eXtreme Liquid Cooling - Build Log + Overclocking

[Jokester]

He's using a cold plate on both side, so I don't see it being that much of an issue.

I hope you're using it in a well ventilated place Pneumonic, it'll be like having a 3bar fire in your room

 

[zipdogso]

He's using a cold plate on both side, so I don't see it being that much of an issue.

Oh I didn't see he was going to use copper plates between the heat exchanger plates if hes is the issue is less as you say.
But I still think the Dt problem hasn't been viewed properly.

If you get the performance charts for your TECs looking at the volts/amps first find the current for 12v at 30 Dt.
Then go to the chart you had in your post read up from the amp rating for your operating point and stop when you get to the 30º Dt line read to the left to find your maximum Q at your operating point and I am afraid you will find 8 TECs totals rather less than 12xx watts you calculated.

Your rough calc of 65% Q at 12v is only at Dt=0 (the Qmax value is also quoted at Dt=0.) and of course Dt at 0 means no cooling. A Dt of 0 means your hot and cold sides are the same temp hence no cooling.

I hope you're using it in a well ventilated place Pneumonic, it'll be like having a 3bar fire in your room

As the heat coming off the hotside is generally double the cooling even a modest 500w cooling is 1Kw coming off the hotside.

 

[1kiwi]

^^^^ indeed although thats not exactly a bad thing in this weather. I think US2 uses an external radbox. Have you thought about that (good for summer) Pneumonic?

 

[Pneumonic]

Oh I didn't see he was going to use copper plates between the heat exchanger plates if hes is the issue is less as you say.
But I still think the Dt problem hasn't been viewed properly.

If you get the performance charts for your TECs looking at the volts/amps first find the current for 12v at 30 Dt.
Then go to the chart you had in your post read up from the amp rating for your operating point and stop when you get to the 30º Dt line read to the left to find your maximum Q at your operating point and I am afraid you will find 8 TECs totals rather less than 12xx watts you calculated.

Your rough calc of 65% Q at 12v is only at Dt=0 (the Qmax value is also quoted at Dt=0.) and of course Dt at 0 means no cooling. A Dt of 0 means your hot and cold sides are the same temp hence no cooling.



As the heat coming off the hotside is generally double the cooling even a modest 500w cooling is 1Kw coming off the hotside.

Zipdogso,

Yeah now I have done the maths my 1200 watt first impression was a bit off.. hence why I went through the exercise of 'doing' it properly on the maths front..

I am getting near enough the same figures out with my calculations as I get with Ultrasonics calculator (for the parallel stage) so I am reasonably happy with 'ball park' figures.

I am reasonably confident it should work! although until its all put together and some tuning done temperatures are guesstimates

 

[1kiwi]

When do you think you will be able to put it together? Assume your just waiting for the block to come from NZ?

 

[Pneumonic]

I am waiting for the heat exchanger plates to come from the US.. along with some gear for the water loop! Hopefully before I go on holiday at the start of Feb..

 

[Ultrasonic2]

Hi guys and in particular Pneumonic.

It's great to see someone so excited by TECs as Pneumonic

I've Sent your block today so it should be there in a week id say ..


i've read through this tread and i'd like to pic up one one of the graphics you made.

Though this is a great graphic it doesn't seem to be anywhere near the truth.
It seems that you are excluding the reduction of QMAX and dTMAX when the hot side of the lower direct direct die block is cools below 300Kelvin

Adding these factors in would reduce the estimated loaded delta.

Probably the biggest problem with this graphic is your not accounting for a change in the hot to cold side delta once you apply a load to it.

IE a 300 watt QMax TEC ran at Umax and Imax with a hot side of 300k will ether move 0 watts to dTmax (69c) or move 300 watts to a delta of 0

For the same given input power as you increase the load applied to the cold side, the resulting delta will reduce as you increase the load.


In your graphic you are implying that the above is not a factor and you seem to have the deltas between the hot and the cold sides of the TEC's at a constant ish as you increase the load.


sooooo unfortunately your expectations may not be realized

Anyway keep up the great work and i look forward to your coming posts

 

[sambot]

interesting thread

 

[Dominor]

You have honored the overclocking Gods my son.

 

[Pneumonic]

Snip..

Ultrasonic,

Thank you for your input its much appreciated..

The calculations were based on an increasing input power to the units as the heat load was increased, more than anything to be as 'power efficient as possible' (I.e the QMAX of the system @ 125 TDP is a lot less than at 300 TDP). Thats why the graph is showing a constant-ish delta, other than towards the upper TDP's where the hotside temperatures on the chilled liquid loop start to increase as the water loop beings to have 'issues' dispersing the heat, is this incorrect? Assuming that the water loop can keep the hotside of the chilled liquid loop at a constant-ish temperature then I should maintain the same delta at an increasing TDP if I am proportionally increasing the QMAX? That was my understanding?

Edit: To make sense!

 

[1kiwi]

"hotside of the chilled liquid loop at a constant-ish temperature " - at the temps you are predicting what temp would you need to maintain? (assume this depends on the overclock???)

 

[Pneumonic]

I am expecting (from memory) a + 2c to +15c above ambient temperature depending on the load

 

[zipdogso]

Ultrasonic,

Thank you for your input its much appreciated..

The calculations were based on an increasing input power to the units as the heat load was increased, more than anything to be as 'power efficient as possible' (I.e the QMAX of the system @ 125 TDP is a lot less than at 300 TDP). Thats why the graph is showing a constant-ish delta, other than towards the upper TDP's where the hotside temperatures on the chilled liquid loop start to increase as the water loop beings to have 'issues' dispersing the heat, is this incorrect? Assuming that the water loop can keep the hotside of the chilled liquid loop at a constant-ish temperature then I should maintain the same delta at an increasing TDP if I am proportionally increasing the QMAX? That was my understanding?

Edit: To make sense!

I agree with Ultrasonic2

Unfortunately every spec of a TEC is fluid that is why specs are always quoted maximums.
In the general course of running as temps, loads change so do the specs nothing remains constant.
As you increase the Q i.e. apply more heat/load to the coldside the Dt will change, the hotside temp will change and the coldside will also change and consequently the cooling level will change...you can't hope for anything to be constant.

And as another point it is my understanding that TECs generally have a limit of -50ºC I have yet to ascertain whether this is an operating limit or a cooling limit.

 

[Pneumonic]

I have seen documentation on multi-stage TECs and minimum cooling which drops well below -100c for 3/4 stages..

The graphs were never meant to be precise just an estimation, I guess the only way is to try and find out It's keeping me out of trouble at any rate

 

[zipdogso]

I have seen documentation on multi-stage TECs and minimum cooling which drops well below -100c for 3/4 stages..

The graphs were never meant to be precise just an estimation, I guess the only way is to try and find out It's keeping me out of trouble at any rate

Yes but you havent SEEN multistage TEC's....

At 4 stage they have a Dt of 111ºC that's why they go so low but they are absolutely tiny and I have yet to see a 3/4 stage in any manufacturers catalogue anywhere in the world bigger than about 45w Qmax. The reason is that there is an anomaly with stacking TEC's that people often just cannot grasp...I am not going into now but just take it from me one CANNOT stack TECs of the size we use for PC's to produce multistage TEC's to get really low temps.

A single stage TEC has a limit in the region of -50ºC now I think about it, it must be an operating temp because the cooling temp is determined by the Dt. So now thinking about it again it must be the limit of the hotside.


With regard to calcs your main problem is Dt. At an operating point of 12v your TECs have a theoretical maximum of Dt 36ºC. Theoretical because a.) the reduction of Dt is not linear so it might be a bit more and b.) in practice you will not achieve the figure what ever it is....same as in practice you will not achieve the 72ºC Dtmax of the TEC. (that is right isn't it, didn't check.)
If your planning hotside temps of ambient +2ºC to +15ºC...I don't know what your ambient is so at a guess i'd say 20ºC your hotside is 22 - 35ºC.
whatever your theoretical Dt is at the operating point of 12v (possibly 36-40ºC.) it will fall from this figure as your load increases and rise back up to it as the load decreases. So when your load is low and hotside is 22ºC you will go subzero but when the load is high, your theoretical 35ºC hotside the Dt will of fallen, lets say for arguements sake the fall is only small to 30ºC your coolant is now +5ºC.
If you are using that said coolant to cool the hotside of the TEC on the CPU you are compounding your problems, you will notice significant lack of cooling which since at that point you have a high load it is not what you want.

That's why generally if you want to go for very low temps hotside cooling is paramount as I said earlier and to be honest i don't think will be sufficient.

I appreciate the need for "ballpark" figures I run TECs myself but I feel your graphs are rather on the high side of optimistic and so too does Ultrasonic2.

 

[1kiwi]

Way to **** on his camp fire.

 

[TaKeN]

Insane! I love it!

Looking forward to more updates

 

[Pneumonic]

Yes but you havent SEEN multistage TEC's....

At 4 stage they have a Dt of 111ºC that's why they go so low but they are absolutely tiny and I have yet to see a 3/4 stage in any manufacturers catalogue anywhere in the world bigger than about 45w Qmax. The reason is that there is an anomaly with stacking TEC's that people often just cannot grasp...I am not going into now but just take it from me one CANNOT stack TECs of the size we use for PC's to produce multistage TEC's to get really low temps.

A single stage TEC has a limit in the region of -50ºC now I think about it, it must be an operating temp because the cooling temp is determined by the Dt. So now thinking about it again it must be the limit of the hotside.


With regard to calcs your main problem is Dt. At an operating point of 12v your TECs have a theoretical maximum of Dt 36ºC. Theoretical because a.) the reduction of Dt is not linear so it might be a bit more and b.) in practice you will not achieve the figure what ever it is....same as in practice you will not achieve the 72ºC Dtmax of the TEC. (that is right isn't it, didn't check.)
If your planning hotside temps of ambient +2ºC to +15ºC...I don't know what your ambient is so at a guess i'd say 20ºC your hotside is 22 - 35ºC.
whatever your theoretical Dt is at the operating point of 12v (possibly 36-40ºC.) it will fall from this figure as your load increases and rise back up to it as the load decreases. So when your load is low and hotside is 22ºC you will go subzero but when the load is high, your theoretical 35ºC hotside the Dt will of fallen, lets say for arguements sake the fall is only small to 30ºC your coolant is now +5ºC.
If you are using that said coolant to cool the hotside of the TEC on the CPU you are compounding your problems, you will notice significant lack of cooling which since at that point you have a high load it is not what you want.

That's why generally if you want to go for very low temps hotside cooling is paramount as I said earlier and to be honest i don't think will be sufficient.

I appreciate the need for "ballpark" figures I run TECs myself but I feel your graphs are rather on the high side of optimistic and so too does Ultrasonic2.

Zip,

Ok thank you I think I understand now..

Does this look more accurate?

So lets assume that I keep the hotside temperature the same and we ignore loss of cooling due to contacts etc... I should expect the following..

So I would be better to consider the following, running 12 19911 to cool the 62mm such as:

So according to Martins Lab I can keep to under a +4c temp rise on the water temp on option 2, so -52c/-36c cold side @ 20c ambient and 100/200 TDP respectively less loss from contacts, etc..

 

[surge]

Something seems to have happened to the images in your last post at my end....

 

[Pneumonic]

Yeah I can't upload them from work will have to do it when I get home