Learning heatsinking from computer overclockers

Status
Not open for further replies.
Jump to Latest
#23
Hi,
What sort of thermal compound do you use between the components and the heatsink? This seems to be the only thing left out of this discussion at the moment, and I've found it to be one of the most important things in getting a successfull and reliable overclock. The CPU manufacturers own heatsinks and fans, which come with not very good thermal compound already on, benefit greatly from removing this and fitting with some arctic silver or similar good compound. I only have experience with Athlons however, I don't think it's quite so crucial with Pentiums as they have a heat spreader fitted to the CPU die already to increase the efficiency of the CPU die/heatsink heat path. With the Athlons though I've noted drops of around 15 - 20degC just from using decent stuff. Fit a decent copper skived heatsink though and the effect of the thermal compound can be even more pronounced. Also do you all go to the same attention to detail of the quality of the mating surfaces? I understand that the energy intensity may be greater with a CPU (though SiSoft has never measured more than around 85 - 90W temp dissipation from the CPU, even with the core voltage increased to 1.8v) but still, all these details must help in getting the heat accross to the heatsink.

Steve
 
#24

Attachments

  • pour4.jpg
    pour4.jpg
    21.5 KB · Views: 379
#25
A simple aluminum heatsink and fan is probably going to be the best and cheapest way.

The reason copper base aluminum fin heatsinks are used is because copper conducts heat better, but disperses it slower.

However all copper, thin fin heatsinks, tend to be the best (but cost a lot).


Using heatpipes is good if you want to use a larger heatsink but it wont fit. Or if you want to increase the efficiency of a fairly large heatsink (have it loop from the bottom to the middle).
 
#26
thomas997 said:
A simple aluminum heatsink and fan is probably going to be the best and cheapest way.


Using heatpipes is good if you want to use a larger heatsink but it wont fit. Or if you want to increase the efficiency of a fairly large heatsink (have it loop from the bottom to the middle).
Click to expand...


Fans are generally not for audio equipment. Using a fan is a contradiction of the aim most of us have...to lower the noise floor.
Thats the reason all those other ideas have come up...to avoid a fan.
I guess most people can figure by themslves to mount a 120mm fan on each heatsink, but i dont think anybody wants it.

Heatpipes are also not really for this application for two reasons.

1) you will find it hard to get a heatpipe that can handle the amounts of energy we are dealing with. Here 150W is not a whole lot of powerhandling.

2) The price. If you should actually happen to find a heatpipe that can handle the power, or you choose to mount a number of 150W heatpipes in paralel, youll find that this is a very expensive approach.

So thats back to the copper busbars!


Magura
 
#34
OK, submersion will have to be the next topic then.... I have loads of ideas that could be used to make efficient and CHEAP, I mean REALLY cheap! I mentioned it in the DIY heatsinks thread.

Oh, and I had seen the guy johnferrier mentioned a long time ago. Although the idea is not that bad, the way he does this is not exactly brilliant.... spending LOTS of money on a fluorinert that is NOT meant for the temperature range he wanted is rather ...not wise (to bee a little polite here).

Here is someone that seems to know a little more of what he is doing: http://forums.2cpu.com/showthread.php?threadid=32859&perpage=30&pagenumber=11

I haven't time to write something long now, just thought I'd mention it as it seems to be where this is going. And hopefully I'll explain some more details for use in amps. After all there are some differences from computers where as low temp as possible is wanted. And the ability to make it totally silent and really cheap at the same time. The disadvantage is that it is a bit job if you want to make it look good at the same time. But I’ll get back to that when I find the time to write it down.

And there seems to be interest in this as well! Cool!

Anders
 
#36
It can be done fairly simple, with a reasonable result. Its also possible to take it all the way....like done in idustrial matters or over the limit...like in nuclear power plants 🙂

Basicly its a matter of phase change of water. The simple solution is to mount the mosfets on the downside of a copper pot, fill the pot with water...and there you go. It takes a lot of energy to heat the water above approx. 70C, cause the water evaporates, so thats where the temp would stop rising significantly.

If lower temperatures are desired, you make a fountain in the pot...then its possible to get even below ambient temp.

This is a simplified explanatin, just for people to get the understanding of what condesation cooling is. I am aware of the fact that some sort of exhaust to get the evaporated water out of the room is a good idea. There are a few other considerations to make as well. This is just a general description.

Magura🙂
 
#37
akb1212 said:



We need to make people start to think a little more alternative.

Anders
Click to expand...


Not really.... For industrial matters copper for heat transfer and condensation cooling are 99% of the cooling and heat transfer you see. There is a reason for that !!!

Since condensation cooling isnt practical for home use, we are left with passive cooling of a heatsink (i for one dont bother listening to a fan).

As for heat transfer. Heatpipes are expensive and not easily obtainable, plus the fact that a heatpipe capable of transferring 500W would be hard to deal with. So also here we are left with the known values....loads of aluminium, or if we need a lot of heat removed, copper.


I am currently working on a x-SOZ, its gonna include more than 10 kilograms of copper and roughly 60 kilograms of aluminium (per monoblock). That is the only ticket to get an even temperature, and to direct the heat where you want it to go.


Magura🙂
 
#38
Magura said:


As for heat transfer. Heatpipes are expensive and not easily obtainable, plus the fact that a heatpipe capable of transferring 500W would be hard to deal with. So also here we are left with the known values....loads of aluminium, or if we need a lot of heat removed, copper.

Magura🙂
Click to expand...

And that is exactly what I wanted to make people understand that is NOT the only way. I know you have access to lots of tools and materials (as do I for that matter, although not as much as you…), but many DIY'ers doesn't. And by thinking a little more like OC'ers we could do it so much easier. For some reason OC'ers aren’t afraid of thinking alternative ways of cooling, even if their hardware most often are more expensive (in case they blow up their CPU it most often costs more than replacing a matched set of IRFP's). Most of them have a completely different view on what is noisy in case of fan noise, but there are also quite a few now starting to make them noiseless (I for one, this is how I started finding out about all this).

Now if we take your example and calculate the price on the raw materials for an average DIY'er..... 60 kg aluminium and 10 kg copper PER AMP isn't exactly what most would go out and get just like that. And then there is the machining.....

And this is where a little alternative thinking comes in handy. First of all there are lots of shortcuts to take when it comes to manufacturing. But most importantly: to be able to cool that amount of heat that these amps are putting out obviously demands MASSIVE (as you obviously have found out given the amounts of metal you are using....) heatsinks if using normal metal heat transfer only.
One doable alternative is to use submersion in combination with radiator (like those found in homes used for central heating) way of designing. The idea is to either copy the design of large oil filled transformers (like the ones on several hundred to thousands of VA used for power distribution), or copy the design of oil filled electrical ovens.
If we use the first method (transformer design) all that is needed of materials is a large plate of aluminium and the equipment to bend it to the correct shape (I have to find a way to draw some sketches to show what I mean). The sealing of bottom and top will have to be special care of, but it is by most means way cheaper than your way of doing it, and can possibly be scaled to about ANY size you want without much trouble. And since the heat is transported out to most of the surface by mass (oil) transportation there will also be no severe penalty in loosing efficiency by increasing size the same way a normal heatsink will have.

As for the last way to do it I went out and bought a 1200 W (the cheapest I could find, I paid 250 kr (about $30) for it) oil filled electrical oven. Without many modifications I could have used that as heatsink. After all the way it work is by using a small surface heating element to heat oil and spreading that oil out over a metal with large surface area by convection. Now I discovered that the metal in those ovens are normally steel, and steel is a very bad heat conductor (as in many times worse than aluminium). But never the less, it would probably be possible to use it. The fact that they normally are white doesn’t make them more suited either, but a layer of black paint would probably fix that. But since it is possible to dissipate the amounts of heat it is with these without high temperatures there is really a potential here if better materials were used. An all aluminium version of this would make one killer heatsink capable of more bang for the buck than any other cooling system I’m aware of, (except for getting stuff for free) especially the one you are making 🙂 Your will most probably be a lot cooler looking though….

There are several advantages in this approach. First of all oil is way better as isolator than air (and doesn’t cause the problem water do). And there is no need for electrical isolation to the heat spreaders (but they have to be electrically isolated… an approach possible to use in normal heatsinks too, but gives rise to lots of more problems in normal heatsinks). There have to be heat spreaders to increase the surface area, but they have to be made to suit the purpose. And since we are disposing those we are already given a head start. As this heat resistance is really close to the die it is also really important to get as low as possible, or taken away completely as we are almost able to!

Ok, my plan was to make a small assembly of the output devices attached to a number of copper heat spreaders. This whole assembly was to be putt in the oven in place of the normal heating element. The way it is made makes it really easy to make the copper heat spreaders electrically isolated, thus avoiding the problems normally faced by this approach in normal heatsinks.

Having said that, my current A-40 is built in a chassis that uses normal, although made for forced cooling heatsinks that is electrically isolated. I have to use 4 120 mm fans (the chaises are made for this set-up) set at as low speed as it is possible to make them run reliably. This amp has a increased voltage to about 45 V making it idle at about 400W on the 8 TO3 devices, and they are not even 40 deg 🙂 (but then again these chaises are meant for dissipating a LOT more than that) . I hadn’t even thought about that this is a lot to dissipate per device before now, and I’ve not had any failures due to this. Must be because they are kept cool.

Ok, well…. My point is that it is possible to make this kind of amps without the need for HUGE amounts of metal and special tooling as you have access to. Of course if you do then why not, but if you want an amp like that without your access to materials and equipment/skills there is no other way than to start thinking alternative. And yes, readymade heatpipes are expensive, but that is not the only alternative… as I hopefully have managed to explain.

I’ll come back with some more on submersion (the much more DIY-friendly alternative) soon….

Anders
 
#39
I see your point Anders, but you have to keep in mind that what we are talking about is something thats to be situated in peoples livingroom.

Thats 90 % of the reason that ive spent (and still am spending) so much time and effort/-money on making a good cabinet that is also good looking.

The idea about oil is good if it wasnt for the danger it represents. If something bad happens, it turns EXTRORDIARY bad with a few liters of oil in the equation. Ask anybody whos been working with electromechanics 40 years ago. Motor run capacitors back then were often oil filled. A very dangerous approach, hence the fact that they are all replaced by now by film caps.


Magura🙂
 
Status
Not open for further replies.
Top Bottom