Recapping for increased stability

[noorman]

If they measured the spacing between the 'so-called' legs, being the terminals (or leads) , they would call that the 'pitch' of that 'leaded' component (part) ...



In my previous post I specifically named the diameter of the caps (10 mm); that was of course the outside case diameter of this cylindrical component ! (a diameter couldn't be the space between the leads, nor could it - in this case - be the diameter of those leads themselves)

[trodas]

Re: Recapping for increased stability

This thread contains a lot of funny errors, so, I took the oportunity to pick few of them.

Kako - I have two DFI LP B boards and never ever (except first power-on) experienced cold boot problem - since I updated the outdated bios (preferably use Hellfire rev. 3 one) and the mobo is working like charm from this time.
Ultra Infinity is not striped down version of Lanparty (LP), it is exactly the same, just w/o the gigabite network port and w/o all the cool black PCB and fancy green UV stuff That it is. People even using LP B bioses in Infinity - just remember to disable the gigabite network
As for you question - Samxon GA or Rubycon MCZ is the way to go. I recently recaped my LP B with Panasonic FM ones, using higher capacity and things looking pretty good. I let you know how far the OC go, once I get to it. Expecting over 2800Mhz and over 250Mhz for FSB on air, but we see.
With Samxons GA it could be far better, check out my thread for table of caps:
https://www.badcaps.net/forum/showth...?t=1345&page=2


Topcat - well, while I can't agree more with you about how ghetto it look and while I have significant doubts how the added caps helped things (the voltage regulation circuit sure benefit a little from increased capacitance, but there are limits when the circuit can't benefit anymore from increased capacitance and I believe Last Viking stepped into it.
I leave your comments about the capacitance alone - others already corrected you (caps in parallel - adding capacity, reducing ESR by 1/R = 1/R1 + /1R2 + 1/Rx and also adding ESL) but I took at the part where you warn about danger for the mobo when cap resistance is reduced to half. (the easy math for resistance decreasing works only in case, where both caps have the same ESR - then you can halve the ESR of single one to get the resulting ERS for the two in parallel) But back to the "danger". In fact, this is a benefit for the mobo and not a danger. Increasing ESR put the mobo in great danger as I experienced. (and 2 mobos and one GFX card pay for it with their life) Decreasing ERS is what all of us want for better OC and more stability in general. That is why Samxon GA caps has so outstandingly low ESR and I want them!
And yes, you are again right - crappy caps leave too much noise - but to get all the noise out, one need also bridge the big caps with small, 47uF ceramic ones. The big ones can't filter hi-frequency ripple, as one happy user with osciloscope and Infinity pointed out well already. Of course, Samxons GA or Rubycons MCZ do far better job that anything else, no question about it


Kako - go for a bit higgher capacity, if you aim overclock the thing to death 4700uF 6.3V and 2200 - 2500uF 16V are fine. Replace all five of the first and all there of the second. Don't forget the cap down there, like Lask Viking recommended. I put 2200uF one there too


kc8adu - you dubt the "ghetto rigged mess will work better than stock" Okay, I do so. However keep in mind that the guy hold WR in FSB on AXP - 297Mhz. So, it should help somehow, tough I also dubt it. I recently have a word with him and the mobo still kicking in. On 24/7 operation he use 284Mhz FSB, IIRC. Yes, it is a ghetto crap. But somehow... working well?
Or maybe the Chemi-con KZG ones are simply overrated and overboosted and did not meet their specs, witch is why the mobo kick then pretty good deal better with other caps and he won? Dunno.
I checking out Panasonic FM in this mobo now and seems they are good. I have power problems on AGP / PCI tough, so no aim to test how far the FSB go with them now yet. Before it was not a problem to get CPC ON, RAM : FSB 1:1 to 242Mhz with 2-2-2 timings


Kako - link for soldering iron did not work, however anyone do the job. I used normal transformator soldering iron and mobo working well Just keep the basic static stuff rules on and on/of the soldering iron farer away from mosfets and you will be fine.


Topcat -

As for butchering up that board, sorry... We'll disagree on that... That was bad... If you want to increase the capacitance, don't solder caps to the bottom of the board. DO IT RIGHT and remove the original cap and install a higher valued cap in its place.

I can't agree more!
There also are't any problem using higher voltage rated caps, yet there should be one problem. Not electrical, but mechanical. Dimensions. It sometimes could prevent installing a heatsink/or GFX card if you put there notably bigger cap. Other than that - not a problem at all.
But about the Vcore part of the Infinity mobo, I would put there (and I did) 4700uF caps. My mosfets and toroids run considerably cooler, yet this could be because there is active fan and the clock is ATM very low (2100MHz, more the damn oldie Bardon can't go, but at least with 1.850Vcore, witch is a lot...)


noorman -

electrolytic caps need to be at or near their working voltage to 'produce' the capacitance value that is 'nominal' in the data sheets.

Good point. Difference between 6.3V and 10V aren't big, tough. Yet on the Vcore stuff, replacing 6.3V ones with 16V ones could do a bad thing - eg. there are no benefit, as the properties aren't optimal.