Load Regulation, Line Regulation and Cross-Loading
DC Output Load Regulation
Of course one of the first things we want to see is how well this PSU can regulate the DC outputs and maintain rock-solid voltages. To simulate real world and maximum loading conditions, the ProXStream 1kW PSU was connected to the load testers and supplied with a constant 115 VAC. In this test we are interested in seeing how well a PSU can maintain the various output voltages while operating under different loads.
The ATX12V V2.2 tolerance for voltages states how much each output (rail) is allowed to fluctuate and has tighter tolerances now for the +12V outputs.
The following table lists the DC voltage results at the different loads for the ProXStream 1KW PSU while operating on 115 VAC, 60 Hz.
The PSU produced very good load regulation on all of the outputs across a broad range of loads; even when delivering 1,000 watts of DC power. It is important to note that as the load increases, the output voltages generally decrease. This is primarily due to fixed resistance in the power supply cables (Voltage drop = IR).
DC Output Line Regulation
In this test we are interested in seeing how well a PSU can maintain the various output voltages while the AC input line voltage changes. In the previous Load Regulation test, the AC line voltage was held constant at 115 VAC. Now we will look at how much the DC outputs change as the AC line voltage is changed from 120 VAC down to 90 VAC.
The Line Regulation test was performed with the combined DC loads set to 600W. The AC input voltage to the power supply (via the Extech power analyzer) was adjusted using a Powerstat variable autotransformer. As expected, very little change in the DC outputs was observed.
PC switching mode power supplies provide multiple DC output voltages. Ideally, the total load should be distributed across all the main outputs (+3.3V, +5V, +12V). This means that the combined +3.3V and +5V load should be proportional to the combined +12V load — as one increases, so should the other. Unfortunately, this is not always the case, especially in newer PCs that predominately use +12V and may put only minimal loads on the +3.3V and/or +5V rails.
Cross-loading refers to imbalanced loads. If a PC pulls 500W on the +12V outputs and only 50W (or less) on the combined 3.3V and +5V outputs, the resulting voltage regulation may suffer. A few years ago the reverse was true. It was common for some motherboards (especially AMD) to pull a heavy load on the +5V rail and a minimal load on the +12V rail. At the time, one common practice was to place a dummy load on the +12V output to help balance the overall load, which often resulted in better voltage regulation. (Increasing the +12V load caused the +5V output voltage to increase!)
In the first test we put a heavy load on the +12V outputs (64A/768W) and a light load on the remaining outputs. Even with this large imbalance, the voltages all look very good.
In the second test we reversed the cross-load and placed a heavy load on the +3.3V and +5V outputs (142W) with a light load (4A/48W) on the +12V rail. Once again, the ProXStream passed without problems.
In both tests the measured AC ripple remained consistent with the values observed during the other tests and stayed well under control. It’s interesting to note that the overall power supply efficiency for the first cross-load test (heavy +12V load) was much higher (81.0%) than for the second cross-load test (74.4%). This illustrates a common trait of most PC switching power supplies; the +12V section is more efficient than the +3.3V/5V section.