Jetway V266B with GSC caps

[trodas]

My stepbrother has a refined taste to pick the worsts boards ever made, I ques. He has sort of "thing" for Jetway mainboards. Anyone familiar with GSC caps must know already how it end
Well, the board was first used with Duron 700 (slightly overclocked to 866Mhz - 133x6.5) and worked great for some time. Yet when my bro bought Palomino 1700+ AXP, the problems started. Board was not stable anymore, and re-flashing and bios updating won't cure a thing.

http://www.slibe.com/fullimage/bd1bf..._V266B_1_j.jpg

Funny thing - the Vcore voltage settings mentioned in manual as well, as available in bios disappeared after one flash update. We never get it back, even flashing the older bios back won't help, or so my bro says.

http://www.slibe.com/fullimage/a4952..._V266B_2_j.jpg

We all at BadCaps know now why and how, but at the time even I was clueless. We decided that "the board simply did not like AXP, or the Palomino" and he bought another board - Jetway N2PAP Ultra. What happen to this is another story. This V266B suxxka (yep, VIA KT266 chipset, so FSB133 is max) was sort-of borrowed to one friend and then it finally get to my sis. At the time the board won't like anything more that 800Mhz (133x6) and yep, Duron only. Yet still problems continue. For example if you copy anything over network + run a WinAMP = sure crash. Copy or WinAMP alone = all good.

http://www.slibe.com/fullimage/8375f..._V266B_3_j.jpg

Sooner or later it get to the point when we had to back from 133Mhz overclock and get to 100x7 default to make things at least not crashing often. There are some remarkable things like enabling fast writes in bios ended up with instant crash as soon, as the Radeon 9000, witch support fast writes flawlessly, start to draw any 3D on screen

http://www.slibe.com/fullimage/0a9d3..._V266B_4_j.jpg

Around at this time I noticed the caps close to the CPU, serving as Vcore caps, are pretty bulged. We cleared the leaking, but it still look kinda bad...

http://www.slibe.com/fullimage/6653a..._V266B_5_j.jpg

It was obvious what to do, but I looked at the board and started to like it. I mean - it have everything - even holes around CPU, so, why so crappy GSC caps have to ruin it?
The board support either SDRAM (2 slots) or DDR ram (2 slots) and we managed to increase stability (speed increase is not noticeable or measurable tough) replacing the SDRAMs with DDR rams. Probably the DDR ones utilize cap's that was not utilized before? Or we just dreamed up the increased stability in all the instability issues?

http://www.slibe.com/fullimage/23d95..._V266B_6_j.jpg

Around that time I got first optimistic results from recapping my DFI LP B boards with Panasonic FM caps, yet there was also a bit of argument on the China (okay, Hong Kong precisely) made caps Samxon's and how they are or are not good. And since space is valuable and Panasonic ones are big I contacted Joe, alias Big Pope and asked about the Samxon caps. Simply if I could test them out.

http://www.slibe.com/fullimage/2de2c..._V266B_7_j.jpg

[trodas]

The idea was to overclock this board high, and stress it with folding - on top of that, it will still serve my sis as gaming/working/browsing machine and sort of "general public computer" to access net at my home :

http://www.slibe.com/fullimage/5115e..._V266B_8_j.jpg

Joe agreed and unlike my bro suggested (replace just the bulging ones...), I replaced ALL the caps on the board. Everything over 100uF simply goes for Samxon's

http://www.slibe.com/fullimage/b9a93..._V266B_9_j.jpg

The results.

Forget semi-stable 100x7 with poor 64k L2 Duron! Now there is a Barton 2800+ inserted and run at 155x12.5 That give PR2500+, witch is nice performance kick! I tried even 166x12.5 to get the 2800 rating the CPU is made for and it was on Prime CPU test stable, yet the used rams can't handle 2-2-2-5 at more that a little over 155Mhz, so I decided that the few Mhz up it not worth decreasing the latency settings. Also fast writes is possible now, yet when I later replaced the R9000 with FX5600XT they has to be disabled, as the XT crap won't support them at all. Not a big deal, tough. All other bios options regarding speed could be now maxed out w/o worries and they work!
I also used Revoltec passive nice cooler on the NB and the standard NB cooler put on the SB to gain stability when overclocking the oldie poor VIA KT266 chipset to 155Mhz
It also have 2x512MB DDR ram's and at 2-2-2-5 I bet they stress the Vdimm powering circuit well!

So, now the mobo run fully on Samxon's, fully stable and under 100% load by folding all the time (run as service) :

http://www.slibe.com/fullimage/46cfe..._V266B_10_.jpg

It is about two months now, so I posted these result as sort of statement that the Samxons aren't about fail soon. We see how long the Samxon's hold up - my bet is, that as long, as we want the board to run, but to know for sure, a stress test is need. Unlike DFI recap work I did not increased any of the caps sizes.

So, stay tuned for longer test, yet so far I managed excellent results, I think. Especially compared to what the board was able run before, it won't even post with AXP CPU, even with 256k of L2 only... And the Samxons are still mostly GD and few GC ones. So, only the more average Samxons are being used.

No Samxons GA used, yet these should be (according to specs) mortal threat to Rubycons MBZ or Nichicons HZ by their cheap price yet still stellar specs - simply the absolute hi-end top.

Samxon GA 1800uF 16V - impedance 0.007 - ripple current 4140 (2000 hours) 10x25
Nichicon HZ 1800uF 16V - impedance 0.006.5 - ripple current 4140 (2000 hours) 10x25
Rubycon MCZ 1800uF 16V - impedance 0.009 - ripple current 3230 (2000 hours) 10x25

Dunno if they could make difference in such oldie mainboard, but in new ones and in case of extreme overclocking - well, that could be different story! :-P

And for the record, the case is fan less, so they get proper thermal stressing as well. Pictures on demand possible ;-)

Samxon caps are available from Big Pope.

[Gizmo]

Nice post Trodas! Thanks for the info!

[Favu]

Trodas seems to be part of a dieing breed of Hands-on overclockers : looking back at old sites and articles people where happy to solder and desolder things, to build their own watercooling settups, heatsinks and such.

Presently it seems like people would rather not get hands-on at all and prefer to be able to buy heatsinks and waterblocks. Very very few would ever go so far as to consider re-capping a motherboard.

to you, Trodas!

[Gizmo]

Err......psst! Hey, Favu! Look here

[Favu]

I've seen it

You're also one of the few, what with your direct-die waterblock

but that doesn't change the fact that most people don't feel safe being hands-on with hardware anymore

And that article is almost a year old, and I haven't heared much talk of hardware voltmods or mobo re-capping or such since then.

[Áedán]

Mostly because it's more work than people are willing to do. If I look at how long it would take and the cost of the capacitors and the value of my time, it is usually cheaper just to buy a new motherboard!

[Gizmo]

Originally Posted by Áedán

Mostly because it's more work than people are willing to do. If I look at how long it would take and the cost of the capacitors and the value of my time, it is usually cheaper just to buy a new motherboard!

Bingo. For most people, by the time recapping becomes an issue, they are probably thinking about upgrading the hardware anyway.

However, Favu also has a point. A lot of people are just afraid of mucking about with their hardware too much.

[trodas]

Thanks, guys Yep, I'm a little old-fashioned guy and I'm proud of that. It is not any secret why the highest O/C was done by old-fashioned japansee modder - for the same reason why Fuger won't manage it - he did not mod the HW enought

I just wanted to show that it is possible and with good caps it can give great results