Thane, The higher the voltage rating, the larger the capacitor - usually this means longer rather than larger in diameter. ESR sometimes drops a little as the working voltage increases, but its not worth the gain just for that fact.
Aluminum electrolytic caps usually fail in PC power supply bypass applications due to excessive heat - heat generated internally by ripple current passing through the internal resistance, heat from external heat buildup outside the cap, or a combination of both. To reduce internally generated heat, you want low internal resistance (lowest ESR). For optimal low-frequency power bypassing performance, you want the lowest ESR from DC to about 100 kHz from your aluminum electrolytic caps. At mid-frequencies, the tantalum chip caps kick in. At high-frequencies, the ceramic chip caps take over. At the highest frequencies the power and ground-planes, feedthroughs, trace sizes and spacing take over. More than you wanted to know, eh? For voltage rating, you don't need more than regulated bus voltage plus 25% or 33% (5V bus gets caps with 6.3 WVDC rating: 5V * 1.25 ~= 6.3 WVDC), 12V bus gets 16V WVDC (12V bus gets caps with 16 DCWV rating: 12V * 1.33 ~= 16 WVDC), etc. Aside from the manufacturer logo, part series/number/type, DCWV, service temperature and capacitance, the really important specs that affect PC power bypassing - like ESR and ripple current rating - have to be read from the manufacturer's datasheet. When replacing PC bypass caps you DEFINITELY do not want to use audio bypass caps (like the Philips 037 below seems to be). The Rubycon ZA-series is the way to go. hth, _jim -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Thane Sherrington (S) Sent: Thursday, October 27, 2005 7:17 AM To: The Hardware List Subject: Re: [H] Replacing capacitors At 12:20 AM 05/10/2005, Jin-Wei Tioh wrote: >I probably shouldn't have made a blanket statement. For those caps, >replacing them with caps of non-equivalent value can be done, >so long as both the voltage and charge ratings are >= the original. In RF >paths though, you usually want to keep the same exact >capacitance value. More reading on this: This page: http://active-hardware.com/english/hardinfo/upmobo.htm says to go with 16V or 25V capacitors because you need the headroom for 5V and 12V lines, respectively. This page: http://www.overclockers.com/tips469/ doesn't agree on anything. The general consensus appears to be higher voltage is fine, as is higher capacitance, but don't overdo it. Here are my questions: If I put a 25V on everything, will I cause problems on the 5V traces? Is there any real advantage to upgrading any of these? I found this comment: " Capacitors on motherboard, Rubycon ZA-series 1000uF 6.3V. This capacitor has max ripple current at max operation temperature (105C,100kHz) of 1650mA and impedance of 24mOhm (at 20C, 100kHz) Some general purpose, cheap-ass-capacitor, Phillips 037-series 2200uF 16V. This capacitor has max ripple current, at max operation temperature (85C,100Hz) of 1150mA and impedance of 150mOhm (at 20C, 10kHz)" This seems to suggest that I'd be better off using the same voltage and capacitance, but getting better max ripple and impedence. If this is the case, is there a way I can find out these rating from looking at a capacitor? T
