Technology

After we published the review of the Turbo Cool 1200W, we talked about the topology a bit. Somehow, our opinions differ in regards to what the best approach is. Why do you think the design is fine the way it is now?

Doug Dodson (DD): In order to get the UL and TUV certificates we had to test and qualify all of the components in the power supply. Every component passed the temperature tests with plenty of margin, so we don't think the topology has any flaws as you stated in your review.

From the topology side of view the design is very sleek; I am just concerned about all those components that are blocking the airflow.

DD: We actually have several wind-tunnels going through the power supply that leave enough space for the air to go through. That is enough for cooling all of the components.

But the temperatures you showed were reached with one of the noisiest fans in the industry.

DD: Yes, but the Turbo Cool 1200W was designed for servers, workstations, and triple-SLI setups that dissipate a lot more heat than simple home PCs. The Turbo Cool 1200W is rated for full load operation at 50C ambient temperature. As an industrial-rated PSU, it obviously requires a higher capacity fan than a consumer PSU rated at 25C-35C ambient. In other words, it's not the topology that demands a high-capacity fan as much as it is the highly-reliable 50C rating.

Speaking of PSU Myths published on your website, what is your opinion about modular cables today since you found them unstable? The fact is that most of the users want modular cables just to be free to remove them when they're not needed.

DD: Just because users like them doesn't make them good. The fact is that with lower wattage consumer PSUs you can probably get away using cable management because the voltage losses are relatively minor and reliability isn't critical. With higher end power supplies above 800W, it would be careless to use such a design because the voltage drop through the connectors is significant and the applications tend to be mission critical.

What is your opinion on the constantly increasing wattages of power supplies? Why there are so many power supplies today offering more than 1000W? There is obviously no need for them at the moment.

DD: The need for high wattage PSUs was created by platforms with multiple CPUs and GPUs. When you see the market of power supplies in general, you will see lots of lower end manufacturers claiming high wattages. In some cases they can only reach half of the stated output, so consumers need to buy a PSU labeled as 1000W to get a continuous 500W. In the high-end sector, you can choose a quality PSU with a more moderate maximum output, because it can actually deliver that amount of power.

But I don't see many power supplies with lower outputs and more cables. I actually had to ask you to build me a custom power supply with all the connectors I wanted.

DD: You asked for connectors to support 3-way SLI (6 PCI-E) from an 860 watt power supply. According to NVIDIA, that setup requires an 1100W PSU. We built the custom Turbo Cool 860 anyway to show you how conservative our ratings are. The reason the 860 doesn't come with six PCI-E standard is because we can't market the product for 3-way SLI without NVIDIA certification and they won't certify a PSU under 1100W for 3-way SLI, no matter how well it works. For users running 3-way SLI, we have our NVIDIA certified Turbo Cool 1200.

Why are the high-end power supply manufacturers still pushing such high wattages?

DD: Systems with multiple CPUs and GPUs can actually draw around 1000W. In other cases, users want the advantages associated with operating at 40-70% capacity. These include wider input operating range, longer hold time, lower noise and ripple, cooler, quieter operation, and longer product life.

And this is good for the companies of course because they make higher margins from higher wattage models?

DD: That's not necessarily true. The margin on the Turbo Cool 1200 is below average because the unit uses expensive low-volume components. The real advantage of building high wattage power supplies, besides bragging rights, is that it moves our proprietary technology forward and that knowledge can then be used to improve the performance of the high volume midrange products.

That sounds reasonable. This is also for example how the single 12V rail got into place. Why exactly is one 12V rail better than separate rails, and why is this not a safety issue for the user?

DD: One 12V rail is better because all of the power supply's capacity is available to the system. With a multi-rail 12V design, as much as 30% of the PSU's capacity can be trapped on under-utilized rails. For example, if one 12V rail rated at 18A is for the CPU, and the CPU only draws 8A, the remaining 10A cannot be utilized by other components in the system.

That's true. I had an AnandTech reader writing me about a problem with his setup powering up 18 hard disks at the same time. I suggested he either gets two power supplies, using the second just for the HDDs, or that he use a power supply with a massive single 12V rail. Eventually he ended up with one of your Silencer 750W power supplies and that completely solved the problem.

DD: That is one of the problems you can solve with a single rail, yes.

So what about safety concerns with let's say 90A on one rail?

DD: The safety agencies wouldn't approve our units if there was a risk to consumers. I've done tests using my own body to prove it's not an issue.

(That's something we really would have liked to see ourselves!)

Index Markets and Work
POST A COMMENT

49 Comments

View All Comments

  • mindless1 - Monday, February 11, 2008 - link

    It applies just as much to generic manufacturers who sell 500W PSU for $15, knowing you'd have to pour some fuel in and light it to get 500W out for more than a moment. Reply
  • Olaf van der Spek - Monday, February 11, 2008 - link

    > for ~98% of the desktop pcs, a quality 400W unit is more than enough.

    I think 150 W is enough if you use onboard video and 300 W is enough for a decent (8800 GT or Radeon 3870) video card.
    Reply
  • mindless1 - Monday, February 11, 2008 - link

    Those figures are only enough if you're an OEM with control over a standardized platform so you know the current per rail the system use and can then spec the PSU current per rail you need.

    Otherwise, consider some recent P4 CPUs produced 110W+ themselves, running off a (roughly) 93% efficient VRM subcircuit on a motherboard that would already be 118W. Will 32W be enough to run the whole motherboard, a hard drive, optical drive, maybe one expansion card? Doubtful, many motherboards use this much by themselves.

    300W is a lot more breathing room, since most with a video card don't buy 300W PSU if it's a higher end card and why should they really? If their budget includes $150+ for a midrange or better video card they shouldn't really be thinking along the lines of paying only $30 or less for a PSU to power it all, especially when the PSU is the least likely to be obsolete within 18 months so it might be reused.
    Reply
  • Olaf van der Spek - Tuesday, February 12, 2008 - link

    > Otherwise, consider some recent P4 CPUs produced 110W+ themselves,

    Those P4s have never been a good choice.
    Reply
  • tynopik - Monday, February 11, 2008 - link

    b/c it only has a pathetic 6 peripheral and 4 sata connectors

    bump it up to 10 and 6 and i might consider it
    Reply
  • JarredWalton - Monday, February 11, 2008 - link

    It goes back to marketing. NVIDIA and AMD have requirements for SLI/CrossFire certification, and you can't easily market a PSU as a high-quality unit unless it has SLI/CF support. So, you need to meet their requirements (1100W for Tri-SLI and 550W? for SLI), which means pretty much all PSUs these days worth considering start at 550W or more. Reply
  • Olaf van der Spek - Monday, February 11, 2008 - link

    > So, you need to meet their requirements (1100W for Tri-SLI and 550W? for SLI),

    Looks like a requirement for the marketing department, not for the technical department.

    Don't AMD and nV have any sense when it's about power?
    Reply
  • JarredWalton - Monday, February 11, 2008 - link

    Doug explains this in the article. A crappy PSU from a generic brand that says 450W isn't going to work with SLI, and many users don't know the difference between good and lousy PSUs. So, to be safe they figure a crappy 600W PSU can run SLI, so they "require" a 600W PSU for SLI. The reality is that your typical SLI rig won't even draw more than 300W, and triple-SLI is under around 600W (without overclocking).

    You may not be susceptible to marketing, but about 90% of the market is unfortunately.
    Reply
  • NicePants42 - Monday, February 11, 2008 - link

    You hit Doug for noise on the Turbocool 1200, and he fires back with 50C ambient requirements - isn't it time someone mentioned water?!

    OCZ did hybrid air/water cooling for RAM, and RAM is already silent. The guys blowing money on Triple-SLI rigs are the same guys blowing money water cooling them for silence. How long will PCP&C keep designing 80mm jet engines?
    Reply
  • mindless1 - Monday, February 11, 2008 - link

    There are many different PSu configs out there to suit many different needs. It would be terrible if everyone changed their designs to suit your one ideal because that one ideal is subjective to your needs.

    If your system uses little enough power that it can run a PSU at acceptible lifespan using low air volume, the Turbo 1200 wasn't right for your needs.

    Just moving up to 120mm fan doesn't magically keep noise low when talking about even an optimistic 85% efficiency at 1200W which is close to 210W of heat it has to remove, and that when at it's rated 50C ambient. Your CPU isn't likely to be in 50C ambient, 210W heat, and run ok from a quiet heatsink fan either.
    Reply

Log in

Don't have an account? Sign up now