April 4, 2009

How to Measure Efficiency of a Multiple-Output Power Supply

On March 27, 2009, the US Department of Energy (DOE) published an amendment to its test procedures for battery chargers and external power supplies (EPS). The amendment includes provisions for measuring standby mode and off mode energy consumption and adds a test procedure for testing switch-selectable external power supplies. Since the document basically clarifies existing test procedures, which DOE originally could not formulate clearly, it is mainly clerical and would not be worth noting if not for one funny thing. The amendment still does not include a test procedure for multi-output power supplies because of "the number of stakeholder comments and the limited timeframe for this rulemaking." OK, here is a little background. The Energy Independence and Security Act of 2007 (EISA 2007) set Federal efficiency standards for certain types of adapters and single-output EPS for consumer electronics. DOE has a test procedure to verify this efficiency. However, this procedure [which is basically: multiply output volts times amps and divide by the input power under various loading conditions] was written for single-output power supplies. It has therefore left out a significant group of products, powering applications such as radio transceivers, video game consoles, and printers. On August 15, 2008, DOE proposed a procedure for the testing of multiple-voltage external power supplies. On page 6 of the Proposed Rule we read: "DOE is not aware of any existing test procedure developed specifically to measure the efficiency or energy consumption of multiple-voltage external power supplies." Huh? Millions of multi-output power supplies both external and internal including PC PSUs are being built around the world and tested for Energy Star® and 80 PLUS® efficiency compliance. Folks, how about this: take a 5th grader, ask him/her to add up volt-amps of all outputs, divide by the input power, and you are done! Seriously, the main obstacle for DOE seems to be the loading profile. The PSU's nameplate power is often smaller than the sum of the nameplate power of each individual output. So what's the big deal? Derate (scale down) the loads of all the outputs proportionally, so that their sum does not exceed the limit for the entire power supply. Of course, it would not be the worst case efficiency (which you would probably encounter if you shift the bulk of the load onto the lowest voltage outputs), but it would be kind of typical efficiency. Back in 2006, at the APEC conference, EPRI Solutions and Ecos Consulting presented their Enhanced Proportional Allocation Method for Loading Power Supply. Their method is incorporated by Intel in its PC power supply guides. DOE did propose to adapt the ‘‘proportional allocation’’ method. Nevertheless, the DOE 2009 final rule still reads: "Due to the limited time provided by EISA 2007 and limited resources available prior to the publication of this final rule, DOE was unable to address the large number of stakeholder comments received and decided to defer action on multiple-voltage EPSs to a 2009 rulemaking." In a private sector usually a new power supply will be designed, built and tested in this timeframe or in a shorter one. It's entertaining to watch how long will it finally take for US DOE to make up their mind on how to test the efficiency and to move to something else.

No comments: