Swiftech mcw6500 w/ pics and benches

[seedomo]

After a week or so I testing I've finally gotten stable at 3.8 using VCore set in bios to 1.6875 (after VDroop = 1.63) I'm Orthos stable as I write this for 1 hour. My Core temps started at -20c at peeked at 40c. Not to bad, I think with all those volts. I'm sure this chip needs these volts since I crash in Orthos at any lower setting. SHOULD I PUSH HIGHER? I would probably need to set Vcore at 1.75 since it would drop to around 1.68 to get 3.85 stable.
http://www.domophoto.com/3802 16875.jpg
http://www.domophoto.com/_A8L7107.jpg
http://www.domophoto.com/_A8L7092.jpg
http://www.domophoto.com/_A8L6850.jpg
http://www.domophoto.com/_A8L6854.jpg
http://www.domophoto.com/_A8L6859.jpg

[eva2000]

wow nice.. that's alot of hdds there :D

[TopherTony]

Yeah seriously, I dunno about the TEC, if there si one, I am captivated by the TEN HDD'S!!!!!

Sorry that was PC pron!!! FTW!

[seedomo]

SHOULD I PUSH HIGHER? I would probably need to set Vcore at 1.75 since it would drop to around 1.68 to get 3.85 stable.

[Pete]

Why dont you do a v droop mod to slove the issue of poor over clocking?

Set ram to the slackest timings going, set your v core your happy with, up the fsb, then find the stable point there, im sure you can hit 4ghz, then back off the vcore till it fails too boot/poost then add a little. Then stress that, then tune your ram tinmings

[seedomo]

Why dont you do a v droop mod to slove the issue of poor over clocking?
I have the equipment to do the mod but don't know how to do it on this motherboard EVGA 680i.
Set ram to the slackest timings going, set your v core your happy with, up the fsb, then find the stable point there, im sure you can hit 4ghz
Since I'm running my memory unlinked to the FSB, if I understand this stuff correctly, this memory shouldn't be the problem at this point. Aren't I eliminating the memory from being the problem. Can't it only be the chip at this point?

[Solarfall]

wow nice temps, put your multiplier to 8 and test if can you clock that chip any higher. you never know, it just might work.

[seedomo]

Just changed multiplier to 8x and raised FSB to where it added up to 3.8Ghz in the bios... Then I get these strange conflicting screens, attached. Some say I'm at 4.2Ghz and another says I'm at 3.8Ghz
http://www.domophoto.com/4277.jpg

[Duh]

looks very very nice. swapping the 226 for the 437 watter would slap some arses

[Solarfall]

Just changed multiplier to 8x and raised FSB to where it added up to 3.8Ghz in the bios... Then I get these strange conflicting screens, attached. Some say I'm at 4.2Ghz and another says I'm at 3.8Ghz

i experienced that same thing with some settings (cant remember what they were). so have you tested did that multi changing allowed you to gain higher clock speeds? or are you still stuck with 3.8ghz

[Orangeblast]

Yes, How do you do the Evga 680i Vdrop mod?

I'd be interested as I have the same exact setup going myself, I'm lucky to hit 3.7ghz...and thats at 1.6250 before Vdrop. If I can fix that vdrop stuff then i may have a better chance at 3.8!

[seedomo]

did that multi changing allowed you to gain higher clock speeds? No, I was able to lower the voltage a hair, literally just one increment. After testing it was clear that there was no significant difference in performance. As I write this I'm at 4.0GHz using VCore 1.7375 at idle and my temps are -20c. Should I see if its stable by running Orthos? Those could me my famous last words...

[erwinz]

very nice setup you got there.. :)

[Holst]

Very nice indeed.
neat and tidy installations are pretty difficult to acheive with a TEC setup.
I could never get my TEC kit to fit into a case.

What voltage are you running the TEC at?
What TEC PSU?

I hope that you can get 4gig+ stable in the end.
I cant give any real advice on intel settings, but goog luck.
What are you water temperatures (load and unloaded)?

[seedomo]

Very nice indeed.
neat and tidy installations are pretty difficult to acheive with a TEC setup.
I could never get my TEC kit to fit into a case.

What voltage are you running the TEC at?
What TEC PSU?

I hope that you can get 4gig+ stable in the end.
I cant give any real advice on intel settings, but goog luck.
What are you water temperatures (load and unloaded)?

I'm running my tec with the meanwell at 13.8Volts, max. I'm using the 226w Swiftech mcw6500-t. Water temps under are about 30c. Can;'t remember if that was under load or not.

[Holst]

As you can see there is a huge difference between your loaded and unloaded temps (a massive 58*c)

If you can measure your water temps at idle (after say 30 min of using the net) then at load (after gaming when overclocked for 30 min) it will help to understand where your system can be improoved.
Let us know ambient as well.
If you dont have a "propper" thermomitor you can use a medical one if you tie string to it and drop it into your resevoire. You dont need super accuracy for this test, just a basic idea is ok.


I found that dropping water temps 5*C at load gave almost an identical drop in CPU temp.. so the colder the water the better.

If your water temps are ok then the rest of the system needs optimising if you want lower temps.

TBH I dont knwo much about the C2D's power usage but im guessing that at 4ghz its going to be a big number. Can anybody give an approximate idea of what wattage you will be putting out with a C2D @4ghz at 1.75v
(or just tell me stock wattage and ill work it out)

[seedomo]

32c water temps under heavy load at 1.6875 volts @ 3.8

[theelectic]

32c water temps under heavy load at 1.6875 volts @ 3.8

There's your problem. If your water temp is 32C and your CPU load temps is 42C, the difference between your water and CPU is only 16C. You're overshooting your TEC and it can't keep up. The difference between the hot and cold side should be a lot more than 16C Get rid of it and use a straight water setup or a phase setup.

[trans am]

Just changed multiplier to 8x and raised FSB to where it added up to 3.8Ghz in the bios... Then I get these strange conflicting screens, attached. Some say I'm at 4.2Ghz and another says I'm at 3.8Ghz
http://www.domophoto.com/4277.jpg


You're at 3.8ghz. the same thing happens to me when I run 7x multi on a e6400.

[seedomo]

:fact: :toast: :banana:
We will see what it does with some better cooling and more volts. I'm pretty sure I'll have better results now!
The code is L627B115 :clap:
Here is a screen he sent me from HIS setup using a regular fan for cooling.
http://img443.imageshack.us/img443/7721/l627b36lfftnp8.jpg

[Holst]

Is that CPU temp correct?

I think you will probably do better without the TEC on your CPU.

I think your CPU is much more powerfull than the TEC can cope with.

A bigger TEC will almost definatley do better but I think a 226 is too small for your CPU.

[seedomo]

Are those temps correct?

Those screens aren't mine. :slapass:
They are from the guy I'm buying the chip from, lol. He's using a Fan. My temps are more like -20c at idle and 40c under load at 1.685 volts at 3.8Ghz. Since this chip needs so little voltage 1.34 for 3.6Ghz I will probably be able to run it at 1.65, maintain good temps and hit much higher speeds. Imagine 1.34 for 3.6Ghz, seriously.
It might go like this:
3.7 @1.45
3.8 @1.50
3.9 @1.57
4.0 @1.68
I'd love to play with Phase and might just get one since I'm having so much fun learning to overclock! Kind of like building a hotrod, just a lot easier.

[Holst]

Oh im sorry. I didnt understand you post.
I thought you had just received the new chip.. I thought you might be testing yourself on air.

Hopefully you will do much better on this new chip as your temps will be much better if you can keep the voltage down.

Let us know how you get on when you receive it.

However... I dont think the TEC is giving you any great gains over straight watercooling at load..
But changing to a bigger TEC isnt easy at all :(

[serialk11r]

That MCW6500, if the TEC isn't doing well, you can always take the TEC out, the 6500 should be a great block.

[Siron]

What is the red stuff all over your board?

[Mankz_91]

Nail varnish or polish i guess.

[Solarfall]

Nail varnish or polish i guess.


yep it has to be something like. what i don't understand is that why does it has to so much of it, then again it's allways better to be safe that sorry :rolleyes:

[theelectic]

What is the red stuff all over your board?

Red conformal coating, like this:

http://www.crazypc.com/products/9349.html

[oshox]

I want your case! What is it?

[TopherTony]

i dunno exactly what it is but it is to keep condensation from ruining the board. Maybe non conductive paint or such.

[Holst]

Looks like red nail varnish to me.
Not sure why there is so much, it looks like the sort of thing you would do for LN2 benching or dry ice (where you get condensation everywhere)

[seedomo]

This was my first MB with below ambient temps so I kinda went extreme with the conformal coating. It comes in handy at unexpected times... like when you remove the pelt assembly to change the chip. Since I kept the tubing attached, I had to hover the tec above the motherboard and the condensation on the cold plate dripped onto the motherboard. I wasn't expecting that, just caught me off guard. Glad to have overdone it!

[seedomo]

Very interesting results as I try to get to 4GHz with my new Week 29 chip. Since it only takes 1.35 volts to get 3.6Mhz, you would think that this would be a great overclocker... you'd be wrong! I'm confused by the results.
Cooling = Swiftec MCW6500 TEC 226 watt TEC
MB = EVGA 680i
Pelt PSU 300watt Meanwell at 13.8 Volts

Strange results in temps:
Week 44 chip @ 3.7 GHz:
Needs 1.6 Volts for stablity. temps 8-35 (idle/load)Celcius Using Orthos Small FFTs
Week 29 chip @ 3.7 GHz
Needs 1.45 Volts for stablity. temps 9-41 (idle/load)Celcius Using Orthos Small FFTs

Sooooo... why are the temps the same? If the needed voltage is less on the Week 29 chip, shouldn't the temps be lower as well? Can the amperage be different on the two chips? Amps*Voltage=Watts

One more thing is that the Week 29 won't even post any higher than 3.75 MHz even at 1.685 VCore but will run stable at 3.50 using only 1.5 volts!

[theelectic]

Sooooo... why are the temps the same? If the needed voltage is less on the Week 29 chip, shouldn't the temps be lower as well? Can the amperage be different on the two chips? Amps*Voltage=Watts

One more thing is that the Week 29 won't even post any higher than 3.75 MHz even at 1.685 VCore but will run stable at 3.50 using only 1.5 volts!

You're looking at it the wrong way. There is usually very little correlation between two chips, even if they came off the same wafer right beside each other. Adding more voltage does not scale linearly, and the fact that a chip can run faster at a lower voltage just makes it cooler, it doesn't mean it will reach higher speeds. If anything they are often WORSE overclockers as you're finding now - just like the EE AMD chips, which can run at a lower voltage than standard chips, but top out at a much lower speed.

Also going by your temps YOU'RE OVERSHOOTING YOUR TEC! SWITCH TO A STRAIGHT WATER SETUP!

[seedomo]

One more thing is that the Week 29 won't even post any higher than 3.75 MHz even at 1.685 VCore but will run stable at 3.50 using only 1.5 volts!
Typo: I meant to say that it will run stable at 3.75 with only 1.5 Volts but will NOT even post at 3.80 with 1.685 Volts

Thanks for your responses, I'm learning very quickly thanks to everyone here.

[serialk11r]

I think its time for you to get a new TEC ;) Those 320W (460something power consumption) TECs on ebay look...juicy ;)

[davekusa]

Some chips also have a wall. No matter how much voltage you thow at it.

[Holst]

You have to remember that its not only the chip that matters.

PSU will often have an effect as you increase the wattage drawn.

Memory can also be a factor.

IMO this is one reason that some people get consistently better overclocks than others.. they try multiple boards, ram and PSU untill they get a combination that works.

I think your TEC is definatley adding an extra layer of complexity.

as your temps are higher with the new chip there are a few possibilitys.

Chips temperature sensors are not calibrated.. hence the difference.
Bad mounting of TEC block.. Have you checked the mount and remounted?
This chip just pulls more amps at the same voltage (I think the difference is too much for this to be true, but im not a CPU expert)

How much effort is it for you to change to straight watercooling.. got a spare CPU block?

[theteamaqua]

so hows the insulation?? i mean do i need to do anything to insulate it?? or does the kit has all u need? to insulate

and yeah for that 4.2Ghz vs 3.8Ghz .. the 4.2GHz is prolly ur speed if it were 9x multi, which is not. the 3.8GHz is real

[Vinas]

Windows messes up your clock with e6600 at 8x multi for some unknown reason... Trust your bios or your mutiplier.

You might try upgrading your PSU. I'm not saying this is the problem, but it quite possibly could be holding you back! Before I upgraded I was stuck at 3.5GHz using the same setup in my sig, except for the PSU which was a (somewhat) crappy bfg 650watt. After upgrading to the PCP&P 750 silencer I was able to achieve 3816MHz.

Vdroop is a problem for me too. At 3800MHz I had to run at 1.700v only to turn it up even further to 1.725 to get 3816MHz!!! vdrop puts the vcore down to about 1.625... A vmod is in order here, but like you I have NO CLUE how to do it on my mainboard.

I'm also limited by my high air temps using the Scythe Infinity -- dual fan mod + pressure mod. I think the key is plenty of clean power + lower stable voltage. I think vdroop could be killing both of our max clock.

My sig says 3.9GHz, it's not Orthos stable! Had to go back down to 3.8GHz (477*8) until we can figure out how to go further.

Very interested to see what you're able to find out here.

[Roger_D25]

seedomo,
I just ended up reading through this whole thread and was very interested in it as I've considered TEC setups in the past but don't know a whole lot about them. With that said I think you have done a great job with it, especially since this was your first go around with a TEC setup! Obviously it can be a bit frustrating when results arn't fully understood or expected ( like the results you had between your old chip and the new chip you got). Although it stinks since you just bought this chip, theelectric was right about the processors. Unfortuenlty voltage needed for certain clockspeeds isn't a good indicator for max overclocks (I've also made that mistake in the past)!

Anyway there have been alot of good suggestions given as to what might be holding you back. I don't want to sound like a broken record but you might want to try without the TEC if only for testing purposes. This will help you to get a better understanding on how your TEC is effecting and is being effected by the CPU voltage/load. Another important aspect that a few people have brought up is your PSU. Although your PSU might be more than enough for the components your running, once you really start to push things like your CPU (especialy with the high voltages your using) the PSU starts to become a much more important peice to the puzzle. Plus you have alot of extra hardware on this rig (namely 12x high-end 74GB Raptors, a decent graphics card, and a power hungry CPU!

Since it is somewhat easy to test, I would really consider testing out another power supply if you can (either borrow one from a friend or just buy another one). You can always benefit from a better supply and the more you push your system the more important it becomes. I have read through threads were people have seen big gains by upgrading their powersupplies and I really think you'd also benefit highly! By the way does your Antec 550w have the 8-pin plug for the ATX_12v connector or just the 1 4-pin? If your supply only has the 1 4-pin this could quite easily be seriously effecting your CPU stability as that is exaclty what that connector does.

Anyway this is just something to consider as I'd like to see you get this sorted. If you end up selling that Swiftech TEC block of yours let me know as I'd happily take it off your hands (along with the Meanwell supply)!

EDIT: Anytime you change the default multipler your monitoring software might get messed up. I find that certain proggys can deal with it and some can't.

[seedomo]

Wow, thanks for the thoughtfulness of your message. Its very helpful. I will be getting a regular water block and additional PSU based no your suggestions. I'm also getting another motherboard since I didn't use enough grease in the socket and water got under the socket and shorted it! I must have changed the chip ten times without a problem and then it happened. I've dried it out with a floor heater in the bathroom overnight, and it didn't help. Just a quick question, should you put grease directly on the 775 socket pins before putting the chip in? Just want to make sure the grease won't prevent the pins from forming a connection, ya know.

[Roger_D25]

Take this answer with a grain of salt because I'm not sure. With that said I would think it's certainly not a problem putting greese onto the actual CPU BSEL pads (the little contact points on the backside of CPU are called BSEL pads) because the pressure used to hold the CPU into socket is substantial, enough to allow a perfect connection despite having greese on the CPU. By the way thanks for the kind words, happy to help out! For me the fun part is learning about all this stuff and testing it! I give you credit because I havn't gained the confidence yet to try TEC cooling myself! On a separate note what part of the city do you live in? I occationally stay with my brother and his wife down in Chelsea (no laughing), lol!

[serialk11r]

Well you can take the TEC and coldplate off the block and use the top part as a normal waterblock. I can imagine it would perform well...saves some money as well.

[Vinas]

Go to autozone like I've done in the past and ask for dieletric grease. It's cheaper than ordering online and the same as this:
http://www.frozencpu.com/products/2245/ex-swi-11/Luberex_Dielectric_Grease.html?tl=g30c105&id=9LdgfLtV

Apply liberally to the CPU socket trying to eliminate as much air pockets as possible. Clam the cpu in the socket, notice the excess comes out. Either take the cpu back out and reapply, or just wipe the rest up. I like to take the CPU out and reapply just to make sure there's no air pocktes.

I use the above method on my test rig cooling system. -9C water.

[Roger_D25]

Also going by your temps YOU'RE OVERSHOOTING YOUR TEC! SWITCH TO A STRAIGHT WATER SETUP!

theelectic,
When you posted the above to seedomo, what exaclty did you mean? I totally agree with what your saying, I just want to understand the concept behind it? Thanks for taking the time to reply!

[serialk11r]

theelectic,
When you posted the above to seedomo, what exaclty did you mean? I totally agree with what your saying, I just want to understand the concept behind it? Thanks for taking the time to reply!

What he meant was, the heat the processor was putting out was EXCEEDING the TEC's capacity, which causes the HUGE temperature increase (TEC cannot pump any more heat, acts as insulator. Heat has no place to go, processor heats up). I think seedomo needs a 320W (460W power) TEC undervolted to 85% volts or so. Better, a 437W undervolted to 75% volts (about 50% power).

[Roger_D25]

Bingo, that is what I thought so thanks very much for clearing it up for me!

[theelectic]

theelectic,
When you posted the above to seedomo, what exaclty did you mean? I totally agree with what your saying, I just want to understand the concept behind it? Thanks for taking the time to reply!

Everything is explained far better than I can do in the TEC FAQ sticky:

http://xtremesystems.org/forums/showthread.php?t=38367

*) A 40watt TEC cooling a modern processor generating 60-watts of heat.
Delta T = -34C (negative 34C). That’s an INCREASE in temperature on the cold side, so the processor would have been warmer than before! Using an under powered TEC can destroy the processor. Cooking your CPU with your TEC is NOT a good thing, avoid this common mistake.

Swiftech states on the MCW6500-T (http://www.swiftnets.com/products/mcw6500-T1.asp) page (in the lower colored graph) that at a 150W heat load with a MCW6500-T and 13.8V PSU, the IHS temp should be around 15C over the ambient temp, same as straight water. Now look what happens when the heat load passes that - at 175W the IHS temp is around 30C over ambient, compared to straight water (yellow line) where it's around 20C. With that high of a heat load, the straight water setup is better than the TEC setup.

Why is this the case? The 226W TEC is not 100% efficient, meaning it cannot move 226W of heat from cold side to hot side and maintain its rated temperature difference. Most likely it's around 50% efficient at best.

In one of seedomo's many previous threads, he stated his CPU on-die temp was something like 48C and his coolant temp was 32C (sorry if the numbers are off, I don't remember them exactly and I don't want to slog through a bunch of threads to look it up). That would mean two things: if coolant is 32C then the hot side of the TEC has to be higher than that. If CPU on-die is 48C, then the IHS surface (cold side of TEC) has to be slightly colder than that - remember heat moves hot to cold. Do the math: if the IHS is supposed to be below 48C, and the hot side above 32C, and the TEC is rated for 67 temperature difference, there's something wrong since the temp difference is only 16C, when the TEC is rated for 67C delta T. If the TEC is working properly in its range, then the temp difference should be more than 16C.

[Holst]

Just to clear up on the dielectric grease.

As long as its propper di-electric stuff then its safe to put it inside the CPU socket and all over the CPU pins.
You cant really use too much, you just make a mess if you do.

Sorry to hear that the motherboard is dead :(
But i hope the new one will be a better overclocker.

[seedomo]

Installed the new MB and packed the socket with grease this time!:clap:
Also pick up a Maze block to get rid of the NB fan and also added a VGA block for the 6600 cause the fan was loud as hell. And the circle begins... my temps are now 12c higher so I just bought a thermochill PA120.3 to run with my Swiftech 220 and am wondering if I need to run them on separate loops or use my existing pump to run them both. Currently I use the 665. The price just keeps adding up. I guess I'll have plenty of rad cooling but where to mount it all, I don't know, best to probably sell the 220 and get a thin 120 for the VGA and NB and run them separately. :mad: More cash!

Southbridge heat... I thought I could just cut off the heat pipe and leave the SB heatsink on, but temps were 60 to 70c, with a fan on it! So I took off the NB heatsink fan off an A8n-sli board and it fit barely. I've ordered the EVGA SB fan (I hate fans damnit) but what else am I gonna do. Seems like I'm chasing my tail here... buy two blocks to get rid of the fans, and because of that I'm gonna have to add a fan. :mad:

I'm running right now at 3.6 at 1.5 volts without issue. CPU temp is 0 at idle but quickly goes beyond 50 on load. I've also measured the load for the PSU which is a Antec True Power 550. It is 270 idle and 375 under load.

[theelectic]

I'm running right now at 3.6 at 1.5 volts without issue. CPU temp is 0 at idle but quickly goes beyond 50 on load. I've also measured the load for the PSU which is a Antec True Power 550. It is 270 idle and 375 under load.

While I'm sure lubing up, playing with grease and all that tubing is fun, if you actually want to get better temps GET RID OF YOUR TEC! You want lower temps yet you add even more restriction and heat with more blocks and another rad? You're making the problem WORSE, not better. You do understand what you're doing here, right? It's like you want your car to go faster and you keep putting premium gas in and then stick a banana in the tailpipe.

[seedomo]

I knew I would see an increase in temps, just how much, I didn't know. Now we all know.

[theteamaqua]

thanx for the lesson ... i was gonna get swiftech TEC block ... but then i realized that it might not be enough .. in which case startight water will do just fine

PS: if u have 2 PSUs.use the best one for ur comp, and the 2nd one for TEC... just use a u shape wire to short the 24pin connector. thats what i do ... i have my pump + fans on my Thermaltake 750W, and rest of my Thermaltake 850W (heh 750 cost me $150 last June i believe, and 850W cost me $189.99 @ frys)

[Duh]

TT ain t not good. for that prices I would have grabbed two pcpower ( 510 w turbo cool ). Not trying to be mean but don t use as a source for an argument that guy johnny bravo..err johnny guru :p

[theteamaqua]

huh? two pcpower only 510w?? dont know how much current it has over 12v rails , but 510w is just too small

also not sure what ur trying to say on that 2nd sentence

nvm just checked .. only 38A over 12v rails .. never see that PSU here.. frys do have the PC Power & Cooling 1000W though for a wooping $530

[Roger_D25]

seedomo,
I have been offline the last couple days rebuilding my case and just got back a few minutes ago. You have been a busy guy since I last posted, lol! New motherboard, new chipset blocks, and a new radiator! And I thought I was busy just building up a new case, lol! Anyway it sounds like your results were a bit dissapointing? Sorry to hear that but as you probably know now theelectic is right!

Fortuenlty you have a few different routes to try out, you could just switch to straight watercooling to test out. You could also buy a more powerful TEC until (although it is more money they are quite cheap thankfully), having the extra headroom with a stronger TEC unit might help out because its' quite clean your now overpowering your current TEC plate. Although I've never run a TEC until it should be quite easy to replace the current plate with a more powerful until right? I am curious to see how this works out for you, since you have already spent so much money and spent alot of time on this I would probably opt to trying a more powerful TEC, but only after testing with straight water first, just to get a baseline (useful for future testing)!

It sounds like you spent some time trying to figure out if your current PSU was capable of powering your system and it sounds like you should be alright (by the way what kind of amps are your rails pusing on that Antec supply, do you know)? With that said my only other suggestion would be to scrap your old radiator and stick with the PA120.3, I believe it is certainly capable of removing enough heat for your system. Plus the added benefit of additional cooling capacity might be overshadowed by the additional restriction to flow (probably have to test to make sure but from expereince I think you'd be better off with a single rad)? Well good luck m8 and keep us posted, keep up the good work!

[serialk11r]

thanx for the lesson ... i was gonna get swiftech TEC block ... but then i realized that it might not be enough .. in which case startight water will do just fine

PS: if u have 2 PSUs.use the best one for ur comp, and the 2nd one for TEC... just use a u shape wire to short the 24pin connector. thats what i do ... i have my pump + fans on my Thermaltake 750W, and rest of my Thermaltake 850W (heh 750 cost me $150 last June i believe, and 850W cost me $189.99 @ frys)

Actually, you will need to take the PSU apart and swap out the wires to lower gauge wires (usually PSU wires aren't rated for the large currents TECs draw).
Also, you want some load on the 5V and 3.3V rails so the 12V rail is stable. This is exactly why you want a meanwell or something like that.

[Duh]

huh? two pcpower only 510w?? dont know how much current it has over 12v rails , but 510w is just too small

also not sure what ur trying to say on that 2nd sentence

nvm just checked .. only 38A over 12v rails .. never see that PSU here.. frys do have the PC Power & Cooling 1000W though for a wooping $530

nope, I meant to grab 2x 510w pcpower ( 650w peak) with 38 amps in the 12v line...or you could grab the one you mentioned, I am pretty confident it won t let you down.

another option would be to buy a silncer 750w which peak is at 825w and 60amps in the 12v

I think there is a 510w version in the FS section. Here I can get the same model at 3 times the price the guy is selling it :S

[theteamaqua]

nope, I meant to grab 2x 510w pcpower ( 650w peak) with 38 amps in the 12v line...or you could grab the one you mentioned, I am pretty confident it won t let you down.

another option would be to buy a silncer 750w which peak is at 825w and 60amps in the 12v

I think there is a 510w version in the FS section. Here I can get the same model at 3 times the price the guy is selling it :S

2 x 510 = 1020W .. right now i have 750 + 850 = 1600W , plus the TT 850W has 4 x PCI-E, total of 60A + 62A = 122A over 12v rails

but yes for TEC i think a single rail PSU would work better, then again TT 750W has 18A over 12V rails which is what swiftech suggest

[serialk11r]

Look, the wires will most likely burn, because they are not rated to pull the 24A that the TEC requires.

[theteamaqua]

swiftech says 18A
http://www.swiftnets.com/products/MCW6500-T_775_installation_guide.pdf
Recommended installation for coldest temps: Connecting to a dedicated auxiliary power supply
�� The minimum requirements for a dedicated power supply are 25A @ +12V.
�� Your TEC module has been measured to draw 18 amps at 12 volts. We recommend using the “Meanwell S320-12” auxiliary power supply, which is adjustable at up to 13.8Volts, and will boost the cooling performance. The S320-12 is available on our website in the Thermoelectric accessories section.
�� You will need to cut and strip the wires of your TEC module in order to connect them to the terminals of the power supply
�� Connect the red wire from the TEC module to the +V terminal, and the black wire to the –V terminal as shown in figure 4 below.


is cutting the wire needed for normal PSU.. from the picture it seems like thats only needed for aux PSU

hey frozencpu has the 400w pelt ... should i get that if 226w is not enough ?? although it is 60x60 .. a bit bigger than the 50x50 from swiftech

[IanY]

If I use a 6500T and two MCW60Ts, does it mean that I should buy three Meanwell S320s ? And if I were to mount the Meanwells in the optical bays, how would I exhaust the PSU heat?

[serialk11r]

If I use a 6500T and two MCW60Ts, does it mean that I should buy three Meanwell S320s ? And if I were to mount the Meanwells in the optical bays, how would I exhaust the PSU heat?

They won't handle a kentsfield, just to let you know ;) The PSUs have fans built in, and I hear that if you take a small 80mm fan or something, they will run cooler and quieter (the internal fan is probably noisy).

[IanY]

So, you are saying that a MCW6500T cannot handle the heat of a Kentsfield. Philly Boy is saying that it can get a Kentsfield to over 3.9 Ghz. I just manage 3.75 Ghz, and I thought its already very good, but I don't have a TEC.

In that case, can a MCW60T's 188 watt TEC handle a 8800GTX ?

[IanY]

Another thing that I'm very confused. They say that an Enermax galaxy 1kW is ok to use, but a PC Power Cooling 1kW-SR single rail cannot handle the TEC? (I'm talking about using two more 1kWs that are separate from the PSU for the rest of the system).

[theelectic]

So, you are saying that a MCW6500T cannot handle the heat of a Kentsfield. Philly Boy is saying that it can get a Kentsfield to over 3.9 Ghz.

Put it this way, who would you believe, someone who has actually tried and used quad with that block, or someone who's talking out of thin air. Even Swiftech's own data shows that the 6500T will handle overclocked quads, albeit barely, and you're pushing into a range where straight water is better. You may end up like seedomo where you're overshooting the TEC. There's not much headroom to play with.

Another thing that I'm very confused. They say that an Enermax galaxy 1kW is ok to use, but a PC Power Cooling 1kW-SR single rail cannot handle the TEC?

No, this is exactly what Swiftech says about using an "existing power supply":

The unit can be connected directly to high-end computer power supplies. At the time of release, we have validated in our labs the Enermax Galaxy 1K power supply.

You CAN connect it directly to a high-end PSU. They have only validated the Enermax Galaxy themselves. Obviously, if the PCP 1k is similar/better is should be able to handle the load.

Again like in the other thread you started, you're better off using the Meanwell - it's smaller, cheaper, more efficient, etc. As Gabe from Swiftech has stated on many occasions, all you have to do is add another quiet 80mm fan on top of the existing one. That way more air is pulled out and the noisy internal fan never starts up. The extra heat that is generated should be easily handled by your case exhaust fan.

[IanY]

Do you know how many TECs a Meanwell S320 can handle? If I were to use a MCW6500T and two MCW60Ts, do I need two or three Meanwells ? How am I supposed to mount three Meanwell PSUs inside a case and yet exhaust all the pSU heat if they are stacked on top of each other?

Many thanks.

[Vinas]

You CAN connect it directly to a high-end PSU. They have only validated the Enermax Galaxy themselves. Obviously, if the PCP 1k is similar/better is should be able to handle the load.


The PCP&P 1KW outputs a max of 1100watts and continuous 1KW. Hard to believe there's anything it can't handle. :eek:

[theelectic]

Do you know how many TECs a Meanwell S320 can handle? If I were to use a MCW6500T and two MCW60Ts, do I need two or three Meanwells ? How am I supposed to mount three Meanwell PSUs inside a case and yet exhaust all the pSU heat if they are stacked on top of each other?

Do the math. The MCW6500T requires 18A @ 12V. Each MCW60-T requires 9.3A @ 12V. These figures are all from the Swiftech website. The S320-12 can supply 25A @ 12V. So you could get away with using two @ 12V, not three.

As for case mounting - better get a bigger case, mount them outside the case, or cut some big holes. Be creative. Whatever you do don't stack them on top of each other.

[seedomo]

So I swapped out the Swiftech 220 rad and put in the mother of all rads, the Thermochill PA 120.3. What a difference! I'm looking pretty good now at 3.9 using only 1.525 volts 1.46VDroop. CPU temp doesn't go above 37c. Cores top out at 51c. But whats great is that I'm stable at 3.9! Pretty close to my goal of 4.0. I think I won't need to run separate loops. I also don't really want to try the water only option since I've seen people using this rad running temps up to 70c. Can that be right? What are average temps to expect for people running straight water at 3.8-3.9Ghz? I can't image that they are able to keep their temps under 37c (CPU reading, not core), right?

[seedomo]

It sounds like you spent some time trying to figure out if your current PSU was capable of powering your system and it sounds like you should be alright (by the way what kind of amps are your rails pusing on that Antec supply, do you know)?
36A on the 12v rail.
I've also determined that I'm pulling 410 watts at 1.535Volts running at 3.9

[Duh]

2 x 510 = 1020W .. right now i have 750 + 850 = 1600W , plus the TT 850W has 4 x PCI-E, total of 60A + 62A = 122A over 12v rails

but yes for TEC i think a single rail PSU would work better, then again TT 750W has 18A over 12V rails which is what swiftech suggest



1- it s not 2x510 as its peak its over 600 w
2-psu s as many other things have a variety of qualities and TT is not top notch nor nothing similar

If I use a 6500T and two MCW60Ts, does it mean that I should buy three Meanwell S320s ? And if I were to mount the Meanwells in the optical bays, how would I exhaust the PSU heat?
hey pug boy :P no one said you should buy meanwell you can buy whatever you like at ebay for cheap like L33t who bought a 1kw psu for aprox $100+S&H IIRC

to cool down your psu you can use something like this

http://www.petrastechshop.com/suovcokitsi.html

cheers

J

[seedomo]

I'm teetering on the edge of stability. Orthos for 32 mins. using SMALL FFTs and then error. So I'm adding a pinch of voltage at a time to see what it takes to get me stable running it for an hour. With each .0125 increase temps rise about 2c. My CPU temp is now peaking at 38c. Fingers crossed.

[Duh]

I would change TEC and make it run confortably with the amount of wattage, amps and volts and would not play too much with voltage.. when you play too much with voltage, :banana::banana::banana::banana: happens

[Roger_D25]

Yeah I'm very happy with my Thermochill PA120.3, great radiator. It is nice to see a big drop in temps for you, running at under 40C underload is pretty decent! If I'm running at 3.8Ghz I need about 1.48v (actual votls) to run Orthose stable (small fft's), at those settings I usually average about 54C-58C depending on ambient temps. I only have one loop which I cool my CPU, GPU (full coverage G80 block), and my SPP (chipset). Your right though, many people get even higher temps then me on their C2D (60c-70c underload) so Im' pretty happy with that, althoug sub 40C temps would be much better!

[Holst]

I'm teetering on the edge of stability. Orthos for 32 mins. using SMALL FFTs and then error. So I'm adding a pinch of voltage at a time to see what it takes to get me stable running it for an hour. With each .0125 increase temps rise about 2c. My CPU temp is now peaking at 38c. Fingers crossed.

Fantastic :)

Im supprised that such a relatively small drop in temps has given you some extra stability.
At least we know that the chip is capable of more but you just need to get the temps down a bit.

I think you are better off with the TEC than without it. On straight water I think your temps would be higher at load.. BUT your not gaining a huge ammount (considering the extra power your using)

Can you get any further improovement in water temps (louder fans)
Just to see what happens if you can get down to 360c (for example)

[theteamaqua]

1- it s not 2x510 as its peak its over 600 w
2-psu s as many other things have a variety of qualities and TT is not top notch nor nothing similar



oh my bad so 600w x 2 = 1200w peak load , while i get 750w + 850w = 1600w continuous load with 6 PCI-E connecter

back to topic:
yeah .. i agree with holst ... its very unlikely that u will get better temps without TEC , but be careful though as huge increase can cause instability (vs say ur idle is -13C and load is 20C)

i think serialkiller is right .. the E6600 @ 1.6+v vcore is a bit too much for the 226w TEC

[seedomo]

Thanks for all the help everyone.

[seedomo]

http://www.domophoto.com/4100@16.jpg
I think this new board is making a difference in my ability to overclock. Remember the first one I shorted out. However it could be the additional cooling on the NB block. Don't really know. I am sure that I would have never been able to boot at 4.1 using only 1.6 volts. I've run some Photoshop benches for bragging rights but doubt it will be stable for very long. Look at those temp for idle: negative.

[theteamaqua]

yeah to sum up:

straight water < MCW6500 < 400+W TEC

heh

waht board is that?

btw what did u do for insulation?? i see that u have the red stuff near the socket

[seedomo]

straight water < MCW6500 < 400+W TEC
Don't understand what you mean by this. I did say earlier that my WHOLE PSU was pulling around 400 watts, but by no means does that mean my processor is pulling that amount.
waht board is that?
Nvidia 680i
btw what did u do for insulation
Used the provided foam with the kit and used the red stuff they sell at crazypc.com

[theteamaqua]

Don't understand what you mean by this. I did say earlier that my WHOLE PSU was pulling around 400 watts, but by no means does that mean my processor is pulling that amount.

Nvidia 680i

Used the provided foam with the kit and used the red stuff they sell at crazypc.com


but if ur TEC is 226W .. even after u overvolt/power it ... its not going to pump 400W of heat to cool it?? is it?? cuz if u can just overvolt/power it, whats the point of getting higher wattage TEC?? not sure though

[serialk11r]

but if ur TEC is 226W .. even after u overvolt/power it ... its not going to pump 400W of heat to cool it?? is it?? cuz if u can just overvolt/power it, whats the point of getting higher wattage TEC?? not sure though

NO! A 226W TEC draws 320W of electrical power, and is moving 226W of heat. Terrible efficiency I know, but all TECs are like that. And in turn, his meanwell which gets about 87% efficiency max I think is wasting a bit of energy as well, making it a grand total of 400W.
You SHOULD NOT OVERVOLT A TEC, THEY RUN A HIGH RISK OF GETTING DESTROYED AND/OR CATCHING FIRE OR SOMETHING WORSE/SIMILAR. There is a reason why the manufacturer gives the TEC a max voltage and max amperage rating...

[theteamaqua]

so if it has 87% efficiency and draws 320W then its 320W/87% = 367W ... how u get 400W??

[seedomo]

400 watts is the psu that's running the cpu and hard drives NOT the tec.

[theteamaqua]

ahh that explains that

[serialk11r]

so if it has 87% efficiency and draws 320W then its 320W/87% = 367W ... how u get 400W??

But a PSU won't always be at max. efficiency ;) usually lower.

[seedomo]

A little update:

First a thing about temps I learned recently after a ton of reading. Fact: Core temps based on Intel's TAT are the baseline for all other temps. This means that from the tat temps you can figure the other temps. Motherboard manufacturers use different ways of estimating the temp of the top of the chip, but it will always be 15c less than the core temps. After discovering that fact I used Speedfan's offsetting feature to correct for my "CPU" temp. It was reading some 5c higher than the core temps which can't be right since it is theoretically farther away from the two cores of the E6600. All that said, all my previous temps were incorrect to some degree if I used the "CPU" temp as the reading. Now that i have corrected it my new temps (using my L627B155 chip which needs very little voltage for a TON of current:mad: ) are 3.6Ghz @ 1.41 Volts (bios) Cores peak at 25c and CPU is now properly reading 12c. Since most people say you shouldn't be above 50c if you are running a TEC, they are talking about the CPU temp NOT the Core temps, so as you can see 12c is pretty respectable, but I can push it further, I've just been a little busy to play lately. The one time I pushed it to the max of 4.1 using 1.6Volts it wasn't stable for long, and my temps were nearing CPU=55c so I got uncomfortable. I still want to see what happens when I isolate the 120.3 rad with the tec. Right now I have it cooling the whole loop, GPU and NB chipset. One more thing, I have tested 3 e6600s (week 29, week 44 and week42) and all 3 hit at least 3.8 at reasonable temps, so you can be assured to get this kind of clock speed using my setup and ANY e6600, I would guess. Mind you, you don't have to get the 120.3 since I hit the same speeds using the Swiftec 220 rad while using air to cool the GPU and chipset. Hope this helps.
Biggest thing learned: Make sure to pack the dielectric grease into and around the socket before you seat the processor.
http://www.domophoto.com/3.6@1.4125.jpg

[seedomo]

Had a little time to play around a bit. Compare this result to the one with the Swiftech RAD! HUGH difference AND the 120.3 is running a GPU and a chipset as well. :D
http://www.domophoto.com/3.75@1.4312.jpg

[theteamaqua]

But a PSU won't always be at max. efficiency ;) usually lower.

what r u talking about?? if he measure the power draw from outlet is 400W, then its outputing 400 x 0.8 = 320W ... i still dunno how a 226W TEC can draw 400w from the outlet .. he was talking about the whole computer with CPU . mobo. RAM .. exclude TEC i believe

226/400 = 66.5% ... thats a lot of power wasted

hey seedmo how high can u go?? for suicide shot , and for pi 1m , and for pi 32m ...

[seedomo]

For the last time, the only PSU I've discussed the wattage for is the PSU that runs the computer, NOT the Tec, damn you guys are making me get a headache! Why would I concern myself with the wattage of the Tec PSU, I know it supplies 320 watts if needed, but since the Tec only draws 226, its probably pulling 226 watts. The only reason I am writing about the wattage of the computer PSU wattage is to use the numbers to find out how much the e6600 is using. I want to know if I'm exceeding the wattage of the Tec,s cooling capacity of 226. If we know the total wattage when I run it at 2.4 at default voltage and Intel says 65 watts of that is the CPU then we can figure how much wattage it pulls at 3.8 using 1.5 volts since the only change is CPU wattage. Understand? Oh hell, I'm getting tired of trying to explain...

Ive done 4.1 using 1.6 volts and ran a few photoshop benches, but when I rebooted, it wouldn't post so I dropped the fsb, still wouldn't work, it would hang just before the option to boot from CD, "dmi pool..." message. So I kept dropping the fsb until I was at 2.4 Ghz. Still no dice. So I pulled the plug to see if totally taking away any energy would reset it. Thankfully, it worked. That will probably be the last test at such extremes.

[serialk11r]

what r u talking about?? if he measure the power draw from outlet is 400W, then its outputing 400 x 0.8 = 320W ... i still dunno how a 226W TEC can draw 400w from the outlet .. he was talking about the whole computer with CPU . mobo. RAM .. exclude TEC i believe

226/400 = 66.5% ... thats a lot of power wasted

hey seedmo how high can u go?? for suicide shot , and for pi 1m , and for pi 32m ...
A 226W TEC draws 320W of electricity to move 226W of heat ;) Get it now? the 226W is the heat pumped by the TEC. I believe TECs get <10% efficiency or so (because they use a ton more power than is theoretically needed to move that much heat. the perfect heat pump, the carnot cycle, does that).
Now seedomo, ya I know you're talking about the PC itself, but I'm telling him about another issue.

[seedomo]

Got it, thanks. I am, however, more interested in finding out if my chip is exceeding the cooling ability of the TEC, that's seems of utmost importance. Plus I find it curious that more people aren't interested in comparing the temps of this tec setup to straight water cooling. That is certainly what most want to find out, in the end. Earlier people were saying that I was exceeding the cooling capacity of the tec, but now it seems that isn't the case at all since moving to the 120.3 Rad. This is the kind of useful information that others might be able to use in determining what that want in their own setup.

[theteamaqua]

can u go to 4.2GHz?? btw i dont think that TEC's efficiency is 70% as not even swiftech's site says it

[seedomo]

LOL theteamaqua, nice edit, if you know what I mean, just a little nicer way of wording... Just put the meter on the PSU for the TEC and it pulls 330 to 340 @ 13.5 Volts. The result is to move 226 watts of heat on the TEC. So now I don't want to ever talk about this damn thing again. I'd rather talk about something more interesting, that's all.

can u go to 4.2GHz?? btw i dont think that TEC's efficiency is 70% as not even swiftech's site says it
I don't know. I've got the wrong chip in currently. This chip gets stuck at 3.75 no matter what. I'm NOT looking forward to changing the chip out AGAIN! It becomes painfully tedious once you've got 3 blocks in the loop. I must drain the system to change the CPU. But I'm sure this is not my best chip.

[Holst]

Ok, lets try and estimate the CPU wattage at 4.1ghz.

At stock 2.4ghz and stock voltage (1.3525v) intel claim 65watts power usage

At 4.1ghz seedomo is using 1.60v

So we can do a simple calculation to estimate the wattage.

65x(4.1/2.4)*((1.6/1.325)*(1.6/1.325))
=161.92watts!!!!

Thats a serious ammount of wattage.
I should say that this is an ESTIMATE based upon intels own estimates. Processors will not necicarally scale exactly with frequency or voltage (v^2)
I think this assumption is going to be a bit higher than actual wattage.

The guys on procooling recomend 80% of the calculatd estimate so thats 130watts

[shimmishim]

ahh, thanks holst i've been looking for a way to estimate cpu output wattage.

so for the mcw6500, as long as the estimated wattage is below 155W (assuming you're running off of a meanwell psu rated at 320W, you're okay with a TEC but once you bust 155, straight up water should be better, is this correct?

[Roger_D25]

Holst,
Just wanted to thank you for posting that easy to use equation for calculating CPU wattage, good to know! Even though it might not be 100% accurate it's certainly good enough to get a decent ballpart! With that said will this equation work with Quad Cores if we make sure to multiply by all 4 cores?

I did the calculation for my Xeon 3060 which is currenlty running at 3.6Ghz on 1.4v (actual voltage). Well I end up getting about 109W, considerably less than in Seedomo got. This just goes to show how much more heat is produced to get those last few hundred Mhz in clockspeeds!

[Holst]

ahh, thanks holst i've been looking for a way to estimate cpu output wattage.

so for the mcw6500, as long as the estimated wattage is below 155W (assuming you're running off of a meanwell psu rated at 320W, you're okay with a TEC but once you bust 155, straight up water should be better, is this correct?

Not as simple as that.

Its a not like you add another 5 or 10 watts and the CPU says "ahhh help im going to explode"
Its more like a sliding scale where another few watts adds a few more *C

Im not doing anything much now so I will retype some of the sticky for you :D

If you dont understand read the sticky!

Delta T (difference between hot and cold side) is idrectly proportional to wattage applied to the TEC.

226watts is the wattage needed to totaly overwhealm the TEC (Qmax)
In this case Delta T will be 0 (you gain absolutely nothing having the TEC there)

At 0watts the TEC will have a delta T of 70*C.
This means that the cold side is perfectly insulated (which is impossible) and is 70*c colder than the hot side

We can make this into a nice graph :D (see below)

As you can see at 150watts you would expect a delta T of about 24*C below the temperature of the [b]TEC hot side[b]
It is iportant to note that this is only true when running the peltier at VMAX (which seedomo is NOT)

The problem for us is that this calculation, although interesting isnt very helpfull :(
Why?

Because our system has many other unknown variables, and inefficencys.

The biggest of these is the waterblock and the thermal interfaces.

Firstly we have ambient air cooling the water,
Then the water cooling the waterblock,
Then the waterblock cooling the TEC (TIM)
Then the TEC cooling the coldplate (TIM)
Then the coldplate cooling the IHS (TIM)
Then the IHS finally cooling the CPU core (TIM)


If you can estimate the CW of your waterblock you can make an estimate of your hot side temps (once you calculate the TEC wattage (output)) and from there you can estimate your CPU temp.
You can even use your waterblock CW to estimate your temps without TEC at the same CPU speeds..
The maths is supprisingly simple once you understand the terms, but if you dont understand the terms you just end up getting confused! as many of you seem to in this thread and others.
If you cant understand the sticky, or want something explained start a thread, its better not to spam up this one any more, seedomo is trying to talk about HIS system, this isnt [H]ocp!!

Anyway.. thats all I have time for now.

[shimmishim]

thanks for the explanation... it makes a lot more sense and i'll stick back to the topic... but for the record the 155 value i threw out was from the swiftech website which had a nice graph showing case T vs. W and showed that once you get over 155W, water is better. :)

[seedomo]

First I need to figure out what my wattage is for my system PSU minus CPU wattage:
I'm using a "Killa-Watt" plugged in between the PSU and the wall.

If I set my system to 2.4 @ 1.35, intel says that 65watts of the total wattage will be for the CPU, however is intel saying this is for load or idle? For example:
2.4@1.35 = 297 watts idle (Actual reading)
So... 65 watts of this 297 is the CPU.
The rest of the wattage we can call the System Wattage (MB, Disks, Fans, Pump)
System Wattage w/o CPU is therefore 232 = 297-65 @IDLE

From now on I will use 232 to represent the system wattage and the only variable will be the voltage and fsb change in the bios.
CPU E6600 L627B155 (This batch needs very little voltage for a TON of current, very different than current batches)
Test 1 result: 2400@1.3563 Idle = 297 watts / CPU wattage 65 watts
Test 2 result: 2400@1.3563 Loaded = 343 watts / CPU wattage 111 watts
Test 3 result: 3750@1.4312 Idle = 315 watts / CPU wattage 83 watts
Test 4 result: 3750@1.4312 loaded = 415 watts / CPU wattage 183 watts

Here are the screens to show I haven't made any mistakes.
Test 1 2400@1.3563 297watts idle
http://www.domophoto.com/2400@1.3563 297wattsIdle.jpg

Test 2 2400@1.3563 343watts loaded
http://www.domophoto.com/2400@1.3563 343wattsLoaded.jpg

Test 3 3750@1.4312 315watts idle
http://www.domophoto.com/3750@1.4312 315wattsIdle.jpg

Test 4 3750@1.4312 415watts loaded
http://www.domophoto.com/3750@1.4312 415wattsLoaded.jpg

[Roger_D25]

seedomo,
Wow thanks for taking the time to do all that work, very impressive! Since CPU wattage and system wattage is all new to me this is very interesting. It goes to show there is a big difference between idle/load CPU wattage right? Anyway great job man, keep up the good work!

[serialk11r]

theteamagua, the TEC's efficiency is NOT 70%.
You do not calculate the efficiency by heat moved/energy consumed...
You compare it to the ideal heat pump, which gets 100% efficiency...The ideal heat pump/heat engine (carnot cycle) uses only as much energy as is theoretically needed to move heat.
Go on wikipedia, and look up "Carnot cycle".

[Roger_D25]

I don't imagine anyone cares but your absolutly right serialk11r about needing to compare the actual efficiency of the TEC unit to the Idealized Case (or Carnot Cycle)! Of course there is a bit more to calculating the actual efficiency of the TEC unit but that is the general idea.

I've spent the last few minutes looking through my Thermo notes on the 2nd Law of Thermodynamics trying to figure out the actual equation to calculate the efficiency of seedomo's TEC unitl and also figure out exaclty how much energy the TEC unit is using based on the CPU's heat output (in watts) but I'm having some difficulty. Anyway it's quite possible to do and if I figure it out I'll post back guys.

Incase anyone cares I think the equation for calculating the Carnot Efficiency of seedomo's TEC unit would be: Carnot Efficiency = 1- Temp(Low)/Temp(High) (Again, could be wrong just having fun trying to figure it out)!

[serialk11r]

I really need to go over some advanced mathematics, the equations for the thermodynamics things, well, my brain just isn't cutting it...
To calculate the efficiency of seedomo's TEC, we need to know the temperature of the hot side of his TEC first, don't we?
OH and BTW, Roger_D25, you say "I don't imagine anyone cares" but you're wrong :D I'm a crazy kid who enjoys complex crap so I do care :D

[seedomo]

Wow, you guys are really getting into this! I was really trying to nail down just how far away I was from the TECs 226 watt threshold. For THIS chip it seems that I'm a fair distance from maxing out my tec, since my CPU is only using pushing 183 watts at its max OC, right? Am I on track here?
BIG NOTE: I'm interested in finding out if Intel's spec for this chip (65w) is loaded or not. I figured that Intel used 65w at idle since its better for marketing, than say 100w, loaded. PLEASE, someone enlighten me.

[seedomo]

don't imagine anyone cares but your absolutly right serialk11r about needing to compare the actual efficiency of the TEC unit to the Idealized Case (or Carnot Cycle)! Of course there is a bit more to calculating the actual efficiency of the TEC unit but that is the general idea.

I've spent the last few minutes looking through my Thermo notes on the 2nd Law of Thermodynamics trying to figure out the actual equation to calculate the efficiency of seedomo's TEC unitl and also figure out exaclty how much energy the TEC unit is using based on the CPU's heat output (in watts) but I'm having some difficulty. Anyway it's quite possible to do and if I figure it out I'll post back guys.

Incase anyone cares I think the equation for calculating the Carnot Efficiency of seedomo's TEC unit would be: Carnot Efficiency = 1- Temp(Low)/Temp(High) (Again, could be wrong just having fun trying to figure it out)!
Can I help you in any way? I've got a Multimeter and a Killa-watt here.

[Roger_D25]

Cool, I'm glad your not upset seedomo about me and seriali11r are talking about this stuff. It's pretty late here and I'm gonna call it a night but hopefully before the weekend is over we'll pin it down! Figuring out the hot/cold side of TEC shouldn't be too difficult and figuring out (as close as we can) the heat output of your processor (in watts) seedomo should also be possible. Heck you have pretty much got it figured out above!

There is a good chance that someone really smart will make me look stupid but hopefully we can get this figured out!

[serialk11r]

Say, seedomo, got a decent thermometer?

[seedomo]

Yes, a good digital one, in fact.

[serialk11r]

Hehe, could you take some temps of the cold side and hot side of the TEC? I know that's not possible, so like, base of water block temps, and cold plate temps... Damnit, seems as if my math skillz aren't good enough for that thermodynamics stuff...I need to study some calculus :P :D

[Holst]

Wow, you guys are really getting into this! I was really trying to nail down just how far away I was from the TECs 226 watt threshold. For THIS chip it seems that I'm a fair distance from maxing out my tec, since my CPU is only using pushing 183 watts at its max OC, right? Am I on track here?
BIG NOTE: I'm interested in finding out if Intel's spec for this chip (65w) is loaded or not. I figured that Intel used 65w at idle since its better for marketing, than say 100w, loaded. PLEASE, someone enlighten me.

Firstly the CPU wattage.
The 65watt figure comes from here - http://download.intel.com/design/processor/datashts/31327803.pdf (page 78)

"This table shows the maximum TDP for a given frequency range. Individual processors may have a lower TDP."

This TDP figure is NOT the maximum wattage!!

"Intel recommends that
complete thermal solution designs target the Thermal Design Power (TDP) indicated in Table 27 instead of the maximum processor power consumption."

This is why all the processors in the range are listed as 65watts.
As such its very hard to know if this 65watt figure is an accurate representation of the CPU wattage.

But, in section 2 of the PDF it lists the electrical specs, which I might be able to use to work out the max instantanious wattage. Let me have a read.


"Am I on track here?"

No, no, your off track :(

I still think that a bigger TEC will give a significant gain.

The important thing to look at IMO is the difference between idle and load. Even with your improoved water temps this is still 40*C!!

The threashold of the TEC is NOT 226watts!!!!!!!!!!!!!!!!!!!!!
Not 226watts!!!!!!!!!!!!!

226 is the point at which the hot side will be the same temperature as the cold side... i.e No gain from the TEC.
BUT its very very important to remember that the TEC is massively increasing the load on the waterblock, and as such the waterblock temp is much higher with the TEC than without...
So... there will be a wattage of CPU that will give the same temps with or without the TEC at load.

[Roger_D25]

Great points Holst, you just cleared up alot for me, thank you!

[seedomo]

could you take some temps of the cold side and hot side of the TEC
Just too hard to do with the neoprene foam around the socket. I would also probably need to get some flat probes as well. Until then I will report some water temps for ya in a bit. I'm probably going to add the Swiftec 220 rad to the loop and see if it affects the overall results.

[serialk11r]

Just too hard to do with the neoprene foam around the socket. I would also probably need to get some flat probes as well. Until then I will report some water temps for ya in a bit. I'm probably going to add the Swiftec 220 rad to the loop and see if it affects the overall results.

OH, more rads, that's good new :) More rads=lower water temp=lower temp on hot side=lower coldplate temp=lower CPU temp! At the flowrates the MCW6500 allows, I don't think the extra restriction will actually be hurting flow that much anymore...that 6500 is a monster.

[Holst]

CPU wattage in seedomos system.

In the intel PDF they claim that idle will be halt wattage will be 20watts (this is idle wattage)
And they claim that load (TDP) will be 65 watts.

So lets look at seedomos results

Stock speed idle = 297watts
Stock speed load = 343watts

If at load we are using 20watts as intel claims then the rest of the system is using 297-20 at idle (277watts)
And at load we are using 343watts, and CPU power will be 342-277 = 66watts :D

So seedomo's numbers are practically the same as intels, load at 66watts.
This is extremely interesting, as the practical numbers are backed up by intels own specs :D :D

So.... now we can look at the overclcoked wattages and try and estimate the CPU wattage.

overclocked idle is 315 watts
And overclcoked load is 415 watts

Using the formula to estimate the overclocked wattage at load we get

65*((3.75/2.4) *((1.432/1.356)*(1.432/1.356)) = 113.266watts estimated load

We can also do the same at idle
20*((3.75/2.4) *((1.432/1.356)*(1.432/1.356)) = 34.85watts estimated

So how do seedomos tests compare to my estimates?

Assuming that the system allways uses 277 regardless of CPU load then we have.
Idle overclocked wattage = 315-277 = 38watts
Idle load wattage = 415 - 277 = 138watts

So the practical numbers seem to be higher than the estimates.. why?

Well the problem is that when overclcoked we are not just overclocking the CPU. FSB has gone up from 267 to 417mhz
So we have also overclocked the chipset.. and the ram..

I think this may account for the extra wattage.

How can we remove this error?

We can try running the CPU at 2.4ghz and 417fsb (or as close as you can manage) and see if the idle and load wattages go up even though the CPU speed is the same.

So for the moment I still cant accurately estiamte the CPU wattage, I can only say thats its somewhere between 114 and 138watts

Is this helping or not?

[seedomo]

Thanks everyone for all the interest in my project. Holst, hope you're not traveling too soon! Here is some more interesting results. I've added another 665 pump (now 2) and my old Swiftec 220 rad. My water temps went down dramatically.

One Rad Loaded for one hour
Air Temp 18
Water Temp 28
Max Core Temp 23

Two Rads, Two Pumps Loaded for 1 hour
Air Temp 18
Water Temp 24
Max Core Temp 19

[Holst]

Ive stopped traveling actually.
Im living in Russia with my girlfriend at the moment, trying to learn Russian LOL.

Its interesting to note that the delta between water and CPU at load is 5*c regardless of water temp. (As expected)

But there is still room for improovement on the water temps. With an excelent ambient of 18*C you might still get the water temps a bit lower.

I dont suppose you can test with just one pump, I suspect the second pump isnt helping. (But if it is helping its even more interesting)

have you ever heard of a bong for cooling water?
(same thing as they use to cool water from power stations)

[seedomo]

I just went throught the bong that guy made in Australia. What a project. I'm guess I don't have the willingness to got through the extra effort at this point. It's very interesting, thanks for pointing it out.

[seedomo]

I dont suppose you can test with just one pump, I suspect the second pump isnt helping. (But if it is helping its even more interesting)
Sure, I'll test it tomorrow, as well as changing the chip out to test two others. I do notice a huge difference in flow rate, although I don't know if that will translate to better temps. One thing I can do is turn the voltage on the 665 (its got a variator) to see if the decreased flow makes a difference, if not, then I will take it out.

[seedomo]

Its interesting to note that the delta between water and CPU at load is 5*c regardless of water temp. (As expected)
Yea, so I guess every bit of radiator helps.

[theteamaqua]

theteamagua, the TEC's efficiency is NOT 70%.
You do not calculate the efficiency by heat moved/energy consumed...
You compare it to the ideal heat pump, which gets 100% efficiency...The ideal heat pump/heat engine (carnot cycle) uses only as much energy as is theoretically needed to move heat.
Go on wikipedia, and look up "Carnot cycle".

so ur saying that it draws 320W and output 226W?? and what is the efficiency then? isnt it 226/320 ~ 70%

also where did u get the temp to calculate the efficiency for the TEC and go on to calculate the power draw by it?

how did u know the TEC draws 320w from the PSU??

seedmo,

do u think that i should get dual radiators?? for mcw6500t?? i already have a black ice gt 360 , i might get a 2 x 120 rad

[serialk11r]

so ur saying that it draws 320W and output 226W?? and what is the efficiency then? isnt it 226/320 ~ 70%

also where did u get the temp to calculate the efficiency for the TEC and go on to calculate the power draw by it?

how did u know the TEC draws 320w from the PSU??

seedmo,

do u think that i should get dual radiators?? for mcw6500t?? i already have a black ice gt 360 , i might get a 2 x 120 rad

The specs of the 226W tec are, 24A drawn at 15V. Multiply and that gets you 320. The efficiency is NOT 226/320, it is MUCH more complex than that and involves calculus and a comparison to the ideal heat pump. Efficiency cannot be calculated by simple division, it involves integration, and involves temperature difference. If it was that simple, then a phase system could be getting 400% efficiency. Now where did 400% come from? 100% is max...:P Sure, the way you said can be a way to get some feeling of efficiency, but it is not nearly that simple.

[seedomo]

do u think that i should get dual radiators?? for mcw6500t?? i already have a black ice gt 360 , i might get a 2 x 120 rad
I'm just playing around at the moment; and, I think it is pretty obvious that I can get 3.8Ghz on probably most 6600 chips and that is with one 120.3 running through GPU, NB and the CPU. At this point it hardly seems worth the money to shave 5c from overall temps by adding another rad. Especially when I'm already around 25c at max overclock, loaded with orthos, with just one. Don't forget that these are the most extreme temps I can create. Typically, I'm more like 0c at typical work loads using other programs.

[theteamaqua]

The specs of the 226W tec are, 24A drawn at 15V. Multiply and that gets you 320. The efficiency is NOT 226/320, it is MUCH more complex than that and involves calculus and a comparison to the ideal heat pump. Efficiency cannot be calculated by simple division, it involves integration, and involves temperature difference. If it was that simple, then a phase system could be getting 400% efficiency. Now where did 400% come from? 100% is max...:P Sure, the way you said can be a way to get some feeling of efficiency, but it is not nearly that simple.

no swiftech says 18A @ 12v

http://www.swiftnets.com/products/MCW6500-T_775_installation_guide.pdf

Your TEC module has been measured to draw 18 amps at 12 volts.

also if a device like a PSU for instance takes in 320w and output 226w .. its efficiency is 70% .. thats all i care .. b/c thats the power it takes and the power it outputs ...

also the 400w seedmo was talking about was computer , not TEC

seedmo,
do u think that 2 pumps helps?? i am going to have that and a MCW30 chipset block and a black ice gt stealth 360 rad in my loop.. and yes i already have 2 pumps

btw do u think i should get thermochill?? cuz i cant really buy it here.. i only go to svc and petras to buy wc stuff

[seedomo]

no swiftech says 18A @ 12v

http://www.swiftnets.com/products/MCW6500-T_775_installation_guide.pdf



also the 400w seedmo was talking about was computer , not TEC

I've put the Killa-Watt on it and read 318-340 watts. Kinda jumps around a bit.

[seedomo]

seedmo,
do u think that 2 pumps helps??

Its certainly helps flow rate but I doubt it helps temps much since after I put the second rad w/ pump temps only went down on the CPU 5c. I imagine that was mostly cause of the rad, but sure yet.

btw do u think i should get thermochill?? cuz i cant really buy it here.. i only go to svc and petras to uby wc stuff

I would say hell yea! Granted I went from a 2x120 to a 3x120, still it was a hugh difference. Its not obvious when you look at the product pics but the Thermochill is almost twice as thick as the others. Its like having a standard 3x120 plus half of another one, something I didn't expect til I saw the beast.

[theteamaqua]

I've put the Killa-Watt on it and read 318-340 watts. Kinda jumps around a bit.

the computer or TEC??

[seedomo]

The TEC PSU.

[seedomo]

Originally Posted by theteamaqua View Post

seedmo,
do u think that 2 pumps helps??
I just throttled down the second pump while running Orthos to see if the temps changed any. Went from 4800 rpm to 1200 and the temp rose 1c. I guess that answers that. Not much help.

[serialk11r]

no swiftech says 18A @ 12v

http://www.swiftnets.com/products/MCW6500-T_775_installation_guide.pdf



also if a device like a PSU for instance takes in 320w and output 226w .. its efficiency is 70% .. thats all i care .. b/c thats the power it takes and the power it outputs ...

also the 400w seedmo was talking about was computer , not TEC

seedmo,
do u think that 2 pumps helps?? i am going to have that and a MCW30 chipset block and a black ice gt stealth 360 rad in my loop.. and yes i already have 2 pumps

btw do u think i should get thermochill?? cuz i cant really buy it here.. i only go to svc and petras to buy wc stuff

Um, no. The heat it pumps depends on the temperature...
You really need to look up more of this stuff before you make random assumptions like that...
If the TEC is 18A at 12V, then its consuming 226W of electricity, but a TEC simply cannot pump 226W at these temperatures, so the cooling power that TEC is providing is LESS than 226W. I'm talking about maximum cooling power.

[theteamaqua]

Um, no. The heat it pumps depends on the temperature...
You really need to look up more of this stuff before you make random assumptions like that...
If the TEC is 18A at 12V, then its consuming 226W of electricity, but a TEC simply cannot pump 226W at these temperatures, so the cooling power that TEC is providing is LESS than 226W. I'm talking about maximum cooling power.

ok shoudl i trust swiftech or some yahoo in xs forum

again the 400w he was talking about was the computer NOT the TEC

http://www.xtremesystems.org/forums/showpost.php?p=2125948&postcount=93

A 226W TEC draws 320W of electricity to move 226W of heat

talk about random assumptions

swiftech never said anything about TEC draws 320w, and therefore ur assumption about TEC draws 320W from PSU, then the PSU draws 400W power from the outlet is invalid

[serialk11r]

ok shoudl i trust swiftech or some yahoo in xs forum

again the 400w he was talking about was the computer NOT the TEC

http://www.xtremesystems.org/forums/showpost.php?p=2125948&postcount=93



talk about random assumptions

swiftech never said anything about TEC draws 320w, and therefore ur assumption about TEC draws 320W from PSU, then the PSU draws 400W power from the outlet is invalid

The TEC is rated at 15V MAX and 24A MAX...if you run it at 12V and it draws 18A, then well duh, it will have less than 216W cooling. Assuming seedomo is giving the TEC 12V, then it should be drawing 216W, (sorry last time I crunched my numbers incorrectly) and pulling something like 270W from the wall socket. The TEC there is a 226/320TEC...no denying that. But that doesn't mean it will always draw that much.

[theteamaqua]

The TEC is rated at 15V MAX and 24A MAX...if you run it at 12V and it draws 18A, then well duh, it will have less than 226W cooling.

yeah and where does it say 15V and 24A MAX?

[serialk11r]

yeah and where does it say 15V and 24A MAX?

Um, take a closer look: Thermoelectric module characteristics
U max I max DT max Q max L ×W×H(mm) R(Ω)
15.2 VDC 24 A >67 ° C 226.1 W 50 × 50 × 3.10 0.48
Measured power draw at 12 VDC: 18A
So, at 12V, it should draw 18A, but at 15.2V, it draws 24A (because at max volts it WILL draw max amps)
BTW if you didn't know, Umax is max voltage, Imax is max amperage, Qmax is highest possible cooling power.
The above characteristics ARE that of the 226/320 TEC. And please do not blab about the 70% efficiency thing (it is incorrect wording and that pisses me off), if you want to say 70%, then call it cooling power to power consumed ratio.

[seedomo]

I can maybe help a little with this measurement:
Multimeter applied to the two leads going to the tec reads 13.6Volts. I really don't want to test Amps since I would have to wire the meter in line with the tec.

[serialk11r]

I can maybe help a little with this measurement:
Multimeter applied to the two leads going to the tec reads 13.6Volts. I really don't want to test Amps since I would have to wire the meter in line with the tec.

So, you turned it to 13.6V? That way, its drawing a little more than 18A then. Probably about 20A.

[seedomo]

So, you turned it to 13.6V? That way, its drawing a little more than 18A then. Probably about 20A.
BTW, that's with the dial turned to max.

I'm pulling it apart, again, to change the chip to one that I know will clock higher. Hopefully I will get to a 24/7 4Ghz with the add cooling now.

[serialk11r]

Um, around 200W of cooling max (will get you to ambient if you give it 200W load)...you sure you want to give it that high of a load? I don't think the chip will hit 4ghz.

[seedomo]

Um, around 200W of cooling max (will get you to ambient if you give it 200W load)...you sure you want to give it that high of a load? I don't think the chip will hit 4ghz.

That what Swiftec recommended from what I gathered. Their own graph shows the Meanwell set at max voltage of 13.8 (while that is theoretical max voltage, I measured more like 13.6) The CEO from Swiftec was around here in another thread where he confirmed that recommendation as well. For the record, I've had it maxed since I installed it weeks ago.

Also, I've already got the other chip to 4.1 GHz albeit not very stable with 1.6 volts using only one 2x120 rad. Now with the added cooling I'm hoping to push the voltage for added stability.

[seedomo]

One more thing: if the measured wattage of the Meanwell A/C cord is 320 and the measured amps for the tec is 13.6, then cant we figure the amps to be 23.5? watts=axv

Um, around 200W of cooling max (will get you to ambient if you give it 200W load)...you sure you want to give it that high of a load? I don't think the chip will hit 4ghz.
Thermoelectric module characteristics
U max I max DT max Q max L ×W×H(mm) R(Ω)
15.2 VDC 24 A >67 ° C 226.1 W 50 × 50 × 3.10 0.48
Measured power draw at 12 VDC: 18A

I'm not following you here. What is your fear in giving the tec 23.5 Amp at 13.6 Volts since its within the spec of the tec? If you wondering why they included the "Measured power draw at 12v", its probably for people that are going to use a PC PSU on their tec.

[serialk11r]

One more thing: if the measured wattage of the Meanwell A/C cord is 320 and the measured amps for the tec is 13.6, then cant we figure the amps to be 23.5? watts=axv

Uh, how do you know the TEC is drawing 320W? It draws 320W at max volts, but at 13.6V (about 90% voltage) it should draw around 90% total amps as well, which is...24X90%=21.6. The PSU doesn't just stuff 320W into the TEC LOL.

[theteamaqua]

Um, take a closer look:
So, at 12V, it should draw 18A, but at 15.2V, it draws 24A (because at max volts it WILL draw max amps)
BTW if you didn't know, Umax is max voltage, Imax is max amperage, Qmax is highest possible cooling power.
The above characteristics ARE that of the 226/320 TEC. And please do not blab about the 70% efficiency thing (it is incorrect wording and that pisses me off), if you want to say 70%, then call it cooling power to power consumed ratio.

where did u find that then? and yes w/e u call it it draws 320 and ouputs 226 thats 70%

[serialk11r]

where did u find that then? and yes w/e u call it it draws 320 and ouputs 226 thats 70%

Go to swiftnets.com click on thermoelectric cooling, then click on MCW6500.

[seedomo]

Uh, how do you know the TEC is drawing 320W? It draws 320W at max volts, but at 13.6V (about 90% voltage) it should draw around 90% total amps as well, which is...24X90%=21.6. The PSU doesn't just stuff 320W into the TEC LOL.
I used the Killa-Watt plugged into the AC cord for the Meanwell. It says 320 and sometimes bounces a bit higher.

[theteamaqua]

Go to swiftnets.com click on thermoelectric cooling, then click on MCW6500.

yeah and how come it says it draws 18A at 12v when its less than 226W?

[serialk11r]

I used the Killa-Watt plugged into the AC cord for the Meanwell. It says 320 and sometimes bounces a bit higher.

Remember, the PSU wastes some electricity in converting the voltage!

[serialk11r]

yeah and how come it says it draws 18A at 12v when its less than 226W?

The TEC can run at higher voltages...they just decided to measure the current draw at 12V

[seedomo]

My head is starting to spin, should I just measure the damn voltage going to the tec so everyone is happy.

[theteamaqua]

The TEC can run at higher voltages...they just decided to measure the current draw at 12V

so without a aux PSU it sux right? ... cuz 216w only gets u like 70% of that which is like 151W .... or i can just get a single rail PSU with 30+A over 12v rails or is it not going to draw more than 18A ??

how do u measure power draw by PSU ? whats that thing called?

[theteamaqua]

My head is starting to spin, should I just measure the damn voltage going to the tec so everyone is happy.

yea try it on a single rail PSU with 30+A over 12v rail ... cuz i relaly dont wanna spend 150 on a aux psu

yeah if it eats 320/12 ~ 26A over 12v rail then its running at max , but if its only 18A then ... it means i gotta spend 150 on a aux psu

[seedomo]

I'm hooking up multi meter to the Tec to measure the amps, its a bit tricky cause I don't want to connection to slip since that wouldn't be so good for the CPU.

[theteamaqua]

I'm hooking up multi meter to the Tec to measure the amps, its a bit tricky cause I don't want to connection to slip since that wouldn't be so good for the CPU.

dont hurt urself :P

anyway can u tr it on a single rail PSU with 30+A over 12v rail?? and compare it to the maxwell aux PSU .. see if u get the same result ... temperature wise

[serialk11r]

A normal PSU's rails will probably be unstable because there isn't load on the 3.3v and 5v rails.
And remember you must swap out the wires for low gauge ones.

[theteamaqua]

A normal PSU's rails will probably be unstable because there isn't load on the 3.3v and 5v rails.
And remember you must swap out the wires for low gauge ones.

really?? how can i load 3.3v and 5v rails?? lol i jsut dont wanna spend 150 on a aux psu just for TEC

[seedomo]

Soldering gun in hand.gonna solder the line to the miltimeter. Blackberry in the other hand if your wondering.

[theteamaqua]

hey i also have a Laing D5 and a DDC in a single loop ... is that bad?? i heard it somewhere that u should only have 2 the same pumps in the loop..

lol i bought that DDC cuz it has pretty good max head .. 22ft vs 10ft of D5 i think

[serialk11r]

hey i also have a Laing D5 and a DDC in a single loop ... is that bad?? i heard it somewhere that u should only have 2 the same pumps in the loop..

lol i bought that DDC cuz it has pretty good max head .. 22ft vs 10ft of D5 i think

I think that causes problems...try putting them seperately.

[seedomo]

Before I push the power button, my meter says 10A max, so I guess I can't test it, right? This is a serious question, so please answer only if you know!

[theteamaqua]

I think that causes problems...try putting them seperately.

why? maybe i should start a thread in lc section

[serialk11r]

Before I push the power button, my meter says 10A max, so I guess I can't test it, right? This is a serious question, so please answer only if you know!

I would NOT try turning it on. Wires inside the meter are probably too thin.

[theteamaqua]

Before I push the power button, my meter says 10A max, so I guess I can't test it, right? This is a serious question, so please answer only if you know!

probably not ... i am not sure, but i think that means it cant measure anything above 10A ...

[seedomo]

What a piece of crap, $99. Ive already soldered to the Tec. Should I try it, just for a second?

[serialk11r]

I wouldn't try that, it being 99$...

[seedomo]

BOOM! jj. I tried it and the meter just read of I think the letter were.

[serialk11r]

BOOM! jj. I tried it and the meter just read of I think the letter were.

How fast did you turn it off? I'm hoping you did it fast enough so that the components on the inside didn't get damaged.

[seedomo]

Just a few seconds.
Meter still works.

[seedomo]

I changed the chip out to my E6600 L644G295. This is the one I THOUGHT I used when i went to 4.1 using only 1.6 volts. Hmmm... I've got a third chip, maybe its that one.

Temps are pretty reasonable for giving it 1.7 volts :shock: These chips must not be pulling the same amount of amps (I have a very limited understanding of elec.:D ) since the loaded wattage was 430, which is similar to the wattage on my other chip that was only using 1.43 volts.

In other words:
L644G295 @ 3.825 Ghz @ 1.70V = 430 watts
Core temp 34c, CPU temp 21c
L627B155 @ 3.750 Ghz @ 1.43V = 415 watts
Core temp 22c, CPU temp 9c

http://www.domophoto.com/3825@1.7 430wattsIdle.jpg

[seedomo]

Estimated use of the 12V rail
Fans .22 x 11 =2.4
Pumps 2 x 2 =4
10 Raptors HD 0.375 x 10 =3.75
2 Segate HD 2.8 x 2 =5.6 (I think this is the total accross all rails but I'll use it anyway.)
DVD 1.6 x 1 =1.6
6600GT 3.0 x 1=3
MB .3 x 1=.3
Total w/o CPU is 20.65
If my CPU gets up to 225 watts then it will be using 18.75Amps

CPU 18.75
System 20.65
Total 39.4
PSU rating for 12V rail is 40A

Looks like I will be getting a new PSU, damn this is starting to cost! I've learned much more than I ever thought I would about processors and electricity as well.

Another thing, I've been monitoring the voltage on the 12v rail and at idle it 11.79 and under heavy load it goes to 11.77. Should I understand this to mean that I DONT need a new PSU cause its not dropping much, is that drop in voltage something to worry about? I guess I'm wanting to know if I can use voltage measurements as an indicator that I'm approaching max amperage.

[shimmishim]

I think you're right about the amperage being pulled by the chip.

Seeing as it that volt x amp = wattage and the B chips have a VID of or around 1.25 while the A's (and maybe the G's not sure) have VID of or around 1.325 but the same amount of wattage at 65, my guess is that the amps are different.

[Holst]

If you read the PDF I posted the voltages are speced per-chip.

As I dont overclock on C2D I dont exactly understand whats ajustable, and whats not.

LOL @ seedomo with 10amp multimeter. Your lucky you didnt blow the fuse, it must have some overcurrent protection in there.
Multemeters for measuring more than 10amps DC are expensive, you might be able to buy a clamp meter attachment for your current meter that will read higher amps if your interested.

You cant really use VxA=Watts for processors as they are not fixed resistances. And there are voltages other than vcore to take into account as well.

[theteamaqua]

Estimated use of the 12V rail
Fans .22 x 11 =2.4
Pumps 2 x 2 =4
10 Raptors HD 0.375 x 10 =3.75
2 Segate HD 2.8 x 2 =5.6 (I think this is the total accross all rails but I'll use it anyway.)
DVD 1.6 x 1 =1.6
6600GT 3.0 x 1=3
MB .3 x 1=.3
Total w/o CPU is 20.65
If my CPU gets up to 225 watts then it will be using 18.75Amps

CPU 18.75
System 20.65
Total 39.4
PSU rating for 12V rail is 40A

Looks like I will be getting a new PSU, damn this is starting to cost! I've learned much more than I ever thought I would about processors and electricity as well.

Another thing, I've been monitoring the voltage on the 12v rail and at idle it 11.79 and under heavy load it goes to 11.77. Should I understand this to mean that I DONT need a new PSU cause its not dropping much, is that drop in voltage something to worry about? I guess I'm wanting to know if I can use voltage measurements as an indicator that I'm approaching max amperage.

how well do u think the TEC being run from a regular PSU with single rail 30+A??

is that gonna hurt TEC's performance??

[serialk11r]

how well do u think the TEC being run from a regular PSU with single rail 30+A??

is that gonna hurt TEC's performance??

What do you mean by performance? A 12V computer PSU will give it 12V (more or less) so you'll be getting around 70% power, which isn't exactly a bad thing (higher efficiency). Just remember, with a normal PSU you can't change voltage, and you need to switch out the wires.

[theteamaqua]

well if it only eats 18A @ 12v .. thats barely 216w ... thats not even close to 226W , what the TEC pumps to cool off ...

so if it eats say 26.66A ... over 12v rails that will be better

[serialk11r]

No lol, the wires that you connect it to most likely will not be rated for 18A current... You'll have to do some modding.

[theteamaqua]

No lol, the wires that you connect it to most likely will not be rated for 18A current... You'll have to do some modding.

what? so ur saying i cant get all 226W with a normal PSU??

[Holst]

This should be in your OWN thread, not spamming up this one.

Ive allready answered some of your questions IN YOUR OWN THREAD and you didnt reply.

I have also posted a reply about the PSU wires ALLREADY, which its pretty obvious neither of you have read.

[seedomo]

Holst what do you think about my assumption about getting a PSU? I've chosen the Silencer 750, do you think I'm waisting more money? My reasoning is this: It seems that as my overclocks are stopping around 3.8 and they all seem to be around the same wattage.

[Holst]

Holst what do you think about my assumption about getting a PSU? I've chosen the Silencer 750, do you think I'm waisting more money? My reasoning is this: It seems that as my overclocks are stopping around 3.8 and they all seem to be around the same wattage.

Powersupplies are very difficult to understand when overclocking.

Its important to measure voltages using a good quality multimeter. I allways prefered measuring at the ATX connector by soldering pins to a couple of probes and poking them into the back of the wires. (measure 12v on the AUX connector)

If your 12v is running at 11.75 then its a bit low (I would run my 12v at 12.5 or so) but its not dropping much.

Can you ajust the voltages on this PSU?

Buying a new PSU may help your overclock, but it may do absolutely nothing at all :(
When buying one make sure you do plenty of research (I would always copy what MrIcee or Macci was running if I couldnt decide :P )

Do you have any other PSU you can try?
Even a "worse" PSU might indicate if PSU is affecting your OC.
If with a worse PSU your overclock is just as good then its unlikely a new PSU will help.

Best bet is to borrow a PSU from somebody local for a quick test, swapping PSU's for testing is pretty quick and easy. And will save you allot of money (or cost you allot when if you get a higher OC with a new PSU)

Good luck.

[Pete]

PSU can and will vary your overclock. Had a 600w coolermaster thing, max clock i could get was 3.8, got a PC+P 510w and got 4.3 and my volts never droped or anyhting, totaly soild all the time

[Roger_D25]

Holst,
I couldn't have said it better myself! Choosing a new power supply can be tricky seedomo as the numbers on the side of the supply don't tell the whole story! You could have 10 different 800w power supplies and they will all probably behave differently. As Holst said it would be best if you could properly test your current supply before making another large investment because there is a chance your supply may be absoluty fine? Although I think that Antec is a reputable company there are not many serious overclockers/benchers who run them?

Just do as much research as you can before doing anything (researching is free)! As hardware has become more and more power hungry and enthusiest have become more and more knowledgable about clocking, power supplies have become very important, infact could be one of the most important pieces of hardware you can buy (this is just my opinion).

Holst brought up a great point, if you have a difficult time deciding on a new power supply (assuming that you even need a new one) take a look at some of the other members supplies who you look up to. I think you'll notice a trend on the makes and wattage ratings among the members who are serious about overclocking!

[serialk11r]

Well I do see people bashing Antec PSUs but I love mine :D the fan is so damn quiet...

[Philly_Boy]

Do you know how many TECs a Meanwell S320 can handle? If I were to use a MCW6500T and two MCW60Ts, do I need two or three Meanwells ? How am I supposed to mount three Meanwell PSUs inside a case and yet exhaust all the pSU heat if they are stacked on top of each other?

Many thanks.
I could only hook up my MCW6500T to my SP320-13.5. If I tried to add a MCW60T then the Meanwell became unstable.
Just too hard to do with the neoprene foam around the socket. I would also probably need to get some flat probes as well. Until then I will report some water temps for ya in a bit. I'm probably going to add the Swiftec 220 rad to the loop and see if it affects the overall results.
You'll get better results from placing the two pumps like this: res>pump 1>pump 2>rad>MCW6500-T. Use the second rad along with another pump (DDC+/MCP355's are nice with Petra's top) to do the gpu and NB/SB. I did something similar and with two MCP355's in series and my water temps came down to 23c/25c (Core0/Core1 TAT loaded on my E6300 doing 3.5ghz @ 1.55vcore) cooled with ambient air running over my PA120.3

PA120.3 with 2x MCP355's with petra's top into MCW6500T:
Ambient=19c
No cold air induction box (it was a warm night)
Boot up CPU idle= -9c (bios)
CPU Core temp= 23c (TAT loaded 1hr)
CPU temp= 11c (my own flat digital K style temp probe on IHS perimeter of top)
H2O temp idle/loaded= 24c idle/28c loaded

yea try it on a single rail PSU with 30+A over 12v rail ... cuz i relaly dont wanna spend 150 on a aux psu

yeah if it eats 320/12 ~ 26A over 12v rail then its running at max , but if its only 18A then ... it means i gotta spend 150 on a aux psu
I tried to hook up the 6500T to my PC P&C Turbo 1K SR and couldn't get the same boot up idle or loaded temps on the CPU core. The Meanwell does an excellent job, albeit on just one TEC block per PSU. My power consumption using a killa-watt wat 322-340 watts (low to high on the Meanwell's voltage adjustment)

I have had the best results with the 120.3 with 2x DDC+ pumps with a 18 quart cooler running 10 ft of tubing thru ice laden salted water. the water temps hover around 23F. Not quite done gathering data yet on this set up...but I have had boot up idle temps in the range of what a single stage phase unit is capable of...and the ice lasts for 2 days if the lid is properly closed. A poor man's phase unit...

[serialk11r]

I could only hook up my MCW6500T to my SP320-13.5. If I tried to add a MCW60T then the Meanwell became unstable.

You'll get better results from placing the two pumps like this: res>pump 1>pump 2>rad>MCW6500-T. Use the second rad along with another pump (DDC+/MCP355's are nice with Petra's top) to do the gpu and NB/SB. I did something similar and with two MCP355's in series and my water temps came down to 28c (TAT loaded on my E6300 doing 3.5ghz @ 1.55vcore) cooled with ambient air running over my PA120.3

PA120.3 with 2x MCP355's with petra's top into MCW6500T:
Ambient=19c
No cold air induction box (it was a warm night)
Boot up CPU idle= -9c (bios)
CPU Core temp= 23c (TAT loaded 1hr)
CPU temp= 11c (my own flat digital K style temp probe on IHS perimeter of top)
H2O temp idle/loaded= 24c idle/28c loaded


I tried to hook up the 6500T to my PC P&C Turbo 1K SR and couldn't get the same boot up idle or loaded temps on the CPU core. The Meanwell does an excellent job, albeit on just one TEC block per PSU. My power consumption using a killa-watt wat 322-340 watts (low to high on the Meanwell's voltage adjustment)

I have had the best results with the 120.3 with 2x DDC+ pumps with a 18 quart cooler running 10 ft of tubing thru ice laden salted water. the water temps hover around 23F. Not quite done gathering data yet on this set up...but I have had boot up idle temps in the range of what a single stage phase unit is capable of...and the ice lasts for 2 days if the lid is properly closed. A poor man's phase unit...

Wow, pretty nice results... Poor man's phase? You call that poor man's phase when you spent how much? :D I'm still saying yuck at the bad efficiency but those are pretty low temps.