Another great review. Maybe the unusually high ac ripple is because of the different manufacturing company?

Will a PS such as this with the fan on top work well in the P182 case.

All PS I have used in the past have the fan placement in the back.

I just want to make sure it will get the proper airflow in the case before I make a pruchase.

I like the price of this one under $100.

Yes.

The article mentioned it could be found for $75 AR, what site was this as I have looked but the lowest I see it is $100 AR.

@MikeC: nice PSU built by Seasonic and of course nice review by SPCR, as always.
But I believe it would be a lot more interesting to review the CWT built PSUs that sell under the Corsair brand.
Some newbies might not know the difference and buy those ones, believing they are as good as the Seasonics.

When the high load is removed, and the system goes back to using ~200W, does the fan eventually slow down again?

Yes, the fan controller is smoothly responsive -- ie, it doesn't change speed too fast, but fast enough. There is the right amount of hysteresis. (Not to be confused with hysteria. )

Do most PSUs have fan control that goes up & down? Or do some of them stay on high - i.e. not slow down after a gaming session

1. yes
2. only if the internal temperature of the case/psu stays high.

"Corsair's five year warranty remains the longest for any computer power supply we know of."

Correct me if I'm wrong, but doesn't Seasonic itself offer a 5 year warranty on its PSUs?


Verbatim from Seasonic,

umm..... drives? Maybe your thinking of seagate which does offer 5 years on Hard Drives... but what does that have to do with PSUs

Buy.com has it for $97 shipped, $73 AR.

It may be a good idea to update the list and include a Core 2 Quad and Radeon 3870 X2 for example.

Nope the post said drives, but meant PSUs, apparently it is new:

From the RMA page at http://www.seasonicusa.com/rma.htm


The service page still says 3 years but you know how slow companies are to update web pages.

http://www.seasonicusa.com/S12II.htm mentions 5 years warranty.

http://www.seasonicusa.com/M12II.htm mentions 5 years warranty.

http://www.seasonicusa.com/m12.htm mentions 5 years warranty.

http://www.seasonicusa.com/s12plus.htm mentions 5 years warranty.

The old S12 series still says 3 years on http://www.seasonicusa.com/s12.htm but that link isn't present on some pages of the website so it was probably meant to be removed and someone missed a link or two.

Oh and there is this


Notice the absence of the old S12 on that list and the mention of Nov 1st 2007 again.

Oh and as to Corsair apparently they went for 5 year warranties on Seasonic AND Channelwell Tech (CWT) units.

from http://www.corsair.com/warranty/default.aspx

5 year warranty mentioned on http://www.corsair.com/products/vx.aspx

5 year warranty mentioned on http://www.corsair.com/products/tx.aspx

5 year warranty mentioned on http://www.corsair.com/products/hx.aspx

Normally, one would describe it as having the fan at the bottom, as most cases have the mounting position in such a manner. If you point it upwards, it's actually upside-down. It will work just as well though, as mentioned above.

The author comments on how high the MTBF is and wonders how it is calculated.

The MTBF comes from any or all of the following:

A design based prediction (MIL-HDBK-217F, Prism, etc)
Field Failure Date
or Accelerated life test

The meaning is as follows:

First, manufactures rarely state MTBF properly. MTBF is a population statistic (i.e. applicable only to populations of items not individual items) This means that it has a particular distribution (which should be stated) and has variance (which should also be stated).

If we assume a exponential distribution (the usual assumption) then an MTBF of 100,000 hours means that when the population of power supplies accumulates 100,000 hours there should be a failure. So for example if your population of power supplies is 1,000 and if all of them are turned on at the same time then after this population operates for 100 hours then there should be 1 failure. (+/- the variance)