If geeks love it, we’re on it

Corsair TX750 Power Supply Review

Corsair TX750 Power Supply Review

Testing

Corsair offered us a TX750 before we were able to complete our new Power Supply testing methodology. After reading very positive reviews of the VX550 and owning one for myself over the last few months, I was eager to get the new TX750 in the lab regardless. For the record, I’d like to state that this review is not characteristic of what future Icrontic PSU reviews will look like. At any rate, I will be testing with tools and hardware available to me.


Testing Configuration

]

  • Intel Core 2 Quad Q6600 (Overclocked to 3.5GHz at 1.45V)
  • Asus P5K-E Mainboard (P35 chipset)
  • 2x1024MB Buffalo Firestix PC6400 DDR2 (2.3V)
  • BFG 8800GTS OC 640MB Graphics Card (Overclocked to 650/850)
  • Maxtor DiamondMax 10 SATA HDD
  • Pioneer SATA DVD drive
  • Thermalright SI-128 SE w/120mm Scythe S-FLEX 63CFM fan
  • 120mm Scythe 46CFM rear case fan and NMB 120mm front case fan.

Idle/Load: For system idle measurements, the PC was left at the windows desktop for approximately 20 minutes until the exhaust temperature stabilized. CPU/Memory load was achieved by running four instances of prime95 using large FFTs and 1300MB of system memory. Launching ATItool, and opening the 3D window to display the “spinning fuzzy cube” achieved GPU load while prime95 kept the CPU/memory loaded in the background. All measurements were taken after exhaust temperature stabilizes to ensure efficiency losses due to heat are accounted for.

Corsair TX750 Testing

Voltage Stability/Regulation: ATX specification states than any of the +12, +5 and +3.3V rails should not be any more than 5% out of specification. A voltage reading on the +12V rail should be between 11.4 and 12.6V. Anything higher or lower is out of specification and could potentially cause system instability, or worse, damage hardware. A quality PSU should provide output at or just slightly above the expected value. Generally speaking, a PSU is more likely to have its rails ‘sag’ when load is placed on them—i.e. during gaming or when the CPU cores are heavily utilized. It is important that voltage measurements be taken at system idle (for a reference point) and then compared to measurements taken during heavy system load. This is important to paint an accurate picture of the PSUs ability to keep voltages in check and to ensure that it is able to reliably power an appropriately sized system. Measurements were taken at idle, CPU/Memory load and CPU/Memory/GPU load with a digital multimeter.

As you can see above, voltage deviated less than 0.04V and was just slightly higher than spec. Excellent!

Efficiency: Efficiency is a measure of how well a PSU can convert 120/220V AC power into the various DC voltages required to power a computer. Unfortunately, this process is not perfect and some of this energy is lost in the conversion process in the form of heat. For example, the components in a PC may be drawing exactly 400W of power in DC form, but the PSU could be drawing 480W of AC power from the wall. Those missing 80 watts are wasted and converted to heat. Many new PSUs are able to convert over 80% of the AC power into usable DC voltages for the PC. A more efficient PSU runs cooler, saves electricity and is better for the environment.

Although it is very easy to determine how much power a system draws from the wall, it is not straightforward to determine how much DC load is actually on the power supply. As such, we’ve taken the middle ground and decided to make a comparison to what I consider a legend in the PSU industry—the PC Power and Cooling 510. It is definitely not an “apples to apples” comparison and this is done intentionally. The TurboCool 510 is a fantastic PSU that has won countless awards, but it employs a slightly older design that is simply not as efficient as their newest ‘Silencer’ PSUs. PC Power and Cooling (now OCZ) rates the 510 Express for about 75% efficiency with a typical load. In my opinion, this is an excellent efficiency ‘standard’ that should be met or exceeded by a current generation PSU. This is exactly what we’ll be comparing. AC load measurements are taken with a UPM EM100 meter. See below for the results.

As you can see, the Corsair TX750 is approximately 10-12% more efficient than our reference PSU. If we assume a 70-75% efficiency on our reference PSU, we are looking at efficiency in the 80-85% range. Very good! Again, exact efficiency figures are speculative based on these results but it is certainly a very efficient PSU as you can see. Picture three CFL light bulbs being illuminated whenever your computer is on—hard to believe that much energy can be saved simply by investing in a current generation 80%+ efficiency PSU like the TX750.

Interestingly, there is a bit of phantom power draw with the PSU completely off but left plugged in—approximately five watts. About 16W is consumed when the PSU is turned on, but the PC is left off. These numbers are still slightly lower than our reference PSU and somewhat expected. For those who are energy conscious, it would be wise to leave your PC unplugged when away for extended periods. Better yet, turn off your power bar/surge protector. I’ll be authoring another article on phantom power draw in the near future—stay tuned.

Thermal Considerations: As mentioned in the previous point about efficiency, about 15-25% of the power converted from AC to various levels of DC voltage is lost as heat in the conversion process. Many of the components in a PSU are only rated to operate until a given temperature so keeping the PSU cool is important to prevent failure. Preventing failure is really only the minimum requirement of cooling the PSU. Keeping the PSU cool ensures that it is operating efficiently and performing consistently. Observations on the internal construction of the PSU, heatsinks, fan placement and specifications are all very important. It is also important that in a PSU with a temperature-controlled fan, fan speed is appropriately increased to keep the internal temperature at an acceptable level.

I took temperature reads using two external probes. One was placed just a few centimetres from the intake fan, and the other was taped to the exhaust grille. Ambient temperature was maintained at between 18.5 and 19.5’C. Below are the results.

Exhaust temperature did increase noticeably after extended periods above 300W AC. It actually dropped slightly at GPU load. I’m guessing this is due to a slight increase in fan speed to compensate for internal temperature increase. Overall, these temperatures are well within acceptable tolerances and were maintained at very low noise levels. Even with the ~415W AC load, the fan was still spinning slowly. According to Corsair’s noise charts, you can see that the fan really doesn’t start spinning quicker until a 375W load is placed on the PSU. This is not far off from our observations.

Acoustic Considerations: The fan(s) used to cool PSUs inevitably produce noise. Although a high output fan at full speed would do a great job keeping the PSU cool, it would be producing its maximum sound level at all times. Many modern PSUs increase the speed of the fan as the internal temperature increases. This is ideal because the fan only produces as much airflow as is necessary to keep the PSU at an acceptable temperature. As I mentioned in the thermal considerations point above, noise level was very low during all of the tests. Fan noise did increase very slightly during extended periods of GPU load but it was still so quiet that I could not hear it over the other fans in the system without getting very close. I think the choice of fan for this PSU was a good one—nice and quiet at low RPM and gobs of cooling power on tap if it is ever required.

Ripple: DC voltage output should technically not have any ripple at all. This is unfortunately not always the case as the conversion process from AC to DC can leave a small amount of waveform present. Ripple can be thought of as ‘dirty’ DC power. The ATX specification states that less than 120mV p-p ripple should be maintained. This is important because high levels of ripple can cause stress to just about all components in your system. This is especially important on the +12V rail, as modern systems utilize +12V for the CPU and graphics card(s). Ideally, measurements should be taken during high utilization or GPU load as ripple often increases along with the load on the PSU. This is one area that I was unfortunately unable to test. We hope to have a quality oscilloscope in the lab for future PSU reviews.


« Previous Next page »

Comments

  1. Zuntar
    Zuntar Nice read Mike! ;)
  2. Winfrey
    Winfrey Very good article Mike! Hopefully more people are realizing that quality systems these days mean starting with quality power. And this PSU definitely seems to deliver!

    BTW:
    We hope to have a quality oscilloscope in the lab for future PSU reviews.

    get on that prime!

    I also look forward to that article on phantom power draw. Very interesting.
  3. Leonardo
  4. lemonlime
    lemonlime Glad you guys liked the review :)

    I happened to stumble across a yet to be released HX1000 model on Corsair's product page: http://corsair.com/products/hx1000.aspx so even higher power solutions are on their way :eek:
  5. jared
    jared trackbacks ftw lol
  6. Tong Is this psu safe to power three nvidia gts 250 cards? Each of my gts 250 requires two 6-pins power connectors and this psu comes with only 4. I have the molex adapters, but I hear that it isn't safe to use molex adapters. Help.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!