Heatkiller 3.0 poorly made? - Page 5 - Overclock.net

typoknig · 3 Weeks Ago

Quote:

Originally Posted by cooper_inc1

And my point is, if that difference is negligible as you stated for a part of the block that you DON'T see, then why does it matter?

We are technical people here who strive for performance. We de-throne OC kings due to 1 MHz difference in CPU speeds... everything matters

Shrimpykins · 3 Weeks Ago

Quote:

Originally Posted by cooper_inc1

And my point is, if that difference is negligible as you stated for a part of the block that you DON'T see, then why does it matter?

There is a slight increase in performance, not one that can be measured but I assure you it is most likely there. It's basic laws that govern the universe. More surface area = more transfer of heat when dealing with a situation like this a shiny finish will have a minute bit more surface area contacting the CPU in most cases. No the shine won't make a temp difference you can see, but that's not the point. People here strive for absolute perfection when it comes to that stuff.

As for the person who "doesn't buy into lapping". Lapping is a proven method of getting lower temps. Yeah, you can attribute some tests to remounting issues but when someone takes an unlapped block and mounts/remounts it 10+ times then records the avg temp and then does that with a lapped version of that block and gets lower temps... The data is hard to ignore.

B.Steven · 3 Weeks Ago

To combat some of the confusion:

Lapping a HK 3.0 will likely result in temperature increases if you have not lapped your CPU. If you have lapped your CPU, then you will get lower temps with a lapped HK. The difference between a flat waterblock lapped with 800 grit compared to 2000 grit is about 0.01 degrees Celsius, and the only thing you are going to notice is the difference caused by your TIM application. Any change is because of that. The surface area increase from that is incredibly small and will barely improve anything because your bottleneck with temps at that point is waiting for the heat to go through the CPU side of the copper to the water side.

The heatkiller is intentionally warped, like most heatsinks nowadays, to groove into the convex shape of the Intel/AMD IHS's. Granted, there is some slight variance between IHS's so the convex groove of the waterblock will not exactly fit, but it is a lot better than a flat surface. The centre of the block touching the centre of the IHS is the most important. Two perfectly flat surfaces is obviously optimal, but a flat surface against a curved surface is terrible. Don't do it unless you do both.

Shrimpykins · 3 Weeks Ago

Quote:

Originally Posted by B.Steven

To combat some of the confusion:

Lapping a HK 3.0 will likely result in temperature increases if you have not lapped your CPU. If you have lapped your CPU, then you will get lower temps with a lapped HK. The difference between a flat waterblock lapped with 800 grit compared to 2000 grit is about 0.01 degrees Celsius, and the only thing you are going to notice is the difference caused by your TIM application. Any change is because of that. The surface area increase from that is incredibly small and will barely improve anything because your bottleneck with temps at that point is waiting for the heat to go through the CPU side of the copper to the water side.

The heatkiller is intentionally warped, like most heatsinks nowadays, to groove into the convex shape of the Intel/AMD IHS's. Granted, there is some slight variance between IHS's so the convex groove of the waterblock will not exactly fit, but it is a lot better than a flat surface. The centre of the block touching the centre of the IHS is the most important. Two perfectly flat surfaces is obviously optimal, but a flat surface against a curved surface is terrible. Don't do it unless you do both.

Partially true but the real reason the HK 3.0's are bowed is to make sure that you at least have contact with the area directly over the die, not to fit into the convex shape of the CPU.

Clockadile Dundee · 3 Weeks Ago

A good point, especially because the term "lap", and to say something is "true", or flat across it's marrying plane, are two different things.

I also don't know how people are measuring temps with such accuracy, as to say anything with certainty about the effectiveness of anything.

Certainly not with the dts.

However, for some reason, I feel comfortable throwing out almost all data, and still choosing
HK3.0, or a couple other blocks, for my own use.

RA, or roughness average, is a measure of smoothness. IIRC, RA of ~20 is a mirror finish.

It is not a measure of flatness.

A sphere could be perfectly smooth, without ever having the "same flatness" over a local
area.

And this sphere could be lapped with a socket, but neither would be getting flatter.

I don't claim to have the answers either, but I know about engines and machining a little, and this is where I have heard the term "lap".

So, I may be wrong about some of this, but I think that there are in fact many misconceptions and misnomers with regards to this subject.

If you rubbed an IHS against a water block, with a compound in between them, that is what I would call "lapping". Now, I don't think that would give you as good of a result as what is normally done, and so it may be semantics, but I like semantics.

If you were a robot though, actually "lapping" them would achieve a superior result, possibly.

A robot, or a person with a simple fixture.

A fixture like this, I would call, a "jig". This will get you a better result, and this is how a professional would do it.

I just would like to suggest these things, until someone more of an expert, clarifies.

(And, you really wouldn't use a jig to lap valves, but that would be necessary in this case.)

Shrimpykins · 3 Weeks Ago

We use the term lap here to describe the process which you go through to achieve a more flat surface on either a heatsink/cpu/gpu... Well, really anything.

I don't think anyone is saying that because something is shiny that it will get you better temps. We are saying if it is shiny because it is flat then yes, you will get better temps. It may be so minute that you can't measure it but the temp drop in theory is there.

You don't need a jig, a good lap job can be accomplished through a flat piece of something (most of the time glass) some tape and some sand paper. You don't need a robot either. Just apply as little pressure downwards as you can and you will end of with a flat surface if you are lapping on a flat surface.

Clockadile Dundee · 3 Weeks Ago

Quote:

Originally Posted by Shrimpykins

We use the term lap here to describe the process which you go through to achieve a more flat surface on either a heatsink/cpu/gpu... Well, really anything.

I don't think anyone is saying that because something is shiny that it will get you better temps. We are saying if it is shiny because it is flat then yes, you will get better temps. It may be so minute that you can't measure it but the temp drop in theory is there.

You don't need a jig, a good lap job can be accomplished through a flat piece of something (most of the time glass) some tape and some sand paper. You don't need a robot either. Just apply as little pressure downwards as you can and you will end of with a flat surface if you are lapping on a flat surface.

I know, I know, I just mean, if you wanted a gain beyond that, it will be immeasurable.

**ira-k** · 3 Weeks Ago

I think the point Fred and others have tried to make in this thread is that shiny does not equal flatness, you can lap a block or IHS and it may be shiny but it can be far from flat.

Most block finishes even if they aren't shiny should be flatter then what we can do by hand. So if you do lap something just make sure your keeping it flat and not rounding off edges from dragging them or getting long bevels in them.

Most CPU blocks since the V1 Fuzion have a bow in them so they will make better contact with s/775 cupped IHS, take the base out of the block so it will lay as flat as possible before you lap or you can remove to much from the center of it. Personally I wouldn't lap a blocks base unless it had some kind of major defects.

I hold some graph paper, home-made usually..

, at a 90 degree angle to my work every now and then as I'm lapping. If the lines start looking wavy or distorted you know your not keeping it flat. Flat is much more important then shiny, but you can get both if you really want to.

On a IHS I don't even lap them anymore, I just file the cup or convex off it with a mill file, I get just of good a temp's by doing that as when I lapped them.

runeazn · 3 Weeks Ago

Quote:

Originally Posted by ira-k

I think the point Fred and others have tried to make in this thread is that shiny does not equal flatness, you can lap a block or IHS and it may be shiny but it can be far from flat.

Most block finishes even if they aren't shiny should be flatter then what we can do by hand. So if you do lap something just make sure your keeping it flat and not rounding off edges from dragging them or getting long bevels in them.

Most CPU blocks since the V1 Fuzion have a bow in them so they will make better contact with s/775 cupped IHS, take the base out of the block so it will lay as flat as possible before you lap or you can remove to much from the center of it. Personally I wouldn't lap a blocks base unless it had some kind of major defects.

I hold some graph paper, home-made usually..

, at a 90 degree angle to my work every now and then as I'm lapping. If the lines start looking wavy or distorted you know your not keeping it flat. Flat is much more important then shiny, but you can get both if you really want to.

On a IHS I don't even lap them anymore, I just file the cup or convex off it with a mill file, I get just of good a temp's by doing that as when I lapped them.

what is that gray stufff?