Under Construction Hard Hat Area Over the past few years a situation has developed that many are unaware of while we have all enjoyed the benefits of ever increasing performance out of our computers, several notable changes have crept into how those are powered. Today many have experienced intermittent stability problems, and or outright damage to components because of poor or inadequate power. Power supply selection is rapidly becoming as important as selecting compatible RAM, and equally complex. Basically what has occured is that while older PSUs (Power Supply Unit) output voltage on the +3.3 Volt and +5 Volt rails to power the Motherboard, Graphics Card and CPU, with the +12 Volt rail for the Hard Drives, Optical Drives and Fans. That has now changed. Modern boards heavilly rely on the +12V rail to power more powerful processors and graphics cards, and with the introduction of the PCI-Express bus, this will only get worse. In short an older PSU of yesteryear is likely woefully inadequate on the +12V rail irregardless of its overall watts rating, since a large portion of the power its producing simply isnt need on those rails anymore, but on the +12V rail instead. This article is going to address the pitfalls and tradeoffs you need to be aware of when selecting a PSU so that you are able to make an informed decision as to what level of performance you need, so lets start with the basics. Power Delivery is a chain, and your PSU is but a link in that chain if you make it a strong link it can actually makeup for some on the shortcoming of the other links, unlike a real chain. Power quality from a utility varies greatly over the world, from utility to utility and even with the seasons, and while there really is no substitute for power conditioning or a UPS (Uninteruptable Power supply) a good PSU is able to effectively deal with a larger range of AC power than a cheaper PSU, and when the range its able to deal with is exceeded, safely shutdown without damaging components, something cheap generic supplies are infamous for not doing. 1.Source Power Brown outs, blackouts, spikes\surges ect. In this category I would also place power issues due to pilot error, hard restarts and shorts, Shutdown properly and pay attention when mounting your motherboard and routing power cables. 2. Under Power: Basically too many components for the power supply, dont be decieved by wattage figures, its the amount of amps per rail that is really important. 3. Voltage Stability how clean the power is one of the problems right now is there are alot of PSUs out there built to the ATX12V v1.3 spec which calls for the dual rails, but they arent documented by the retailers or manufacturers often youll find the appearantly very same supply with a different wiring harness as an EPS12V (which has at least three rails) or ATX12V v2.0 and they might have the +12v rails documented the same goes for the main if you veiw the connector on the mobo with the clasp on the right you would need roon to hang the extra 4 pins off the top there are 24 to 20 pin cconverters readilly available but there arent 8 to 4 pin coverters (so EPS12V to ATX12V v2.0 or v1.3 is problematic as well as the possibility that you wont have access to the power off the 3rd +12V rail, and certainly not if its a quad rail) in addition there will be a new 6 pin video connector soon and of course SATA for a short course in the wiring harness difference ATX12V v1.3....20 pin main 4 pin aux ATX12V 2.0.....24 pin main 4 pin aux EPS12V.........24 pin main 8 pin aux the pins are backward compatible if there is clearance on the mobo so you can connect an 8 pin auxillary to a 4 pin header if there is room above the connector for basically double the pins when viewing the mobo header with the clasp on the right in short there are supplies that meet the latest specs but are described by the wiring harness they have, and they are mixed up in the retail marketplace ATX pot with ATX12V 1.1 supplies that dont have dual rails confusing aint it? add up the rails with this calculator http://takaman.jp/D/index.html?english and compare it to the specs listed on the PSU then you build in a safety margin of from 1\2 to 1\3rd by deducting 1\2 to 1\3 the value of the PSU's rated amps and see if it still fits it actually varies with the distribution ratios your likely to need more +12V than +3.3V or +5V (CPUs now being powered by the +12V primarily) possibly more if it a long term infrastructure investment and there is growth built in of if the veracity of the manufacturer is in question (generics tend to lie like dogs) most 250 or 300 watt PSUs will actually run most configs, but stability has become an increasing concern with the tighter tolerances onboard (FSB) There is a decrease in total capacity with the rise in temperature , which reduces your amps, the rated amp values where taken at 25C while your likely operating temperature will be 40C (especially if the PSU is in the top of the case exhausting the CPU HSF) and that is roughly a 30% decrease That is offset by the additive nature of the calculator, employing all tha maximum draw figures for the assorted components, something that will never occur However it gets even more complicated if you have alot of drives and fans, those are typically given a "run time" draw value in a calculator, there "spinup" draw can be 4 to 5 times as much and they greatly contribute to transient response overshoot and undershoot in some supplies at startup if there isnt enough +12V The way you torture a power supply is to give it a fluctuating AC feed to deal with (from surge to brownout), at the same time you ask it to deal with a really dynamic internal load change (like spinning up alot of drives) while still keeping the rails stable enough for the onboard voltage regulation components of unknown quality Failure anywhere along the chain from too big a spike at the source to too long or high an overshoot or undershoot to the mobo, with too much ripple or noise for the onboard regulator to deal with and RAM or other components can go bye bye, ideally the power supply will trip off and protect your components, the operable range it has is largely what the difference in one PSU to another is about. And when it comes to a comparision of a flyweight generic, the whole protection scheme of shutting down in time really comes into question. And of course how stable the rails can be maintained, how low the AC Ripple and noise. so if you elect to get a supply in the near future you have to ask yourself if its going to be a true infrastructure investment in the past that was typically true, now it a little tougher with more power hungry devices PCI Express, video cards ect. They recently added 4 more pins to the main connector from 20 to 24, and an additional 4 pin +12V auxillary power connector, (ATX12V 2.0) and the spec keeps jumping the total amps on the +12V rail (ATX12V 2.0 & 2.2), actually there are now two +12V rails and Ive seen power supplies that have Quad rails ATX12V v1.1 (ATX 2.03 standard) is a 20 pin PSU with a 4 pin +12V connector but if your considering a long term investment the most important thing for you to determine is the number of pins the mobo connector has, and get a ATX12V v2.2 compliant power supply (unless you need an SSI Compliant EPS12V), and if you have a 20 pin connector see if it can be attached directly with a 24 pin PSU or if an adapter is needed (cap clearence) as far as what your rails are reading in a monitoring program, for starters, you cant observe that during startup with software or the BIOS here is a Codgen300X1 under a dyanmic load from > http://terasan.okiraku-pc.net/dengen/tester/index.html and > http://terasan.okiraku-pc.net/dengen/tester2/index.html (but hosted independently) note the instabiliy at spinup (an extreme example, but thats common if you review the links) those are logged Digital Multimeter readouts off a Sanwa PC510 (0.08% accuracy) via RS232C or USB port to PC Link the logging software you see there the test system is a Tyan Tiger 133 (S1834D).w\ Dual P3 650s overclocked to 865MHz (I think ) Dual sticks of 128MB PC133, a Matrox G400SH, Xwave 6000 sound card, NIC and a single HDD, FDD and CDROM measured from startup, booting into Windows 2000, and Running 3DMark 2000 to conclusion, measured in 5 second intervals, which is definately less than ideal, but the best he could manage not exactly a power hungry system that would be "pushing" those supplies too hard I use those as illustrative of the instability that spinning up drives can produce those supplies are hardly at the edge of their capacity, nor are they dealing with less than ideal AC power Ripple & Noise, not being specifically addressed. and while they arent "out of spec" they arent being taxed either compared to a PC Power & Cooling 450ATX same source so Ideally youd like to use a Digital Multimeter to read your rails directly, watch the spinup, and then if you can find a realtively stable voltage state use it to calibrate the voltage your reading on the DMM to the software (easy to do in MBM) a bit more cut and paste ----------------------------------------------------------------------------------------------- "the majority of damaged RAM returned to memory manufacturers is destoryed by fluctuations in the voltage." http://www.anandtech.com/showdoc.html?i=1774&p=8 Winbond Launches New Bus Termination Regulator April 4th 2003 "Winbond Electronics Corporation, a leading supplier of semiconductor solutions, today launched the W83310S, a new DDR SDRAM bus termination regulator. The solution, new to Winbond's ACPI product family, is aimed at desktop PC and embedded system applications with DDR SDRAM requirements. Computer systems architectures continue to evolve and are becoming more complex; CPU and memory speeds continue to increase ever more rapidly with every technology turn. More and more high current/low voltage power sources are required for PC systems. This is particularly true for high-speed components such as CPU, memory, and system chipsets. The performance of these components is highly dependent upon stable power. Therefore, motherboard designers require accurate, stable, low-ripple and robust power solutions for these components. Many system designs use discrete components to implement bus termination functions. This approach creates several problems including poorer quality load regulation; higher voltage-ripple, increased usage of board space and inconsistent designs when different discrete components are used." the transient response is the critical internal measure, unfortunately its not a metric that is commonly supplied with the PSU specs (this seems to be slowly changing, as some manufacturers are supplying the transient response now) Transient Response: As shown in the diagram here, a switching power supply uses a closed feedback loop to allow measurements of the output of the supply to control the way the supply is operating. This is analogous to how a thermometer and thermostat work together to control the temperature of a house. As mentioned in the description of load regulation above, the output voltage of a signal varies as the load on it varies. In particular, when the load is drastically changed--either increased or decreased a great deal, suddenly--the voltage level may shift drastically. Such a sudden change is called a transient. If one of the voltages is under heavy load from several demanding components and suddenly all but one stops drawing current, the voltage to the remaining current may temporarily surge. This is called a voltage overshoot. Transient response measures how quickly and effectively the power supply can adjust to these sudden changes. Here's an actual transient response specification that we can work together to decode: "+5V,+12V outputs return to within 5% in less than 1ms for 20% load change." What this means is the following: "for either the +5 V or +12 V outputs, if the output is at a certain level (call it V1) and the current load on that signal either increases or decreases by up to 20%, the voltage on that output will return to a value within 5% of V1 within 1 millisecond". Obviously, faster responses closer to the original voltage are best."