A closer look
We were immediately impressed by the presentation of the HX620W. While some power supplies come with a boring brown box and rubber-banded cables, Corsair’s unit arrived neatly-organized in a striking box.
It's rare that we see such orderly presentation in a power supply.
Removing the unit from its packaging, we can see that Corsair has stuck with their traditional look by opting for texturized black metal accented with a single color. The following images show the attractive jet black and red combination that Corsair has chosen for their unit.
The rear of the HX620W. There is no 110v/220v switch as the power supply is auto-switching.
The label containing safety certifications and power output dominates the top of the unit.
The interior-facing side of the HX620W contains a suite of jacks for modular molex and PCIe cables.
The assortment of connectors on the HX620W are provided in a swank nylon case. This is much better than the single-use plastic bags that many other manufacturers use.
We really appreciated that the HX620W's molex cables are flat, as opposed to heat-shrunk or loomed.
The HX620W also provides a excellent variety of connectors which are detailed in the table below:
We were especially impressed with the flat cables. While the power connectors are rounded, the device cables are all flat. For many system integrators and airflow-conscious enthusiasts, this is a serious coup. Beyond the obvious benefits for the system’s ability to breathe, they make a great impression in cases with a window.
Lastly, it wouldn’t be a proper review if didn’t tear the sucker open to get a look at its innards. Who cares about a pesky sticker?
OOPS.
The quality of the internal components is evident at first look.
Regarding the components
Corsair has recently obtained its own Underwriter Laboratories number which has made it difficult to identify the company responsible for producing the guts. In the past, Corsair products like the HX1000W and the TX750W have been known to source components from Channel Well Technology (CWT). Research has led us to understand that the HX series’ internals are designed by Seasonic.
Testing methodology
We don’t mess around with BIOS readings at Icrontic. We came to the table armed with a DMM, digital oscilloscope and a hand-built load tester, all of which you can read about below:
Digital oscilloscope
We used a Velleman HPS40 digital oscilloscope to check the quality of the output from the HX620W. This particular unit has a sample rate of 40 megasamples per second. This means that our equipment is capable of analyzing variations in the quality of the HX620W’s output at up to 40 million times per second. Compare this accuracy to the typical fluctuations in quality which occur at a tiny fraction of that potential.
We’ve also configured the oscilloscope to measure the root mean square (RMS) voltage of the HX620W’s output. This means that the DC components of the signal are filtered by the Velleman, allowing us to analyze the quality of the AC component at an accuracy of 5 millivolts per division of the oscilloscope’s output.
Digital multimeter
While the oscilloscope busies itself with the accuracy of the AC components of our output, we recruited a Fluke 73 III DMM to test our DC voltage. This device will ensure that our voltage readings are accurate where the BIOS may not be.
Load testing
Our load tester is built using a 16 gauge nickel-chromium resistance wire. This nichrome wire is connected to two load generators:
- 1kW load with 12V applied to a 0.5 ohm resistor for 18A.
- 1kW load with 5V applied to a 0.25 ohm resistor for 20A.
In this case, both load sources are connected directly to the power supply with a standard molex connector.
Lastly, we triggered the power supply to run using a CoolMax PC/SPS tester. This unit will allow us to turn the power supply on and off without being connected to a motherboard, and it’s way more glamorous than a paperclip.
Results
Power supplies are designed to output steady DC current, which should be pegged at a constant voltage at all times. Sensitive electronics require the current to be steady else they may be damaged if the power is too erratic. Any voltage that deviates from a perfectly steady output is known as ripple.
We can calculate the ripple with a value called mVDC_RMS, or the average amount the line’s quality wanders away from our perfect world as measured in millivolts. Higher values for mVDC_RMS indicates a more inconsistent output from our power supply.
+5V rail quality
As testing began, the +5V rail started below the desired 5V at 4.96V. The application of a 20A load further toppled the HX620W to 4.69V. While the voltage did dip, the ripple was scarcely stronger than the smallest value our oscilliscope could measure; this performance was both exceptional and encouraging.
The quality of the output from the +5V rail at idle. This oscillopscope image shows a nominal peak-to-peak fluctuation of 3mV with outlying transients of up to 20mV.
The quality of the output from the +5V rail under load. This oscillopscope image shows a nominal peak-to-peak fluctuation of 6mV with outlying transients of up to 20mV.
+12V rail output
Meanwhile, our +12V rail started out strong at a solid 12.13V. Unfortunately for the HX620W, the application of a 20A load brought the output down to 11.8V. While slipping under a consistent 12 volts is unappreciated, it is not outside of the ATX specification.
You’ll also notice that the ripple on the +12V rail is higher than what was observed on the +5V rail. While this may seem like cause for concern, it’s performance that could best be described as “good.”
The quality of the output from the +12V rail at idle. This oscillopscope image shows a nominal peak-to-peak fluctuation of 10mV with outlying transients of up to 100mV.
The quality of the output from the +12V rail under load. This oscillopscope image shows a nominal peak-to-peak fluctuation of 30mV with outlying transients of up to 200mV.
The HX620W’s +12V rail under load definitely demonstrated instability that would appear in the BIOS or in a voltage monitoring application. While this kind of variation is not cause for great concern, we would prefer to see a more consistent output on the +12V.
Articles RSS