Cooling a pentium 4 3.4GHz CPU

[Santalum]

Continuing on from my previous post about this but pursuing a different method.......

 

I have decided to have a go at making a water cooling system.

 

My idea is to use a piece about 2.5 inch diameter copper pipe to make a cylinderical water filled chamber, i.e. silver solder the ends shut with sheet copper with an inlet and an outlet at the top.

 

Would generating turbulence in the chamber result in better heat transfer into the total volume of the water in the chamber, i.e. as in stirring a pot of water on the stove, or would it be better to generate a venturi effect in the chamber, i.e. with a flatter chamber?

[CaptainPanic]

From your other thread, I understand that you want to cool the computer processor (and in fact, the whole computer case) with a cooling element. I.e. put something cold into the computer, so it will absorb the heat from the air inside the computer. Correct?

 

Dedicated systems for computers

You're about to spend some money anyway, so my primary advice would be to go to a computer store, and ask them how you can improve cooling. There are dedicated systems for computers! That's better than tinkering with ammonia (toxic, smelly!) inside your living room.

 

Google for [Liquid Cooling Processor], and you will instantly find solutions. My google tells me it can be as cheap as $57, which is probably the same as what you're looking at with your plans. Shop well, and find it even cheaper.

 

Water condensation

Secondly, you risk condensation of water (moisture from air) on the copper cooling tube you are about to make. That can drip onto your computer hardware. Water close to electronic hardware is a bad idea. Also, leakage is a risk, but you may be a good plumber so I will not comment on that any more than this.

 

Cool most where it's needed.

I am assuming that the air in the computer case can easily get to 50 degrees Celsius, otherwise you wouldn't be worried.

But the processors get a LOT warmer - they have their own small fans and radiators. Apply your cooling directly to the processor, and solve your problem at the root.

There exist dedicated liquid coolers for processors.

 

Delta T (temperature difference)

You are proposing to have a second heat exchanger to cool the coolant, like any refrigerator. But why not simply bubble some air through the hot water? Then you can leave out the ammonia. Some water will evaporate, and that will give cooling. Ok, your coolant will be around 20-30 degrees Celsius, instead of 10 or lower. That only means you need to increase your water flow by 50% or so (just an order of magnitude estimate - no calculation). If you're building it yourself, and you don't really know what you're doing: simple is best.

Btw, you probably want to calculate how much air you need.

 

The heat exchange bottleneck

The heat will travel the following path:

The bulk of the air --> a thin stagnant layer of air around the cooler

a thin stagnant layer of air around the cooler --> the copper wall of the cooler

the copper wall of the cooler --> a thin stagnant layer of water inside the cooler

a thin stagnant layer of water inside the cooler --> the bulk of the water

 

Without even calculating, I can guess that the bottleneck of your heat exchanger is the thin stagnant layer of AIR around the heat exchanger. And that will get worse as the copper pipe gets dusty/dirty.

Making the water turbulent will reduce the thickness of the thin stagnant layer of water inside the heat exchanger, but that's not the bottleneck.

To make this thing more efficient, you need a fan on the outside, blowing hot air against your heat exchanger.

 

Concluding

If this is just a hobby project on an old PC, then go for it. It's fun to play with heat transfer, although it can be difficult.

If you're actually trying to improve a good (expensive?) PC, then don't. Go for the professional solutions at your local computer-hardware store. I'm not a computer expert, but 3.4 GHz sounds like it's not a really old crappy PC. I wouldn't play with water near such a PC.

[Santalum]

From your other thread, I understand that you want to cool the computer processor (and in fact, the whole computer case) with a cooling element. I.e. put something cold into the computer, so it will absorb the heat from the air inside the computer. Correct?

 

Dedicated systems for computers

You're about to spend some money anyway, so my primary advice would be to go to a computer store, and ask them how you can improve cooling. There are dedicated systems for computers! That's better than tinkering with ammonia (toxic, smelly!) inside your living room.

 

Google for [Liquid Cooling Processor], and you will instantly find solutions. My google tells me it can be as cheap as $57, which is probably the same as what you're looking at with your plans. Shop well, and find it even cheaper.

 

Water condensation

Secondly, you risk condensation of water (moisture from air) on the copper cooling tube you are about to make. That can drip onto your computer hardware. Water close to electronic hardware is a bad idea. Also, leakage is a risk, but you may be a good plumber so I will not comment on that any more than this.

 

Cool most where it's needed.

I am assuming that the air in the computer case can easily get to 50 degrees Celsius, otherwise you wouldn't be worried.

But the processors get a LOT warmer - they have their own small fans and radiators. Apply your cooling directly to the processor, and solve your problem at the root.

There exist dedicated liquid coolers for processors.

 

Delta T (temperature difference)

You are proposing to have a second heat exchanger to cool the coolant, like any refrigerator. But why not simply bubble some air through the hot water? Then you can leave out the ammonia. Some water will evaporate, and that will give cooling. Ok, your coolant will be around 20-30 degrees Celsius, instead of 10 or lower. That only means you need to increase your water flow by 50% or so (just an order of magnitude estimate - no calculation). If you're building it yourself, and you don't really know what you're doing: simple is best.

Btw, you probably want to calculate how much air you need.

 

The heat exchange bottleneck

The heat will travel the following path:

The bulk of the air --> a thin stagnant layer of air around the cooler

a thin stagnant layer of air around the cooler --> the copper wall of the cooler

the copper wall of the cooler --> a thin stagnant layer of water inside the cooler

a thin stagnant layer of water inside the cooler --> the bulk of the water

 

Without even calculating, I can guess that the bottleneck of your heat exchanger is the thin stagnant layer of AIR around the heat exchanger. And that will get worse as the copper pipe gets dusty/dirty.

Making the water turbulent will reduce the thickness of the thin stagnant layer of water inside the heat exchanger, but that's not the bottleneck.

To make this thing more efficient, you need a fan on the outside, blowing hot air against your heat exchanger.

 

Concluding

If this is just a hobby project on an old PC, then go for it. It's fun to play with heat transfer, although it can be difficult.

If you're actually trying to improve a good (expensive?) PC, then don't. Go for the professional solutions at your local computer-hardware store. I'm not a computer expert, but 3.4 GHz sounds like it's not a really old crappy PC. I wouldn't play with water near such a PC.

 

Thanks captain but I am no longer going down the route of a refridgeration type element for the whole case as it is impractical and possibly dangerous given the easily available refidgerants are highly flammable - plenty of source of sparks if any vapour was to leak out.

 

Am going down the route of a water cooled system similar to the professional ones. Problem is that the Pentium 4 is a socket 478 and socket 478 water cooling systems are difficult to come by - the ones easily available are socket 775 and more recent.

 

Besides it is not my primary PC and I thought it would be an interesting challenge to try and create a system as I love making things.

 

So returning to my question which would be better for heat transfer? Turbulance or venturi?

 

If I transfer heat from the cpu outside the case then the interior of the case will not heat up so much so no need for a case cooler.

Edited January 24, 2012 by Santalum

[CaptainPanic]

So returning to my question which would be better for heat transfer? Turbulance or venturi?

As I explained briefly in my previous post: the heat transfer resistance is in the air around the copper pipe, not the water inside it. So, the answer to your question: it doesn't really matter.

 

Just make the water flow a little with a fan ("forced convection" is your scientific term of the day).

 

Balance the water flow to the heat you need to absorb with the water (i.e. the computer's power). Make sure the water flows so much that it has heated up noticeably, but not too much. 10 degrees Celsius or so... this is a designer's decision. I'll leave it to you.

 

How much power is your computer? Peak power of computers can be in the order of magnitude of 500 W (= 500 J/s). That's probably the CPU and also the video card all running at max. power.

If you allow the water to heat by 10 degrees, and you assume that you remove ALL the heat with the water (which is not true), and you assume your computer is always running at max. power (again not true), your total water flow should be:

 

P = F*Cp*dT

or:

F = P/(Cp*dT)=500/(4180*10) = 0.011 liter/s, or 0.7 liter/minute (about 0.19 gallon/minute).

 

In reality, you will probably need less water, or heat up your water less.

 

Again, doing fancy stuff to the water of the heat exchanger will not make it noticeably more efficient. A fan on the outside will though. I'm sorry not to give you your answer... but I cannot change heat transfer phenomena.

[Santalum]

As I explained briefly in my previous post: the heat transfer resistance is in the air around the copper pipe, not the water inside it. So, the answer to your question: it doesn't really matter.

 

Just make the water flow a little with a fan ("forced convection" is your scientific term of the day).

 

Balance the water flow to the heat you need to absorb with the water (i.e. the computer's power). Make sure the water flows so much that it has heated up noticeably, but not too much. 10 degrees Celsius or so... this is a designer's decision. I'll leave it to you.

 

How much power is your computer? Peak power of computers can be in the order of magnitude of 500 W (= 500 J/s). That's probably the CPU and also the video card all running at max. power.

If you allow the water to heat by 10 degrees, and you assume that you remove ALL the heat with the water (which is not true), and you assume your computer is always running at max. power (again not true), your total water flow should be:

 

P = F*Cp*dT

or:

F = P/(Cp*dT)=500/(4180*10) = 0.011 liter/s, or 0.7 liter/minute (about 0.19 gallon/minute).

 

In reality, you will probably need less water, or heat up your water less.

 

Again, doing fancy stuff to the water of the heat exchanger will not make it noticeably more efficient. A fan on the outside will though. I'm sorry not to give you your answer... but I cannot change heat transfer phenomena.

 

 

0.7L/min seems to be in the ball park of the pumps I have seen recommended for cpu cooling systems.

 

Also of the order of 2W or electrical power and roughly 200mA draw.

Edited January 24, 2012 by Santalum

[CaptainPanic]

0.7L/min seems to be in the ball park of the pumps I have seen recommended for cpu cooling systems.

That's because the water flow can be minimal. No need to recirculate it to induce turbulence.

 

Anyway, at 0.011 liter/s, or 0.000011 m3/s, and a diameter of 2.5 inch (0.0635 cm diameter, area perpendicular to flow = 0.0031 m2), your linear velocity will be 0.0035 m/s. Your Reynolds number (to see if it is turbulent or not) is:

 

[math]\mathrm{Re} = {{\rho {\bold \mathrm v} L} \over {\mu}} = {{1000 \cdot {0.0035} \cdot 0.0635} \over {0.001}} = 220[/math]

 

... which to make a long story short means it will be laminar flow. There will probably be some turbulence around inlets and outlets though.

But the lack of turbulence will be no problem.

 

You have all kinds of other problems I already mentioned.

[Santalum]

That's because the water flow can be minimal. No need to recirculate it to induce turbulence.

 

Anyway, at 0.011 liter/s, or 0.000011 m3/s, and a diameter of 2.5 inch (0.0635 cm diameter, area perpendicular to flow = 0.0031 m2), your linear velocity will be 0.0035 m/s. Your Reynolds number (to see if it is turbulent or not) is:

 

[math]\mathrm{Re} = {{\rho {\bold \mathrm v} L} \over {\mu}} = {{1000 \cdot {0.0035} \cdot 0.0635} \over {0.001}} = 220[/math]

 

... which to make a long story short means it will be laminar flow. There will probably be some turbulence around inlets and outlets though.

But the lack of turbulence will be no problem.

 

You have all kinds of other problems I already mentioned.

 

I was toying with the idea of soldering two inner smaller tubes vertically beneath the inlet and out let and with holes in the bottom on opposite sides. That should create a bit of turbulance in the chamber above the CPU at least, don't care about the tubes connecting it all together or course.

 

From what I have been reading on the computer forums etc, P4 above 3 GHz are very prone to over heating if you don't get the fans etc just right in case. I have a P4 3GHz and have had no problems with a stock standard Al CPU heat sink and fan plus pretty ordinary case fans. Above 3GHz P4 are useless for lap tops since they drain the battery too quickly.

 

With the P4 3.4GHz CPU I have, I can play around with over clocking it if I wish. But to do that you need an efficient water cooling system to keep the CPU within the prescribed temp limit.

 

With socket 775 and newer sockets, intel has gone back to around 2.6GHz but are now making CPUs with dual and quad cores. Effectively chips with multiple processors in them, and bigger memory caches, that gives the same or better performance as P4 but with lower power consumption.

 

Have recently purchased a Core 2 duo PC and the Core 2 socket 775 series will be the next to become redundant technology before too much longer. P4s have been redundant for a while now. The cream of the intel crop are now the I3, I5 and I7 systems.

 

All the nuff nuffs prefer to pay top dollar for the latest technology, but I prefer to stay a few steps behind the latest as you get everything very cheaply off ebay etc.

 

P.S. I have been reading about pipe coolers for P4s. These consist of vertical oriented pipes that contain probably water at low pressure and a heat exchanger and fan at the top. The water vapourises in a chamber above the CPU and caries the heat up the tubes to the heat exchanger where it cools and condenses and runs back down to the chamber. But these don't work so well in a tower case where the mother board is vertically oriented.

Edited January 24, 2012 by Santalum

[CaptainPanic]

I was toying with the idea of soldering two inner smaller tubes vertically beneath the inlet and out let and with holes in the bottom on opposite sides. That should create a bit of turbulance in the chamber above the CPU at least, don't care about the tubes connecting it all together or course.

I'm sorry if this sounds rude, but did you read what I wrote at all?

 

Turbulence inside your copper tube does not matter. You will not be able to see it, you cannot feel it, and you will not notice any effect. So, why bother with it?