Sounds like your PSU got damaged during the recapping. Test the card in another machine or try a different PSU.

The PSU was properly tested in another machine, scope checked the output and it was much better that anything you can buy (way under ATX ripple limits under load) It is good to be skeptic about mods, but the PSU powered friend rig for like 6 months and entierly stable, there was not one single crash or problem, so I would not question the PSU at all, based on measuring AND real life testing.
Sadly I did not have any other reasonably powerfull one ATM handy, but as you can see from my edit I made after you post the reply, the machine is up and running fine w/o any troubles, when I removed the eVGA GTX 660 card. With it it just does not post anymore...
 
Makes me wonder, if there is not some failure to speed up the fan, when things are getting way too hot...? One way or another, what now with the card that just turn the fan at 100% and does not post at all? Any way to "reset" some stuck temperature setting or RMA it?

The GPU fan can spin up to 100% like that if it's not receiving proper power.

No, I did not flashed anything modified to the card, I just take at as it goes and put it up in build. No mods, no flashes, no nothing. I did not even checked if there are any more up-to-date official bios for the card, nothing like that. The power the card give was well enought for me, even in slow WoT it manage around 41fps (all tanks on start on screen) in worst cases, usually around 60 to 75fps (over 100 near the edge of map) and WarThunder go over over 140fps many times. So no reason to tweak the card in any way, shape or form. After all, it is the "Superclocked" edition, witch is basically pre-overclocked, so there will be a little overclocking potential, as I see it.
 
(there is too many missing caps on the back side of the card anyway... There is even a spare place for a huge tantal-polymer SMD cap directly into the GPU core and the used cap is just 330uF one, where I bought for my recapping and overclocking experiements 470uF 2.5V Tantal-polymer Panasonic cap (2R5TPF470M6L) - AND there are 4 big places for similar caps into the power section card empty and there is another 17 (!) positions for smaller ceramic caps empty on the power section - so IMHO too much stolen caps (marketing owned design - eg. design is very good, but pricey, so marketing removed all caps it can to cut the price) for any serious overclocking anyway)
 
 
Sajin -

The GPU fan can spin up to 100% like that if it's not receiving proper power.

 
Good thing you say it "can" ... because I dubt that this is the case for there simple reasons:
 
1) that would mean that the PSU is damaged, yet the PSU is not only "not smoking", but also it is running cold even under load (Prime95 multicore ATM testing how much I need to push Vcore for some more overclock) - yes, the air that come out of the PSU is cold -> that is what good caps can do any can make a difference :)
 
2) if the GPU is not getting their 12V, then the CPU voltage regulator is not getting their 12V too, resulting in not working PC at all, because there is no substitute to the 8pin CPU 12V power... and there is only one rail on the Enermax 620W Liberty PSU (regardless what marketing is telling anyone, there IS usually only ONE rail for the 12V, just open the PSU and check for yourself, there are no more rails in 99.99% of mass sold PSUs, period)
 
3) if the GFX card is not getting it's juice, then it is likely that it will beep and complain that way, that the required power is just not there... dunno if this card have the beeper (I did not peek under the hood of the card), but that is something we can try - I will put the card back in and I will not connect the 6 point power cable on purpose and we see what happens. Sounds like a good test to you?
 
BTW, you could be right in a way... if the voltage regulators get damaged somehow by excess heat or whatever, then the card might really not getting the juice. Maybe a little suggestion from eVGA, where to measure and what voltages should be there, could be helpfull. The GPU voltage is damn obvious - measuring at the spare cap place. But for rams...?

Test w/o power cable = same behaviour. No change. No beep.
 
However much more interesting are the measurments I did. The PSU is giving out 12.12V at all times. On the back of the card, there are eight very thick wires, in the power section near the end. That IMHO translate into 4 phase PWM and these thick cooper wires are from the coils of the voltage regulator. So I measured on them and there are 0.09 to 0.1V, witch does not sound good at all.
Same voltage are on the GPU chip itself (the +pin is the marked one with tantal caps - they have the marking reversed by some weird historic way... at normal cap, where are mark, there are the cathode) - 0.1V.
 
That do not seems to be enought, not by a long shot. Not even 20nm or 14nm technology can give reliable working transistors at 0.1V, so obviously, that is a dead card :( Damn, damn, damn!
 
 
PS. as the RMA goes, I'm currently in Czech republic (middle of Europe) - what time I could expect to get a working card and will someone tell me what went wrong? Or is there anything I could do to make the card live again?

First contact EVGA indeed, but it also reminds me of this ..
 
I heard about strange methods of baking a video card. These are old I guess, and I don't know if they really work. WARNING !!! I'm not responsible for what might happen !!!
 
http://lifehacker.com/5823227/save-dying-video-cards-with-a-quick-bake-in-the-oven
 
Don't know if it's relevant to your problem in any way. But, I remembered this from long ago watching tech videos on YouTube.

Have u tried putting the card in the other pci-e slot?

Sajin - 12.12V measured on the end of the PCIE power cable w/o load, todays I measured on the card (the topmost there pins are +12V, the lower are ground, for safety - to protect shorting the measuring pins by accident - I used ground connection on the PSU in the another free 12V 6pin red rail the PSU offer and it is empty) when the cable is connected to the eVGA GFX 660 12.11V. Fan on full speed, no post, of course :(
So I would say that the PSU voltage is more that okay. Won't you agree?
 
BasSchilperoord - contacting eVGA? Okay, got it. What should I tell? After 4 days of happy usage now this...? Or perhaps redirect to this thread?
Baking videocard sounds insane to me. These caps usually are rated at 105°C, probably can take like 140°C, but not much more w/o any ill effects. Also baking the card does sure require striping it down, eg. voiding the warranty. And on top of that, to help a cold contact, it must melt the tin, witch goes over 200°C, so I dubt about the effectivity of such actions in the first place.
However the card is not acting like it suffer from cold connection. That means - the card work, until it heat up. In that case it might work. In this case the card is cold as death and it still does not work and that is because there is no voltage for the main chip passing thru the regulators. 0.1V is pathetic, my i5 require 1V (even it can overclock a lot with it, it still require it and will not work at 0.1V :) )...
 
So based on the measurments and observations - hey, the card worked so well for me for 4 days - there is no cold connection.

Trodas, your card is still under warranty, but it is a secondhand card, so you will need to go to http://eu.evga.com/support/guestregister.asp to request your RMA.

From your description of the card's behaviour, it sounds like the card is quite dead. Even though you only had it four days, and it was new to you, it is a secondhand card that has had a good run it seems. Your best option is to contact EU support for an RMA.

It's rather uncommon for a detailed explanation of a card's failure to be offered, since the fault can often be deduced by the troubleshooting process.

RMA card eVGA GTX 660 Superclocked recieved todays - as eVGA GTX 660 FTW - and it have much nicer cooler!



It is very nice fast RMA from eVGA and also the card is a bit faster, because the Superclocked version:
http://www.evga.com/Produ...aspx?pn=02G-P4-2662-KR
...had "only" 1046/1111MHz core clocks, while the FTW version:
http://www.evga.com/Produ...aspx?pn=02G-P4-3063-KR
...does feature a 1072/1137MHz core clocks. The more the better and I willing to bet that this card will be also MUCH more quiet in the run.



 
 
Card seems to work quite all right (and very quietly too!):
http://hwbot.org/submission/2589403_
...and while registering it, it offer me to buy extended warranty - 10 euro to 5 year, 16 euro to 7 years. Really I could get this warranty offer? I mean... I better ask about it first. I will pray that I never need eVGA RMA anymore, but still, to be safe, I would like to purchase that plan. Can I?