Switching Power Supply Hold-Up Time Testing: Principles, Standards, and Methods

Switching power supply hold-up time testing verifies that a power supply can maintain a regulated DC output for a specified duration after the AC input is lost, protecting sensitive equipment from data corruption and uncontrolled shutdowns. As a critical performance and reliability requirement for server, computing, industrial, and medical power supplies, this test is typically performed by accredited third-party laboratories in accordance with IEC 61204, IEC 62368-1, GB 4943, and related international standards. This article explains the technical principles, test setup, measurement procedures, acceptance criteria, and practical considerations involved in hold-up time evaluation.

Technical Overview

Hold-up time is defined as the interval between the loss or removal of the AC mains input and the moment the DC output voltage of a switching power supply falls below its specified regulation limit, commonly set at 95% of the nominal output voltage. The measurement is conducted under defined conditions of load (typically 100% of rated load), input voltage, frequency, and ambient temperature, and it directly quantifies the ability of the power supply to ride through brief power disturbances.

The engineering purpose of hold-up time is straightforward: during a momentary mains interruption, grid disturbance, or transfer to a backup power source, the power supply must continue delivering energy long enough for the connected system to detect the event, save critical data, execute an orderly shutdown, or ride through the interruption entirely. At 50 Hz mains frequency, one full AC cycle lasts 20 ms; at 60 Hz, one cycle lasts 16.7 ms. For this reason, many procurement and platform specifications require a minimum hold-up time of at least one complete AC cycle.

In an off-line switching power supply, hold-up energy is stored in the high-voltage bulk electrolytic capacitor located between the input rectifier (or PFC stage) and the DC-DC converter stage. When the AC input disappears, the converter continues to draw energy from this capacitor and the DC bus voltage decays. The approximate relationship is expressed as:

t(hold-up) ≈ C(bulk) × (V1² − V2²) × η / (2 × P(out)), where V1 is the bus voltage at the moment of input loss, V2 is the minimum bus voltage at which the converter can still regulate, η is the efficiency, and P(out) is the output power. Because stored energy scales with the square of the bus voltage, both load level and bulk capacitance strongly influence the measured result.

Relevant reference frameworks include IEC 61204 for performance characteristics of low-voltage DC output switch-mode power supply units, IEC 62368-1, UL 62368-1, and GB 4943.1 for the safety context of audio/video, information, and communication technology equipment, and IEC 61000-4-11 and IEC 61000-4-34 for immunity to voltage dips and short interruptions. Platform design guidelines for desktop and server power supplies additionally define widely adopted hold-up time classes that buyers frequently reference in procurement specifications.

Testing Samples

Information Technology and Server Power Supplies

Redundant server power supplies, desktop ATX-class units, and storage system power supplies represent the most common test articles. These products are typically specified with hold-up requirements in the range of 16 ms to 20 ms or more at full load, because the hosted systems must flush write caches, park storage media, and signal power-fail interrupts before the outputs collapse. Redundant and hot-swappable configurations are additionally evaluated for output transfer behavior when one module loses input.

External Adapters and Open-Frame Power Supplies

External AC/DC adapters for networking equipment, printers, and set-top boxes, as well as open-frame and encapsulated modules intended for integration into end products, are tested to confirm that the completed system achieves its intended ride-through performance. Designers of the host equipment frequently specify adapter hold-up time as a component-level requirement so that the end product can pass system-level voltage dip and interruption immunity testing.

Industrial, Medical, and Telecommunication Power Supplies

DIN-rail mounted industrial power supplies, programmable logic controller power units, medical-grade power supplies, and telecommunication rectifiers are evaluated against application-specific ride-through requirements. In industrial automation, adequate hold-up time prevents controller resets and machine faults; in medical systems, it supports continuity of monitoring and life-sustaining functions during transfers to backup power; in telecommunication installations, it coordinates with battery backup systems.

Testing Items

A comprehensive hold-up time evaluation typically covers the following measurement items, with the exact matrix defined by the product specification, the applicable standard, and the customer requirements:

• Hold-up time at 100% rated load and nominal input voltage
• Hold-up time at minimum and maximum rated input voltage, at both 50 Hz and 60 Hz where applicable
• Hold-up time at partial load conditions, such as 50% and 25% of rated load
• Hold-up time under cold ambient conditions, where low-temperature performance is specified
• Power-fail detect (PFD) or power-fail warning (PFW) signal timing relative to output dropout
• Captured output voltage waveform quality during the dropout event, including absence of glitches or out-of-specification excursions
• Recovery and restart behavior after AC input restoration, including re-establishment of the power-good signal
• Repetitive dropout verification at multiple input phase angles to identify the worst-case interruption point

Testing Methods

Test Equipment and Setup

A standard hold-up time test bench comprises the following instrumentation, configured to provide safe, isolated, and repeatable measurements:

• A programmable AC power source capable of interrupting the output at a user-defined phase angle, and of simulating dips, dropouts, and brownout waveforms
• A programmable electronic DC load operating in constant-current mode at the required load levels
• A digital storage oscilloscope, typically 100 MHz bandwidth or higher, with isolated or differential voltage probes for the output rails and a probe channel monitoring the AC input
• An optional current probe and power analyzer for characterizing input behavior and efficiency at the test operating point
• An environmental chamber when low-temperature or temperature-dependent hold-up verification is required

Step-by-Step Test Procedure

The generally accepted test sequence, consistent with IEC 61204 methodology and common industry practice, is as follows:

• Install the unit under test on a non-conductive bench and connect the AC source, electronic load, and measurement probes with appropriate safety isolation.
• Set the AC source to the first specified condition, for example nominal input voltage and frequency, and set the electronic load to 100% of rated output current.
• Allow the unit to reach thermal steady state at the specified ambient temperature so that internal components, particularly the bulk capacitor, are at realistic operating conditions.
• Trigger an AC input dropout lasting longer than the expected hold-up time, for example 30 ms to 100 ms, at a defined interruption phase angle.
• Capture the input voltage and output voltage waveforms simultaneously on the oscilloscope and measure the time from the start of input loss to the point where the output crosses the specified threshold, typically 95% of nominal output voltage.
• Repeat the dropout at several phase angles, commonly 0, 45, and 90 degrees, and record the minimum measured value as the representative worst-case result.
• Repeat the full measurement across the defined matrix of input voltages, load levels, and temperatures, including minimum rated input line, which usually produces the shortest hold-up time.
• Restore the AC input and verify that the output voltage and status signals recover within specification without abnormal behavior.

Measurement Criteria and Acceptance Thresholds

The unit passes if the minimum measured hold-up time, taken across all specified conditions, equals or exceeds the specified limit. Commonly referenced acceptance values include approximately 17 ms at full load for mainstream desktop platform design guides, 20 ms or more for server and telecommunication power supplies at 230 VAC 50 Hz input, and 10 ms class requirements for many external adapters. Where a power-fail warning signal is provided, its assertion must lead the output dropout by the specified margin, frequently on the order of 1 ms to 5 ms, so the host system has time to react. The pass threshold for the output voltage is defined by the product specification and is most commonly expressed as 95% of nominal DC output.

Worst-Case Test Conditions

Hold-up time is shortest at minimum rated input line, maximum load, and low ambient temperature, and it further degrades over product life as the bulk electrolytic capacitor loses capacitance and exhibits increased equivalent series resistance. Rigorous qualification programs therefore evaluate the unit at minimum input and full load, and either verify performance with a deliberately reduced bulk capacitance sample or demonstrate sufficient design margin through calculation, ensuring compliance is maintained at the declared end of useful life.

Applications

Hold-up time test results are used across the product life cycle and across many industries:

• Data centers and servers: confirming that storage controllers, RAID cards, and compute nodes can flush caches and log state before shutdown during a mains anomaly
• Networking and telecommunications: verifying uninterrupted operation of switches, routers, and radio units during brief grid disturbances and during transfer to backup systems
• Industrial automation: preventing unplanned resets of PLCs, motion controllers, and safety systems that could halt production lines or create hazardous conditions
• Medical equipment: demonstrating continuity of monitoring and diagnostic functions during power transfers in clinical environments
• Consumer electronics: ensuring set-top boxes, recording devices, and smart home hubs tolerate momentary power interruptions
• Research and development: validating bulk capacitor sizing, PFC brownout thresholds, and control firmware behavior early in the design cycle
• Quality assurance and production: screening for component substitutions and manufacturing defects that silently reduce ride-through capability

FAQ

What is a typical hold-up time requirement for a switching power supply?

Most specifications fall between 10 ms and 20 ms at full load. Requirements of 16 ms to 17 ms correspond roughly to one full cycle at 60 Hz, while 20 ms corresponds to one full cycle at 50 Hz, which is why server-grade and telecommunication products commonly demand at least 20 ms of ride-through.

Why is minimum input voltage combined with full load considered the worst case?

The energy available for hold-up is stored on the bulk capacitor, and at low input line the achievable bus voltage and stored energy are reduced, while full load draws that energy at the highest rate. The combination of low line, full load, and cold ambient therefore produces the shortest measurable hold-up time.

Does hold-up time degrade over the life of the product?

Yes. The bulk aluminum electrolytic capacitor gradually loses capacitance over its service life, particularly at elevated operating temperatures, which directly reduces the stored energy available during an input dropout. Qualification programs account for this by testing with end-of-life capacitance values or by demonstrating adequate design margin above the specified minimum.

What is the difference between hold-up time and power-fail warning time?

Hold-up time measures how long the output remains in regulation after input loss, whereas power-fail warning time measures how far in advance of output dropout the status signal asserts. The warning must arrive early enough, typically by at least 1 ms to 5 ms, for the host system to execute protective actions while the output is still valid.

What voltage threshold defines the end of the hold-up interval?

The most common definition is the point at which the output falls below 95% of its nominal value, but the exact threshold is always set by the product specification or the system-level requirement, since downstream regulators and loads define the true functional limit.

How is the AC interruption applied during testing?

A programmable AC source interrupts the input at a controlled phase angle and for a controlled duration. Because the residual charge and converter behavior differ depending on where in the AC cycle the interruption occurs, the test is repeated at several phase angles and the minimum measured value is reported as the worst case.

Do safety standards mandate a specific hold-up time?

Safety standards such as IEC 62368-1 and GB 4943.1 generally do not prescribe a universal hold-up value; the requirement originates from performance standards like IEC 61204, platform design guidelines, customer specifications, and system-level immunity needs addressed by IEC 61000-4-11 and IEC 61000-4-34. Manufacturers and buyers therefore define the applicable limit explicitly in the product specification.