In the course of shopping for parts for a new build, the object of my obsession today has to do with power supplies.
First and most important, I'd like to discuss power ratings briefly. Make sure you understand the difference between peak or max power capacity and sustained power capacity. In case it's not obvious, the first number will always be higher and is the power that can be delivered only for brief surges, such as the excess power during device power-up.
Next, do the math. Basic Ohm's Law, Volts times Amps = Watts. Let's take a certain 750W power supply, whose label says "750W MAX" and completely neglects to mention sustained power capacity
: Its label shows 3.3V at 30A, 5V at 28A, and four 12V "rails" at 18A each, plus a few smaller numbers. That's a total of well over 1100W. So yeah, that doesn't add up. In fact, the 864W at 12V alone exceeds the 750W rating. A wise buyer will make sure not only that total demand is met by the power supply choice, but that the power distribution across the rails is sufficient for the needs of the individual components.
Modular power supplies are all the rage now, and I'm a big fan of anything that permits a neater case interior. How much neater it'll be I'm not 100% sure. Maybe it'll save a cable or two. And most of the modular supplies aren't entirely modular, having motherboard cables permanently attached, and terminating with all the different motherboard connectors, not just the ones you need. And some of the modular cables will probably have more terminations than you'll use. Not so neat after all. Personally, I've found that cable ties and self-adhesive cable tie mounts make a huge difference in keeping things where they belong, out of the way and neat. Sleeving also helps with appearance, and twisting unsleeved cables is very helpful too.
Active PFC is and 'green' efficiency are also big selling points now. I'm delighted, but efficiency still varies widely. Not only do efficiencies vary, but also the efficiency curves, and the curves aren't always published or even easy to find. It is however easy to find as much as a 15% difference in efficiency between two similarly rated Active PFC supplies, and at 300W draw from the computer's components this difference works out to a 75W difference in draw from the electrical outlet. If you run that box 24x7, at a cheap residential 10 cents per KW/h, the difference will cost you more than $65 over the course of just one year. The curve is also something to think about.
Then there's the subject of noise. I hate noise, and I'm glad that the engineers are finally paying some attention to reducing it, in power supplies and elsewhere. When looking at specifications the only thing that's pretty much standard is that the numbers will be in dB-A (meaning decibels, on an A-weighted scale to be similar to the frequency range and sensitivity of human hearing), at a distance of 1 meter. But that number will ignore certain subjective characteristics of the sound, as well as how those sounds are transmitted through and resonated by the case the power supply will be used in. While it really is entirely unfair to criticise that aspect of the measurement, it is entirely fair to expect some consistency in how the numbers are reported, now that most PS fans are variable-speed. Some specs will report the noise at the lowest fan speed and only the most daring will report the noise level at the highest fan speed.
One final thought is that of scaling your power supply power capacity. Most power supplies are most efficient close to the middle of their load rating, by a difference of up to 5% or more. And most power supplies with variable speed fans don't ramp up the fan speed until the load increases beyond roughly the middle of their load rating. So there are some benefits in both energy cost and noise, in being generous with power supply capacity.
Shop carefully!
First and most important, I'd like to discuss power ratings briefly. Make sure you understand the difference between peak or max power capacity and sustained power capacity. In case it's not obvious, the first number will always be higher and is the power that can be delivered only for brief surges, such as the excess power during device power-up.
Next, do the math. Basic Ohm's Law, Volts times Amps = Watts. Let's take a certain 750W power supply, whose label says "750W MAX" and completely neglects to mention sustained power capacity
: Its label shows 3.3V at 30A, 5V at 28A, and four 12V "rails" at 18A each, plus a few smaller numbers. That's a total of well over 1100W. So yeah, that doesn't add up. In fact, the 864W at 12V alone exceeds the 750W rating. A wise buyer will make sure not only that total demand is met by the power supply choice, but that the power distribution across the rails is sufficient for the needs of the individual components.Modular power supplies are all the rage now, and I'm a big fan of anything that permits a neater case interior. How much neater it'll be I'm not 100% sure. Maybe it'll save a cable or two. And most of the modular supplies aren't entirely modular, having motherboard cables permanently attached, and terminating with all the different motherboard connectors, not just the ones you need. And some of the modular cables will probably have more terminations than you'll use. Not so neat after all. Personally, I've found that cable ties and self-adhesive cable tie mounts make a huge difference in keeping things where they belong, out of the way and neat. Sleeving also helps with appearance, and twisting unsleeved cables is very helpful too.
Active PFC is and 'green' efficiency are also big selling points now. I'm delighted, but efficiency still varies widely. Not only do efficiencies vary, but also the efficiency curves, and the curves aren't always published or even easy to find. It is however easy to find as much as a 15% difference in efficiency between two similarly rated Active PFC supplies, and at 300W draw from the computer's components this difference works out to a 75W difference in draw from the electrical outlet. If you run that box 24x7, at a cheap residential 10 cents per KW/h, the difference will cost you more than $65 over the course of just one year. The curve is also something to think about.
Then there's the subject of noise. I hate noise, and I'm glad that the engineers are finally paying some attention to reducing it, in power supplies and elsewhere. When looking at specifications the only thing that's pretty much standard is that the numbers will be in dB-A (meaning decibels, on an A-weighted scale to be similar to the frequency range and sensitivity of human hearing), at a distance of 1 meter. But that number will ignore certain subjective characteristics of the sound, as well as how those sounds are transmitted through and resonated by the case the power supply will be used in. While it really is entirely unfair to criticise that aspect of the measurement, it is entirely fair to expect some consistency in how the numbers are reported, now that most PS fans are variable-speed. Some specs will report the noise at the lowest fan speed and only the most daring will report the noise level at the highest fan speed.
One final thought is that of scaling your power supply power capacity. Most power supplies are most efficient close to the middle of their load rating, by a difference of up to 5% or more. And most power supplies with variable speed fans don't ramp up the fan speed until the load increases beyond roughly the middle of their load rating. So there are some benefits in both energy cost and noise, in being generous with power supply capacity.
Shop carefully!
Facebook
Twitter
Instagram