measure with the same bandwidth and probe method as the rating, instead of judging by eye from a noisy-looking trace. See switching DC supplies and computer-programmable DC supplies; the full range is in the DC power supply category.

Related Articles

All News

Frequently Asked Questions

Read the measurement conditions first?
Ripple and noise figures are only comparable under the same method. Datasheet numbers usually carry conditions: measurement bandwidth (for example 20 MHz), load, whether an end capacitor is fitted on
Common measurement mistakes?
MistakeConsequence Long ground lead on the probeForms a loop antenna, picks up ambient noise, inflates the reading No bandwidth limitCounts high-frequency interference that is not from the supply Poor
Ways to reduce noise?
Pick the topology: linear supplies usually have far lower noise floor than switching types; prefer them for very noise-sensitive loads, see linear DC supplies. Output filtering: where allowed, add an
Selection and acceptance?
Define the load's tolerance and bandwidth first, then compare figures; at acceptance, measure with the same bandwidth and probe method as the rating, instead of judging by eye from a noisy-looking tra