I'm realy curious to see how it look's - system like this, and it's ok for everyday usage? Some special power-up procedures etc? Just curious

 

How can I say this......it's effective but not pretty....it's fine for everyday use, I've only just in the last day upgraded to my 2700K, I've been testing the system out for several months with my old Q6600 @4Ghz before I put my expensive new hardware in there.

 

I'll post some pic's up later if you're interested, but it basically uses a sealed chamber and a stripped down air con unit.

 

The only special start up procedure is starting the ac unit before the pc and gradually pulsing the chamber air circulation fans on and off a few times until sub zero temps are reached takes no more than 1 minute or 2 to do.

This gives any water vapour in the contained air the chance to condense out on the cold ac unit radiator, if you cool too quickly the vapour can dump out on any surface....but really no problem at all.

 

Ps...thanks for the advice on safe vcore...most helpful.

Anything above 1.5v and there is going to be degradation of the chip. Pure and simple! Rather or not you're going to keep the chip long enough to notice is a different story altogether. I would venture to say that 1.45v is pushing it, but I my personal max is 1.36v with this current processor. From 4.5Ghz to 5.0GHz, I observed a maximum of 2FPS increase under heavy load with a saved flight at FSDT's KFLL with AI traffic and the NGX. The difference just wasn't worth the extra .05v needed to maintain 5.0GHz stably (IMO).

 

I wanted to add my .02. Carry on! ^_^

It can be maxen out , or you hit the Sandy cold bug it can be on differnt multi

For ex multiplier 53 not work 54 ok 55ok 56 not 57. Ok

Mine Did the same with chiller 5.5ghz no more Did not test it on the phase change.

Went for Ivy 3770 with have no cold bug runs at 5.4ghz 24/7 bencable at 5.6ghz

But booth at 5.7 can run super pi but not stable needs more cold Ln2 to get higher

How can I say this......it's effective but not pretty....it's fine for everyday use, I've only just in the last day upgraded to my 2700K, I've been testing the system out for several months with my old Q6600 @4Ghz before I put my expensive new hardware in there.

 

I'll post some pic's up later if you're interested, but it basically uses a sealed chamber and a stripped down air con unit.

 

The only special start up procedure is starting the ac unit before the pc and gradually pulsing the chamber air circulation fans on and off a few times until sub zero temps are reached takes no more than 1 minute or 2 to do.

This gives any water vapour in the contained air the chance to condense out on the cold ac unit radiator, if you cool too quickly the vapour can dump out on any surface....but really no problem at all.

 

Ps...thanks for the advice on safe vcore...most helpful.

 

Ok, so is not really complicated. How is fsx performance, for example 4.9GHz vs 5.3, 5.4? I am curious becouse i noticed visible improvements for each +100MHz.

For what i said about max safe vcore, here is some links, i find them very good and interesting:

http://www.overclock...-i7-2600k-2700k

http://www.overclock...es-and-voltages

 

Sandy Bridge guides, tips about BSOD 124, freezing, PLL, VCCIO&VCCSA: http://www.overclock...inc-spreadsheet

 

 

But you should talk with westman and TechguyMaxC here on forums, i didn't use anything more than air/water for cooling, and there is some differences about voltages

 

From 4.5Ghz to 5.0GHz, I observed a maximum of 2FPS increase under heavy load with a saved flight at FSDT's KFLL with AI traffic and the NGX.

In my experience, i noticed improvements in fps(and stutters elimination/reduction)with each +100MHz, as i stated above. Maybe you should try more than one scenario for testing to notice fps improvements. Just my opinion, if you are ok with 4.5GHz, keep it there, you can't have any problems at that vcore :smile:

It can be maxen out , or you hit the Sandy cold bug it can be on differnt multi For ex multiplier 53 not work 54 ok 55ok 56 not 57.

 

Thanks for that, I'd heard of the SandyB cold bug but wasn't sure what it meant...as I'm stable at 5.4Ghz with the InteBurn test I feel there is probably 200-300mhz more to be had with lighter loads like wprime.....l'll try a 57 multiplier.

 

As regards the return you get in fps on pushing the overclock......I can't really comment on fsx...just found this post via a google search on "2700k max vcore".

 

But in general terms, if your on 5.0GHz with a core i7, I doubt there is much in the way of games atm that would be limited by that in any realistic way, so if your system is delivering 80fps at 5.0GHz and 85fps at 5.5GHz .....both are well within playable limits......but I guess it's down to the individual....some are more sensitive to frame rate drops than others.

 

Anything above 1.5v and there is going to be degradation of the chip. Pure and simple! Rather or not you're going to keep the chip long enough to notice is a different story altogether. I would venture to say that 1.45v is pushing it, but I my personal max is 1.36v with this current processor.

 

Thanks for that, yeah degradation is a concern. One of my reasons for building the chill box was to get more out of the components so they would be powerful enough to last me longer between upgrades....hence the Q6600 at 4.0GHz..lol.

So I don't want to shorten the cpu's life span.

 

I don't intend to run it 24/7 at the max possible overclcock, I'll have several bios profiles set to use depending on how cpu demanding the game is.

 

I'm quite happy at 5.0GHz and 1.38vcore, that is stable with intel burn test so I could probably shave off some more voltage and still be stable for lighter loads. A good thing with active sub ambient cooling is you are able to achieve a given clock speed with less voltage, there is less leakage in the silicon at lower temperatures so it's more stable and requires less voltage.....I think 5.0GHz will be ample for most of my requirements......it's just the overclocker in me wants to see how far I can push it.

 

Ok, so is not really complicated.

 

True, but as with most things the devil is in the detail, my idea at the start of the project was to make a cooler that allowed you to use air cooling as it is simple and easily/cheaply upgradeable, and yet produce sub ambient, better than water cooling results.

 

There are some caveats though:

 

Noise....I'm running a 12000btu air con unit....it makes a noise but I wear headphones so it doesn't bother me.

 

Power consumption....it's a 2.5KW unit and while I don't think it draws that continually while under lighter loads, it does eat some juice.

 

Heatsink limitations......most modern cpu heatsinks use heat pipes.....they do not work.....they contain water as a working fluid/evaporant and that freezes and so cannot move the heat as intended.

For my cpu I use a solid copper Zalman 7500cu led.

Currently both my GTX 460's have solid heat sinks and so are fine, but most gpus now use heatpipe or vapour chamber (work on the same principle) designs....so I may have problems finding a suitable graphics card upgrade.

Oh, almost forgot,any fan must use ball bearings not fluid bearings as the fluid freezes.....like I said devils in the detail.

 

Size.....the chamber and ac unit together is about 4ft long and 2-3 foot wide.....you dont want this in your living room!!

 

Difficulty and expense....although simple on the face of it, it actually took a lot of planning and hard work. The chamber has to be completely sealed and air tight to prevent ingress of water vapour....no mean feat when you have to run cables from inside the chamber from the mobo to outside of the chamber to the psu, hard drives, optical drives, usb cables, lan cables, audio cables, monitor cables....etc etc...every individual cable has to sealed..even the psu bundles so that air cannot track into the chamber along their length or in between the cables.

Expense...it all adds up.....£400 for the ac unit and at least another £400 for materials and cable/extenders.

 

Having said that now its done its done and I have my very own fairly unique frankensteins monster of a cooling system.....like I said....it's not pretty but its effective......so here's somepictures.

 

Heres the ac unit all stripped down exposing the evap (radiator) to be built into the chamber.

 

 

Uploaded with ImageShack.us

 

Heres the chamber, outer acrylic box and inner insulation box, both independantly sealed. The evap/rad and fan stack by the side of it seal on their upper edge against the lid making two chambers, the fans push air into the back chamber through the evap/rad back into the front chamber where the computer is housed.

 

 

Uploaded with ImageShack.us

 

Heavy duty lid seals via two gaskets on the upper edge of both the two boxes and evap/fan stack. I used rachet/webbing ties to clamp the lid on tightly to make air tight....not pretty...but effective.

 

 

 

Uploaded with ImageShack.us

 

And with the mobo inside....exerything else has to be outside.

 

 

Uploaded with ImageShack.us

 

And finaly a very important part......seriously......connected by a 19mm tube to the inside of the chamber...it acts as an expansion/contraction chamber. When you chill the box down the air contracts making a suction, in my system it contractsby about 9 litres....that would suck in through any tiny hole or imperfection in the seal...nothing is 100%

and so take water vapor into the chamber....the rubber bag equalizes the pressure and still keeps water vapour out....as a belt and braces I also have large sachets of dessicant in the chamber.....I leave it sealed for weeks...infact for over a month with a slight positive pressurein the rubber sac toprevent atmospheric ingress.

 

 

Uploaded with ImageShack.us

 

This is still a work in progress and although I'm very happy with the results I'm always looking at ways to improve.

 

Using air cooling parts is nice and simple but its not as effective as other cooling methods, air cooled heat sinks often have a temp delta of around 50 Deg C at full load......I still have that same temp delta but its working off a -30c ambient in effect.

 

To reduce that delta I'm thinking of changing the chamber air for a gas that is more effective at cooling, I'm toying with the idea of using R134a refrigerant gas although it's only got 85% of the specific heat capacity of air it is 5X denser...so it may cool about 4X more effectively than air for a given cfm air/gas flow.

 

Also toying with the idea of putting a mixture of two gases in there, one with a boiling point that would allow it to liquify at chamber temps so I could pump that onto the components for phase change cooling....but thats all for the future.

1.36v with this current processor

 

What?! I've heard plenty of people run CPUs at 1.45+ volts for over a year without ANY degradation!

What?! I've heard plenty of people run CPUs at 1.45+ volts for over a year without ANY degradation!

 

That's fine and dandy! I need this processor to last longer than that!

 

Seriously, I can't back my statements up with hard evidence. I just really hate the idea of telling folks 1.45v is A-OK to pump through their hardware. That and that's a buncha voltage..

Ps....just one more caveat I forgot to put in to my project notes for anyone considering something similar.....safety.....don't forget the air con unit runs at full mains voltage....and there is at least one big capacitor in there that will hold charge long after the power is turned off.

 

As pc enthusiasts it's easy to forget as we are used to dealing with 12volt max.

Seriously, I can't back my statements up with hard evidence. I just really hate the idea of telling folks 1.45v is A-OK to pump through their hardware. That and that's a buncha voltage..

 

That's a bunch of Malarkey! :lol:

 

Intel Garuntees CPUs will run at MAX stress for AT LEAST 3 years at default settings etc... Now, if you use your computer for maybe 4-6 hours daily maybe only 10-20 minutes of that time was at MAX stress.

 

If you run an OC, you might cut that time down to maybe 1.5-2 years. That would still take decades of daily use to wear it down....

Degradation is a fact.....it is going to get worse as process sizes get smaller due to electromigration, Nvidia are concerned and restricting their AIB partners from excessive overvolting and they are at 28nm.....makes you wonder how much worse it is on the 22nm process used with IvyB?

 

The trouble is nobody really knows where the safe limit is......perhaps there is no such thing.....perhaps degradation in continual and just increases it's rate with increased voltage...so you make you choice.....higher clocks/voltage : shorter life ......lower clocks/voltage : longer life