That's awesome jwt27! Really impressive work.

Thanks! 😀

As you may have noticed, I haven't replaced the caps in the corner yet. These are on the 12V rail, and I'm pretty sure 6.3V caps would go right up in smoke here. For the big purple caps I don't have suitable replacements on hand, but then these are polymers already so replacement may not be necessary.

jwt27 wrote:

So now, this is what I have so far. Yeah it's a mess. Still don't have a suitable case for it. […]
Show full quote

So now, this is what I have so far. Yeah it's a mess. Still don't have a suitable case for it.

Now that's what I call a workbench! 😁 If only I had that much room to play with.......

(And I'm pretty sure I have a 3-button mouse like yours stashed somewhere ^^).

Found a stand for the 5151... Yay more space, and my neck no longer hurts!

Yeah I should start looking for a proper case...

may be i missed something, but what is reason to recap v5? i have many vsa100 cards and never had any issues with caps. think much better mod for v5 is to replace chip to 320 and memory to 4ns. cutting fan connector is bad idea. you can easily remove plastic case up from pins and then desolder pins if it necessary.

here is my variant how to cool quietly

anthony wrote:

may be i missed something, but what is reason to recap v5? i have many vsa100 cards and never had any issues with caps. think much better mod for v5 is to replace chip to 320 and memory to 4ns. cutting fan connector is bad idea. you can easily remove plastic case up from pins and then desolder pins if it necessary.

No specific reason, I had the caps laying around and just wanted to make sure it'll last forever. Replacing the GPUs sounds like a nice idea but I'd have to upgrade my soldering gear first.

Some cool heatsinks you got there! Where did you find those?

http://www.ebay.com/itm/Aluminium-Adjustable- … =item1c3139cd5a

Copper one exists too

jwt27 wrote:

http://i.imgur.com/4x2m4EE.jpg […]
Show full quote

These are all 6.3V, 1000µF Nichicon FPCAP ("Functional" Polymer, whatever that means). Looks funny, doesn't it? 😀
Yep, that's less than the original 1500µF, but I'm hoping the massive reduction in ESR will make up for the loss in capacitance. These supposedly have an ESR of only 0.007 Ohm! 😳 ...

From a distance it looks like there are 15 Vcc caps in two rows of 12+3 . Thus with the original 1500uF lythic caps the total Vcc capacitance was 15x 1500 = 22500uF.

22500uF is a HUMONGOUS, WAY BEYOND OVERKILL amount of Vcc capacitance for the relatively meager demands of a ~35W tops Pentium III.

For comparison, a P4P800 came OEM with just 6x 1500uF (=9000uF), and that's empirically enough to handle a P4 Preshott OCed above 4GHz (well beyond 100W TDP).

What were the Abit engineers after when they chose to deploy a full regiment of caps for Vcc?

When you place 15 caps in parallel, the total ESR is = (ESR of one cap / number of caps). So total ESR is fifteen times lower. And this low total ESR, which is the alpha priority for Vcc caps, was their objective. The humongous capacitance is just an unnecessary by-product.

So how much total capacitance does a PIII mobo really need for Vcc caps? My Tekram P6Bx 440BX came OEM with just 5x 1500uF (=7500uF total), and that's empirically enough to handle a Tualatin 1400S OCed to 1.6GHz. As a matter of fact 7500uF is most probably still overkill for the real needs of any PIII.

So no sweat, with 15x 1000uF polys for 15000uF total you're still well inside hugely overkill territory. 😀

We both reached Rome by different roads: You reduced from 22500 to 15000 uF, but due to the much lower ESR of your polys (7 mΩ / 15 = 0.46 mΩ total) your mobo came out much better.

I augmented from 7500 to 16500 (5x 3300uF), and due to the lower ESR of the larger Nichicon HM (the larger the lythic cap's can, the lower the ESR) mine also came out better than before (12 mΩ / 5 = 2.4 mΩ total), though not as good as yours.

Those Nichicon caps are TOP NOTCH!!! Great recap job! I would have ordered them for my Asus TX-97XE but these were not available either in Farnell or Mouser UK.

I'm jealous of all these good soldering jobs! to me this is more impressive then over the top overclocking

Eventually, you'll need to solder. Eventually, you can't stop throwing away boards and getting new ones, you'll need to start fixing them when the supply dries up.

The vintage Mac scene, amongst others, has had to solder for a while. The x86 PC side, other than very rare or early machines, (or the plague) almost never does - and they'll need to do soon.

TELVM wrote:

http://s17.postimg.org/bz1begprz/Necro.gif […]
Show full quote

Oh please, it's only been three months, let's save that for the guy who bumped a five-year-old thread last week. 😉

Any updates on this build?

To further illustrate on the real capacitance needed by CPU VRMs, pic of a ~2009 vintage ASUS M4A89GTD Pro mobo (the VRM heatsink has been removed for clarity):

Just 810uF total for the VRM high caps, and just 6560uF total for the Vcc caps.

And yet this mobo has been running a six-core Phenom II OCed to 4.1GHz (= TDP above 150 watts, CPU-Z shows '161W') for five years like a champ.

This diminutive mini-ATX ASUS AM1I-A mobo comes OEM with just ONE SINGLE 270uF for VRM high, and 5x 820uF = 4100uF total for VCC:

And yet it manages to handle an AM1 Athlon 5350 OCed to 2.5GHz, where its TDP is similar to the higher-end Pentium IIIs of yesterday. 😉