Together with an APC SmartUPS SC 620 this power supply worked at loads up to 385W when powered by the mains. The pair switched to the batteries normally but was not stable then. It is only at a load of 300W that the UPS could work for more than a minute.
Thus, the power supply is compatible with UPSes but the UPS must have a reserve of wattage in order to ensure stability.
Although the PSU offers two independent +12V power rails, the latter are designed in the same way and have similar parameters. Therefore I will only show you the test result for the 12V1+12V4 rail below. The voltage on the 12V2+12V3 rail differs by only a tenth of a volt under the same load.
Our testbed can ensure loads up to 1300W, with up to 1100W on the +12V rail. That’s why I did not test the PSU at its maximum load, limiting myself to the capabilities of the testbed.
The PSU was connected to the testbed with a mainboard connector, one SATA power plug (the ground and +3.3V pins only), two PATA connectors, three 6-pin graphics card connectors and an 8-pin CPU connector. All the pins in every connector were used, save for the SATA plug.
Thermaltake’s power supplies that used the same dual design from CWT had very poor results in my earlier tests. Their voltages sagged quickly under load, making it hard to get even 1000 watts from the PSU. And it did not seem to be a defect of the particular sample because both products I tested then had identical results.
Channel Well’s engineers seem to have introduced some innovations since then. At least, the Xigmatek NRP-HC1501 has very stable voltages. The +12V rail boasts excellent stability, staying within a 1% deflection through the biggest part of the diagram. The +5V voltage fits within a 3% deflection, and the +3.3V voltage is the only one to violate the permissible 5% deflection – but only at near-1300W loads. The latter thing may be partially our testbed’s fault, though. I guess such a high-wattage power supply should be connected to the testbed with even more connectors in order to reduce the voltage drop on the “ground” wires.
Summing it up, the Xigmatek NRP-HC1501 has no problems with delivering stable voltages.
Output Voltage Ripple
The high-frequency pulsation on the +12V rail is within the norm. On the +5V rail the ripple is occasionally as high as the permissible maximum of 50 millivolts. There are periodic spikes above the 50-millivolt limit on the +3.3V rail but they are not as high as to cause any problems.