Who's Who In Power Supplies: Brands, Labels, And OEMs

Great article with lots of data! Thanks for the comprehensive list.

I had a few comments to share that can help some. I am a power electronic engineer who designs switching supplies (PSUs) and inverter systems, so I see these technologies every day. One misconception is that PFC correction improves efficiency. It CAN improve efficiency if implemented right, but often it is not. PFC is needed to keep the current draw from the utility low in harmonics (only sinusoidal current) and in phase (phase angle power factor). This helps keep the utility system efficiency highest - but does not guarantee anything about the end power supply efficiency. That is depended upon the PSU design itself and how thorough it is done. Often, the simple passive diode-bridge and cap bank front end PSU is very efficient versus its PFC counterpart.

Another point - the IC pictured in one of the slides is the control IC, not a security device. There are tons of manufactures of these from Asia, Europe, and the US. The cheapest low efficiency supplies will use hard switching chips. The highest efficiency PSUs will use some form of resonant controller(s) which requires more cost and more parts - which is why higher efficiency costs more.

Last is a comment on what size PSU to use. Too many builders are using huge supplies "just to be safe" and are paying the price on a lower efficiency build when it is all done, not to mention too much cost. A larger PSU will have larger transformers, MOSFETS, diodes, and other parts. Larger parts require more overhead losses, just like having a V8 motor in a car will always have lower mileage than a 6 cylinder motor. The efficiency curve of most PSU designs falls off quickly as you go below 20% load because of losses in the larger parts and other extra overhead. That is why the Plus 80 curves don't go below 20%! It gets ugly below that. So size you PSU closely to the needed load. I have a 10TB server using an AMD 790GX motherboard and IGP graphics. There are a total of 12 WD Green series hardrives in this server. It has an Antec Earthwatts unit rated at 380W, and it only has to work hard during power up - the plus power peaks around 290W. Nominal power is around 115W, which is about 30% rated for this PSU - perfect.

One anecdote - I had a server several years ago that kept killing drives. We thought it was the SATA controller, but got the idea to scope the PSU. We found that the 12V rail was going over 20-21V during power up!! Ah, that's out of spec. Kinda.

- Charles

Great article with lots of data! Thanks for the comprehensive list.

I had a few comments to share that can help some. I am a power electronic engineer who designs switching supplies (PSUs) and inverter systems, so I see these technologies every day. One misconception is that PFC correction improves efficiency. It CAN improve efficiency if implemented right, but often it is not. PFC is needed to keep the current draw from the utility low in harmonics (only sinusoidal current) and in phase (phase angle power factor). This helps keep the utility system efficiency highest - but does not guarantee anything about the end power supply efficiency. That is depended upon the PSU design itself and how thorough it is done. Often, the simple passive diode-bridge and cap bank front end PSU is very efficient versus its PFC counterpart.

Another point - the IC pictured in one of the slides is the control IC, not a security device. There are tons of manufactures of these from Asia, Europe, and the US. The cheapest low efficiency supplies will use hard switching chips. The highest efficiency PSUs will use some form of resonant controller(s) which requires more cost and more parts - which is why higher efficiency costs more.

Last is a comment on what size PSU to use. Too many builders are using huge supplies "just to be safe" and are paying the price on a lower efficiency build when it is all done, not to mention too much cost. A larger PSU will have larger transformers, MOSFETS, diodes, and other parts. Larger parts require more overhead losses, just like having a V8 motor in a car will always have lower mileage than a 6 cylinder motor. The efficiency curve of most PSU designs falls off quickly as you go below 20% load because of losses in the larger parts and other extra overhead. That is why the Plus 80 curves don't go below 20%! It gets ugly below that. So size you PSU closely to the needed load. I have a 10TB server using an AMD 790GX motherboard and IGP graphics. There are a total of 12 WD Green series hardrives in this server. It has an Antec Earthwatts unit rated at 380W, and it only has to work hard during power up - the plus power peaks around 290W. Nominal power is around 115W, which is about 30% rated for this PSU - perfect.

One anecdote - I had a server several years ago that kept killing drives. We thought it was the SATA controller, but got the idea to scope the PSU. We found that the 12V rail was going over 20-21V during power up!! Ah, that's out of spec. Kinda.

- Charles

Maybe add something recommending quality PSUs?
Also, you missed XFX and its seasonic PSUs.

Replacing the psu resolves many more of the hardware issues brought to Tom's than any other part . . . probably more than half.

It isn't possible for the average user to determine whether or not a specific psu is worth using. When we recommend a psu to someone in the forums, it's based on overall corporate reputation (arguably two companies) or specific reviews done by qualified reviewers. That gives us a relatively limited range of safe recommendations.

The more psus are reviewed, the more information that's made available, the better.

Every single SeaSonic PSU I have owned is still running.
I have had Antec and countless others die. Obviously the Antec wasn't the SeaSonic model.

Now I only buy SeaSonic.

Article doesn't have any stats on PSU usage, market share, or failure rates.

Very nice article, though most of the info was available on the net, but not in one place, and this article fixes that...Do update the tables often...