Sign in with
Sign up | Sign in
Your question

stupid question

Last response: in Overclocking
Share
May 16, 2008 7:55:02 PM

Does anyone know if it would be possible to solder a heatsink directly to a CPU without damaging the CPU?

More about : stupid question

May 16, 2008 7:59:50 PM

Why would you want to?
May 16, 2008 8:03:19 PM

The amount of heat you'd need (even assuming the solder would form an inter metallic bond..or "stick") would fry your cpu like bacon, period.
Related resources
May 16, 2008 8:04:08 PM

Short answer, no
The heat would cause the very delicate joins within the CPU to melt/fuse (this isn't even counting the damage to the silicon that you could do) However if you are looking for a permanent solution for attaching a HSF to the CPU, look at this: http://www.newegg.com/Product/Product.aspx?Item=N82E168...
Thermal Adhesive.
May 16, 2008 8:04:35 PM

1) There are no stupid questions.

2) Yes I believe you could but it would have to be a somewhat mild solder (not with a propane torch!). Maybe a tube of multicore electronics solder? You would have to get some thermal paste between the cpu and heatsink though for it to be functional, and I am not sure how that part would work.
May 16, 2008 8:12:50 PM

IH8U said:
Short answer, no
The heat would cause the very delicate joins within the CPU to melt/fuse (this isn't even counting the damage to the silicon that you could do) However if you are looking for a permanent solution for attaching a HSF to the CPU, look at this: http://www.newegg.com/Product/Product.aspx?Item=N82E168...
Thermal Adhesive.


Damaging silicon is a rather hard thing to do. You can break it, but it needs extreme heat to actually damage the structure of the silicon itself, you would melt the metal inside the silicon long before you melted the silicon itself. To head off any arguments on this point. I am a silicon crystal grower, and currently work for Solarworld where we make silicon crystals for solar panels. We have to heat the silicon to approximately 1300 degrees celcius in order to make it molten. We use the charkozsky method for making our crystals that are 155 mm in diameter. We have the capability of making processor grade silicon crystals, but there is more money in the Solar side of things these days then the processor side. Anywho hope that clarifies any misconceptions.
May 16, 2008 8:14:45 PM

njalterio said:
1) There are no stupid questions.

2) Yes I believe you could but it would have to be a somewhat mild solder (not with a propane torch!). Maybe a tube of multicore electronics solder? You would have to get some thermal paste between the cpu and heatsink though for it to be functional, and I am not sure how that part would work.


Sorry it would not work, both surfaces have to be at solder temp for it to bond...and the very lowest solder melt's at...ready...

183 °C or 361.4 F

Some of us soldered for a living, and were taught by an aerospace engineer....and ain't no way to be delicate when you'd have to bring a relativley huge mass up to that temp.

Plus there is no guarantee that the solder would bond, depending on the 2 metals and or fluxes used.
May 16, 2008 8:16:34 PM

@ blacksci, tru, but what about the other parts of the chip such as gates and other doo dads ...
May 16, 2008 8:21:21 PM

Thats why i said the metal inside the silicon itself would melt before the silicon lost structure, also including gates and such, but the integerity of the silicon itself would still be there. Silicon makes a great insulator for electricity, but heat not so great.
May 16, 2008 8:23:22 PM

Ya, my old eyes ran it together and I missed that....

...the op had a good idea though...it'd beat ANY thermal compound for sure,l the weak link would be the bond to the heat spreader from the chip then.
May 16, 2008 8:24:30 PM

blacksci said:
Thats why i said the metal inside the silicon itself would melt before the silicon lost structure, also including gates and such, but the integerity of the silicon itself would still be there. Silicon makes a great insulator for electricity, but heat not so great.


Yeah, but it makes for awful boobies ;) 

I hate those jello mold ones ...hehe
May 16, 2008 8:25:35 PM

Well if you could bond the heatspreader to the metal part of the chip-b4 they installed it with the rest of the chip, might be onto something there.
May 16, 2008 8:27:19 PM

royalcrown said:
Sorry it would not work, both surfaces have to be at solder temp for it to bond...and the very lowest solder melt's at...ready...

183 °C or 361.4 F

Some of us soldered for a living, and were taught by an aerospace engineer....and ain't no way to be delicate when you'd have to bring a relativley huge mass up to that temp.

Plus there is no guarantee that the solder would bond, depending on the 2 metals and or fluxes used.


Funny you mention aerospace engineering, I am an aerospace engineer. :-)

I believe there are techniques that aren't so heat intensive. Did you know that all PCB's are soldered at some point in manufacturing? It is called wave soldering.
http://en.wikipedia.org/wiki/Wave_soldering

I assure you it is possible, my only question is how practical this would be.
a b à CPUs
a c 330 K Overclocking
May 16, 2008 8:30:55 PM

I believe a TIG welder would do the trick nicely.

Did you want it to work after you were done?

;) 
May 16, 2008 8:33:21 PM

Wave soldering is like a 100,000 dollar machine
May 16, 2008 8:34:51 PM

^ Exactly, which is partly why I question the practicality
May 16, 2008 8:34:59 PM

njalterio said:
Funny you mention aerospace engineering, I am an aerospace engineer. :-)

I believe there are techniques that aren't so heat intensive. Did you know that all PCB's are soldered at some point in manufacturing? It is called wave soldering.
http://en.wikipedia.org/wiki/Wave_soldering

I assure you it is possible, my only question is how practical this would be.


Yes, I know what wave soldering and SMT is:

I also know that that bringing a discreet component lead up to temp is not the same as bringing a large mass up to temp, besides the fact that the large mass is designed to cool off as fast as possible. And that doesent take into account the smaller surface area and mass of the cpu. Guess which one is gonna take the heat...

There may be ways to do it, I believe Intel has tried it underneath their spreaders, but it's not gonna be done at home without cooking it.
May 16, 2008 8:35:54 PM

TIG weilding is quite hot.....
a b à CPUs
May 16, 2008 8:41:11 PM

IF you really want to try it, find an old system and use it as a test. Overall, I say it won't work.
a b à CPUs
a b K Overclocking
May 16, 2008 9:06:50 PM

Wave soldering, not an option even if it was cheap. Heat speader / HSF block to large of a surface area and the HSF Cu block mass is to great. Kind of like tring to solder 2 ga copper wire to a post using a 25 W soldering pencil. The large solid Cu block (Normal solder does not solder to Al) would be a have a large thermal energy. As BlackSci pointed out the Si can handle that, But not the fine delicate interconnects.

Editted - To Slow, Sorry for the wasted space.
May 16, 2008 9:10:23 PM

if it still works afterwards let us know your temps :) 
May 16, 2008 9:28:29 PM

How is the heat spreader connected to the cpu? Is it direct metal contact to the silicon, or is there another layer of TIM in there?
May 16, 2008 9:49:28 PM

They use an adhesive, i believe toms actually took one apart just to show folks, anyhow, its a glue.
May 16, 2008 9:49:38 PM

There used to be tim in there thermal adhesive , I think... and I think intel soldered a few on, but it's not the same as a whole heatsink. Not sure what they are using now though. :) 
!