0PSU Efficiency: worth the upgrade? part 3

This is the third install­ment in a series of art­icles address­ing PSU effi­ciency. Art­icle one focused on a [int­link id=“76” type=“post”]theoretical new server[/intlink], art­icle 2 on a [int­link id=“77” type=“post”]typical desktop PC[/intlink].  This art­icle will focus on a cur­rent real serv­er, and intro­duces a new PSU — an 80+ plat­in­um model.

The sys­tem spends a lot of time semi-idling (serving net­work and disk requests but not doing too much CPU work).  I estim­ated that it would pull around 150W on aver­age, and then cal­cu­lated it below. It really did come out exactly like that!

Out of interest, assum­ing a cost of £0.10 per kWh this sys­tem would cost a total of £131.50 a year to run even with a per­fectly effi­cient PSU!

The ATX spe­cific­a­tion requires 70% effi­ciency. Below is a com­par­is­on between a bare min­im­um ATX PSU and a few highly effi­cient PSU’s.

The fig­ure for the gen­er­ic PSU is based on the min­im­um require­ment and is there­fore a worst case. I have cal­cu­lated a best-case yearly sav­ing, based on a kWh charge of 10p and assum­ing 24.7 usage.

Con­clu­sion: This is the first scen­ario where new high­er-effi­ciency PSUs are reas­on­ably price com­pet­it­ive based only on power usage. For such a sys­tem, a high effi­ciency PSU will pay for itself with­in approx­im­ately 3 years. If you plan to keep the sys­tem for more than 3 years a high effi­ciency PSU is a worth­while invest­ment as its TCO will be lower. As energy prices con­tin­ue to rise this TCO advant­age will widen.

Update (25-Dec-2011)

Added the HuntKey Jump­er R90. This PSU changes the con­clu­sion con­clus­ively. It pro­duces sav­ings of over £40, which is at least the price of the PSU!  A pay­back peri­od of 1 year, with an expect­a­tion that elec­tri­city prices will con­tin­ue to rise is well worth the investment.