Who needs a big case when a small one can hold big components? That could be a solid question for anyone who likes putting their PC on their desk for the world to see, particularly when the case in question is short in both height and depth:

Those who love RGB will have to pick one of those colors when deciding between:

• The Red-lit,
• The Green-lit, or
• The Blue-lit

version of Rosewill’s 15x15x10” Cullinan PX, but filling it with your choice of lighted components should result in a spectacular-looking build, thanks to all that darkly tinted tempered glass. The hardest part will be getting one before the sale ends, whereupon the price goes back to $150.

Photos courtesy Newegg

Rosewill has a new keyboard, the RK-9300, which comes in Cherry MX Blue and Cherry MX Brown varieties.

The RK-9300 has a fairly austere look, with an all-black design and an aluminum top panel that extends out and down at a slight angle, like a mock wrist rest.

The keys are backlit but have only blue LEDs--a feature omission that some won’t love, but it’s slightly more of a “grown up” look than all-red backlighting. (Even cleaner would be white backlighting.) There are three lighting modes: one with all the keys backlit, one with all but the numpad area lit, and one with only the WASD and arrow keys lit. The fully-lit option has a breathing mode, and there are four brightness levels. (Presumably, there are actually five, assuming one of the levels is “off.”)

The underside of the keyboard offers a pair of flip-up feet and a cable routing trough with three openings across the top of the keyboard--left, center, and right. On either side of the RK-9300, at the top, are a pair of what our own Thomas Soderstrom would call “cupholder”--some bit of decoration that are kind of snazzy looking but ultimately appear to serve no purpose.

The upper right side of the keyboard has a bank of notification lights. The F keys double as media and volume controls, including brightness, Windows lock, and more. It also features n-key rollover. The RK-9300 features no software, and there’s no way to create macros.

The two versions of the RK-9300 list for $99, but both are available for $85 on Amazon (Blue switches and Brown switches) and Newegg (Blue switches and Brown switches).

Rosewill Tokamak 1500 Power Supply Review

Rosewill chose a weird name for its new high-end line. According to Wikipedia, a Tokamak is a device that uses a powerful magnetic field to confine plasma in the shape of a torus. The purpose of this device is to assist in producing controlled thermonuclear fusion power, which is seemingly safer than nuclear fission. If a fusion reactor suffers any damage or loses even a small control percentage, fusion reactions and heat generation rapidly cease. Quite the contrary for nuclear fission reactors. We don't know how Rosewill came up with this name, but it's at least interesting since it describes a device that isn't widely known to the public.

The Tokamak line consists of only two members: one model with 1.2kW and another with 1.5kW capacity. Both feature 80 PLUS Titanium efficiency and are based on a sophisticated platform provided by Enhance Electronics. It's difficult to combine top-end efficiency with high capacity because energy losses increase under significant loads, especially as operating temperatures go up. Nonetheless, even using analog controllers, modern platforms increasingly satisfy the strict 80 PLUS Titanium requirements.

According to Rosewill, the Tokamak family is designed for gamers, even though today's GPUs are the power-hungry beasts they were a generation or two ago. In our opinion, PSUs with over 1kW capacity are mostly for aggressively overclocked systems that can push energy consumption sky-high. With 1.5kW max power, the flagship Tokamak will probably never hit its upper ceiling, especially since it easily delivers 110% of its rated capacity for prolonged periods. During our protection features testing, we stopped increasing the load at 1908W, with the power meter showing over 2kW of consumption. Thankfully we have a 3kVA AC source that doesn't have a problem delivering such high amperage. And the PSU comes with a heavy-duty cord sporting a C19 coupler, which can handle up to 16A.

Specifications

Again, the Tokamak 1500 is 80 PLUS Titanium-certified, and its maximum temperature for continuous full power delivery is limited to 40°C. Normally it should be 50°C, but because this is a very high-capacity PSU we'll cut it some slack. We'll still perform our full-load tests at 45°C and see how they go, though. Since the Tokamak is covered by all necessary protection features, including over-temperature protection, nothing should go wrong.

The 135mm diameter fan promises quiet operation. However, as we found out, once you push the PSU hard, you'll almost need earplugs. This is one of the loudest PSUs we've ever reviewed and the lack of a semi-passive mode makes the acoustic situation even worse.

Size-wise, the Tokamak is fairly normal given its massive capacity. And this is a heavy PSU because Enhance uses large heat sinks for cooling inside.

A five-year warranty is long enough to convey some confidence, though the competition in this wattage category does offer up to 10-year coverage.

Power Specifications

The single +12V rail is powerful; it can deliver 125A according to Rosewill's specs and even more if you push the Tokamak 1500 beyond its official limits. The minor rails have enough capacity to handle every modern system, while the 5VSB rail looks weak. In a 1.5kW PSU we expect to see at least 4A max current output at 5VSB.

Cables And Connectors

Naturally, Rosewill includes a huge number of cables and connectors, including 12 PCIe and two EPS ones, all available at the same time. In addition, there are 16 SATA and six four-pin peripheral connectors. Cable length is satisfactory, although it's strange that the EPS cables are shorter than the main ATX one. Normally it is the other way around.

The distance between connectors is great enough to avoid problems with parts that aren't close to each other. We aren't fans of the fixed floppy drive connector on a modular cable, though. Rosewill should have provided that as an adapter instead. Thankfully, there are so many cables with peripheral connectors on them that you can avoid this one altogether.

All of the cables use 18-gauge wires (high-capacity power supplies normally use 16-gauge wire). But because our testing shows minimal voltage drop, we don't have a problem with Rosewill's cable choice, particularly since thicker wires add rigidity, complicating cable routing in your chassis. 

Power Distribution

Since this PSU features a single +12V rail, we do not have anything to say about its power distribution.

MORE: Best Power Supplies

MORE: Power Supplies 101

MORE: How We Test Power Supplies

MORE: Picking The Right Power Supply: What You Should Know

MORE: All Power Supply Content

Packaging, Contents, Exterior, And Cabling

Packaging

There's a strange image on the front of Rosewill's box that appears to depict a Tokamak. Around back, a table provides the power specifications information and a graphical list shows all available cables and connectors. We would really like to see cable length specs as well.

The 80 PLUS Titanium badge sits in the bottom-left corner next to a five-year-warranty icon. Two lists cover the product's most interesting features and the package contents. Rosewill says that the Tokamak 1500's mean time between failures is 100,000 hours at 25°C and 80% of maximum-capacity load. This is a satisfactory MTBF period.

Contents

Once you remove the box's top cover, you're greeted by a large bag that holds the modular cables. Underneath it, the PSU is protected by packing foam.

The bundle includes a user's manual, a heavy-duty AC cord, and the bag described above. The bag's quality is only average, so treat it with care. Besides the modular cables and AC power cord, Rosewill's packaging also houses a small nylon bag with a set of fixing bolts and several zip ties. We'd at least like to see some Velcro straps in there as well.

Exterior

There are no decals on the Tokamak's sides. This might look weird to some, but it's fine with us. Rosewill went with a more professional approach, so it avoided gaudy stickers.

We're less enthusiastic about the lack of a power switch, though. This is something Enhance likes to do, and we're disappointed that the practice carries over to such a high-end PSU. Really, a power switch should be mandatory regardless of price; it's the only way to totally remove power from the 5VSB circuit. Without it, you have to pull the cord or use a power strip with a switch of its own. Most manufacturers already use power switches. Enhance is the only one that breaks this fine tradition.

The sides look empty without stickers covering them. This also makes it harder to identify the product's model number. Fortunately, that information is provided on the specifications label, affixed to the bottom.

Around back, the modular PCB includes a number of sockets and several LED indicators that start to blink when load on the corresponding rails exceeds 90% of the max-rated capacity. There is also a Power Good LED indicator, which lights up green when everything is in order.

The fan grille is nice. However, it does look restrictive, limiting airflow inside of the Tokamak 1500. We'll use our IR camera to check this. 

Cabling

The cables use dark wires and are flat, blocking less airflow when they're routed properly. Unfortunately the main ATX cable is a mess. It's not even particularly functional since its wires aren't held together. At least that's not the case for the other cables.

A Look Inside And Component Analysis

Parts Description

Before proceeding with this page, we strongly encourage you to a look at our PSUs 101 article, which provides valuable information about PSUs and their operation, allowing you to better understand the components we're about to discuss. Our main tools for disassembling PSUs are a Thermaltronics soldering and rework station, and a Hakko FR-300 desoldering gun.

The Tokamak 1500 uses a sophisticated platform made by Enhance Electronics. Besides the typical analog controllers, we also find a micro-controller (MCU) that's probably used to provide amperage information to the circuit controlling the LED indicators, located on the PSU's rear side. The PCB uses large heat sinks, which is weird for an efficient PSU that doesn't dissipate much waste heat. Then again, given a 1500W ceiling, even small energy losses can create a significant thermal load. Plus the PCB is loaded with components, so it looks like Enhance simply chose to be safe.

The primary side's APFC converter uses two chokes, which would normally indicate an interleaved PFC where two converters work in parallel with a phase difference between them. This isn't the case though, since the PFC controller is a Champion CM6502S, which doesn't support interleaved PFC operation. The primary switching FETs are arranged into a full bridge topology and an LLC resonant converter is used to boost efficiency. On the secondary side, the FETs that regulate the +12V rail are installed on the PCB's solder side, so they're mainly cooled by the unit's chassis through a thermal pad. Two DC-DC converters are installed on the modular PCB, a method popularized by Seasonic. By putting them on the modular board, energy losses are limited since power takes a shorter route with less resistance.

A majority of the electrolytic caps aren't provided by a Japanese manufacturer, which we prefer to see in this prestige category.

The first part of the EMI filter starts at the AC receptacle with lots of parts, including three X and two Y caps. It is very rare to see so many X caps installed in this stage. As usual, the same filter continues on the main PCB with two more Y caps, another two X ones, two CM chokes, an MOV, and a CM02X installed on the solder side of the PCB, blocking current through an X cap's discharge (bleeding) resistor once AC is connected.

The bridge rectifiers cannot be identified without battling the primary heat sink.

The APFC converter uses four Infineon IPP50R140CP FETs and a couple of CREE C3D10060A boost diodes. Obviously, Enhance used a large number of parts to facilitate increased amperage and minimize losses, since each FET handles less current this way. Two parallel chokes are used in this converter, and we had to remove one of them to clear line of sight to the components mentioned above. The bulk caps are four Chemi-Cons with a combined capacity of 1560uF. Although that's a lot, it's not enough for this unit; we measured a <17ms hold-up time.

The PFC controller, a Champion CM6502S, along with a CM03X Green PFC controller, are installed on a small vertical daughterboard. This is a familiar combination in high-efficiency PSUs.

A large NTC thermistor handles inrush current suppression and a bypass relay supports it.

Four Infineon IPP50R140CP FETs arranged into a full-bridge topology are the main switchers. Their drivers are a couple of Silicon Labs Si8230BD ICs. A Champion CM6901 is the LLC resonant controller; this part is becoming more popular as PSUs get more efficient.

On the secondary side, the +12V FETs are installed on the PCB's solder side. In total, 16 Infineon BSC014N04LS are utilized for this purpose. As in the APFC converter, a lot of FETs are used to minimize power losses under high loads. Those FETs are mostly cooled by the PSU's chassis.

A large number of Unicon caps filter the +12V rail, and they're all rated for 105°C. Unicon is a Japanese company with pretty good products which are made though in either Taiwanese or Chinese factories. However in a $360 PSU we'd rather see high quality Japan-made caps. We also find several KY Chemi-Cons, along with few Rubycon caps, all with the same 105°C rating.

The supervisor IC is a SITI PS223, which supports OTP out of the box.

The couple of DC-DC converters use four Infineon BSC018NE2LS FETs and two ANPEC APW7073 PWM controllers.

The standby PWM controller is a Sanken STR-A6053M, while the rectifier that handles the 5VSB rail is a PFR10V45CT SBR.

The -12V rail is generated with the help of a STMicroelectronics L7912CV regulator.

A large number of Unicon polymer caps are found on the modular board's front side, providing additional ripple filtering on all rails. This isn't enough, though; at least on the +12V rail, ripple is much higher than we'd expect.

On the back of the modular board, we also find a STMicroelectronics STM32F030F4P6 responsible for sending data to the LED indicators' control circuit.

Enhance's implementations usually feature good soldering quality. The Tokamak 1500 is no exception.

The unit's top cover is made of aluminum and, according to Rosewill, it improves cooling efficiency. We think the fan grille should be less restrictive to allow better airflow, which would definitely lower temperatures inside the Tokamak. The cooling fan (Young Lin Tech DFB132512H) is powerful; at 100% duty cycle, it makes a lot of noise. Moreover, its minimum speed is pretty high, so don't expect silence under light loads. A semi-passive mode would have come in handy.

Load Regulation, Hold-Up Time, And Inrush Current

To learn more about our PSU tests and methodology, please check out How We Test Power Supply Units. 

Primary Rails And 5VSB Load Regulation

Load Regulation testing is detailed here.

Hold-Up Time

Our hold-up time tests are described in detail here.

The hold-up time we measured is lower than 17ms, and to our disappointment, the power-good signal is inaccurate. It drops when the rails are already out of spec.

Inrush Current

For details on our inrush current testing, please click here.

The inrush current results are a little higher than the competition in this category.

Load Regulation And Efficiency Measurements

The first set of tests reveals the stability of the voltage rails and the Tokamak 1500's efficiency. The applied load equals (approximately) 10 to 110 percent of the PSU's maximum load in increments of 10 percentage points.

We conducted two additional tests. During the first, we stressed the two minor rails (5V and 3.3V) with a high load, while the load at +12V was only 0.1A. This test reveals whether a PSU is Haswell-ready or not. In the second test, we determined the maximum load the +12V rail could handle with minimal load on the minor rails.

Load regulation is tight, especially on the minor rails. In addition, the Tokamak has no problem delivering more than its full power under high operating temperatures. The only issue we see so far is the fan's loud noise under loads in excess of 600W. The minimum fan speed could also be lower.

With 10% and 20% of its maximum-rated loads, this PSU easily meets the 80 PLUS Titanium requirements. However, in the middle range it falls 1% short of the specification's target. In addition, it's not able to hit 90% under a full load. As we're always forced to mention, the 80 PLUS organization doesn't test under real-world conditions inside of a case. It instead chooses to measure at 23°C +/-5°C ambient.

Efficiency, Temperature, And Noise

Efficiency

Our efficiency testing procedure is detailed here.

Using results from the previous page, we plotted a chart showing the Tokamak 1500's efficiency at low loads, and loads from 10 to 110 percent of its maximum-rated capacity.

Efficiency-wise, this platform performs well, taking over the top spot under normal loads and scoring second place under light ones.

Efficiency At Low Loads

In the following tests, we measure the Tokamak 1500's efficiency under loads significantly lower than 10 percent of its maximum capacity (the lowest load the 80 PLUS standard measures). The loads we dialed were 20, 40, 60, and 80W. This is important for representing when a PC is idle, with power-saving features turned on.

We would like to see a >70% reading with 20W load, but we're getting picky there since this is a 1.5kW unit and 20W is only 1.3% of its rated capacity. The main problem we have is with fan noise, which could be a lot less intrusive if Rosewill used a lower minimum or implemented a semi-passive mode.

5VSB Efficiency

The ATX specification states that 5VSB standby supply efficiency should be as high as possible, recommending 50 percent or higher efficiency with 100mA of load, 60 percent or higher with 250mA of load, and 70 percent or higher with 1A or more of load.

We take four measurements: one each at 100, 250, and 1000mA, and one with the full load the 5VSB rail can handle. 

The 5VSB rail is highly efficient, as you can see in the table and graphs above. Only Thermaltake's TPG-1250D-T achieves an overall higher 5VSB efficiency score.

Power Consumption In Idle And Standby

In the table below, you'll find the power consumption and voltage values of all rails (except -12V) when the PSU is idle (powered on, but without any load on its rails), and the power consumption when the PSU is in standby mode (without any load, at 5VSB).

Standby power consumption is very low, giving Rosewill a big advantage.

Fan RPM, Delta Temperature, And Output Noise

Our mixed noise testing is described in detail here.

The first chart below illustrates the cooling fan's speed (in RPM), and the delta between input and output temperature. The results were obtained at 34°C (93.2°F) to 45°C (113°F) ambient temperature.   

The next chart shows the cooling fan's speed (again, in RPM) and output noise. We measured acoustics from one meter away, inside a small, custom-made anechoic chamber with internals completely covered in sound-proofing material (be quiet! Noise Absorber kit). Background noise inside the chamber was below 18 dB(A) during testing, and the results were obtained with the PSU operating at 34°C (93.2°F) to 45°C (113°F) ambient temperature. 

The following graph illustrates the fan's output noise over the PSU's operating range. The same conditions of the above graph apply to our measurements, though the ambient temperature was between at 28°C (82.4°F) to 30°C (86°F).  

This is a noisy PSU, no doubt about that. With more than 1.1kW of load you're going to want earplugs. Given our high efficiency measurements, Rosewill could stifle some of the noise by using a more relaxed fan profile. Fortunately, up to around 900W in normal ambient conditions, the fan's noise is low enough in the 31-34 dB(A) region.

Protection Features, Evaluated

Check out our PSUs 101 article to learn more about PSU protection features.  

Our protection features evaluation methodology is described in detail here.

OCP on the minor rails is set pretty high, while at 5VSB it's more reasonable. Still, it should probably be lower since we measured more than 50mV ripple with 4.2A load.

Over-temperature protection's threshold seems to be configured properly, at around 125°C on the +12V heat sinks.

OPP is set very high. We had to stop increasing the load at 1908W, although our loaders are good at up to 2.4kW. We didn't want to destroy the PSU or our equipment. Enhance should set the OPP triggering point lower. The Tokamak 1500 may be able to dole out more than 1908W under normal operating temperatures, but under tougher conditions this could lead to a disaster for the PSU and the hardware behind it.

As mentioned, the power-good signal is inaccurate and the inrush current protection should be more efficient.

Cross-Load Tests And Infrared Images

Our cross-load tests are described in detail here.

To generate the following charts, we set our loaders to auto mode through our custom-made software before trying more than 25,000 possible load combinations with the +12V, 5V and 3.3V rails. The load regulation deviations in each of the charts below were calculated by taking the nominal values of the rails (12V, 5V and 3.3V) as point zero.

Load Regulation Charts

Efficiency Chart

Rosewill's Tokamak 1500 does achieve over 94% efficiency, though not in the middle region as the corresponding 80 PLUS certification demands. The sweet spot falls between 370W and 590W with load on the minor rails staying below 35W.

Ripple Charts

Infrared Images

Toward the end of the cross-load tests, we took some photos of the PSU with our modified FLIR E4 camera that delivers 320x240 IR resolution (76,800 pixels).

Temperatures inside the PSU stay pretty low thanks to a combination of high efficiency and a powerful fan.

Transient Response Tests

Advanced Transient Response Tests

For details on our transient response testing, please click here.

Ιn these tests, we monitor the Tokamak 1500's response in two different scenarios. First, a transient load (10A at +12V, 5A at 5V, 5A at 3.3V, and 0.5A at 5VSB) is applied for 200ms while the PSU works at 20 percent load. In the second scenario, Rosewill's contender is hit by the same transient load while operating at 50 percent load. In both tests, we use our oscilloscope to measure the voltage drops caused by the transient load. The voltages should remain within the ATX specification's regulation limits.

These tests are crucial because they simulate the transient loads a PSU is likely to handle (such as booting a RAID array or an instant 100 percent load of CPU/GPUs). We call these tests "Advanced Transient Response Tests," and they are designed to be very tough to master, especially for a PSU with a capacity of less than 500W.

Advanced Transient Response at 20 Percent

Advanced Transient Response at 50 Percent

The +12V rail's transient response is excellent! Of course, we didn't expect anything less from such a high-capacity PSU. Performance is pretty good on the other rails too, though we'd like to see deviations within 3% during both tests.

Here are the oscilloscope screenshots we took during Advanced Transient Response Testing:

Transient Response At 20 Percent Load

Transient Response At 50 Percent Load

Turn-On Transient Tests

In the next set of tests, we measured the response of the PSU in simpler transient load scenarios—during its power-on phase.

For the first measurement, we turned off the Tokamak 1500, dialed in the maximum current its 5VSB rail could output, and switched the PSU back on. In the second test, we dialed the maximum load the +12V rail could handle and started the 1500W supply while it was in standby mode. In the last test, while the PSU was completely switched off, we dialed the maximum load the +12V rail could handle before switching it back on from the loader and restoring power. The ATX specification states that recorded spikes on all rails should not exceed 10 percent of their nominal values (+10 percent for 12V is 13.2V, and 5.5V for 5V).

The 5VSB rail registers a fairly smooth spike, and during the second test we notice a small wave. It's only in the third test that we see a notable spike, though it's way below the limit defined by the ATX specification.

Ripple Measurements

To learn how we measure ripple, please click here.

The following table includes the ripple levels we measured on the Tokamak 1500's rails. The limits, according to the ATX specification, are 120mV (+12V) and 50mV (5V, 3.3V and 5VSB).

Ripple on the +12V rail is terrible. We didn't expect such lousy results from an ultra-high-end PSU like this one. It looks like Enhance focused on efficiency and load regulation at the expense of ripple suppression.

Our results on the other rails aren't as bad, though they're still not able to match the competition.

Ripple Oscilloscope Screenshots

The following oscilloscope screenshots illustrate the AC ripple and noise registered on the main rails (+12V, 5V, 3.3V and 5VSB). The bigger the fluctuations on the screen, the bigger the ripple/noise. We set 0.01 V/Div (each vertical division/box equals 0.01V) as the standard for all measurements.

Ripple At Full Load

Ripple At 110-Percent Load

Ripple At Cross-Load 1

Ripple At Cross-Load 2

Performance, Performance Per Dollar, Noise, And Efficiency Ratings

Performance Rating

The following graph shows the Tokamak 1500's total performance rating, comparing it to other units we have reviewed. To be more specific, the tested unit is shown as 100 percent, and every other unit's performance is shown relative to it.

The overall performance score would be notably higher if ripple suppression at +12V was better. After all, this platform is highly efficient and offers tight load regulation.

Performance Per Dollar

The following chart may be the most interesting to many of you because it depicts the unit's performance-per-dollar score. We looked up the current price of each PSU on popular online shops and used those prices and all relative performance numbers to calculate the index. If the specific unit wasn't available in the United States, we searched for it in popular European Union shops, converting the listed price to USD (without VAT). Note that all of the numbers in the following graph are normalized by the rated power of each PSU.  

Efficient, high-capacity power supplies usually score terribly when it comes to value. However, the Tokamak 1500 shows up in the middle of the pack, pretty much tied with Corsair's old (yet highly competitive) AX1500i.

Noise Rating

The graph below depicts the cooling fan's average noise over the PSU's operating range, with an ambient temperature between 28°C and 30°C (82°F to 86°F).

Aside from its weak ripple suppression, this platform's other major downside is an overabundance of noise when it's pushed hard. Rosewill should have Enhance use a more relaxed fan profile.

Efficiency Rating

The following graph shows the PSU's average efficiency throughout its operating range, with an ambient temperature between 28°C and 30°C.

The Tokamak beats Corsair's AX1500i in an efficiency comparison. However, it's behind Super Flower's 1.6kW Titanium platform, used by the EVGA SuperNova T2 1600.

Pros, Cons, And Final Verdict

Rosewill's super high-end Tokamak 1500 leaves us with mixed feelings. On one hand, it demonstrates very high efficiency, has tight load regulation, its 5VSB rail is one of the most efficient we've seen in this category, and the +12V rail's transient response is amazing. On the other hand, its ripple suppression at +12V is terrible and the fan is ridiculously noisy. We can understand the excess ripple a little bit since Enhance clearly prioritized top-shelf efficiency and tight load regulation. However, we cannot fathom why the company didn't use a more relaxed fan profile and/or a semi-passive mode. 80 PLUS Titanium PSUs don't put out a ton of heat, and we saw this first-hand when we had to constantly use the heating elements in our hot box to keep the temperature up, even with the Tokamak under full load. Normally those elements aren't needed at all during high-load tests since the PSU itself affects ambient temperature. The large heat sinks that Enhance uses inside would easily facilitate semi-passive operation. What we get instead is acoustic readings that start above 30 dB(A).

Fortunately for Rosewill, this is a high-capacity power supply aimed at enthusiasts and overclockers who aren't always picky about noise, so long as performance is solid. Should you find yourself looking for high capacity and quiet operation, though, alternatives like Corsair's AX1500i and EVGA's SuperNova 1600 P2 come recommended.

The Tokamak 1500 has enough good aspects to make it an appealing PSU, including 80 PLUS Titanium efficiency, good build quality, a nice finish, and the cool external design. If ripple suppression was better, we might have considered an award. As far as noise is concerned, we think that's an easy fix since Enhance only needs to modify the aggressive fan profile. Since the thermal loads we see aren't that high, lower fan speeds wouldn't compromise reliability over time.

We'd also like to see Enhance stop putting out platforms without power switches. This is a major omission from any PSU. The fact that the Tokamak 1500 is top-end only adds insult to injury.

Finally, some of you might frown on the idea of non Japan-made capacitors inside of a flagship PSU, even though Unicon caps are considered to be of good quality in general. We must note here that Unicon's HQs might be in Japan, however two of its factories are located in Taiwan with the third located in China. We admit that in such a high-end PSU we would like to see only quality Japan-made caps being used, however the five year warranty that Rosewill provides shows that this unit's caps won't be a problem even on the long run.

MORE: Best Power Supplies

MORE: Power Supplies 101

MORE: How We Test Power Supplies

MORE: Picking The Right Power Supply: What You Should Know

MORE: Computer History: From The Antikythera Mechanism To The Modern Era

MORE: All Power Supply Content

Follow us on Facebook, Google+, RSS, Twitter and YouTube.

Introducing The Cullinan

Many brands got started as “house brands” for major sellers, and while some have stuck (Craftsman tools anyone?), others have spread out. Beginning life as the better-value brand of Newegg, Rosewill quickly picked up several other sellers as it added premium parts to its product line.

The Cullinan makes a bold statement about Rosewill’s premium product strategy by using tinted glass panels on three sides. Thick 0.20” (5.2mm) tempered side panels add rigidity without the need for frames, while the moderate weight of the front panel indicates a thinner glass insert.

The front panel ports are more typical of high-end cases, with two USB 3.0 and two USB 2.0 ports, headphone and microphone jacks, and a two-speed fan controller. Both the power and reset buttons have black-anodized aluminum caps, and the rings surrounding these light up in blue (power) and red (HDD activity).

Apart from the position of the removable warning sticker, both side panels are identical and secured identically with slotted thumb screws upon rubberized locator pins.

The rear of the case shows off the 2.2” of above-motherboard radiator space, along with a pair of knock-outs for external radiators. The included 120mm rear fan can be upgraded to 140mm, and slotted fan mounts allow vertical adjustment to ease the installation of various rear-mounted radiators.

The Cullinan’s top panel and front panel use magnetic filter sheets with rounded edges to provide an inset appearance. The top panel is shorter and only able to support two 140m (or three 120mm) fans, while the front supports three fans in either size. Builders who would like to place two enormous radiators in both locations should consider possible tank clearance issues at the Cullinan’s upper-front corner before selecting radiators.

A power supply intake filter slides out from behind the bottom panel.

The Cullinan has only 0.5” of cable space behind the motherboard tray, and a large portion of that space is consumed by three 2.5” drive trays. A 2.5”-wide support section between the motherboard tray and front panel is inset a full 1.0” to provide additional clearance for the main power cable, but routing in that way forces it under the two 3.5” hard drive trays. And that’s where the third 3.5” drive mount resides, bolted through rubber grommets to the bottom panel via shoulder screws.

Powered by a four-pin ATA-style power lead, the two-speed fan controller is delivered with two empty headers. We used one of those for the rear fan. The L-S-H markings of its top-panel switch correspond to low-stop-high, and both its low-speed and stop settings disable fan LEDs.

MORE: Best Deals
MORE: Hot Bargains @PurchDeals

Building With The Rosewill Cullinan

A power supply shroud covers cables and stops exactly 1.5” short of included fans, so that any 1.5”-thick radiator placed there will touch it. That shroud is also riveted in place, further restricting access to a modular power supply’s cable connections.

A plastic shroud helps builders conceal even more cable length. Removing it provides space for full-EATX (13” deep) motherboards, and the installation kit even includes the extra standoffs.

Located 10” forward of the power supply mounting flange, the lower drive cage does not have any backplane connector, nor rear access to drive cables, so anyone who needs to reach their cables will need to point the interface towards the right-hand panel, where cables are visible through the Cullinan’s tinted side panel. These trays also have 2.5” mounting holes, although these are inset too far to work with right-angle cable connectors.

A cardholder bracket slides open 1/2” to allow card installation, which is unfortunately 1/8” less than the space required to slide a PCIe graphics card over its slot. After wiggling and jamming our test card through that access hole, I was happy to find regular slot screws in addition to the clamping mechanism.

Cables include an ATA-style 4-pin power connector for fans, USB 3.0, USB 2.0, HD Audio, and the switch/indicator cluster. The Power LED connector is split to fit both standard 2-pin and Asus/legacy AT 3-pin spacing. Builders must be advised to attach an ATA power cable to their modular power supplies before shoving it into the limited-access power supply bay.

Roswill was kind enough to include both thumb screws and standard #6 case screws for mounting a power supply, where the thumb screws match those of the Cullinan’s side panel. Shoulders on the 3.5” HDD screws prevent them from being over-tightened against vibration-damping grommets.

One thing Rosewill doesn't mention in the manual is that the upper cable cover also acts as a card brace. The circuit board of our test card wasn’t long enough to reach it, but it still mated nicely with the card’s cooler support frame.

Other than minor cable access difficulties and a card tab access hole that wasn’t large enough to fit a card, component installation went smoothly. I mounted our 2.5” test drive behind the motherboard tray, and removing another 2.5” tray made space for the EPS12V lead.

MORE: Best Cases
MORE: All Case Content
MORE: In Pictures: 40 Unusual Computer Case Mods

Testing Methods, Results And Final Analysis

In spite of any installation woes, the finished product looks great! FYI, the grey reflections are cast by the table it’s sitting on, and the off-white reflection is from a light under that table.

Great looks and photographic augmentations aside, we really want to see how Rosewill’s Cullinan compares to other cases. To the point, I found two vastly different cases of similar size with which to compare the excellence of its glass panels and the inconvenience of its riveted-in power supply shroud/drive cage.

Test hardware and methods remain consistent from the earlier review, and have been kept that way for over a year so that readers can compare data from any of our recently-tested mid-towers and full-towers.

Test Results

Similarities between Cullinan and Z9 Neo temperatures can probably be attributed to fan arrangement. The Z9 Neo may be cheaper, but it does have one more (low-cost, non-controllable) fan.

The Cullinan’s fans aren’t exactly quiet, either. Its tempered glass side panels do a fairly good job of blocking in graphics noise, but so does the Z9 Neo’s plastic panel.

Focusing on performance similarities between the Cullinan and Z9 Neo hasn’t put much light on Enthoo Evolve ATX Tempered Glass performance, but we can see in the cooling-to-noise chart that its lower average temperatures and similar average noise levels produce a far better overall performance profile.

Rosewill quotes a Newegg price of $150 for the Cullinan, where the Enthoo Evolv ATX TG sells for around $40 more. The 27% additional cost can’t be offset by a 12% gain in overall performance, so the Cullinan takes the value lead between the two tempered-glass cases. The Z9 Neo, being a far cheaper case with plastic and steel panels, isn’t competing for the same buyers and was only included because of similarities in both size and internal construction.

That last part puts Rosewill in a bit of a bind. While the Cullinan has some high-end features such as the tempered glass side panels, front panel glass insert, better fans, and better port selection, it still has a few design and manufacturing shortcuts that let us know it’s not a complete contender for the tempered glass version of the Enthoo Evolv ATX. For example, I really despised the abuse I had to put my graphics card through to get it past the small opening for its screw tab. I reached maximum frustration after forgetting to add an ATA power cable to my power supply, since there isn’t enough room inside the riveted-on power supply shroud to fit my hand in and insert the cable end. I even had to ditch one of the case’s SATA trays just to make room for the EPS12V cable. While any of these flaws may have been mere annoyances on a case half its price, each of these are potential deal-killers in the high-end market.

And then there’s the missing eighth slot. It’s even missing from the more ergonomically-designed Enthoo Evolv ATX. We all know why some high-end system builders might need it, and that case makers consistently give us additional mounting space in the other direction because motherboard manufacturers keep labeling 10.6”-deep boards as EATX (13”) rather than XL-ATX (which requires an eighth slot). A case manufacturer could indulge a few extreme builders by adding that slot to its EATX designs, but even that wouldn’t have been a big enough move to mitigate all of the Cullinan’s other hardware installation difficulties.

MORE: Best Cases
MORE: All Case Content
MORE: In Pictures: 40 Unusual Computer Case Mods

Follow us on Facebook, Google+, RSS, Twitter and YouTube.

We paid a visit to Rosewill's booth at Computex and we noticed several new products including cases, PSUs, keyboards, mice and a new liquid cooler model.

CULLINAN

The Cullinan is a mid-ATX tower case featuring three tempered glass surfaces (side and front panels) and LED lighting. It supports up to E-ATX motherboards and it is equipped with a fan/LED controller. You can install a water cooling radiator either at its front, or on top (up to 360mm length). Given this case's features and specifications, the $110 MSRP looks pretty decent.

Raptor 22

The Raptor 22 features a slide-down front cover and it supports up to E-ATX motherboards. Triple radiators of 360mm length can be installed into this case at either the front or the top side. In addition, CPU coolers with increased height and very long graphics cards are both supported. Finally, the MSRP is set at $110.

We should also mention that the Raptor 22 has an easily detachable part on one of its side panels, which allows the effortless installation of HDDs and SSDs, without removing the whole side panel. This is a clever idea and we also liked the fact that this part of the side panel is secured in place with magnets, so it can be removed and installed again easily. 

Himars

According to Rosewill the Himars case is ideal for the construction of a multi-HDD system, thanks to the easily accessible hot-swap drive bays. This mid-ATX case supports up to ATX-sized motherboards, along with 160mm max height CPU coolers, and up to 300mm length graphics cards. This is mostly a chassis for server systems where strong (and very long) graphics cards aren't necessary.

Nautilus

The Nautilus is another mid-ATX tower. It features a windowed side panel, LED fans and up to 240mm radiator support. This case can house ATX motherboards and graphics cards up to 360mm in length. Overall this is an affordable case for mainstream/mid-level systems.

SCM-01

The SCM-01 is a pretty compact case featuring a plain design. It addresses the budget-oriented users who don't need side window panels or other extras, like LED lighting. This chassis is for mainstream systems, equipped with up to ATX-sized motherboards. We don't have any information on its price, but we are pretty sure that it will be quite low.

Power Supplies

Rosewill showed us two Titanium PSUs during CES 2016 and at Computex 2016 we were able to see the Tokamak units in action. There was a Chroma load station installed at Rosewill's booth, which was used to load up one of the PSUs to show us one of its interesting features. At the PSU's rear side are several LED indicators, which will start to blink once the load on a rail exceeds 90% of its max rated capacity. According to Rosewill this is a quick way to check if your system manages to put a huge stress on the PSU. With such high capacities the most likely scenario is that you won't see those LED lights blinking, though.

Both Tokamak units are made by Enhance Electronics, according to Rosewill. Unfortunately, we didn't have access at the units' internals in order to confirm this.

The MSRP is very high and the competition is quite stiff at this price point, so the Tokamak units have to register high performance in order to be competitive. The 1.2 kW model offers 8 PCIe connectors while the 1.5 kW unit raises the bar even higher, with 12 PCIe connectors. You will have to spend a fortune on graphics cards and also have the proper chassis to install them in to utilize all available PCIe connectors.

The power specification tables of both Tokamak units follow.

Keyboards and Mice

The RK-9000V2 features RGB lighting, and all of its lighting effects can be set directly on the keyboard without the use of a software application, as is usually the case. We don't have any information on this keyboard's retail price but given the fact that it uses Cherry MX switches, we don't expect it to be lower than $100.

Rosewill also introduced two new affordable mice, equipped with the optical sensors most gamers prefer. The main specifications of both mice are listed below. As you will notice, their only difference is the color of the paint job.

RGM-25

• 5-Button Optimal Wheel Mouse
• DPI: 1200/1600/3200
• Programmable buttons via Software
• Black Model
  

RGM-26

• 5-Button Optimal Wheel Mouse
• DPI: 1200/1600/3200
• Programmable buttons via Software
• Red Model
  

Liquid CPU Coolers

Rosewill also introduced the Charybdis Liquid CPU cooler, which promises high performance. Just for the record, Charybdis was a sea monster which kept company to another sea monster, called Scylla. So sailors that attempted to avoid one monster would fall into the other one. This way both monsters were kept always busy. According to the Greek mythology Charybdis swallowed a huge amount of water three times per day, creating large whirlpools. So calling a liquid cooler Charybdis is interesting to say the least! [Editor's note: having a Greek writer comes in handy. Also, for those not into Greek mythology, the band The Police immortalized Scylla and Charybdis in "Wrapped Around Your Finger."]

The characteristics of the Charybdis RC120 are as follows:

• Dimensions: 120mm x 168mm x 70mm
• Closed top design
• PWM speed control fan

The only different in the characteristics of the RC240 are the dimensions, which are 120mm x 288mm x 70mm.

Follow us on Facebook, Google+, RSS, Twitter and YouTube.

Specifications, Interior And Exterior

The split between liquid-cooling and traditional power user cases has never been starker, as liquid-cooling fanatics scramble for emptied-out boxes. Those builders often go to extremes in their assessments, implying that the retention of a single front-panel bay to hold an Asus OC Panel is an affront to the entire enthusiast community. Nothing (emptiness) has never been so popular! Traditional bay-heavy designs rebel against a neo-enthusiast PC market where trendiness defines conformity. Replete with three external bays and eight (removable) drive trays, Rosewill's Nighthawk 117 is the latest provocateur.

Ten expansion slots can also be seen through the side window, which is the biggest development I've seen in cases since 2008. That's when Ultra ATX was proposed. Yet when we combine the height of Ultra ATX with the depth of EATX, we end up with a case that can hold virtually any enthusiast-market motherboard short of EVGA's HPTX models. Those who need room for HPTX in addition to drive bays will be looking for a much larger case.

Those ten expansion slots are really the only thing that pushes the Nighthawk 117 into full-tower territory, and then just barely at 23.3" tall and 22.5" deep. Top-panel radiator space and the row of four line-passage grommets behind that space notwithstanding.

Just when you thought the entire Nighthawk 117 was stuck in 2008, we turn to the top panel and find two USB 3.0 ports. That brings the design up to at least a late 2010 feature set. There's also a dual-format (2.5" and 3.5") hard drive dock dating from around the same period, two three-speed fan control switches, four USB 2.0 ports, headset jacks, a slide that opens top-panel louvers, a power LED and, in a day where other companies are eliminating it, a hard drive activity LED. What an extraordinary range of conventional features!

Specifications

Interior And Exterior

The Nighthawk 117's power supply intake vent features a dust filter that slides out from under its lower, rear edge.

You'll have to pull off the face panel to remove the front-panel dust filter, which is built into a screw-on fan panel. The Nighthawk 117 lacks room for the end caps of front-mounted dual-140mm radiators, but builders who don't mind sacrificing hard drive cages will find room for a dual-120mm radiator by using alternative screw holes on the same mounting panel.

A fourth (hidden) 5.25" bay is also exposed by removing the face panel. We're not sure what we'd use that space for, other than perhaps a pair of 3.5" hard drive adapter brackets

The top panel appears far more interesting, as it is drilled for both central and offset mounting of both 140mm and 120mm fans. The offset mounts are provided for additional motherboard clearence when using thick radiators.

Rosewill equips the Nighthawk 117 with enough space behind the motherboard for the main ATX cable and even some cable crossing, leaving enough room after installation to stuff a few unused cables. Grommets dress cable passages, except those at the motherboard tray's top edge.

Builders who need extra card space will find that the center hard drive cage slides out after removing two thumb screws, and that the lower drive cage can be pulled out after removing a few Phillips-head screws. There isn't enough room above and below the 140mm fans for the end caps of a radiator, but those same mounts make radiators an option at 2x120mm.

Hard drive trays secure 3.5" drives using integrated side pins and/or bottom screws, but 2.5" drives are secured only by screws. Either method is slightly less convenient than the 5.25" bays, which have slide latches to disengage locator pins.

Each fan controller uses the power connector of a 4-pin ATA drive, and outputs that power to three 3-pin fan connectors. The Nighthawk 117's front fans have both 3-pin fan and 4-pin ATA style connectors, leaving builders to choose the one they prefer. Using the fan's 4-pin connector defeats the function of the case's fan controller.

Other cables include drive dock data (SATA) and power (4-pin), plus interfaces for HD-Audio, USB 3.0, and two USB 2.0 motherboard headers.

The Nighthawk 117 includes a bag of screws, standoffs and cable ties. Also within that bag is a nut driver to #2 Phillips-head screwdriver adapter, to ease standoff installation.

The Nighthawk 117 has cable passages spaced for ATX and XL-ATX motherboards, including those roughly 10.6"-deep by 12"-high models that have been oddly-misnamed EATX. True 13"-deep EATX boards also fit, but without the convenience of cable passages along the tray's front edge.

The finished Nighthawk 117 has the traditional appearance of a high-end gaming case.

MORE: Best Cases
MORE: Cases in the News
MORE: All Case Content
MORE: Cases in the Forums

Test Results And Conclusion

How We Test

We continue to use our reference PC with its reference overclock for case testing, apart from replacing the X99S XPower AC with the X99S Gaming 7. That replacement has allowed us to compare tighter "Standard ATX" cases, since the XPower motherboard is oversized.

Noise is measured .5m from the case's front corner, on the side that opens. The numbers are corrected to the 1m industry standard—used by many loudspeaker and fan manufacturers—by subtracting six decibels.

Drivers And Settings

Comparison Cases

Test Results

The Nighthawk 117 competes directly against the NZXT Noctis 450 in both size and price, but has far more features. Conversely, Azza's enormous GT1 gets included based on price alone. These are all considered to be true full-towers, with none of those "put taller feet on it" hijinks.

We've noticed that some cases require extended heat-soak periods of up to six hours, while others quit escalating temperatures after two. Rosewill's Nighthawk 117 gives us a clue to the reason, as it behaved normally (no long heat soak) with its fans set to full speed. That lets it match the Azza GT1, which doesn't have a fan controller. Meanwhile, its low-speed temperatures approach those of the Noctis 450, which was one of the first cases I tested on this platform to require extended heat-soak.

The Nighthawk 117 doesn't get much quieter at low speed than at medium speed, but the exhaust fans could be part of the problem. I had to leave those at full speed because they would shut off when slowed. I recommend connecting these two to the motherboard's controller in a real-world build, since those can read the fan speed and ramp the voltage up accordingly.

Because the noise difference was far less than the temperature difference between full fans and low fans, the Nighthawk 117 rates a far better temperature-to-noise ratio at full fans. Those low temperatures allowed it to blow way past the Noctis 450, which admittedly is designed to be used with the additional case fans that accompany an advanced liquid cooling configuration.

Better performance at the same price allows Rosewill's Nighthawk 117 to walk away with a top value score. Of course some readers will say that we used the wrong case for the comparison, and I have an answer for that. Check out what happens when we compare the top result for all tall cases that have been tested with this platform.

Only the Silent Base 800 can match Rosewill's Nighthawk 117 in cooling-to-noise ratio, and it costs more. I looked at even more cases and found that I have to get well into budget mid-towers before any manufacturer is able to reach the Nighthawk 117's price-to-performance ratio.

MORE: Best Cases
MORE: Cases in the News
MORE: All Case Content
MORE: Cases in the Forums

Thomas Soderstrom is a Senior Staff Editor at Tom's Hardware, covering Cases, Cooling, Memory and Motherboards. Follow him on Twitter.

Follow Tom's Hardware on Twitter, Facebook and Google+

We visited Rosewill's booth at CES 2016 and had the chance to check out its latest products. Most of them are already on the shelves with only the Gungnir case expected to be released in about a month from now. The Gungnir is a mid tower case supporting VGAs with up to 40cm length and liquid cooling kits with radiators with up to 28cm length. The side window panel features quick release latches and all I/O ports, along with the fan control switch, are located on the top of the case for easier access.

Rosewill has upgraded its monstrous capacity Hercules PSU, offering higher efficiency that meets the 80 Plus Gold requirements; the previous model was 80 Plus Silver certified. All of the other characteristics look to be the same and only through a product review we will be able to point out the actual differences between the older and the new platform, which lead to this efficiency upgrade.

The original equipment manufacturer of the Hercules-1600S is High Power (Sirtec), which also built the previous version of this PSU. The only downside of the new Hercules unit compared to the older version looks to be the packaging, which won't include the fancy aluminum case anymore; this was obviously removed in an effort to lower the price of the unit. 

Recently, Rosewill has been trying very hard to enter the household appliances market and has several interesting products in this category, including an induction cooker, a double wall kettle device, a pressure cooker and a slow cooker. While our focus is on interesting technology products, we all use household appliances, so it is interesting to see a hardware company trying its luck in this tough market.

According to our sources, Rosewill is also expanding to other markets beyond the U.S. So far, Rosewill products have been sold mostly on Newegg and Amazon, however the company is making an effort to enter the European market as well. It will be nice to see it go global since this will increase the competition in the chassis, PSU and peripheral products market, which will be benefit all consumers.

Aris Mpitziopoulos is a Contributing Editor for Tom's Hardware, covering Power Supplies. Follow us on Twitter @tomshardware, on Facebook and on Google+.

Rosewill Capstone-G1200 Power Supply Review

Rosewill initially partnered with Super Flower, which made the company's entry into the power supply market easier, since the first Rosewill PSUs featured very high performance-per-dollar ratios. However, a company cannot rely entirely on a single manufacturer, and the truth is that demand for Super Flower's services weighs on its ability to keep up. In addition, the prices of Super Flower's high-end platforms aren't as affordable as they used to be. When enthusiasts weren't as familiar with Super Flower, good performance and low prices brought new customers in. However, when a company no longer needs exposure, then it makes sense to raise prices, matching the competition.

Higher costs, coupled with the fact that Super Flower cannot meet excess demand any more, probably pushed Rosewill to approach Enhance Electronics for its new Capstone G line (even though the older Capstones employed Super Flower platforms).

While the older Capstones were split between modular models and those with fixed cables, the Capstone G line-up consists only of semi-modular units. Actually, only the main ATX cable is fixed; everything else is modular, so in fact these units are close to fully modular. The Capstone G PSUs come in six versions with capacities ranging from 550W to 1200W, covering much of the enthusiast market. Today we're looking at the 1200W model with a specific emphasis on whether the new Capstone is any better than its predecessor.

Specifications

Rosewill rates its latest effort for 80 PLUS Gold efficiency. In addition, thanks to a couple of DC-DC converters that regulate the minor rails, the PSU is compatible with the C6 and C7 sleep states that Intel's Haswell CPUs introduced. Unfortunately, the maximum operating temperature at which full power can be delivered continuously is limited t 40 °C. The ATX specification recommends at least 50 °C, and we are going to push the PSU above 45 °C during our testing. Thanks to the Over-Temperature Protection (OTP) that all mid- and high-end Enhance platforms feature, we won't have to worry about the PSU's health, even when we push it beyond its limits. Other protection features cover the basics we'd expect.

The physical dimensions and weight of the Capstone-G1200 are normal given its maximum power. We find compatibility with the ATX 2.3 spec since the PSU only employs one +12V rail. The more recent ATX specs require at least two +12V rails. Finally, the warranty is long enough at five years, while Rosewill's price is reasonable for a 1.2kW Gold-rated PSU. 

Power Specifications

The single +12V rail is powerful, specified for 100A and easily supporting multiple graphics cards on an aggressively overclocked platform. The minor rails are strong as well with 150W combined maximum output. Meanwhile the 5VSB rail has enough juice to charge a tablet.

Cables And Connectors

As we expect from a 1.2kW PSU, there are eight PCIe connectors, along with two EPS ones that are all available at the same time. The number of SATA connectors is also huge, while the four-pin Molex connectors are on a single cable, limiting their numbers.

Strangely, the main ATX cable is much longer than the EPS ones. Usually it's the other way around, since the EPS sockets are farther away on most motherboards. Finally, the 24-pin ATX, PCIe and EPS cables use thicker 16-gauge wires for lower voltage drops, while the other leads employ 18-gauge wiring.

Power Distribution

Since this PSU features a single +12V rail, we do not have anything to say about its power distribution.

MORE: PSUs 101: A Detailed Look Into Power Supplies
MORE: How We Test Power Supplies
MORE: All Power Supply Articles MORE:
Power Supplies in the Forums

Packaging, Contents, Exterior And Cabling

Packaging

The box is large and its graphic design fails to impress us. At least it's is sturdy enough to protect the PSU. Up front, two large icons convey the five-year warranty and Gold-rated efficiency. Rosewill makes it clear that this is a gaming power supply, meaning of course that it supports systems with multiple graphics cards.

Around back, a brief list recounts the product's most interesting features and another graphical list shows the available connectors. The power specifications table is located here as well.

Contents

Packing foam surrounds the unit, offering adequate protection. In addition, the PSU is wrapped in a thick plastic bag.

Rosewill adds a nylon pouch to its bundle, which you can use to store whatever modular cables you don't end up needing. The user's manual is common to all Capstone G models, while the rest of the bundle includes a set of fixing bolts and several zip ties.

Exterior

The PSU is quite large and its design looks interesting, mostly thanks to a unique fan grille. The grille appears highly restrictive though, and we believe the platform would be better served by something that'd allow more air to pass through.

Up front, there's a power switch under the AC receptacle. You'll find the specifications label affixed to the chassis' bottom side.

The lack of a grommet around the ATX cable's exit hole left a negative impression on us. The hole is pretty big given the one cable that passes through it, and we would have liked to see some rubber around the edges for increased protection.

This PSU's aesthetic is gloomy in general, especially considering it's destined for gaming systems with windowed panels. A lack of decals on the sides plays some part in this, though again, the top grille does look interesting. However, as we already stated, it restricts airflow, which might affect the unit's reliability and performance under tough conditions.

Although the packaging provides ample protection against damage, we noticed a defect in our sample that probably had nothing to do with shipping, but could have happened during manufacturing. We fixed it with a flat screw and small hammer, but it's going to cost Rosewill some build quality points.

Cabling

The fixed ATX cable features colored wires and is round, while the modular cables are stealth and flat.

A Look Inside And Component Analysis

Parts Description

Before proceeding with this page, we strongly encourage you to a look at this article, which provides valuable information about PSUs and their operation, allowing you to better understand the components we're about to discuss. Our main tools for disassembling PSUs are a Thermaltronics soldering and rework station, and a Hakko 808 desoldering gun.

This PSU uses an Active Clamp Reset Forward (ACRF) topology in the primary side and the usual stuff on the secondary side: a synchronous design for regulating the +12V rail and DC-DC converters for the minor rails. Its heat sinks are huge, similar to every high-capacity Enhance PSU we've seen. Admittedly, the company's capacitor selection is a little strange, since we found a couple of brands that we don't encounter often in this price range. Not that we have anything against Unicon or Dura Tech caps. However, we don't have any solid information about their reliability under tough conditions. We'd prefer to see Chemi-Con or Rubycon caps instead.

The ACRF topology provides high efficiency at lower cost since there's no need for an LLC resonant converter or FETs with very low RDS-on values. In this case, we have a two power switch ACRF design, where a couple of FETs act as the main switchers (Q1 and Q2), while a third serves as the reset switch (Q3). The reset switch's job is to disconnect the APFC converter's bulk capacitors while Q1 and Q2 are active. When Q3 is open, power is transferred from the primary to the secondary side. This design allows for almost lossless switching of the Q1 and Q2 FETs, as their drain voltage is very low while both are turned off. A couple of Infineon SPW60R099C6 FETs act as the Q1 and Q2 switches, and a Fairchild FQPF3N80C is the Q3 reset switch.

The first part of the EMI filter is right behind the AC receptacle and includes two X caps and two Y ones. It continues on the main PCB with three more X caps, a couple of Ys, two CM chokes and an MOV. Overall, the EMI filter of this PSU is more than complete.

The bridge rectifiers are a pair of GBU1506, which receive cooling from the APFC heat sink and a smaller one.

We spotted a large NTC thermistor before the PFC coil and another smaller one after it. The small NTC thermistor isn't supported by an electromagnetic relay, resulting in quicker cooling and lower energy loss. Instead, it employs a diode. On the other hand, the large thermistor doesn't use anything that would allow it to cool quickly. There's no need anyway, since this thermistor is bypassed. On the solder side of the PCB, we noticed a copper wire that shorts the thermistor's leads, rendering it useless. Enhance could have removed it altogether to save a few cents. More than likely, the PSU couldn't hit 80 PLUS Gold efficiency levels with this thermistor in place, so it was bypassed.

There are two Infineon SPW60R099C6 FETs in the APFC converter, along with a single STTH3006DPI boost diode. The bulk caps are provided by Chemi-Con, rated at 450V and 105 °C, and their combined capacity is 780uF, which is quite low for a 1.2kW PSU.

The main switchers are a couple of Infineon SPW60R099C6 FETs. The reset switch is a Fairchild FQPF3N80C. A large vertical daughterboard hosts the PFC controller (a Champion CM6802 IC), and the ACRF controller is a Texas Instruments UC2715D. On the left side of the same daughterboard, we found the supervisor IC, a SITI PS223 capable of supporting all basic protection features. Finally, in the middle of the board are two optocouplers, which offer isolated communication between the primary and secondary sides.

The FETs that regulate the +12V rail are four Infineon IPP023N04N G.

Two thermistors are attached to the large secondary side's heat sink. The first provides information to the fan control circuit, determining fan speed according to the sink's temperature, while the second thermistor sends information to the supervisor IC.

Two small VRMs generate the minor rails. On them, we found polymer caps from Enesol, a respected Korean manufacturer. The other filtering caps on the secondary side come from Chemi-Con, Dura Tech and Unicon. We also found a single Rubycon electrolytic cap. It seems as though Enhance used what it found available and didn't focus on only one brand.

The 5VSB rail is rectified by a PFR10V45CT SBR, while the standby PWM controller is a Sanken STR-A6062H.

At the front side of the modular PCB, some electrolytic Chemi-Con caps and several Dura Tech polymer ones further suppress ripple. On the bottom of the board, Enhance's soldering quality looks good.

Enhance generally demonstrates good manufacturing quality, and this PCB is a good example. On the solder side and under the transient filter area, we found a CM02X IC that blocks current through an X cap's discharge resistor when AC voltage is connected. This way, less energy is wasted, boosting efficiency.

The cooling fan comes from Young Lin Tech, and its model number is DFB132512H (135mm, 12V, 3W, 1800 RPM, 95 CFM, 38 dB[A]). It uses double ball-bearings, so it should last longer than sleeve-bearing fans (though it's not as reliable as pricier FDB- or HDB-based coolers).

Load Regulation, Hold-Up Time And Inrush Current

To learn more about our PSU tests and methodology, please check out How We Test Power Supply Units. 

Primary Rails And 5VSB Load Regulation

Load Regulation testing is detailed here.

Hold-Up Time

Our hold-up time tests are described in detail here.

As we expected given the bulk capacitors' low capacity, measured hold-up time didn't meet the ATX specification's requirements. There simply wasn't enough space on the PCB for larger caps, which would have also had a negative impact on efficiency. The trade-off could have cost Rosewill its 80 PLUS Gold rating.

Inrush Current

For details on our inrush current testing, please click here.

This platform's small NTC thermistor results in higher inrush currents, taxing your home's electrical infrastructure.

Load Regulation And Efficiency Measurements

The first set of tests exposes the stability of the voltage rails and the PSU's efficiency. The applied load equals (approximately) 10 to 110 percent of the supply's maximum in increments of 10 percentage points.

We conducted two additional tests. During the first, we stressed the two minor rails (5V and 3.3V) with a high load, while the load at +12V was only 0.10A. This benchmark reveals whether a PSU is Haswell-ready or not. In the second test, we determined the maximum load the +12V rail could handle with minimal load on the minor rails. 

Our load regulation measurements show that Rosewill fails to match the competition. This doesn't surprise us, particularly when it comes to the +12V rail, because we haven't tested any ACRF topology-based PSU with great load regulation.

Under a 20 percent load, the PSU easily satisfies the 80 PLUS Gold efficiency requirement (87 percent). With a 50 percent load applied, the margin is a lot closer. But given our warm ambient test environment, we'll give the PSU a pass on this one. However, under a full load, the measured efficiency level was significantly lower than the permissible minimum. Obviously our 48°C operating temperature played a big part in this.

The big Capstone unit proved to be very noisy under tough conditions. During the first two tests, the fan rotated slowly. However, it sped up quickly, and with 70 percent load applied, acoustic output approached 52 dB(A). If you need a quiet PSU, avoid this one.

Efficiency, Temperature And Noise

Efficiency

Our efficiency testing procedure is detailed here.

Using the results from the previous page, we plotted a chart showing the Capstone-G1200's efficiency under low loads, and loads from 10 to 110 percent of the PSU's maximum-rated capacity.

The ACRF topology doesn't facilitate impressive efficiency figures, so the Capstone-G1200 doesn't compare well to its competition, falling to last place in both graphs. Under normal loads, the difference isn't as large. However, under light loads, the Capstone unit is left behind. For context, Rosewill's Photon-1200, which bears the same 80 PLUS Gold rating and costs $10 less, outperforms the Capstone-G1200 at normal and light loads levels.

Efficiency At Low Loads

In the following tests, we measure the Capstone-G1200's efficiency at loads significantly lower than 10 percent of its maximum capacity (the lowest load the 80 PLUS standard measures). The loads we dialed were 20, 40, 60 and 80W. This is important for representing when a PC is idle, with power-saving features turned on.

Under light loads, our measured efficiency levels aren't high enough, though at just 80W of load the PSU passes the 80 percent mark.

5VSB Efficiency

The ATX specification states that 5VSB standby supply efficiency should be as high as possible, recommending 50 percent or higher efficiency with 100mA of load, 60 percent or higher with 250mA of load and 70 percent or higher with 1A or more of load.

We will take four measurements: one each at 100, 250 and 1000mA, and one with the full load the 5VSB rail can handle. 

The 5VSB rail yields a pleasant surprise. This is the first time we've seen such an efficient 5VSB rail, and as you can see in the table above, we crossed the 80 percent threshold in three of our four tests, the highest exceeding 85 percent.

Power Consumption In Idle And Standby

In the table above, you'll find the power consumption and voltage values of all rails (except -12V) when the PSU is idle (powered on, but without any load on its rails), and the power consumption when the PSU is in standby mode (without any load, at 5VSB).

Phantom power is very low, improving the 5VSB rail's efficiency, especially under light loads.

Fan RPM, Delta Temperature And Output Noise

Our mixed noise testing is described in detail here.

The first chart illustrates the cooling fan's speed (RPMs), and the delta between input and output temperature. The results were obtained at 38 °C (100.4 °F) to 49 °C (120.2 °F) ambient temperature.   

The next chart shows the cooling fan's speed and output noise. We measured acoustics from one meter away, inside a small, custom-made anechoic chamber with internals completely covered in sound-proofing material (be quiet! Noise Absorber kit). Background noise inside our chamber was below 18 dB(A) during testing, and the results were obtained with the PSU operating at 38 °C (100.4 °F) to 49 °C (120.2 °F) ambient temperature. 

The following graph illustrates the fan's output noise over the entire operating range of the PSU. The same conditions of the above graph apply, though the ambient temperature was between at 28 °C (82.4 °F) and 30 °C (86 °F).  

The graph looks like a mosaic, at least up to 650W, because the fan control circuit frequently made adjustments through the light and moderate load levels. Overall though, this is a noisy PSU that'll get loud enough to annoy you under tough conditions.

Cross-Load Tests And Infrared Images

Our cross-load tests are described in detail here.

To generate the following charts, we set our loaders to auto mode through our custom-made software before trying more than 1500 possible load combinations with the +12V, 5V and 3.3V rails. The load regulation deviations in each of the charts below were calculated by taking the nominal values of the rails (12V, 5V and 3.3V) as point zero.

Load Regulation Charts

Efficiency Chart

The Capstone-G1200 delivers greater-than 90 percent efficiency in a small part of its operational range. Efficiency mostly falls between 85 and 90 percent. This isn't bad until you start drawing comparisons between other Gold-rated PSUs with similar capacity, which typically deliver even better efficiency.

Ripple Charts

Infrared Images

Toward the end of the cross-load tests, we took some photos of the PSU with our modified FLIR E4 camera that delivers 320x240 IR resolution (76,800 pixels).

The temperature inside of our hot box easily rises to high levels due the G1200's waste heat. Inside the power supply, its bridge rectifiers approached their limit after being subjected to a 110 °C operating temperature. However, keep in mind that we pushed the PSU hard with an ambient temperature in excess of Rosewill's specifications.

Despite the elevated temperatures inside, the PSU worked flawlessly, and we didn't encounter any unexpected shut-downs from over-temperature protection. This means that the OTP trigger point is set higher than the levels we saw (strange, given the PSU's 40 °C rating).

Transient Response Tests

Advanced Transient Response Tests

For details on our transient response testing, please click here.

In these tests, we monitored the response of the PSU in two different scenarios. First, a transient load (10A at +12V, 5A at 5V, 5A at 3.3V and 0.5A at 5VSB) was applied for 200ms while the PSU was working at 20 percent of its capacity. In the second scenario, the PSU was hit by the same transient load while operating at 50 percent of its capacity. In both tests, we used our oscilloscope to measure the voltage drops caused by the transient load. The voltages should have remained within the ATX specification's regulation limits.

These tests are crucial because they simulate the transient loads a PSU is likely to handle (such as booting a RAID array or an instant 100 percent load from a gaming platform). We call these "Advanced Transient Response Tests," and they are designed to be tough to master, especially for PSUs with less than 500W of capacity.  

Advanced Transient Response at 20 Percent

Advanced Transient Response at 50 Percent

The +12V rail's performance is very good in these tests; we were also satisfied by the performance of the 5V and 5VSB rails. As usual, the 3.3V rail registered higher deviations. However, in both cases it managed to stay above 3.2V, so we aren't complaining.

Here are the oscilloscope screenshots we took during Advanced Transient Response Testing:

Transient Response At 20 Percent Load

Transient Response At 50 Percent Load

Turn-On Transient Tests

In the next set of tests, we measured the response of the PSU in simpler transient load scenarios—during the PSU's power-on phase.

For the first measurement, we turned off the PSU, dialed in the maximum current the 5VSB could output and switched the PSU on. In the second test, we dialed the maximum load the +12V could handle and started the PSU while it was in standby mode. In the last test, while the PSU was completely switched off (we cut off the power or switched off the PSU by flipping its on/off switch), we dialed the maximum load the +12V rail could handle before switching on the PSU from the loader and restoring the power. The ATX specification states that recorded spikes on all rails should not exceed 10 percent of their nominal values (+10 percent for 12V is 13.2V, and 5.5V for 5V).    

Our scope caught a small spike at 5VSB, though it's nothing to worry about. In the second test, we didn't see any spikes or voltage overshoots. All of that changed during the last test though, where ripple suppression briefly went out of control. More than likely, this has to do with the ACRF topology. We must stress that even short periods of increased ripple can apply huge stress to system components.

Ripple Measurements

To learn how we measure ripple, please click here.

The following table includes the ripple levels we measured on the Capstone-G1200's rails. The limits, according to the ATX specification, are 120mV (+12V) and 50mV (5V, 3.3V and 5VSB).

Ripple suppression on the +12V rail is flat-out bad. We don't want to see anything above 50mV on this rail, especially in PSUs that cost more than $100. Usually, good PSUs register up to 40mV of ripple at +12V and the extraordinary ones stay below 20mV.

At least ripple suppression on the other rails is good enough. Still, what matters most is the +12V rail's performance, since it's used by most system components nowadays.

Ripple Oscilloscope Screenshots

The following oscilloscope screenshots illustrate the AC ripple and noise registered on the main rails (+12V, 5V, 3.3V and 5VSB). The bigger the fluctuations on the screen, the bigger the ripple/noise. We set 0.01V/Div (each vertical division/box equals 0.01V) as the standard for all measurements.

Ripple At Full Load

Ripple At 110-Percent Load

Ripple At Cross-Load 1

Ripple At Cross-Load 2

Performance, Performance Per Dollar, Noise and Efficiency Ratings

Performance Rating

The following graph shows the total performance rating of the PSU, comparing it to other units we have tested. The Capstone-G1200 is shown as 100 percent, and every other unit's performance is shown relative to it.

Rosewill's new PSU takes the lead from the same company's Photon-1200 thanks to better ripple suppression (believe it or not). The older model approached the ATX specification's ripple limit on the +12V, 5V and 3.3V rails during our tests, costing it many performance points.

Performance Per Dollar

The following chart may be the most interesting to many of you because it depicts the unit's performance-per-dollar. We looked up the current price of each PSU on popular online shops and used those figures (and all relative performance numbers) to calculate the index. If the specific unit wasn't available in the United States, we searched for it in popular European Union shops, converting the listed price to USD (without VAT). All of the numbers in the following graph are normalized by the rated power of each PSU.  

In this chart, the Photon-1200 jumps ahead due to its lower price. However, given the lousy ripple suppression we just discussed, we'd favor Rosewill's Capstone-G1200 if we had to choose between the two.

Noise Rating

The graph below depicts average noise over each PSU's entire operating range, with an ambient temperature between 28 and 30 °C (82 to 86 °F).

Without a doubt, the Capstone-G1200 is a noisy PSU. However, the Photon-1200 is even worse.

Efficiency Rating

The following graph shows the average efficiency of the PSU throughout its entire operating range, operating in an ambient environment between 28 °C and 30 °C.

The G1200 doesn't do so well in our efficiency performance metric. However, it isn't far away from the Photon-1200 and other Gold-rated platforms.

We should note that these results were collected under normal operating conditions. In more extreme environments, the efficiency impact varies since it depends on the design of the platform and the parts that it uses.

Pros, Cons And Final Verdict

At a glance, you might wonder why the Capstone-G1200 is worth considering when the Photon-1200 costs $10 less and features full modular cabling. Although the Photon-1200 achieves tighter load regulation on all rails and higher overall efficiency, it has a major flaw: its ripple suppression is particularly bad, especially at high operating temperatures. For this reason alone the Photon-1200 trails Rosewill's Capstone-G1200 in our performance rating chart. On top of that, the Photon generates more noise than the already-loud Capstone-G1200. If we had to choose between both products, we'd pick the Capstone without a second thought.

However, if we weren't restricted to Rosewill's line-up, the high-capacity Capstone probably wouldn't be on our short list. It's an average performer, mostly because it's based on a topology that cannot match more modern designs in efficiency and ripple suppression. To make things worse, Rosewill implements a restrictive fan grille that doesn't facilitate good airflow inside the chassis. That fact, combined with the platform's increased heat dissipation, leads to high operating temperatures once the PSU is pushed. The only way to remove excess heat is spinning the fan faster, contributing to the noise issue. Of course, you can't expect a 1.2kW PSU to be silent, but with the proper design, noise can at least be minimized.

The platform might not be top-notch in terms of design or performance, but it does survive our toughest benchmarks. And we believe it's reliable enough. Otherwise, Rosewill wouldn't guarantee its longevity for five years. The older Capstone PSUs, which were made by Super Flower, used a better platform though. We're forced to wonder why Rosewill deviated from its winning formula. If Super Flower couldn't provide the same platform, why not choose another of the company's designs? Enhance Electronics might be a good OEM, but its high-capacity offerings don't stand up to Super Flower's Leadex platforms.

We are understandably displeased when we encounter a new product that cannot beat its predecessor, and that's what happens here. The older Capstone lines offered good performance along with affordable prices, and helped Rosewill establish its name in this challenging market. On the other hand, the new Capstones offer lower performance and cost about the same. So what's Rosewill's strategy here? If it wants to keep up with its competition, it needs competitive platforms at affordable prices. Even if the company couldn't keep its relationship with Super Flower going for the new Capstones, it should have found a better-performing alternative rather than an inferior platform unable to match other options out there.

MORE: PSUs 101: A Detailed Look Into Power Supplies
MORE: How We Test Power Supplies
MORE: All Power Supply Articles MORE:
Power Supplies in the Forums

Aris Mpitziopoulos is a Contributing Editor for Tom's Hardware, covering Power Supplies.

Follow us on Twitter @tomshardware, on Facebook and on Google+.

Specifications, Interior & Exterior

We don’t often refer to size as a metric of greatness, but when it comes to size the “greatest” form factor enthusiasts have encountered is EVGA’s old HPTX at 13.6” tall by 15” deep. Yet even that wasn’t the tallest board developed for enthusiasts, as Foxconn once attempted to establish a 10-slot, 14.4” form factor called Ultra ATX. Rosewill’s latest big tower, the B2 Spirit, is large enough in both directions to support both standards. More importantly, the tenth “Ultra ATX” rear-panel slot allows HPTX users to put a double-slot graphics card in the motherboard’s bottom slot.

The “Greater Than” symbol is my latest solution to the naming problems of oversized parts, as it’s less likely to be misidentified as a “new form factor” than a plus sign following the form factor’s name. My hope is that readers will ask “how much larger” when they see it in a title. But just how large is Rosewill’s latest “Greater Than HPTX” solution?

Don’t let the slender visual effect of black fool you: Over 26” tall and nearly 26” deep, the B2 Spirit is larger even than Azza’s recently-tested GT1.

Exterior

Potential buyers need not worry that panel rigidity decreases as panel size is increased, as the B2 Spirit is reinforced with nearly 38 pounds of steel and heavy-duty plastic. Worry instead about your back, or whether your reach is broad enough to tote this case without having it slip from your grasp. Spirit of the B2, indeed!

Notice that the top panel has six USB ports, in addition to two audio jacks, under a tiny slide-away cover? That cover is nearly as wide as a dollar bill!

The B2 Spirit is larger than a traditional “Full Tower,” breaking from that tradition by having neither a rear drive rack or second power supply mount. Instead, the extra height is devoted to mounting space for thick top-panel radiators and lower expansion cards.

A pair of fans are mounted to hinged brackets behind the face panel for easy access, replacement, and filter cleaning. Although they look fairly small in the photo, each of these is 140mm square. When it comes to scale, the word “staggering” comes to mind.

A slide-in filter protects the power supply intake plus a pair of 120/140mm fan mounts on the B2 Spirit’s bottom panel. It's over 17.5” long, so most users will probably need to turn their case around to pull it out.

Interior

Marketed both with and without the hyphen in its name, the B2 Spirit includes a 10-drive, and a removable 3-drive drive cage. The smaller cage covers the second bottom fan mount, and the larger cage includes a 2x140 / 2x120mm radiator mount.

The B2 Spirit has access holes for boards both big and huge: The inner cable access holes are spaced for a 12” x 10.7” oversized ATX board, such as MSI’s X99S XPower AC and Asus’ Rampage V Extreme. That means the outer access holes are spaced for something really big!

Rosewill didn’t sneak any extra storage behind the motherboard tray, but the B2 Spirit does have enough space for all your cables.

The B2 Spirit top panel can hold up to three 140mm or 120mm fans, with enough space to sandwich a 1.5”-thick radiator between two layers of 1”-thick fans.

If 3.5” of space above the board isn’t enough for your cooling bits, you could dig into top panel space. Sliding bars that tie together the top panel’s opening mechanism could get in your way, but modders are welcome to customize.

MORE: Best MotherboardsMORE: Best CPU Cooling


MORE: How To Build A PCMORE: All Case Content

The Build, Test Results & Conclusion

The Build

The B2 Spirit includes a reasonable good user’s manual, plus bag of screws with motherboard standoffs and cable ties.

Cable ends include two USB 3.0, one USB 2.0, HD Audio, switch and LED. The power LED lead is split for both standard and Asus (old-school AT) spacing, and the similarly-outdated AC-97 pigtail is eliminated.

Cushioned pins reduce vibrational transmission from 3.5” mechanical drives, and slide out to ease screw mounting of 2.5” drives with the connector positioned in the same place.

Our full-ATX parts look tiny inside the B2 Spirit.

Perhaps the tiny appearance of full-sized parts is why Rosewill decided to use a small window, relative to the huge size of its side panel? That does help the case appear more “filled” in an ordinary configuration.

How We Test

We continue to use our reference PC with its reference overclock for case testing, apart from replacing the X99S XPower AC with the X99S Gaming 7. That replacement has allowed us to compared tighter “Standard ATX” cases, since the XPower motherboard is oversize.

Noise is measured .5m from the case’s front corner, on the side that opens. The numbers are corrected to the 1m industry standard — used by many loudspeaker and fan manufacturers — by subtracting six decibels.

Drivers & Settings

Comparison Cases

Test Results

The closest match in market to Rosewill’s B2 Spirit is Azza’s GT1, though the B2 Spirit has a slightly higher-quality fit and finish. The Define S and MasterCase 5 by comparison are ATX-based, with only the Define S supporting slightly oversized ATX boards, and both of those smaller cases are designed to be used with additional fans.

Equiped from the factory with two intake and three exhaust fans, the B2 Spirit is designed to run well out-of-the-box. It returns excellent temperatures from our overclocked Haswell-E and mid-sized cooler, and likewise beats all competitors in GPU cooling.

The B2 Spirit isn’t just cooler, it’s also quieter than its closest competitor in size. As for the closest in design (having a closed front-panel over front fans), the Define S achieves the same average noise level.

Cool and quiet are both great ways to top an acoustic efficiency chart, and that’s exactly what the B2 Spirit does. Rosewill should be thrilled with these results.

Value seekers on the other hand would point to the Define S’ ultra-low $70 price as a reason not to buy the B2 Spirit. Of course the markets aren’t the same, as the B2 Spirit supports over four times as many hard drives, over three times as many total drives, front-bay devices such as fan controllers and system monitors, three times as many USB devices, four-way SLI, and huge dual-socket motherboards.

Conclusion

Because the Azza GT1 is the only case we’ve tested with the same hardware, and that’s also in the same target market as the B2 spirit, we really only need to figure out if the $170 B2 Spirit is worth $40 more than the $130 GT1. The answer to that question is far tougher than I’d like it to be, if only because I value the B2 Spirit’s higher quality, better performance and flip-out fan holders over the GT1’s removable motherboard tray and somewhat-flimsy card brace. To say that I’d pay the extra $40 for the B2 Spirit would only be meaningful to those who understand my excruciating frugality.

Even though I’d rather call B2 Spirit a “better bargain” for its remarkable step upward in quality, my willingness to make concessions to those who demand removable motherboard trays demands that I treat the cheaper GT1 as having similar value. Since the GT1 already received our stamp of approval, so too proceeds the B2 Spirit.

MORE: Best MotherboardsMORE: Best CPU Cooling


MORE: How To Build A PCMORE: All Case Content

Thomas Soderstrom is a Senior Staff Editor at Tom's Hardware, covering Cases, Cooling, Memory and Motherboards. Follow him on Twitter.

Follow Tom's Hardware on Twitter, Facebook and Google+

Introduction

Rosewill has a rich power supply portfolio that includes a wide range of families covering every market segment. The company is trying hard to establish a good name in the competitive PSU business, which is a good thing not only for Rosewill but for its customers as well, since competition typically lowers prices. This is especially true in the U.S. market where, currently, two companies dominate the field.

One of Rosewill's newest PSU lines is the Photon series, which consists of six units ranging from 550 to 1200W, covering all of the important categories from low- and mid-wattage to the mega-PSU category. A strong selling point of these models is their affordable prices, resulting mostly from Rosewill's choice of OEM. Sirfa usually manufactures platforms with high price-per-performance ratios. Throughout this review, we will determine if this also applies to the Photon-1200.

All Photon units are fully modular and 80 PLUS Gold-certified. These are two features that, even today, are considered high-end. In addition, the Photon-1200 that we're focusing on is a powerful unit with a single +12V rail, able to feed multiple GPUs without running out of reserves. Admittedly, there aren't many enthusiasts who need a PSU with more than 1kW of output, even for a high-end gaming rig. But if you're trying to use three or four big graphics cards, it may become necessary to buy something bigger than 1000W. This is where 1.2kW PSUs come into play.

Specifications

MORE: How We Test Power SuppliesMORE:
All Power Supply ArticlesMORE:
Power Supplies in the Forums

The Photon-1200 PSU is compatible with the sleep states of Intel's Haswell-based CPUs (C6 and C7), where power consumption on the +12V rail is minimal. According to Rosewill, the maximum temperature at which it can deliver full power continuously is 40 degrees Celsius (104 degrees Fahrenheit), although the ATX spec says 50 °C (122 °F). During our tests, we will go higher than 40 °C inside our hotbox and see how the PSU responds, especially under higher loads. Hopefully it survives, despite the lack of over-temperature protection (OTP), which is particularly essential for PSUs with lower temperature ratings.

A double ball-bearing fan handles the unit's thermal load. And with no semi-passive mode, noise output is noticeable even under light loads. In high-capacity PSUs like this one, passive operation when there's not much going on is a nice feature, especially since most of us don't push our systems constantly. Finally, Rosewill offers a long warranty—five years—on the Photon-1200.

Power Specifications

The single +12V rail can deliver 100A. That's a lot of current, easily able to meet the demands of power-hungry gaming systems. Additionally, the minor rails' maximum combined power is high for a contemporary system that uses them only lightly, while the 5VSB rail is significantly stronger than average.

Cables And Connectors

The number of EPS and PCIe connectors is adequate, and the number of SATA and four-pin Molex connectors is high. With 15 SATA connectors, this PSU will easily power a storage server with numerous HDDs/SSDs, and the five peripheral connectors are sufficient for devices like system fans, coolers and LED lights. The cable lengths are good, and the distance between connectors is adequate (although, in some cases, a smaller distance between the SATA connectors could be better). Finally, the 24-pin ATX connector, along with all of the EPS and PCIe connectors, use thicker 16-gauge wire for lower voltage drops at high loads. The rest of the connectors use standard 18AWG wires.

Power Distribution

Because this PSU features a single +12V rail, we do not have anything to say about its power distribution.

Packaging, Contents, Exterior And Cabling

Packaging

The Photon-1200 comes in a large box with a black background. On the front, you'll find the model number and a photo of the PSU with its modular panel exposed. The box also includes an 80 PLUS Gold badge and an icon that indicates the unit's five-year warranty.

On the back of the package, there's a power specifications table, along with a graphical list describing the available connectors. However, the information doesn't specify cable length. Finally, the back of the box includes a brief list of product features.

Contents

The unit is not protected very well inside the box—we found no packing foam on top or on the sides, and the bottom of the box only includes a thin sheet of packing foam. We expect much better protection for a product that costs $170. The bundle includes a pouch for storing the modular cables, a set of zip ties, a set of screws and a really thick AC power cord.

Exterior

The Photon-1200 might not be the best-looking PSU on the market, but it certainly has a distinct look that makes it stand out. The design of the top mesh plays a key role in this aesthetic, though we think it could restrict airflow. On the front, there is a classic, honeycomb-style mesh, and the on/off switch is installed next to the AC receptacle. The power specifications label is placed on the bottom of the unit, and there are decals with the series name on the sides.

The finish is more matte than glossy, and overall, it exudes good quality. The back of the modular panel includes many sockets, which could have been arranged better in order to have more space between them, making them easier to attach and detach. Finally, the PSU is compact considering its 1.2kW capacity.

Cabling

The main ATX cable is fully sleeved and consists of colored, non-stealth wires, so it's not a very good match with the rest of the stealth/ribboned cables. It was wise of Rosewill to not include a flat and/or stealth Berg adapter because it probably wouldn't be needed by anyone shopping for such a high-end power supply.

A Look Inside And Component Analysis

Parts Description

Our main tools for disassembling PSUs are a Thermaltronics soldering and rework station and a Hakko 808 desoldering gun.

The Photon-1200 is made by Sirfa, an OEM that’s mostly experienced in creating budget PSUs. However, Sirfa does have some interesting high-end platforms in its portfolio. The Photon-1200's internals are identical to the High Power Astro GD 1200W, which makes sense since both PSUs are based on Sirfa's HPJ platform. In the primary side, Sirfa doesn't use an LLC resonant converter, so we don't expect this unit to be able to compete in the efficiency section (at least not compared to PSUs featuring LLC resonant converters). In the secondary side, a synchronous design is used where active components rectify all rails, which is typical for this category.

The first part of the EMI/transient filter is on the AC receptacle and, in this case, consists of one X and two Y caps. On the X cap, we noticed a CM02X IC that blocks current through the cap's discharge resistor when AC voltage is connected to avoid energy losses on the bleeding resistor. The EMI filter continues on the main PCB with one X and two Y caps, along with a pair of CM chokes and an MOV. The latter is partially hidden by the PFC input capacitor.

Both bridge rectifiers, a pair of GBJ2506s, are bolted on a small, dedicated heat sink. The amount of current that these bridge rectifiers can handle is huge, easily covering the needs of the Photon-1200.

Three Infineon SPW24N60C3s are used in the APFC converter, along with a pair of CREE C3D08060A boost diodes. In addition, the bulk caps are provided by Rubycon, and their combined capacity is 940uF.

The main switchers are a pair of Infineon 60R125CP FETs. The combo PFC/PWM controller, a Champion CM6800 IC, is installed on a small daughterboard. It's commonly used in Bronze-rated PSUs and lower. We don't have high expectations when it comes to the efficiency levels that this platform will provide, given the plain design and outdated PFC/PWM controller it uses.

Two small heat sinks in the secondary side hold all FETs that regulate the +12V rail, which are eight Infineon IPP034NE7N3 G. The DC-DC converters that handle the minor rails are installed directly on the modular board in order to save some losses on power transfer cables. Each of the aforementioned converters uses four Infineon IPD060N03L G, and the common PWM controller is an Anpec APW7159.

All filtering capacitors in the secondary side, both electrolytic and polymer, are Japanese, provided by Nippon Chemi-Con. Although this is a low-cost Gold unit, Rosewill and Sirfa decided wisely to use high-quality capacitors to increase the PSU's reliability. On the front side of the modular board, several polymer caps by Chemi-Con are used for ripple filtering, while some of them are utilized by the DC-DC converters.

The standby PWM controller is a Power Integrations TOP256EN, which can provide up to 86W with an open-frame design. The SBR that rectifies the 5VSB rail is a PFR 30L45CT.

Housekeeping is done by a SITI PS223, one of the very few relevant ICs that supports OTP (over-temperature protection) out of the box. In addition, this IC supports OCP (over-current protection) for up to two +12V rails. The Photon-1200, however, has only one.

Soldering quality isn't among the best we have seen, but it's good overall, especially for a budget, high-capacity unit. On the solder side of the main PCB we also spotted a Unisonic LM393L dual differential comparator.

The cooling fan is provided by Globe Fan (model number RL4Z B1352512HH, 12V, 0.45A, 129.76 CFM, 1800 RPM, 33.9 dB[A]). It uses double ball-bearings, so it will last for quite a long time. This fan is powerful, capable of pushing a significant amount of air. However, its profile is aggressive. Because this supply's efficiency isn't great, thermal loads inside the PSU are increased, causing the fan to spin faster.

Load Regulation, Hold-Up Time And Inrush Current

To learn more about our PSU tests and methodology, please check out How We Test Power Supply Units. 

Primary Rails And 5VSB Load Regulation

Load Regulation testing is detailed here.

Hold-Up Time

Our hold-up time tests are described in detail here.

The Photon-1200's hold-up time easily surpasses the minimum allowed by the ATX spec.

Inrush Current

For details on our inrush current testing, please click here.

The inrush current that our shiny, new power meter (N4L PPA1530) measured is very high, especially with 115V. Sirfa should make the necessary changes to fix this.

Load Regulation And Efficiency Measurements

The first set of tests reveals the stability of the voltage rails and the PSU's efficiency. The applied load equals (approximately) 10 to 110 percent of the supply's maximum in increments of 10 percentage points.

We conducted two additional tests. During the first, we stressed the two minor rails (5V and 3.3V) with a high load, while the load at +12V was only 0.10A. This test reveals whether a PSU is Haswell-ready or not. In the second test, we determined the maximum load the +12V rail could handle with minimal load on the minor rails.  

Load regulation isn't bad, but it's not top-notch, either. Let's call it somewhere between good and bad. Compared with the more expensive PSUs in this category, the Photon-1200 doesn't score well. The supply does deliver its full power for prolonged periods, even at operating temperatures close to 50 °C (122 °F).  However, this result wasn't achieved without problems, as you will see from the ripple results we measured. So, if you're building a power-hungry system with this PSU, keep its operating temperature below 40 °C (104 °F).

The situation gets worse when we start looking at noise output. Even at light loads, high ambient temperatures cause the fan to get really loud. Also, as you can see from the table above, it has significant bearing noise at around 1600 RPM where its acoustic output peaks. The inefficient platform doesn't leave room for a more relaxed fan profile.

Efficiency, Temperature And Noise

Efficiency

Our efficiency testing procedure is detailed here.

Using the previous page's results, we plotted a chart showing the Photon-1200's efficiency at low loads, and loads from 10 to 110 percent of the PSU's maximum-rated capacity.

The Photon-1200 isn't very efficient under light and normal loads, at least not for a Gold-rated PSU. This is something we expected, given the outdated platform. If you need higher efficiency and less noise, you'll have to spend more money.

Efficiency At Low Loads

In the following tests, we measure the efficiency of the Photon-1200 at loads significantly lower than 10 percent of the device's maximum capacity (the lowest load the 80 PLUS standard measures). The loads we dialed were 20, 40, 60 and 80W. This is important for representing when a PC is idle, with power-saving features turned on.

Efficiency under light loads is mediocre. However, that's not the only bad thing about this unit. If you take a look at the fan noise column in the table above, you will see that its biggest weakness is noise, again, even under very light loads. We should note that we tested at an ambient temperature around 40 °C. Still, we expected much lower noise levels in this section.

5VSB Efficiency

The ATX specification states that 5VSB standby supply efficiency should be as high as possible, recommending 50 percent or higher efficiency with 100mA of load, 60 percent or higher with 250mA of load and 70 percent or higher with 1A or more of load.

We will take four measurements: one each at 100, 250 and 1000mA, and one with the full load the 5VSB rail can handle. 

Efficiency on the 5VSB rail is just terrible. We strongly believe that high vampire power consumption is the cause of this rail's poor efficiency.

Power Consumption In Idle And Standby

In the table above, you'll find the power consumption and voltage values of all rails (except -12V) when the PSU is idle (powered on, but without any load on its rails), and the power consumption when the PSU is in standby mode (without any load, at 5VSB).

Phantom power is well above 0.5W, meaning that this unit isn't ErP Lot 6 2013-compliant. However since it isn't sold in the EU, it doesn't have to be. Nonetheless, the less power wasted, the better. And in this case, increased power consumption at standby leads to poor 5VSB efficiency as well.

Fan RPM, Delta Temperature And Output Noise

Our mixed noise testing is described in detail here.

The first chart below illustrates the cooling fan's speed (RPMs), and the delta between input and output temperature. The results were obtained at 40 °C (104 °F) to 49 °C (120.2 °F) ambient temperature.   

The next chart shows the cooling fan's speed (RPMs) and output noise. We measured acoustics from one meter away inside of a small, custom-made anechoic chamber with internals completely covered in sound-proofing material (be quiet! Noise Absorber kit). Background noise inside the anechoic chamber was below 18 dB(A) during testing, with ambient temperature at the levels noted above. 

The following chart illustrates the fan's output noise over the entire operating range of the PSU. The same conditions of the above graph apply to our measurements, though the ambient temperature was between at 28 °C (82.4 °F) to 30 °C (86 °F).  

Even with normal operating temperatures the Photon-1200 is noisy. Rosewill and Sirfa should offer a more relaxed fan profile, especially at lower loads. If you want to build a silent PC, you should refrain from buying this PSU.

Cross-Load Tests And Infrared Images

Our cross-load tests are described in detail here.

To generate the following charts, we set our loaders to auto mode through our custom-made software before trying more than 1000 possible load combinations with the +12V, 5V and 3.3V rails. The load regulation deviations in each of the charts below were calculated by taking the nominal values of the rails (12V, 5V and 3.3V) as point zero.

Load Regulation Charts

Efficiency Chart

Ripple Charts

Infrared Images

Toward the end of the cross-load tests, we took some photos of the PSU with our modified FLIR E4 camera, which delivers 320x240 IR resolution (76,800 pixels).

Transient Response Tests

Advanced Transient Response Tests

For details on our transient response testing, please click here.

In these tests, we monitored the PSU's response in two different scenarios. First, a transient load (10A at +12V, 5A at 5V, 5A at 3.3V and 0.5A at 5VSB) was applied for 200ms while the supply was working at 20 percent load. In the second scenario, we hit it with the same transient load while operating at 50 percent capacity. In both tests, we used our oscilloscope to measure the voltage drops caused by the transient load. The voltages should have remained within the ATX specification's regulation limits.

These tests are crucial because they simulate the transient loads a PSU is likely to handle (such as booting a RAID array or an instant 100 percent load of CPU/GPUs). We call these tests Advanced Transient Response Tests, and they are designed to be tough to master, especially for a PSU with a capacity of less than 500W. 

Advanced Transient Response at 20 Percent

Advanced Transient Response at 50 Percent

Performance in these tests was pretty good; the +12V rail stayed well below one percent in both cases, while deviations on the other rails were normal for this PSU's capacity. Ideally, we would like to see a deviation within three percent at 3.3V; however, this rail is lightly used by modern systems, so we are satisfied by the 3.56 percent result.

Here are the oscilloscope screenshots we took during Advanced Transient Response Testing:

Transient Response At 20 Percent Load

Transient Response At 50 Percent Load

Turn-On Transient Tests

In the next set of tests, we measured the response of the PSU in simpler transient-load scenarios—during its power-on phase.

For the first measurement, we turned the supply off, dialed in the maximum current the 5VSB could output and switched on the PSU. In the second test, we dialed the maximum load the +12V could handle and started the PSU while it was in standby mode. In the last test, while the PSU was completely switched off (we cut off the power or switched off the PSU by flipping its on/off switch), we dialed the maximum load the +12V rail could handle before switching on the PSU from the loader and restoring power. The ATX specification states that recorded spikes on all rails should not exceed 10 percent of their nominal values (+10 percent for 12V is 13.2V, and 5.5V for 5V).   

At 5VSB, the spike was quite high at 5.46V, reaching close to the ATX spec's upper limit. Sirfa should look into this. On the "standby to full +12V" load test, performance was good, while in the last test we noticed a pretty low voltage overshoot and a small dive at the start of the waveform.

Overall, we'll call this performance mediocre, with the most significant problem being the 5VSB spike.

Ripple Measurements

To learn how we measure ripple, please click here.

The following table includes the ripple levels we measured on the Photon-1200's rails. The limits, according to the ATX specification, are 120mV (+12V) and 50mV (5V, 3.3V and 5VSB).

Ripple suppression isn't good, especially at operating temperatures that exceed 44 degrees C. On the +12V rail, we don't want to see results above 50mV. Clearly, this platform lets us down with ripple measurements close to 120mV. The Photon-1200 cannot operate properly at high ambient temperatures and this is why it didn't manage to keep ripple suppression under control, failing in the 110 percent and Cross-Load 2 tests. Keep this thing's operating temperature below 40 °C, just as Rosewill recommends.

Ripple Oscilloscope Screenshots

The following oscilloscope screenshots illustrate the AC ripple and noise registered on the main rails (+12V, 5V, 3.3V and 5VSB). The bigger the fluctuations on the screen, the bigger the ripple/noise. We usually set 0.01V/Div (each vertical division/box equals 0.01V) as the standard for all measurements; however, in this case we had to increase to 0.02V/Div for the +12V rail.

Ripple At Full Load

Ripple At 110-Percent Load

Ripple At Cross-Load 1

Ripple At Cross-Load 2

Performance, Performance Per Dollar And Noise Ratings

Performance Rating

The following chart shows the total performance rating of the PSU, comparing it with other units we have tested in the past. To be more specific, the tested unit is shown as 100 percent, and every other unit's performance is shown relative to it.

Compared to more advanced and more expensive platforms, the Photon-1200 PSU doesn't stand a chance. It registers a much lower overall performance score. The truth is that if you spend $30 to $40 more, you can get a much better-performing unit, which might also be significantly quieter than Rosewill's noisy offering.

Performance Per Dollar

The following chart may be the most interesting to many of you because it depicts the unit's performance-per-dollar score. We looked up the current price of each PSU on popular online shops and used those prices and all relative performance numbers to calculate the index. Note that all of the numbers in the following graph are normalized by the rated power of each PSU. 

Despite its good price, a low performance score doesn't allow the Photon-1200 to achieve a higher performance-per-dollar ratio.

Noise Rating

The graph below depicts the cooling fan's average noise over the PSU's entire operating range, with an ambient temperature between 28 and 30 °C (82 to 86 °F).

The Photon-1200 manages to surpass even Seasonic's noisy units, which use even smaller-diameter fans. This is quite an achievement, and we mean this in the most negative way possible. 

Pros, Cons And Final Verdict

Rosewill is doing an outstanding job expanding its portfolio with all kinds of PSUs, covering the needs and demands of all categories, from the mainstream to the ultra-high-end. In the case of the 1.2kW Photon, the company's aim was to offer a powerful and efficient unit with some high-end features at a budget price. Admittedly, the Photon-1200 is one of the most affordable 1.2kW PSUs with an 80 PLUS Gold rating. However, if you spend just a bit more, you can get a much better-performing unit with higher efficiency, lower ripple and lower noise output from Rosewill or another vendor.

For many users, a PSU's acoustics can become really annoying, and in this case, they have every right to complain. The Photon-1200 is among the noisiest supplies we have ever tested. Rosewill could easily make it quieter by using a more relaxed fan profile. The existing one was most likely chosen to ensure the unit's cooling would be optimal under all conditions, as Rosewill's top concern was increasing the PSU's reliability to support a lengthy five-year warranty. The fan almost certainly has a greater thermal load to handle compared with other, more efficient PSUs in this category. But under lower loads, the company could make it spin slower, suppressing its noise levels.

If you're looking for a powerful PSU to go in a system that will be isolated (so that its noise won't be an issue), the Photon-1200 could save you some money, so long as you keep it running inside of a cool chassis. Unfortunately, the supply's price advantage isn't very big compared with other high-performance PSUs in this category, so it's equally tempting to simply buy something else.

Besides its noisy operation, the most significant problem with the Photon-1200 is bad ripple suppression at increased ambient temperatures. On the other hand, its strong points are quality Japanese capacitors, a fully modular cabling design and the ball-bearing fan that should last for quite a long time.

If the price came in under $150, we believe this could be an appealing option for folks wanting an affordable, fully-featured 1.2kW PSU. With its current price tag of $170, however, the unit is unlikely to claim a big market share in this category where other companies, including Corsair, Seasonic, EVGA, Super Flower and FSP offer better alternatives.

MORE: How We Test Power SuppliesMORE:
Who's Who In Power Supplies, 2014: Brands Vs. ManufacturersMORE:
Power Supplies in the Forums

Aris Mpitziopoulos is a Contributing Editor for Tom's Hardware, covering Power Supplies.
Follow us on Twitter @tomshardware, on Facebook and on Google+.

Rosewill Quark 1000 Power Supply Review

The big advantage of Rosewill's PSU portfolio comes from the company's affordable prices, which allow it to achieve high performance-per-dollar ratios. The Quark series, however, isn't a stand-out value contender compared to the PSU market's tough competition; this shows that Rosewill isn't afraid of having its PSUs do battle with products from well-established companies in this market segment. The truth is that when you invest in a good platform and quality components, you cannot lower the price of the product below a certain point without taking a loss.

Today we're reviewing the second-strongest unit in the Quark series, which offers 1kW max power along with a fully modular cabling design and 80 PLUS Platinum efficiency for $200. Many PSUs with similar specifications cost a little more than the Quark 1000, and this means that Rosewill's PSU will have a small advantage in the price/performance ratio category. However, the price difference is small, so the Quark still has to perform pretty well in order to keep up with its tough competitors like Corsair's HX1000i, Seasonic's Snow Silent 1050W and EVGA's SuperNOVA 1000 P2.

Since Titanium-rated units aren't very popular yet, the prices of Platinum PSUs remain high enough to make them less appealing to many users, especially those on a tight budget. Rosewill tries to offer good deals on its PSUs, but Gold-class models with similar capacity cost significantly less, making them look like a better deal (especially since the efficiency gap between a good Gold PSU and its Platinum counterpart isn't very big). Having clarified this, we can take a better look at Rosewill's Quark series, which consists of six units with capacities ranging from 550 to 1200W. With this line, Rosewill covers most users and systems, from mid-range to high-end ones.

Specifications

Again, the Quark 1000 meets the 80 PLUS Platinum requirements and can deliver its full power continuously at up to 40 degrees Celsius, according to Rosewill. Of course, we plan to crank up the temperature inside our hot box above that. Rosewill includes all relevant protection features, including OTP (Over-Temperature Protection). A Riffle bearing fan handles the unit's cooling without the support of a semi-passive mode. The PSU's dimensions are normal for its capacity, while the warranty is long enough and the price tag is a little lower than what competitors offer for similar specs.

Power Specifications

The single +12V rail can deliver the PSU's full power on its own. The minor rails can deliver enough power for the needs of a contemporary high-end system; however, we would like to see a stronger 5VSB rail in a 1kW unit.

Cables And Connectors

The PSU has enough cables to deliver its full power efficiently. There are two EPS connectors along with eight PCIe connectors, all of which are available at the same time. The number of SATA and four-pin Molex connectors is more than adequate, and there is even a rarely-used FDD connector. Strangely enough, the main ATX cable is longer than the EPS cables (usually it's the other way around), though in general, all connectors are installed on pretty long cables. In addition, the distance between connectors is sufficient, even if some of you would prefer a shorter distance between the SATA plugs. After all, HDDs are typically installed close to each other. Finally, the 24-pin ATX connector along with the EPS and PCIe connectors use thick 16AWG wires for lower voltage drops, while all the rest of the connectors use standard 18AWG wires.

Power Distribution

Since this PSU features a single +12V rail, we do not have anything to comment on about its power distribution.

MORE: How We Test Power Supplies MORE:
Who's Who In Power Supplies, 2014: Brands Vs. Manufacturers MORE:
All Power Supply Articles MORE:
Power Supplies in the Forums

Packaging, Contents, Exterior And Cabling

Packaging

On the face of the box we find a nice depiction of the quark, an elementary particle and a fundamental constituent of matter, according to Wikipedia. On the front right-bottom corner, we find two badges depicting 80 PLUS Platinum efficiency and the Quark 1000's five-year warranty. On the back of the box, Rosewill provides a brief features list, along with the power specifications table and a graphical list describing the available connectors.

Contents

Packing foam protects the PSU inside the box, ensuring that it will reach customers in pristine condition. The bundle includes a pouch for storing the unused modular cables, the user's manual, some zip ties, an AC power cord and the necessary modular cables.

Exterior

The PSU has a high-quality matte finish. Instead of the classic honeycomb exhaust on the front, the punched fan grill offers a unique look. The large on/off switch is installed below the AC receptacle (with the fan facing upward), while the power specifications label is on the bottom of the PSU.

Around back, we find a fully modular panel and a series of LED indicators, which depict the proper (and improper) operation of all rails. Normally these should be all green; if they're not, something is wrong. The only thing users should keep in mind is that in order to see the LED indicators, a windowed case is required. And even with a windowed case, if the PSU is installed at the bottom with its fan facing down, the LEDs will not be easy to see. Indicators are mostly useful when you use a PSU outside of its case or on a bench table. We're willing to bet that won't be how most folks use this unit, since usually more powerful PSUs are typically found on bench tables. 

Cabling

Rosewill was rather conservative in its choice of cables, since they're neither flat nor stealth. Since this unit is competing in the high-end 1kW PSU category, the company should have offered better cables. Overall, cable sleeving is decent; however, the main ATX cable is rather short, leaving a significant part of the wiring exposed.

A Look Inside And Component Analysis

Our main tools for disassembling PSUs are a Thermaltronics soldering and rework station and a Hakko 808 desoldering gun.

Rosewill's Quark PSU is made by Enhance Electronics--a good OEM, but one we don't often encounter in our reviews. The platform is definitely cutting-edge, with an LLC resonant converter on the primary side and synchronous rectification (SR) and DC-DC converters on the secondary side. Also, all of its electrolytic capacitors are made by Japanese manufacturers. In addition, a massive heat sink cools down the APFC converter's switchers along with the primary switchers. This heat sink will surely lighten the load of the cooling fan, at least on the primary side.

At the AC receptacle, two X caps and a pair of Y caps form the first part of the EMI filter, while on the main PCB we find two more pairs of X and Y caps along with a couple of CM chokes and an MOV. There is also a CM02X IC on the solder side of the main PCB, which is responsible for blocking the current through an X cap's discharge resistor when AC voltage is connected. In addition, the bridge rectifiers are installed between the enormous primary heat sink and a much smaller heat sink. Close to them is a sizable NTC thermistor that provides protection against large inrush currents. A relay bypasses this thermistor once the APFC caps are fully charged in order to provide higher efficiency.

In the APFC converter, two Infineon IPP50R140CP FETs are used along with a single CREE C3D10060G boost diode. The pair of bulk caps is made by Nippon Chemi-Con (450V, 390uF each, or 780uF combined, 2000h @ 105 °C). We believe that their combined capacity is on the low side, so they won't be able to provide the hold-up time that’s defined by the ATX spec. The APFC controller is a Champion CM6502, which is most likely supported by a CM03X Green PFC controller.

There are four main switchers (Infineon IPP50R190CE) that (we believe) are arranged into a full-bridge topology. The LLC resonant controller is a Champion CM6901 IC, which operates the main switchers at PWM mode with light loads, while at higher loads it uses FM modulation. The main transformer is pretty small for a unit of this capacity and it's nearly surrounded by three heat sinks that have nothing bolted on them; these heat sinks play a key role since they dissipate heat from the MOSFETs underneath them. These MOSFETs (8x International Rectifier IRFH7004PbF) regulate the +12V rail and, in addition to the above heat sinks, are also cooled by the PSU's chassis through a thick thermal pad. Finally, two thermistors are attached to the fins of one of the +12V sinks with the help of heat-shrinks. One of them provides data to the fan control circuit, while the other one is used by OTP (Over-Temperature Protection).

On the secondary side, two DC-DC converters generate the minor rails through +12V. An Anpec APW7073 PWM controller is used on each converter, along with a single 86350D MOSFET. On the secondary side, several high-quality electrolytic caps, provided by Rubycon and Chemi-Con, filter the rails along with some Teapo polymer caps.

The standby PWM controller is a STR-A6069H IC, and the SBR that rectifies the 5VSB rail, a PFR10V45CT, is right next to the 5VSB transformer.

At the front side of the modular PCB, some polymer Duratech caps, along with two small electrolytic Chemi-Con caps, provide extra ripple filtering. On the back of the same board, the soldering quality is very good.

The main PCB also features solid soldering quality. All component leads are carefully trimmed, too. Enhance Electronics did an excellent job here.

Besides the supervisor IC (SITI PS223), which supports OTP out of the box, we found several other interesting components on the main PCB, including a pair of Lite-On optocouplers, two Silicon Labs Si8230BD gate drivers and four shunt resistors under the +12V islands. This is an indication that with the proper modifications (since the PS223 supports OCP for up to two +12V rails) this platform can support OCP for up to four virtual +12V rails.

The cooling fan is made by Young Lin Tech and its model number is DFS132512H (135mm, 12V, 3W, 1820RPM, 91.16 CFM). It uses a Riffle bearing, so it will last for quite a long time. However, it is quite noisy at higher speeds and the fan profile isn't as relaxed as we expected it to be, given this unit's high efficiency and large heat sinks.

Load Regulation, Hold-Up Time And Inrush Current

To learn more about our PSU tests and methodology, please check out How We Test Power Supply Units. 

Primary Rails And 5VSB Load Regulation

Load Regulation testing is detailed here.

The following charts show the voltage values of the main rails, recorded over a range from 40W to the maximum specified load, and the deviation (in percent) for the same load range. You will also find a chart showing how the 5VSB rail deals with the load we throw at it.

Hold-Up Time

Our hold-up time tests are described in detail here.

The hold-up time was below 16ms, meaning the PSU didn't successfully pass our hold-up test. That's disappointing from a high-end unit.

Inrush Current

For details on our inrush current testing, please click here.

The registered inrush current at 115V and 230V inputs was low for a unit of this capacity. We would much prefer a higher inrush current for a hold-up time that would exceed 16ms, since the capacity of the APFC bulk cap(s) plays a key role in inrush current.

Load Regulation And Efficiency Measurements

The first set of tests reveals the stability of the voltage rails and the PSU's efficiency. The applied load equals approximately 10 to 110 percent of the maximum load the supply can handle, in increments of 10 percentage points.

We conducted two additional tests. During the first, we stressed the two minor rails (5V and 3.3V) with a high load while the load at +12V was only 0.10A. This test reveals whether the PSU is Haswell-ready or not. In the second test, we determined the maximum load the +12V rail could handle while the load on the minor rails was minimal. 

The load regulation on all rails was within three percent, though we would like to see the regulation on the +12V rail to be within one percent, given its importance. As for efficiency, although this unit wasn't officially certified by the 80 PLUS organization at the time of the review, we found that it meets the Platinum requirements even under extremely tough conditions.

With up to 20 percent of its maximum-capacity load, the noise output was below 40 dB(A), so it wasn't in the annoying region for most users. However, the combination of high ambient temperatures along with increased loads forced the fan control circuit to operate the fan at higher speeds, increasing output noise. Finally, the large heat sinks and the fan's airflow work well together. This is shown by the large delta between the PSU's input and exhaust.

Efficiency, Temperatures And Noise

Efficiency

Our efficiency testing procedure is detailed here.

Using the results from the previous page, we plotted a chart showing the Quark 1000's efficiency at low loads and at loads from 10 to 110 percent of the PSU's maximum-rated capacity.

This Enhance Electronics platform proved to be highly efficient at normal loads. It didn't stand a chance at light loads though, even against the Gold-rated competition. We would like to see higher efficiency readings with less than 100W loads.

Efficiency At Low Loads

In the next tests, we measure the efficiency of the Quark 1000 at loads significantly lower than 10 percent of the device's maximum capacity (the lowest load the 80 PLUS standard measures). The loads we dialed were 20, 40, 60 and 80W. This is important for representing when a PC is idle with power-saving features turned on.

At light loads with 115V input, efficiency is higher than with 230V. However, we expected even better results from a Platinum-rated unit based on a modern platform.

5VSB Efficiency

The ATX specification states that 5VSB standby supply efficiency should be as high as possible, recommending 50 percent or higher with 100mA of load, 60 percent or higher with 250mA of load and 70 percent or higher with 1A or more of load.

We will take four measurements: one at 100, 250 and 1000mA each, and one with the full load the 5VSB rail can handle. 

The Quark 1000 features a highly efficient 5VSB rail. However, it does not compare with Seasonic's high-end platforms.

Power Consumption In Idle And Standby

In the table above, you'll find the power consumption and voltage values of all rails when the PSU is idle (powered on, but without any load on its rails), and the power consumption when the PSU is in standby (without any load at 5VSB).

Phantom power was very low with both 115V and 230V inputs. This surely played a significant part in the 5VSB rail's high efficiency. 

Fan RPM, Delta Temperature And Output Noise

Our mixed noise testing is described in detail here.

The following chart illustrates the cooling fan's speed (RPMs) and the delta between input and output temperature. The results were obtained at 36 to 47 °C ambient temperature. The next chart shows the cooling fan's speed (RPMs) and output noise. We measured acoustics from one meter away, inside a small, custom-made anechoic chamber with internals completely covered in soundproofing material (be quiet! Noise Absorber kit). Background noise inside the anechoic chamber was below 18 dB(A) during testing, and the results were obtained with the PSU operating at 36 to 47 °C ambient temperature. 

Up until around 500W, the PSU showed silent operation. But it quickly reached loud levels after that, and became downright annoying with more than 800W of load. Considering its high efficiency and the large heat sinks, we expected a more relaxed fan profile. We suspect that Rosewill wanted to ensure adequate cooling of the +12V MOSFETs, which are partially cooled by the unit's chassis.

Cross-Load Tests And Infrared Images

Our cross-load tests are described in detail here.

To generate the following charts, we set our loaders to auto mode through our custom-made software before trying more than 1000 possible combinations with the +12V, 5V and 3.3V rails. The load regulation deviations in each of the charts below were calculated by taking the nominal values of the rails (12V, 5V and 3.3V) as point zero.

Load Regulation Charts

Efficiency Chart

Ripple Charts

Infrared Images

Toward the end of the cross-load tests, we took some photos of the PSU as it was being tested using our modified FLIR E4 camera that delivers a 320x240 IR resolution (76,800 pixels).

Transient Response Tests

Advanced Transient Response Tests

In these tests, we monitored the response of the PSU in two different scenarios. First, a transient load (10A at +12V, 5A at 5V, 5A at 3.3V and 0.5A at 5VSB) was applied to the PSU for 200ms while the PSU was working at 20 percent load. In the second scenario, the PSU was hit by the same transient load while operating at 50 percent load. In both tests, we used our oscilloscope to measure the voltage drops caused by the transient load. The voltages should remain within the ATX specification's regulation limits.

These tests are crucial, as they simulate the transient loads that a PSU is likely to handle (such as booting a RAID array or an instant 100 percent CPU/GPU load). We call these Advanced Transient Response Tests, and they are designed to be very tough to master.   

For details on our transient response testing, please click here.

Advanced Transient Response 20%

Advanced Transient Response 50%

Our Advanced Transient Response tests showed that the voltage drops were under control on all rails. However, we would have liked to see smaller deviations on the +12V rail, which handles transient loads constantly inside a real system. The higher-than-usual voltage drops on the 5VSB rail are not serious, since the average deviation was within 1.5 percent.

Here are the oscilloscope screenshots we took during Advanced Transient Response Testing:

Transient Response At 20-Percent Load

Transient Response At 50-Percent Load

Turn-On Transient Tests

In the next set of tests, we measured the PSU's response during the PSU's power-on phase.

For the first measurement, we turned off the PSU, dialed in the maximum current the 5VSB can output and then switched on the PSU. In the second test, we dialed the maximum load the +12V rail could handle and started the PSU while it was in standby mode. In the last test, while the PSU was completely switched off (we cut off power or switched off the PSU by flipping its switch), we dialed the maximum load the +12V rail could handle before switching on the PSU from the loader and restoring power. The ATX specification states that recorded spikes on all rails should not exceed 10 percent of their nominal values (+10 percent for 12V is 13.2V, and 5.5V for 5V).    

We measured a really tiny voltage overshoot at 5VSB and two larger voltage overshoots, but everything was pretty smooth on the other two tests. All in all, the PSU performed well here.

Ripple Measurements

To learn how we measure ripple, please click here.

The following table includes the ripple levels we measured on the Quark 1000's rails. The limits are, according to the ATX specification, 120mV (+12V) and 50mV (5V, 3.3V and 5VSB).