Which of these two base-plates would you have?

[Tyreal]

(Images from Cathar of OCForums. Thread idea is his too.)

Which one?



Once you've picked one, click on the base that you chose.
Left Base
Right Base

Originally Posted by Cathar

Really this is a poor-man's flatness test. The hand-lapped plate on the right is actually not really flat either (it's certainly MUCH better than the block on the left though), it's just that this poor-man's flatness test is unable to show its deficiencies very clearly. It all depends on how far one wants to go in the definition of flatness.

The whole point of this exercise was, of course, to point out the difference between reflectiveness and flatness in a relevant way.

Heck, if waterblock review sites included a test even as simple as this, they would be giving people infinitely more useful information than they do now, which is to show some nice shiny reflective bases and comment on how great they are with absolutely no understanding of the vast difference between the concepts of flatness and reflectivity.

Really, it has to stop now. It has gotten out of hand. It's gotten to the point where anything that is shiny and reflective is perceived as great, yet anything that is non-reflective is perceived as crap, with absolutely no justification or intelligence behind the comments other than, "Ooh - shiny! It MUST be good!"

Originally Posted by Cathar

All I'm trying to do is to effectively explain with visual examples how there is no correlation between reflectivity and flatness.

i.e. one cannot show a picture of a highly reflective base-plate, like most web-sites do, and then make any assumption as to its flatness and hence its suitability to make good thermal contact against a CPU.

I agree 100% with Cathar's statements. I do not understand why review sites often relate reflectiveness with performance. There is no connection. Does that mean if I have a heatsink that is extremely convex, but reflects the image of my pen exactly, that it is a great base? No.
This thread is meant to teach you that, that is a false statement.

In short:
Reflectivity does not equal good thermal contact. Flatness does.

[gvblake22]

Very true, but *usually* if a base is lapped to the point of high reflectivity, it will also be flat. If you think about it, it's pretty hard to lap a base to a mirror finish while being able to follow all of the concave and convex peaks and valleys in the base of the heatsink. I think this is something probably more common in the manufacturing stage as their machining methods are much different than if someone were to lap a heatsink by hand. Still very good information though :thup:

[Capper]

Some of these bases are actually poured, meaning that there is a mold used.....and sometimes these molds are not perfectly flat...which leads to a base that isn't flat.
A machined base is usually not going to be as shiny or "reflective" but unless they leave a bunch of machine marks on it, it will usually be as flat as possible.
Having said all that, and as true as the orginal post is, people overlook the IHS (Integrated Heat Spreader) in all of this, these things are usually horrible quality and rarely ever flat

[Tyreal]

Originally Posted by Capper

Some of these bases are actually poured, meaning that there is a mold used.....and sometimes these molds are not perfectly flat...which leads to a base that isn't flat.
A machined base is usually not going to be as shiny or "reflective" but unless they leave a bunch of machine marks on it, it will usually be as flat as possible.
Having said all that, and as true as the orginal post is, people overlook the IHS (Integrated Heat Spreader) in all of this, these things are usually horrible quality and rarely ever flat

The IHS is usually very concave or convex, but since that will void your warranty or even void the processor, I won't delve into it.

So the original post is true. The point of the post is to state that reflectivity does not mean flatness. Some reviews that i've read on heatsinks just show a heatsink reflecting an object, that doesn't really tell me anything, other than that its reflective.

[KillingField]

The method I go by is to lay a machinest rule across in all directions to check for flatness. It's not the best, but it will give you a idea if it's flat or not.............;) Ploishing a base as some companies do is done on a high speed buffing wheel, this method removes material from the base and makes things worse than before they started. To get a good flat base, a high quality maching is required, some milling machines out of China and such totally suck.
As for the IHS, remove it or hand lap it..........I do the lapping part myself...........:)

[Tyreal]

I haven't tried lapping an IHS yet. Worried about killing the processor and voiding my warranty. I'm sure it's pretty safe and easy though.

How do you go about doing it? The same way you would lap a heatsink? I'd image you hold the CPU upside down so the pins face upward and lap normally.

[KillingField]

That's it, except that I stick the CPU on a chunk of that foam it's shipped in to keep from damaging the pins, and go lite on the water..........;)

[Jokerswild]

After 33+ years as a machinist I can attest to the fact that a mirror finish is not necessarily flat. The real way to test flatness is on a surface plate using a indicator. Most hand lapped surfaces tend to be slightly rolled on the edges. Really want it flat use a surface grinder.

[Tyreal]

Exactly. Thats the whole purpose of the thread. A mirror finish does not represent a flat surface. The grid test does not necessarily represent a flat surface either, but gives you MUCH better idea.

As you mentioned, a surface grinder and an indicator is the only you to have a truely flat surface.

[Lead Head]

I think cathar also said 600Grit gives the best performence with AS5 or somethign