Functional and protection testing of power supplies

Keywords:Power Supply Testing, Power Supply Verification, Functional Testing, Protection Testing.

DC power supplies (PSUs) are often referred to as the "heart" of electronic systems. With the evolution of high-performance computing (HPC) and AI server architectures, power supplies have evolved from simple power conversion components into sophisticated systems that combine ultra-high power density with digital management capabilities.

Evaluating a globally competitive power supply product requires a foundation built upon three key technological pillars: steady-state accuracy, dynamic resilience, and protection mechanisms. Beyond these, it must also transcend the boundaries of regulatory compliance —from electromagnetic compatibility (EMC) and global safety certifications to stringent energy efficiency standards. Compliance is not only a market "passport" but also the ultimate line of defense for brand reputation and user safety.

Based on these four dimensions, this article will delve into the core functional specifications and protection characteristic testing practices of power supplies during the R&D and verification phases.

1. Functional Specification Tests

1.1. Output Voltage Regulation

When manufacturing switching power supplies, the first testing step is to adjust the output voltage to the specified range. This step ensures that subsequent specifications are met. Typically, when adjusting the output voltage, the input AC voltage is set to a normal value (115 Vac or 230 Vac), and the output current is set to a normal value or full load current. Then, the output voltage of the power supply is measured with a digital voltmeter, and its potentiometer (VR) is adjusted until the voltage reading is within the required range.

1.2. Line Regulation

Line regulation is defined as the ability of a power supply to provide a stable output voltage in response to variations in the input voltage. To accurately measure power regulation, the following equipment is required:

• A power supply capable of providing variable voltage, capable of supplying at least the minimum to maximum input voltage range of the power supply under test.
• A root-mean-square (RMS) AC voltmeter to measure the input power supply voltage.
• A precision DC voltmeter with a regulation at least ten times greater than that of the power supply under test.
• A variable load connected to the output of the power supply under test. The typical connection of the test equipment is shown in the diagram below.

The test procedure is as follows: After the power supply under test has reached thermal stability under normal input voltage and load conditions, measure and record its output voltage values at Min (low), Normal (normal), and Max (high) input voltages, respectively. Power regulation is typically expressed as the percentage of output voltage deviation caused by input voltage variations under a normal load, as shown in the following formula:

Line regulation can also be expressed as follows: under changes in input voltage, the deviation of its output voltage must be within the specified upper and lower limits.

1.3. Load Regulation

Load regulation is defined as the ability of a power supply to maintain a stable output voltage in response to changes in output load current.

The required equipment and connections are similar to those for load regulation, the only difference being the need for a precision ammeter connected in series with the output of the power supply under test. See the diagram below:

The test procedure is as follows: After the power supply under test has reached thermal stability under normal input voltage and load conditions, measure the output voltage value under normal load. Then, measure and record the output voltage values (Vmax and Vmin) under low (Min) and high (Max) loads, respectively. The load regulation rate is usually the percentage of the output voltage deviation rate caused by the change in load current under normal fixed input voltage, as shown in the following formula:

Load regulation can also be expressed as follows: Under varying output load current, the deviation of its output voltage must be within the specified upper and lower voltage limits.

1.4. Total Regulation (Cross Regulation)

Total regulation is defined as the ability of a power supply to provide a stable output voltage in response to changes in input voltage and output load current. It is a combination of line regulation and load regulation, and this test provides a more accurate verification of the power supply's performance under varying input voltage and load conditions. Combine regulation is expressed as follows: the deviation of the output voltage from the specified upper and lower voltage limits must be within the specified range under changes in input voltage and output load current.

1.5. Output Noise （PARD）

Output noise (PARD) refers to the periodic and random deviation of the average DC output voltage when the input voltage and output load current remain constant. It is typically expressed in mVp-p peak-to-peak voltage. Generally, switching power supplies specify output noise as less than 1% of the output DC output voltage, with a bandwidth of 20 Hz to 20 MHz. For example, a 5 V output requires output noise to be less than 50 mV. Output noise represents all unwanted AC and noise components on the regulated and filtered DC output voltage.

When measuring output noise, the electronic load must have a lower PARD value than the power supply under test (PST). Simultaneously, the measurement circuit must have good isolation and impedance matching to avoid unnecessary ringing and standing waves on the conductors. Generally, a dual coaxial cable with 50 Ω at its terminals is used. GW Instek's Prodigit 3310 F/G, PEL-2000 series, and ATS-12000 power conversion automatic test system feature electronic loads with low PARD values, and the ATS-12000 power conversion automatic test system can measure the noise values of 8 outputs simultaneously.

1.6. Input Power and Efficiency

1. The input power of a power supply is defined by the following formula:

This is the integral of the product of the input voltage and current over one cycle. It's important to note that P.F. stands for Power Factor. Generally, the power factor of a low-power power supply is around 0.6 to 0.7, while a high-power power supply with a power factor corrector typically has a power factor greater than 0.95. When the input current waveform and voltage waveform are exactly the same, the power factor is 1; depending on the degree of difference, the power factor ranges from 1 to 0.

2. The efficiency of a power supply is defined as follows:

This is the ratio of the total output DC power to the input power. Typically, the efficiency of a PC power supply is around 60 % to 70 %. Efficiency provides verification that the power supply is working correctly. If the efficiency exceeds the specified range, it indicates a problem with the design or component materials. Too low an efficiency will lead to increased heat dissipation and affect its lifespan.

During testing, the GPM-8300 series can be used to measure the input power and power factor of the power supply under test; the 3310 F/G and PEL-2000 series loads can be used to simulate and measure each output voltage, current, and power. The efficiency can then be calculated.

When using the ATS-12000 power conversion automatic testing system, it can measure input and output power, automatically calculate efficiency, and set upper and lower limits to determine pass/fail status.

1.7. Dynamic or Transient Load

A power supply with a constant voltage output incorporates a feedback control loop to continuously maintain a stable output voltage. However, the feedback control loop has a limited bandwidth, restricting the power supply's response to changes in load current. If the phase shift between the input and output of the control loop exceeds 180 degrees with a unity gain of 1, the power supply will exhibit instability, runaway operation, and oscillation.

Actual load currents are dynamic, not constant (e.g., hard drive, CPU, or RAM operation), making dynamic load testing crucial for power supplies. Electronic loads can simulate the most severe load conditions encountered during actual power supply operation, such as the rapid rise and fall of load current, its rate of change, and its periodicity. If the power supply can maintain a stable output voltage under these severe load conditions without overshooting or undershooting, it passes this test.

The 3310 F/G, PEL-2000 series electronic loads can simulate various high/low load cycles, rising/falling current slopes, and high/low load currents, enabling the testing of a power supply's responsiveness to dynamic loads.

1.8. Power Good / Power Fail Signal

The Power Good Signal (PGS) is a signal sent by the power supply to the computer system. It notifies the computer system when its output voltage stabilizes, allowing the power-on process to begin. The Power Fails Signal, on the other hand, indicates that the power supply's output voltage has not reached or has dropped below a specific threshold. Both are typically represented by the PGS signal, indicated by a logic change: a logic value of "0" or "High" indicates a good power supply, while a logic value of "0" or "Low" indicates a power failure. Please refer to the timing diagram below:

The power supply's power-on time is the time from when the output voltage stabilizes until the PGS signal changes from "0" to "1", typically between 100 ms and 2000 ms. The power supply's power-off time is the time from when the PGS signal changes from "1" to "0" until the output voltage drops below the regulated range, typically greater than 1 ms. The ATS-12000 power conversion automatic test system can directly measure the POWER GOOD and POWER FAIL times.

1.9. Start-Up and Hold-Up Time

Start-Up time is the time from when the power supply is connected to the input until its output voltage rises to within the regulated range. For example, for a 5 V output power supply, startup time is the time from power-on until the output voltage reaches 4.75 V.

Hold-up time is the time from when the power supply's input is cut off until its output voltage drops outside the regulated range. Taking a 5 V output power supply as an example, the hold-up time is the time from power-on until the output voltage drops below 4.75 V, typically 17 ms or more. The timing sequence of startup and hold-up time is shown in the following diagram:

The ATS-12000 power conversion automatic test system has the capability to test set-up time and hold time. Upper and lower limits can be set, and the system can determine whether a test is qualified or unqualified during testing.

2. Protection Action Tests

2.1. Overvoltage Protection (OVP)

When the output voltage of a power supply exceeds its maximum rated voltage, it will shut down its output capability, a process known as overvoltage protection.

Overvoltage protection testing verifies whether the power supply can react correctly to the aforementioned abnormal conditions. Overvoltage protection is particularly important for voltage-sensitive loads such as CPUs, memory, and logic circuits, as these valuable components can suffer permanent damage and significant losses if the operating voltage exceeds their rated value. The output voltage waveform of a power supply in an OVP situation is shown in the following figure:

Output voltage waveform when the power supply experiences OVP
t1：Output voltage begins to rise
t2：When the power supply's OVP is triggered, the output voltage drops.

The ATS-12000 power conversion automatic test system can directly simulate the OVP situation of the power supply, and then test the power supply's response to shut down its output. Upper and lower limits can be set, and the system can determine whether the test is successful or not.

2.2. Overcurrent Protection (OCP)

When the output current of a power supply exceeds its rated value, the power supply should limit its output current or shut down its output to avoid damage due to excessive load current. Similarly, if internal components of the power supply are damaged, causing a load current higher than normal, the power supply should also shut down its output to prevent damage.

Overcurrent protection testing verifies whether the power supply can respond correctly when any of the above conditions occur. The ATS-12000 power conversion automatic test system can simulate and test the output overcurrent condition of the power supply. During the test, the load current begins to rise from a preset load current value until the output voltage of the power supply falls below the set critical voltage value.

2.3. Short Circuit Protection

When the output of a power supply is short-circuited, the power supply should limit its output current or shut down its output to avoid damage. Short-circuit protection testing verifies whether the power supply can react correctly when the output is short-circuited.

The ATS-12000 power conversion automatic test system and the 3310F/G and PEL-2000 series electronic loads all have a short-circuit test button to simulate a short-circuit load, allowing direct measurement of the output voltage and current when short-circuited.

Balancing Precision and Resilience: A Guide to Validating Modern Power Supply Performance

In the field of power electronics, DC power supplies are often referred to as the "heart" of a system. With the evolution of high-performance computing (HPC) and AI server architectures, power supplies are no longer simply energy converters, but complex systems with high power density, dynamic response, and intelligent protection. To evaluate the quality of a power system, a deep analysis is necessary from three dimensions: "steady-state accuracy," "dynamic resilience," and "protection mechanisms."

1. Steady-State Accuracy Testing: The Cornerstone of Power Quality

Steady-state testing aims to verify whether the PSU can maintain "absolute stability" of its output under different input and output conditions.

1-1. Total Regulation

Regulation reflects the degree of output voltage deviation under load changes and mains power fluctuations. This includes:

Editor's Note: For modern low-voltage, high-current digital cores (such as the 0.8 V core of a GPU), a voltage drop of even one-thousandth can lead to calculation errors. Therefore, testing the compensation capabilities of remote sensing is crucial.

1-2. Output Noise and Ripple (PARD)

Periodic and random deviations (PARD) determine the "purity" of the power.

2. Dynamic Response and Timing Testing: Facing the Challenges of Harsh Loads

In the era of AI computing, load current switches from standby to full-speed operation within microseconds, posing a significant challenge to the PSU's feedback loop.

2-1. Transient Response

We use an electronic load to simulate high-slew-rate (di/dt) current switching, observing the overshoot and undershoot of the output voltage.

2-2. Timing Criteria: Start-up and Hold-up Time

2-3. Signal Logic: Power Good (PG)

The PG signal must be delayed after voltage stabilization and preemptively turned off before voltage failure. This is a "safety alarm" mechanism to ensure that downstream processors do not operate under unstable voltage conditions.

3. Protection Testing: The Last Line of Defense for the System

Protection mechanism

4. Efficiency and Energy Management: The Core of Green Electricity

Under the 80 PLUS Titanium standard, efficiency testing has become a key technical focus.

5. Conclusion: The Necessity of Automated Testing (ATS)

As product complexity increases, traditional manual measurement methods can no longer handle the cumbersome test combinations. A mature automated testing system (such as the GW Instek ATS-12000 or equivalent) offers three major benefits:

Summary: Measurement is not about getting a "pass," but about understanding the product's "boundaries." Rigorous load stress testing ensures that power supplies operate stably and gracefully under any extreme environment.

Contact Us:

Diana

Digital Service Specialist  

E-mail: diana@goodwill.com.tw