Enthusiast Power Protection: Four-Way 900 W UPS Roundup

Why So Pricey?

Most of our enthusiast-class builds have an approximate value of $2000. Because a $200 UPS represents only 10% of the system’s price, we thought a roundup of those would bolster our “cheap protection” argument. That’s when the realities of modern hardware slapped us in the face.

Problem 1: Real vs. Apparent Power

Have you ever wondered why a UPS of enormous VA rating might be required to support a far more moderate wattage load? Aren’t watts calculated by multiplying volts by amps, the V and A in VA ratings? Getting to the bottom of this required us to think about a few of the things we learned in high-school science classes and apply those lessons to a Web search.

The AC (alternating current) power that comes from our wall jacks is graphically represented as a sine wave, where the center line is equal to ground. Voltage is rated using RMS values, so a 120 V line is 170 V peak (RMS * √2). In a perfect world, current (amperage) would track force (voltage) perfectly, resulting in a peak wattage (at the crest of the wave) that’s far higher than the average wattage for a single cycle. That is to say, even in a perfect world, the circuit must be designed to handle more than its rated power.

Unfortunately, amperage load doesn’t always track voltage perfectly, and this is where VA and watts part ways. When voltage and amperage are slightly out of phase, a portion of the energy from the load (such as a computer’s power supply) returns to the source (such as a UPS battery). Though the cancelled-out power reduces the energy required from the source, all the parts between the source and the load must be beefed-up to carry it.

VA ratings are a way for power supply companies to advertise the current-carrying capacity of those middle parts without discussing the continuous load the battery is able to supply to the PC.

Problem 2: Active PFC and Step Waves

While traditional PC power supplies focus on reducing electrical noise on the output rails, Power Factor Correction-equipped power supplies also attempt to reduce the noise the unit places on its power source. Filters align amperage and voltage cycles, reducing the portion of energy that is returned from the load to the source. That is to say, a power supply with perfect PFC (1.0) would never require the middle parts to carry more VA than the PC consumes in watts. With enthusiast PC power supplies now advertising 0.95-0.99 PFC, it’s time most of us learned to ignore VA ratings and focus on the wattage capacity of our UPS devices.

Yet active PFC power supplies can also be more sensitive to wave form than traditional PC power supplies have been. Dan Farnsworth of APC explains:

“For some basic info, Europe has required that all electronics devices now be made with PFC power supplies for a while now. And they are becoming more common here in the US as time passes. These power supplies are more efficient...but also more expensive. Accordingly, in order to minimize cost increases associated with active PFC power supplies, some manufacturers are cutting corners in their power supply designs, which can impact the compatibility of traditional desktop (modified square wave output) UPSs with computer systems that use these new power supplies. For example, in some cases, the PFCs' "hold up" time is shorter, meaning UPSs, which normally would work fine with a computer, now do not transfer to battery power fast enough to keep the computer running during an outage. Active PFC power supplies can also draw excessively large inrush currents upon initial operation or during switches to battery power, overloading a UPS that would otherwise have been properly sized. Some active PFC power supplies work better when they receive pure sine wave power (Smart-UPS units), rather than the "modified sine wave" output of a Back-UPS unit, although, our testing shows that some active PFC power supplies don't care which type of output they receive.”

The problem is graphically represented at the CyberPower Systems Web site as a zero-output state occurring as a step-wave crosses from positive to negative voltage. While that zero-output state persists only for milliseconds, that’s still far longer than the instant that it occurs on a true sine wave. For most UPS manufacturers, the option is to upsell their customers a true-sine-wave model that costs far more to produce.

CyberPower’s solution appears to clip the peaks on a triangular wave, eliminating the zero-output state at lower cost than pure sine wave equipment. CyberPower calls this its “adaptive sine wave” technology, and one of our goals is to test its capabilities using a mid-priced, “enthusiast market,” active PFC power supply.

Thomas Soderstrom
Thomas Soderstrom is a Senior Staff Editor at Tom's Hardware US. He tests and reviews cases, cooling, memory and motherboards.
  • sudeshc
    Have been using APC for last 2 years now without any issues what so ever.
    Would recommend it although they might be a bit costly but they perform really well.
    Reply
  • hmp_goose
    Soooo you have nothing to say about the TripLite atoll? Dead last? Distant second? Muddled mess? Nice part for not-our-application?
    Reply
  • dEAne
    I have several units of APC and a hundred after-sales issues which remains unanswered, I opted to installed the generator set and a AVR than a UPS.
    Reply
  • Emperus
    Most folks would just be happy to ignore on the need for a backup power source.. Hope this article enlightens them on getting one to protect their precious components and/or be ready if and when the situation arises.. Would be great if Tom's could put up a tier based classification on suitable UPS choices for various PC's (gaming, server, home theatre etc.)..
    Reply
  • super_tycoon
    What about AVR? (Automatic Voltage Regulation) I have a CP1500AVRLCD and I've seen (at least I'm pretty sure) I've seen the AVR feature kick in when one of the local power lines got knocked out and the voltage drooped to ~100. (killed some of my non-dimmable cfl's) I went to CP's website and it claimed the pfc units still have AVR functionality. So here comes the confusion, what's the point of the voltage tolerances? If it has an AVR, why would it matter what voltage it's fed as long as there steps on the transformer? I thought the unit was supposed to take 90-140 without having to engage the battery while still passing ~120v?

    You've also made me want to test my non-pfc ups with my 850hx, but my gaming rig and my workstation are an hour apart....
    Reply
  • aldaia
    This is a quite inefficient & expensive approach to the problem. We take AC and use a transformer to convert to DC and store in a battery, meanwhile another transformer is also taking AC and converting to DC to feed our components. When there is a blackout, we take DC from the battery, convert to AC and feed the computer PS where we convert back to DC to feed our components. Wouldn't be more cost effective and energy efficient to use the same unit to both feed our components with DC and the battery? Most of the time a PS has a surplus of power that can be used to charge the battery. When there is a blackout we could just take DC from battery and feed our components without the inefficient double DC->AC->DC conversion. I think an integrated UPS/PS would be cheaper (most components are replicated). That would be enough for most enthusiasts, that only need a relatively short battery life to be able to save work. A different matter is in professional environments where a UPS feeds dozens of computers that cannot be stopped. ¿Anyone knows if such a think exists?
    Reply
  • wrxchris
    I've been using a CyberPower 1350AVR for the past couple years and it has been great. Gives me approx. 17 mins of runtime at idle / basic tasks (approx. 250 watts between my rig and main LCD). It's good for 810W, and I believe I picked it up on sale for $120.
    Reply
  • while i personally use APC XL series, what about eaton powerware??
    Reply
  • g00ey
    I've always been wondering about the possibilities to replace the small batteries in those UPSes with standard car batteries or deep cycle (marine) batteries. Since the batteries in the UPSes are standard 12V lead cells it shouldn't be a problem and this would be a cheap way to keep the computer alive for days without external power.
    Reply
  • nebun
    g00eyI've always been wondering about the possibilities to replace the small batteries in those UPSes with standard car batteries or deep cycle (marine) batteries. Since the batteries in the UPSes are standard 12V lead cells it shouldn't be a problem and this would be a cheap way to keep the computer alive for days without external power.wrong
    Reply