Hunkey X7 900W Power Supply Review

Recently a new PSU from Huntkey arrived at our lab. As a major power supply maker and the largest power supply manufacturer in mainland China,Huntkeytargets value market segment, and has been popular among entry-level users. Traditionally they build value power supplies with some older designs and a peak wattage rating. Recently Huntkey decides to focus on consumer's ever-increasing green concerns, therefore introduces a brand new green product lineup. The first product named X7 comes with 900W continuous output, SSI EPS12V design, and 80Plus Silver high efficiency. It targets gamers and overclockers who have a multiple GPU rig and care about cost benefits.

Is 900 Watts sufficient for DIY enthusiasts?

OnSLI Zonemost power supplies certified with a triple-SLI system are more than 1100 Watts. However, according to Huntkey, the power consumption of GTX280 cards in 3-way SLI configuration is 520W maximum. One card pulls over 200W peak power, but the second and the third cards won't be fully loaded, unless you're running some power-hungry distributed computing clients. Summing up with the power consumption of CPU and other devices, the whole system consumes about 800W maximum. So Huntkey is confident that 900W is sufficient for even 3-way SLI systems. Higher than that, the gaming performance scales very badly. Up to now 900W is the highest wattage of Huntkey PC power supplies. Perhaps there is a need for an even bigger unit, but it is more expensive and more likely a showcase rather than what a user actually needs.

Those seem like a more realistic attitude towards enthusiast products, but what we expect from a power supply, regardless of its wattage, is its build quality, electrical performance, energy efficiency and operating noise. The specs of X7 900W seems promising, so let's see if it can do something to surprise us.

External Quality Overview

The first thing we're going to look at with the Huntkey X7 900W is its packaging, accessories and documentation. Normally none of these items determine a power supply's quality, the packaging and user's manual can provide us with some information about the product which can be quite helpful, and some useful accessories can make cable management and use more efficient.

click to enlarge

The packaging is a black box in a black cardboard sleeve. On top of the sleeve are the name and wattage (X7 900W) along with an "80Plus Silver" stamp. On the inner side of the lid, a label tells us the unit has a 2 year warranty, which is a bit on the short side. Next to it, are some marketing points given as icons, followed up are some more marketing words:

· Two unique technologies(interleaved PFC and phase-shifted) can guarantee ultra-high efficiency.

What isthatall about? Does this PSUshift a phaseon my motherboard, or does it have abridge inside? Nevermind, please go on to the next line.

· The smart fan of the segmented voltage regulator can create a very quiet environment.

It is better known as a smart fan controller. I know that.

· The pluggable terminal can make the wire cleaning more easily.

Actually it's called as modular cables. Everybody out there is talking about cable management now, and I like that.Yep, I copied those words from theirwebsitepage.

Before introduction at the beginning, I thought you might not be familiar with mysterious codes like "Schottky barrier MOSFET", "advanced hard-switching resonant converter" or "unsymmetric quad-forward design", but you should know about the 80Plus icon. We'll discuss them later. Let's move on.

Here is all you'll get when you purchase a X7 900W PSU, the unit, a bag of modular cables, two handbooks, fix screws and a power cord (which got lost in my big mess of equipment and test samples, fortunately I have replacements). The accessory set is basic. We have just all we need for a power supply, though I'd like to see a couple of cable straps included.

The housing of the unit is nickel-plated metallic grey with a 14cm overhead fan, round fan grid, and has honeycombed venting holes all over the rear panel. There's no AC switch on the rear panel, sadly. The housing is very compact, being 160mm in length. The power label sticks on the side. Go and see details.

This unit has a penta-12V rail design with 18A capacity on each rail, good for a 12A 8PIN PCIe connector, but unhandy if a rail got two of these. The combined 12V capacity is 864W (72A), and combined +3.3V/+5V capacity is 170W. Judging by the color of +12V wires you can tell which connector goes to which +12V rail. The overall capacity of the unit is rated in a standard way: "Rated Power", no "Maximum" or "Continues Power" this time. The unit supports universal input range, as is backed up by its APFC circuitry. In my opinion the specifications have better be shown in a load table.

click to enlarge

All cables but 24PIN ATX12V cable is modularized. X7 900W comes with two 6/8 PIN PCIe, two 6 PIN PCIe, one 4/8 PIN ATX/EPS CPU, two SATA and one HDD modular cable (with floppy connector). The cable length, connector count and rail distribution are demonstrated below.

Cable length is a satisfaction. When the unit is mounted at bottom, the EPS12V 8P cable might be unable to reach the connector in some full-tower cases, but in most mid-towers it's long enough. You can refer to the image below to distinguish 12V rails on modular interface. Connectors and modular plugs are Molex Minifit Jr. Type with Poka-Yoke Design. Each type of modular cable corresponds to a colored interface.

As you can see, rails with PCIe connectors are designed so that no rail exceeds 12A load current, leaving 6A headroom to prevent OCP triggering. However there is one problem if you run triple-SLI on X7 900W. Since you're limited by the connectors, you have to connect the third card to 12V1 rail via peripheral-to-PCIe adaptors, which occupies a quite large portion from that rail and might cause its over-current protection to kick in.

The main features are summarized below.

By the way, the "X7" series was once introduced under a "Cable Master" name. I'm not sure why it is given an "X7" name since there's no "X6" or "X5", but it's definitely better than a "cable master".

On next page we'll take it apart and see its guts.

Internal Build Quality, Overall Look

Now it`s time to take it apart and investigate its guts.

Once we remove all the screws and open the housing we're looking at a fairly traditional yet somewhat dense layout. Two bar-like aluminum heatsinks split the main board into three major zones, APFC & Primary side, power transformers, and secondary side. To fit a 900W power supply into a housing only 160mm in length (including modular PCB), the main PCB gets pretty narrow and packed with components. However components are well arranged on the PCB and carefully secured by caulking. Components such as coils are protected with Heat-shrink tubing, and wires that lead to the modular PCB are neatly routed. Exposed solder joints are protected by Heat-shrink.

The solder side looks even better than the component side. Soldering job is to a very high standard. All solder joints and leads are neatly treated. Some sensitive leads are covered by rubber pads for insulation. The power traces on the secondary side are reinforced by solder stripes with high quality wave soldering. Extensive use of SMD components ensures high mounting density, and the component mounting is also very good. On the PCB, you can see this letter "HK900-11PEP", is printed to this side, the proof of a Huntkey product. To our suprise, the primary side PFC and PWM controller chips, which are usually situated on a daughter board, are mounted directly on the main PCB. We'll discuss them later.

The PCB itself is a double layer gold-plated board.Using gold-plating on solder pads is a good idea which prevents oxidation, keeps the surface solderable, and enhances aesthetics. The board should be using at least 2oz thick copper layers, since it's pretty normal on a unit of this size. Overall this unit is nicely built and product process is taken good care of.

The 140mm diameter 2-ball bearing fan from Yate Loon Electronics (Nice Full) cools the whole unit, model number D14BH-12 (L-S04), rated at 12V 0.70A, which is quite commonly seen in CWT units. Half area of the exhaust side is covered by a hard plastic baffle, which corrects the airflow direction of an overhead-fan structure, but reduces the flow rate and static pressure of the fan almost by half.

The basic structure of an AC/DC power supply (with APFC) is shown as follows:

On X7 900W, the PFC stage uses a technology named "interleaved PFC", and DC/DC converter stage uses the "phase-shifted ZVS full bridge (PS ZVS-FB)" topology, with "synchronous rectification (SR)" on the secondary side, so the figure becomes like this:

The relevant areas to the part aforementioned are marked with colors.

We'll follow the sequence from input to output during the introduction. First let's look at the EMI filtering stage.

Internal Build Quality, EMI filter

EMI filter is the filter on the input side which rejects the conducted electromagnetic interference (EMI) coming into the PSU and going out of PSU to the grid. It also rejects or absorbs the line surges which might harm computer components, thus is also called "transient filter". It is often an indicator about the build quality of a power supply. Due to the parasitic elements of components, a single-stage EMI filter can't provide sufficient filtering, hence at least two-stage filter is required.

First stage of the EMI filter is located on a small card fixed to the AC receptacle, which provides the best noise filtering. It consists a CM choke (in Heat-shrink), a pair of Y caps (blue disc), two X caps (grey box), and an NTC thermistor (in Heat-shrink tubing) which limits the startup inrush current. The wiring leads to the main board wraps around a ferrite bead (covered by Heat-shrink) to reduce CM interference.

The NTC thermistor presents a higher resistance initially and a much lower resistance when heated up, thus limits the startup inrush current during primary cap charging at turn on. After the primary cap is charged, a relay bypasses the NTC thermistor and presents "zero" resistance. It not only helps efficiency a bit, but also ensures that NTC thermistor is ready (becomes cool) for the next cold startup. The use of relay together with NTC thermistor, sometimes referred to as "soft-start circuitry", is especially important in big units with large inrush current. Good job Huntkey.

The second stage of EMI filter, located on the main board, consists of a CM choke, an X capacitor (the grey box), and a large greenish MOV disc for surge protection. A fuse (in Heat-shrink) stands next to the MOV. A red metalized polyester cap (in Heat-shrink) smoothes the input of the APFC circuit.

The EMI filter stage of X7 900W is flawless. Now let's move on to the bridge rectifier(BDR) and APFC circuit.

Internal Build Quality: Bridge Rectifier and APFC

A BDR is fixed to the primary heatsink using a screw, a nut and red thread locker. It is LL25XB60 from Japanese manufacturer Shindengen, rated at 25A/600V -- Almost a luxury for this power supply. The huge derated use of BDR here ensures a safety margin against inrush current.

Now we come to the APFC stage. APFC circuit acts as a current shaper which shapes the input current sinusoidal, and a pre-regulator which boosts the pulsating rectified mains up to a loosely regulated ~380V DC bus voltage. Conventional APFC stage uses a single-phase CCM topology, whereas a 2-phase interleaved CCM topology is used on X7 900W. By making two smaller CCM PFC work out of phase, it reduces the input and output ripple current, current stress on PFC FETs, and improves efficiency by 1 or 2 percent.

As is shown, the PFC stage includes two symmetrical boost PFC circuits, each gets of a MOSFET, a UFRD, a choke and other tiny components. In X7 900W, each phase gets an FCA20N60 (20A 600V) from Fairchild Semi as the PFC FET, and an LQA08TC600 from Q-speed as the boost diode. Two bulk capacitors filtering the output of APFC are Japanese from Nippon Chemi-con, both rated at 85c/450V, with a capacitance of 330uF and 470uF. The voltage rating is sufficient, and the total capacitance of 800uF can meet the hold-up time requirement even under full load.

These two tiny ferrite coils in Heat-shrink are current sensing transformers of each PFC phase.

The PFC controller chip is UCC28070 interleaved PFC controller from Texas Instruments/Ti, mounted on the back side of the PCB. Each PFC FET is driven by a transistor pair (totem-pole) in SO-8 package.

Now let's take a look at the PWM stage of X7 900W which features a soft switching technology.

Internal Build Quality, Primary side

As we mentioned, the DC/DC conversion stage of X7 900W features a topology named "phase-shifted ZVS full bridge", or PS ZVS-FB(PSFB) for short. Traditional topologies such as double-forward or half-bridge are all "hard switching" topologies. In a hard switching topology the switch, aka a MOSFET, turns on with a high voltage (typically some hundred volts) across its drain and source. During the actual switching interval, there's a finite period when this voltage is still high and current flow thru the FET is not zero, generating a significant heat dissipation (aka switch loss) in the FET according to the basic P=V*I law. As switching frequency increases, so does switch loss. The leakage inductance of main transformer is an "unwanted" element in hard switching since it results in a spike across the FETs. As an alternative, a soft switching converter uses a certain mechanism to make voltage or current across the switch reduces to zero before it turns on/off, eliminating the switch loss, thus achieves higher efficiency than a hard switching converter. PS ZVS-FB drives four primary FETs in a phase-shifted way. By using the intrinsic capacitance of FETs and leakage inductance of transformer as resonant elements, all four switches can achieve ZVS with proper design, making PS ZVS-FB a good choice to achieve very high efficiency and power density. The schematic and operating waveforms of PS ZVS-FB are shown in the following figure (thankYang Bo):

Now we know there is a "bridge" inside this unit. I don't care how you shift it, I just wonder if it is efficient. Moreover we've pulled off the primary caps and chokes for ease of view...

On the full-bridge stage, four FQA24N50 from Fairchild Semi (on the left) are used as switching FETs. Each FET is rated at 24A, 500V and presents an RDS(on) of 200mOhm. For 900W usage they're more than sufficient. On the right you can see one of the PFC FETs (FCA20N60) and a boost diode (LCA08TC600). Since the PFC stage is still hard switching, ferrite beads are added to some of the leads of PFC components to suppress turn-on ringing.

Maybe Huntkey is quite confident that primary switches won't get too hot. The heatsink is barely an aluminum bar about 6mm thick.

A tiny inductor connected in series to the primary winding of the main transformer takes the place of leakage inductance as a resonant inductor.

The PWM controller chip, UCC3895DW from Ti, is also mounted on the back side. Huntkey didn't use isolated transformer to drive high-side and low-side FETs. Instead two IR2113S high&low side FET drivers from IR are used. What a feast of chips! (Sorry we don't have potato chips here~)

Internal Build Quality, Secondary side

Follow up is the synchronous rectifier circuit.

The +12V output is rectified using power MOSFETs instead of conventional Schottky barrier diodes (SBD). This is a technology called "Synchronous Rectification" or SR for short. SR can improve efficiency significantly; however the control method presents much more complexity whereas diodes need no control at all. The SR stage consists of four IRFB3207 FETs (75V/180A/3.6mOhm typ.) in paired configuration. The two minor rails are independently regulated, and each rail uses an STPS3045CT Schottky rectifier paired with an IRF1404 FET as a "semi-SR" rectifier, rather peculiar. It can be considered as a modification to the conventional SBD solution to improve efficiency with a slight increase in cost, while maintaining the independent regulating ability of conventional mag-amp post-regulators. I'd love to see DC/DC VRMs used here though. One more thing, (cough) the secondary heatsink is also an aluminum bar. Didn't Huntkey use to screw a piece of scrap metal there?

Let's take a note at a small card between the main and the standby transformer. The two toroid coils are transformers which relays the MOS driver signal from primary side to the gate of SR FETs on the secondary side to control the semi-sync rectification of +5V and +3.3V. By the way, the standby source is handled by a TNY278P switching regulator, a latest TinySwitch-III product from PowerInt good for 21.5W in openframe environment. Later we'll see its efficiency.

The three major outputs are independently regulated using the popular "double mag-amp" topology, so we see three large toroid coils on the secondary side. All magnetic cores are black in Sendust/Koolmu material. +12V is filtered directly by four Teapo capacitors (SC Series, 2200uF 16V 105C). +5V, +3.3V, +5VSB output get a pi filter each (the additional filtering coils also act as current sensors), where filter caps are Chinese from Fcon (2200uF/10V/105C), KSC (3300uF/6.3V/105C) and Fcon+Teapo respectively. A ferrite bead is added to the 5VSB output to reduce the electrical noise. While Teapo is a well-known Tier-1 in Taiwanese suppliers and generally considered safe, the quality of Fcon and KSC caps are unknown to us. Interesting though Huntkey has used quite many Fcon and KSC caps across their entire product lineup. Perhaps they're confident with them on unimportant rails or they might just want to save a couple of pennies. Finally, there're five +12V rails so we can see five shunt resistors for OCP/SCP.

At the modular connector PCB, every rail including 5VSB is filtered by a small Teapo capacitor (SC Series 25V/220uF/105C) as a secondary filter stage, eliminating interference from primary side or transformers. Solder side of this PCB is also exceptional. I have to nitpick here that the exposed solder joints aren't protected with sleeving. They are indeed fragile and can cause a short after vibrations and some thermal cycles.

Housekeeping job is handled by a Weltrend WT7510 supervisor chip together with two LM339 comparators mounted on a separate card on secondary side, providing all necessary protections (OVP/UVP/SCP/OCP .etc) and power good output (PGO).

The thermistor fixed to the secondary heatsink is packaged in such an interesting way. It's used in the fan speed regulator circuit.

Overall the X7 900W power supply is very well built with good assembly and excellent soldering. While we don't like some capacitor choice and the heatsink style, there isn't much we need to complain about. Let's move on now and see how it performs in load tests.

Output Stability Test:

The first part of the load test session, output stability test (or Load Regulation test more formally), is carried out in this way. The AC input is set at 115VAC/50Hz, ambient temperature is about 23 degrees Celsius, and the power supply is run under six different load patterns, 10%, 20%, 50%, 75%, 100% and 110% of its labeled capacity. Load distribution follows theEPRI test method(or better known as "80Plus method"). Since the unit is tested in a cool environment, it shouldn't have problem in the sixth test, we just use it as an indication of the unit's robustness. The result at room temperature represents what you'll get in a modern chassis with separate power supply chamber or a bottom power supply location. Here are the results.

Now here's the good news. The voltage of all rails were very well controlled and maintained inside +/-3% tolerance. +12V rails presented good output accuracies between -2.5%~1.5% with load regulations between 2~3%. Minor rails in general are less stable, but still managed to stay within +/-3% tolerance, with +5V being the poorest with a half-decent 4.1% regulation. In this test, the voltage regulation of X7 900W, if not that spectacular, is still a pleasant surprise.

Please refer to the following spreadsheet for figures in detail:

Cross Loading Test:

No machine is an ideal balanced load. Chances are that a power supply is presented to a load mostly on +12V, or heavy on +5V but much less on +12V. Some power supplies are prone to deliver too high or too low output voltage under unbalanced load, which cause troubles. A Cross Loading Test simulates a series of unbalanced loads according to the cross-loading diagram in ATX12V and EPS12V design guide, and the output voltage should maintain +/-5% tolerance in all cases. Since X7 900W complies with ATX12V V2.3 and SSI EPS12V 2.92, we have adopted the 900W Cross-load diagram in EPS12V specification as follows.

We already know it's an independent regulated design. Here are the results.

The image above shows the maximum deviation of every rail. Amazing isn't it? The following images show the cross-regulation results in a cross-loading chart style. Green plot indicates that the voltage deviation does not exceed ± 2% on that load point, yellow dot indicates that voltage deviation is between 2~4%, the red dot indicates that voltage deviation is between 4% to 5% which is dangerous, black dot is out of spec.

Most dots are green. Peak voltage deviations of all 12V rails are within 2%, deviations of 3.3V & 5V are within 3%. It's almost perfect in this test.

Detailed figures are listed below:

100% load burn-in torture:

Coming up next is our four-hour survival test. Huntkey didn't specify what temperature this unit is labeled at, but we're curious. We've made a hot-box with a couple of wire-wound resistors and fans. Let's meet it:

During the four-hour test session, the unit under test is put into the box and fully loaded continuously. The temperature at PSU's air inlet is controlled at about 45 degrees Celsius. The power supply under torture should feel as comfortable as having a turkish bath, until it survives the whole process, or eventually gives up and makes a "pop" sound. The unit is fully loaded at 115VAC input which puts the most stress. Will X7 survive?

video

Yes X7 made it. During the test process the temperature of exhaust air peaked at about 70 degrees, and voltages on all output were stable. Here are the figures.

Efficiency & Standby Efficiency:

Uh-oh, this time the result is imperfect. It is safe to say X7 900W is very efficient. It boasted very high efficiency from 10% load up to 50% load over the 80Plus baseline. As a bonus it's 80% efficient at a 90W output, meaning it'll be efficient with any rig. However under high load it's not as efficient as we expected. The efficiency dropped quickly as load rose, being only 84% at full load, one percent below Silver line. There are some difference between our test method and EPRI's we think worth mentioning. We don't do the measurement right until UUT is loaded for two seconds. We collect the datum after a 15 to 30 minute burn-in procedure, in a warmer internal temperature. The sample didn't hold on to the entire 80Plus Silver curve, yet quite efficient nonetheless.

The power factor (PF) started at 0.98 at 10% load and rose to above 0.99 from 50% load up. This is very good and quite typical for APFC.

Standby consumption of home electronic devices make up of a large portion of global energy waste and carbon emission. Intel ATX12V guideline recommends that the standby efficiency under 100mA, 250mA, 1A load should be no less than 50%, 60%, 70% respectively. A coming EU regulation and some initiatives require a standby power consumption to be less than 1W. We follow these two criteria and test the power supply at 115VAC/50Hz line input. X7 900W met all these requirements and achieved an astounding number: only 0.1W (barely measurable on our WeiBo power analyzer) at zero-load! In summary, the standby efficiency of X7 900W is outstanding.

AC Ripple/Noise

"Ripple" refers to periodic waveforms in DC output, while "Noise" is periodic or random "spikes" and "disturbances" in output. They're all "unwanted" AC elements, which disturb the operation of digital circuits, or cause filter capacitors to age quickly when it is too high. ATX12V spec requires the peak-to-peak value of ripple&noise on +12V, +5V, +3.3V and +5VSB outputs to be no more than 120mV, 50mV, 50mV and 50mV respectively. Our ripple measurements are taken with a Rigol DS1052E oscilloscope (50MHz/1GSPS) in compliance with Intel ATX12V guidelines, with a 10uF Tantalum and a 0.1uF ceramic decoupling capacitor across the probe head. We collect both the regular ripple waveforms and waveforms at switching frequency for your study. The test method is in such way that at 100% load the output rail measured is fully loaded and the whole power supply is fully loaded simultaneously, so you're seeing the worst case scenario of a rail is delivering its maximum labeled current, unlike under a balanced load when it's less stressed.

		+12V	+5V	+3.3V	5VSB
100% Load	low-frequency
	high frequency
75% Load	low-frequency
	high frequency
50% Load	low-frequency
	high frequency

The period of ripple waveform indicates a switching frequency of approximately 82KHz. Under full load, four rails have about 47mV, 31mV, 24mV, 21mV peak-to-peak ripple/noise respectively. All rails except +5V rail are below 50% of the limit. The ripple on +12V is only 47mV at a massive 72A output. Even +5V rail is well within spec. The ripple suppression of X7 900W is excellent in general.

Acoustical noise impression

The fan curve tells almost everything.

In load regulation test 1 and 2 the noise is inaudible, in test 3(50% load) it's barely audible at a distance of 50cm. Higher is a different story. At 75% load the noise is clearly notable in 50cm distance, and at full load the noise becomes even louder. Sadly, the plastic baffle does have some negative impact on noise control above 75% load. X7 900W is a quiet power supply in general, but can be very noisy when fully loaded. When it comes to cooling performance X7 900W is only so-so despite the high rotation speed. At 75% and 100% load the fan is already very loud but the temperature rise peaks at about 27 degrees Celsius.

Besides, we have collected some load regulation results at 220VAC/50Hz for readers in 220V regions. Let's see.

Load regulation of 12V rails seemed no different, but the minor rails dropped a bit. They were well inside spec.

Next let's see how the efficiency goes.

Thanks to a higher input voltage condition, the whole efficiency curve rose above 80Plus Silver baseline, whereas at 10% load it dropped by a fair bit mysteriously. We've double-checked and this is not coincidence. Our best assumption is that the interleaved PFC stage operated in 2-phase mode during Test 1 of 115V test session, but in single-phase mode during that of 220V test session, so the efficiency of PFC stage is actually higher at 115V input, 10% load.

Final Thoughts and Conclusions

Now we've completed our tests it's time to summarize what X7 900W has brought to us.

Build Quality

Externally the unit comes with a compact housing, a modular cable system, sufficient connectors for most users and pretty good fool-proof modular interface. Internally the unit is very carefully built, with some of the best mounting and soldering quality we've ever seen. The circuit design and most component choices are very good. We're not satisfied with a few details, including some secondary caps and the absence of some heatsink sleevings, but in general it is of high quality. The cooling could be better if Huntkey has used a better heatsink/fan design. The unit came with basic accessories. The 2-year warranty seems too short, given many major PSU brands are providing 5 yrs to 7 yrs warranty with their enthusiast products.

Electrical Performance

The electrical characteristics of X7 900W is awesome. In all load tests (115VAC) the voltages of three major rails stayed rock solid within +/- 3% tolerance, and the load regulation on most rails except +5V (4.1%, still good) is decent. The cross-load test didn't mean a thing to it, and the torture test didn't cause trouble either. As with DC output quality, ripples on all rails are well within spec and the 47mVpp ripple of +12V at a massive 72A output is excellent. Its efficiency has left a bit to be desired since efficiency missed the 80Plus Silver at full load by one percent (As previously mentioned, our testing was more rigorous than 80Plus), but the unit is efficient in general and standby efficiency is again outstanding.

Acoustical Noise

Since our test enviroment isn't that silent we don't have a precise measurement of its noise. What we can say is that X7 900W can be considered as "quiet" at up to 50% load (450W), and becomes pretty noisy under higher load. Considering the type of rig it will be powering, its acoustical performance can satisfy most users.

Final Words

Huntkey X7 900W is a power supply that impressed us. It is indeed a solid power supply which not only succeeded in our full-load torture test, but boasted very good performance in every aspect. The internal build quality is another highlight of this product, being exceptionally good compared to some other popular OEMs. Right now the MSRP in NA market is unclear to us but from price in Australia (about USD $170) our estimation is that it'll be priced competitive. Kudos to Huntkey engineers for their work on this unit and it might be the time that Huntkey revamp their market strategy and bring some real good stuff to computer enthusiasts. Keep honest rating, extend the warranty period and you're on the right track, Huntkey.