Ceramics near CPU socket.

[linuxguru]

This has probably been mentioned elsewhere, but this is what I did to an MSI 7061 (KM4M-V) board to permit a stable overclock of an Athlon XP2000 (266 FSB) to an XP2600 (333 FSB). I populated the six 0603-outline ceramic bypass capacitors directly below the socket with 4.7uF/6.3 X5R ceramics with measured ESR of ~0.4 ohms each. The ESR specs aren't great, but six of them in parallel very close to the pins of the CPU appear to help greatly. Not only does it overclock stably, the idle CPU temperature dropped from 60c (too hot) to 57c (still hot, but better) with no other changes.

I'm not sure if these ceramics bypass Vcore or 3.3Vdd, but they seem to work regardless.

[Per Hansson]

Re: Ceramics near CPU socket.

Yup, but pics please

https://www.badcaps.net/forum/showthread.php?t=1493

[linuxguru]

Re: Ceramics near CPU socket.

Messy soldering, but here is the pic of the underside of the socket:

[Agent24]

Re: Ceramics near CPU socket.

They probably found that those capacitors weren't critical for operation and figured they could save 1 cent per board if they didn't fit them!

[linuxguru]

Re: Ceramics near CPU socket.

Yeah, or they sell a Pro- version of the board priced at 2x, which has better electrolytics, VRM and all these ceramics populated, and validated with higher-FSB processors. I'm fine with that, as long as they're up-front about it - Intel has been doing similar stuff from the era of the 487SX, and later with the Celeron.

[nehal]

Re: Ceramics near CPU socket.

How big would a 820uF 6.3V capacitor be? What kind of ESR would you get?







nehal

[kc8adu]

Re: Ceramics near CPU socket.

it would be a big brick if it was an mlcc.
even a poly would be too big.
btw 2.5v rated parts would do here.if vcore exceeds 2v you got big problems!

[mariushm]

Re: Ceramics near CPU socket.

Next time, use ceramic capacitors rated for higher voltage or use C0G/NPO rated ceramics (which don't have voltage bias issues).

A 4.7uF 6.3v rated capacitor will barely be 2.5-3 uF with 2-3v on it. And those locations are for decoupling capacitors, something like 100nF, not >1uF. High capacitances are unlikely to matter that much there, just the mere existence of something there helped.

As for bulk capacitance on the other side, such boards don't get much benefit from polymer capacitors but if you do want to use some, 680-820uF 4v rated polymers would work just fine.