Welcome to my new hobby... Trying to blow up power supplies!
Recently, quite a few power supply companies have upped the ante with their power supply efforts.
So often, the power supply is an overlooked part of a computer. As power supply requirements increase, so has the specifications of the power supplies on the market. The power supply companies have answered this calling with better power supply units.
This review was going to be a big power supply shootout, but I found that so many power supplies are so different in so many ways, it really wasn't fair to put them "head to head." Is it fair to put a $200 PC Power and Cooling against a $50 Powmax? Is it fair to put a modular Ultra X-Connect against a non-modular Thermaltake Pure Power Butterfly? There's so many different strokes for different folks, I decided to break down the review into several smaller reviews that will be posted over the next several weeks.
So why be concerned with what a power supply can do?
The first thing I need to tell you is: Quit looking at wattage!! Wattage doesn't mean squat! All wattage is is the total capability of all of a power supply's rails. The 5V, 12V, 3.3V, -12V, -5V and 5VSB capability all added up. That total number really tells you nothing about the power supply's actual capability. And then, is that wattage continuous power or maximum peak power? There's also variables that come into play like, what was the temperature at which the testing was performed? For what period of time was the testing performed at the specified wattage? Basically, you should look at the amperage each rail is capable of and then just consider that the power supply's BEST CASE SCENARIO capability.
Now you need to figure out your computer's WORST CASE SCENARIO load.
Once you do this, you'll really find out how unimportant maximum wattage is, and how important the way the manufacturer distributes power across the rails is. If you have a 500W power supply with 40A available on the 5V line and you're using a Prescott with SLI video cards, you might be in trouble because the 5V line alone is using up 200W of that power supply's total power not leaving much else for other rails! Given that most power supplies give you 20 to 30A on the 3.3V (which is way high by today's standards, but even 30A on the 3.3V is only 100W) and split up about 20W for negative voltage and stand by, you're only left with 180W for the 12V rail. That's only 15A! Mind you, we're talking maximum combined peak power, but better safe than sorry, right?
If you don't have the time or resources to do this, then do this instead: Just try to figure out if your PC is going to be 5V heavy or 12V heavy, and then buy the biggest, best quality power supply you can afford with the load balanced most appropriately for your PC. For example: If you have a Pentium III or an Athlon XP board without an ATX12V connector (like Biostar Socket A motherboards never have the 2x2 connector) then something like an Antec or Raid Max with an insanely high 5V is most suitable for you. If you have a Prescott or an AMD64, consider something with a high 12V like an Ultra or an OCZ. If you have PCI Express video card or cards, consider something with a really, really high 12V rail.
All that said, fact of the matter is, if you have a power supply that has a load capability properly balance for your PC, you could actually run your machine with a quite a bit of stability with a mere 300W power supply. If you don't believe me, you might want to consider picking up a Kill A Watt. You might find that you're currently pulling about 200W from the outlet. Given that PC power supplies typically only have an efficiency of 75%, that's only 150W!!
So what will I be testing?
I will be throwing five tests at each power supply.
Test number one is the Zero Load test. This will be performed with no load on the power supply other than the internal load the power supply provides itself to function, the load of any fans that may be installed and running in the power supply and an artificial 5V stand by load of 2A.
The data gathered on this test will be:
After this test is performed, the load tests will begin starting with a synthetic load representative of three builds
The first build is a a pretty common representation of a PC found today. Heavier on the 12V than PC's of the past and lower on the 5V than PC's of the past. We used a Pentium 4 3.2 Northwood, an AGP video card, 1GB of DDR RAM, two hard drives, two optical drives and three case fans. For fun, we threw four USB devices on the machines just to get the 5V load up over 30A.
This is what we came up with:
So we made our load specs for this build....
This next load test was based on an Athlon XP 3200+ on a Biostar M7NCD Ultra 400 motherboard, an AGP card, 512MB of PC2700 RAM, two hard drives, two optical drives and two case fans.
The estimated maximum load for the build was:
To make worst case scenario even worse, we made the load specs this:
The last build is sort of gonzo. We built a Pentium 4 3.4E Prescott with two 6800 Ultra PCI Express video cards, two hard drives, two optical drives, 1GB of RAM, three case fans and came up with:
When I replaced the Prescott with an AMD64 3400+, I had enough ceiling to run four hard drives in a RAID-5 array with room to spare. So we decided the load test for this configuration should be:
We also took the power supply's temperature at this load.
Now, all of these builds had USB mice and keyboards as well as floppy, NIC... the standard fare. Whether 512MB or 1GB of RAM were used, this was always done by using two sticks of RAM, which is why the voltage demand never really changed on any of the builds. Naturally, if you had more sticks of RAM, you'd have more of a load on that rail.
This is what we're looking for with each of the above four tests:
That's right. Same as the Zero load test! I just wanted to make sure you were paying attention. ;-)
We also took a look at the label and cranked the 12V and 5V up to maximum specs as per the PSU's label. We then did the math: (3.3+5V Combined Maximum)-(5V * Maximum Amperage)=(3.3V Testing Amperage Setting) and set the 3.3V accordingly.
At the point where we cranked up the 5V and 12V to label spec, we were usually already beyond the power supply's maximum wattage rating, so it wasn't unusual for a power supply's overload protection to kick in if we tried to compensate a little extra 3.3V. This is why it ticks me off when a review site says "we ran the power supply at maximum load" but doesn't break down how the rails were loaded. It's a given you can't load a power supply to "maximum load" if you juice up each rail, so one is left wondering where "maximum load" really is.
This also brings me to the "power supply blowing up" scenario and what it means to me and my tests. If overload protection trips, that's good. I actually rely on overload protection to kick in to effectively test a power supply. If I run a power supply with a 12V load of 20A and the rail is only rated at 16A and overload protection kicks in, then no big deal. I'm over spec. I write down that the power supply failed the test and move on to the next test. This is also how I do the max load test. I set the 12V and 5V to the maximum on the label and then raise the 3.3V until overload protection kicks in. When it kicks in, I lower the 3.3V and cycle the power on the power supply. If the overload protection fails to function, the power supply blows up. This does not mean that the power supply is inherently a fire hazard. It means I received a defective unit. Unfortunately, it also means the testing process comes to a screeching halt.
When everything is said and done, I lower the load back down to test two and wait for one minute for the power supply to cool down. We then cut the power off and watch the Compu-Nurse. After one minute, an idle temperature is taken.
How will the tests be performed?
Well, I'll be honest... I don't have three builds I can switch power supplies in and out of. Quite frankly, I don't have the time to do this that way either!
For testing power supplies, I am in possession of a SunMoon SM-268, a power supply load tester, and a little Weibo PF1211, which is a mains efficiency reader similar to a Kill A Watt, but easier to read because Volts, Amps, Watts and Power Factor are all displayed on four LED's simultaneously, instead of having to click through one LCD to get all pertinent information. Despite my wife's best attempts to convince me to set up elsewhere, the lab is in the dining room.
Everything I need is all laid out on my dining room table. Thanks to Ultra Products for the big ol' anti-staic mat.
The SunMoon SM-268 can dynamically load a power supply at the push of a button. There's five memory settings and the ability to crank up the amperage while the power supply is already up, running and loaded. I have three memory settings set to settings that can perform the above three tests. The other two tests are set up for dual 12V rail power supplies and I'll explain those later.
In the lower left of the SM-268 is a cluster of power connectors I plug the power supply into and a toggle switch that I toggle for single 12V rail and dual 12V rail power supplies. If I'm testing a dual 12V rail power supply, I make sure I plug the 2x2 12V connector into the SM-268 because that connector gets it's power from 12V2.
Here is where I set the amp load for each rail.
With the press of a button, I can test the output voltage of each rail while it's under load.
Some math is done for me. This mode of the tester shows me how much wattage is being put out on each rail, as well as the total wattage.
The display on the SM-268 has six fields. Each one representing a loaded rail. There's a display for the 12V rail, 5V, 12V2 if I'm testing a dual rail power supply or -5 if I'm testing a single rail power supply, -12V, 3.3V and 5V stand by. With the press of a button, the displays will switch from load in amps, to voltage, to wattage. In wattage mode, I also get a display telling me the total wattage being loaded onto the power supply.
Now let's say we have a power supply plugged into the Sun Moon and it says it's pushing 300W. Now I'll take a look at the PF1211 and see how much wattage I'm getting from the mains. If it says 500W, I divide 500W by 300W and come up with 60%. This would mean the power supply is running at 60% efficiency.
Busting out the old school calculator and notepad for this test.
Along the bottom of the display of the SM-268 is a number of buttons. It's quite intimidating and sometimes to accomplish something you have to push a "shift" button and another button simultaneously. I have most of my settings pre-programmed, so all I have to do is select "memory 1," through to "memory 5." I then have an up and down arrow for increasing and decreasing load while the tester is running. I use these arrow buttons for bringing the juice down for my zero load test and cranking my juice up to see where the limits of a particular power supply may be.
For the record (because someone asked me this the other day,) the preset loads that I have pre-programmed into the SM-268's memory DO NOT ramp up. The load on the PSU is immediate, so not only does the Power Good signal on the power supply have to work correctly, but the power supply also has to be able to accept a sudden, nearly crushing load, in a split second's notice. The only tests that are "ramped" is the zero load (where I ramp down from test two) and the full load (where I ramp up from test three.)
I have a Compu-Nurse plugged in as well so I could see what operating temperatures are. I began doing this after loading up an Antec and Enermax power supply and noticing how hot the air coming out of them were. I also shut down an Ultra PSU abruptly after a 550W load and noticed the housing getting very hot.
What I WON'T be testing:
There's a lot of power supply tests out there, and I'm not here to replace any one of them. I'm going to tell you what I'm already prepared NOT to test in an effort to head the"why didn't you do this or that" emails off at the pass.
Because I do not have a wafeform monitor, I will not be able to test the cleanliness of the rail. This is a shame, as it's very important that a rail is not only powerful and within spec, but that it has little ripple.
I'm also not testing power supplies for noise. I will try to point out when I think a power supply is unusually loud, but "quiet-PC's" is not my forte.
And now... ON TO THE POWER SUPPLY TESTING!
Ultra X-Connect 500W
Here we take a detailed look at the Ultra X-Connect 500W Titanium w/ UV Blue effects. The most distinct difference between the X-Connect and other power supplies is the X-Connect's modular cables.
Ultra's been around for a long time. Starting out as the white box division for Midwest Micro, Ultra is now largly an enthusiast component company, manufacturing all sorts of niche components with that little tweak to it that makes them unique. Ultra sells everything from rechargable MP3 players with built-in email clients to DDR memory with heat spreaders in mahogany boxes. But their most popular items are no doubt their power supplies.
So who actually makes these Ultra power supplies?
Looking at the Ultra X-Connect 500W power supply, you can tell that they it's design is really unlike any other. By looking at the components, PCB and pulling up the UL listing found on this power supply, we can see that this power supply is manufactured by Youngyear.
Here's the inside of the Ultra X-Connect 500W.
Smallish Heat Sinks
Heat sinks in the Ultra power supplies has always been a point of contention for me. Most people say that they "look" too small. I have to admit that they are smaller than almost every other brand power supply I've seen in this price range, but the folks at Ultra stand behind their decision to use these heat sinks. Although the perception that the smallish heat sinks means the power supplies must be "cheap," they were used for a reason. With the larger components required to manufacture a decent 500W power supply capable of "continuous power," smaller heat sinks needed to be used to help maintain airflow between the two fans. This is why the X-Connect isn't available in a 600W model where the X-Finity is. There simply isn't enough room inside your typical PSU housing to make one. Think about this: If this power supply wasn't this beefy inside, say like the Powmax I'm also reviewing this week, there'd be plenty of room for larger heat sinks. What's more important? A well built power supply or large heat sinks on a piece of crap?
There are better designs. Take a look at the inside of a PSU made by Topower like the Raid Max I just got done reviewing. That power supply has fins going in every which direction. There's more surface area to dissipate heat inside the Raid Max. The fans on the Raid Max also spin a lot slower and are a lot quieter.
A little Chinese bird told me that the Ultra power supplies are "carefully tested under load with the fans spinning at no more than 3000 RPM in an effort to determine how large the heat sinks really need to be." I'm wondering if this test is done in the same unrealistic 25 degrees Celsius lab that most manufacturers do their tests at, because although I've yet to have a heat related issue with an Ultra power supply, I've yet to have my own Ultra power supply fans NOT spin right up to full RPM within 15 minutes of being powered up, even when it's only 75 degrees in the house. Would larger heat sinks prevent this?
Here's another angle of the X-Connect with it's top off.
Down to the Wire
You may notice that with an X-Connect power supply, all of your modular cables terminate at a single PCB! And that PCB is fueled by nothing more than two 20AWG wires per rail!
Fear not and whip out the Ohms law calculator. Without going into the exact numbers, I think you'll find that even a single 20AWG wire will handle the load of your entire PC, never mind just the 12V or 5V needs of your PC. As for the X-Connect interface PCB itself, all I need to do is remind some of you how much juice we regularly push through the mere traces of a PCB. Never mind the fact that it's traces that get the juice to your power connector's wires in the first place in a "normal" power supply, think of something much larger, like a hot-swappable drive array where up to 8 drives plug into and get their power from a single Molex connector plugged into a PCB. That's a lot of juice going through some long traces.
Now, I'm not going to argue with logic. Does the X-Connect interface add substantial resistance? Yes. Does the X-Connect interface present another point of failure? Absolutely. Do I foresee a problem? But in my opinion, not any more of a problem than using splitters or adapters.
Here's the back of the X-Connect showing all of it's X-Connect cable interfaces.
The X-Connect definitely wins bling off da hizzy in the def department. Word. This shiznit is pimped out mack daddy style... err... sorry. This power supply is like that big chrome coffee can exhaust tip with the carbon fiberesque trim you just can't wait to clamp onto your Honda 1.5 liter to get you the 105hp your hard earned money deserves. We've got a shiny, mirror-like finish. We've got windows. We've got lights. LED's? Nah.. UV lights and UV reactive components! And then the hook up (literally) comes with the UV reactive stainless steel jacketed modular power connectors... Boo yeah!
Ok. Sarcasm aside, I do like the looks of this power supply. The eerie glow of the UV puts it over the top and the modular cables make the insides of a PC look so neat. On the other hand, you ARE paying for this bling.... cough... excuse me... The X-Connect is one of the more expensive PSU in this line up for a reason.
There's also a less expensive X-Connect that doesn't have the UV reactive windows and UV fans (there's a blue one shown in a picture later on in this review) and at around $100, costs $20 less than the model tested. Which is good if you want the modular cables without all of the bling (I've got to quit using that word.)
Here's a look at the PSU opened up and turned on. Notice how the UV light and UV reactive fans have a completely different look than your typical LED fan?
B B B Breakdown.....
Here's the skinny... Table please! (Don't forget to tip the maitre d'!)
|ULTRA 500W X-CONNECT||+3.3V||+5V||+12V||-12V||-5V||+5VSB|
|Max Output Current||28A||30A||34A||0.8A||0.3A||2A|
|Max Combined Wattage||
Note the heavy on the 12V at the sacrifice of the 5V. This is actually a good thing for newer systems that are regulating CPU voltage from the 12V rail where they used to regulate from the 5V. The -5V and -12V are really low, which is fine by today's standards. Now that ISA is all but extinct, I believe onboard sound is the only thing that uses negative DC current. That and serial port devices still sometimes use -12V, but I'm not sure who uses those anymore either (using Hyper Terminal to get into your MultiTech MultiVoIP does not require -12V)!
Now the X-Connect is sort of tricky to lay out to you when it comes to explaining cables. The power supply accommodates up to however many cables you plug into it. There's an ATX only port and a 2x2 only port and a Xeon/AUX only port, but other than that, there are 5 12V/5V ports that can be used for 5.25, 3.5, SATA and PCIe power connectors. And some of the optional cables split into two, so without even adding "Y" splitters you can hook up 10 drives, 9 drives and a PCIe video card or what have you. So the below table only reflects what CAME WITH my 500W X-Connect sample. If you're willing to spend about $10 a pop for more cables, it's carte blanche, baby.
|ULTRA 500W X-CONNECT POWER CONNECTORS||QUANTITY OF CONNECTORS|
|ATX connector||20 pin|
|2 x 2 12V connectors||1|
|2 x 3 PCIe||0|
|6-pin Xeon/AUX connector||1|
|5.25" Drive connectors||8|
|3.5" Drive connectors||1|
|SATA Drive power connectors||2*|
|Fan only connectors (thermostatically controlled 12V only)||0|
Here's the power supply running with a UV light placed underneath the cables to show how they are UV reactive.
What's my opinion about what the outcome will be?
If we believe the label completely, we'd have to assume that this power supply will be within spec on all three tests. 30A on the 5V and 34A on the 12V is nothing to cough at. But can it really do it?
Let's see how this modular monster pulls it off in the load tests.
As with the last test, red means out of spec.....
|Ultra X-Connect 500W||Zero Load||Test One (373W)||Test Two (304W)||Test Three (341W)||Full Load (536W)|
This is the blue, non-UV version of the X-Connect during the full load test (536W). Notice that the rails are closer to spec here than during any other test.
Analysis of results:
I was surprised the 5V behaved itself as well as it did during test one given the load was greatest on the 5V rail and the power supply is engineered for a high 12V. Typically, we would see the 5V drop as it's loaded if the 12V isn't loaded accordingly.
Test two was a little disappointing. With a power supply so evenly spec'd out for the 5V and 12V rail, one would think such an even load would result in more stable voltage. As you can see, the 12V is above the 5% tolerance.
No surprise that the X-Connect came out of test three with flying colors. This power supply was made for high 12V demands and it pulled out with flying colors.
The full load test allowed me to bump the 3.3V to only 5A before overload protection kicked in. The end result was 536W, which is considerably over spec.
As I expected, the temperatures sky rocketed when the power supply was shut down. The temps were actually a little high to start with in my opinion. The Raid Max ran at only 23.8C and was much quieter with it's slower RPM fans. That will be a point against it.
Sure it's got modular cables, lights, windows and a shiny finish, but the Raid Max has the EMI filters and the fans keep spinning after power off. Trade offs will make two seemingly different power supplies grade exactly the same. For features, on a scale from one to ten, I will give this power supply an 8.
For performance, I give this power supply a 9. It performed over spec, but the 12V should've been within spec during test two.
It's average overall score is an 8.5.
Reviews, Articles, News, All Reviews...
Reviews, Articles, News...
Career Center, News Users, Login...
A Guru's World, CPU/Memory Watch, SLDeals...
Forums, Register(Free), Todays Discussions...
Search, About Us, Advertise...