Thermalright convex base plate

Sinopa

Sinopa

Sinopa

Associate
Joined
23 Nov 2004
Posts
301
In the Anandtech review of the Swiftech H2O-120 Compact and Corsair Nautilus 500 it states this on the Final Words page
The Thermalright designs do use the convex (bowed) base plate which has been shown to increase cooling by 2C to 3C and therefore increase overclocks.
Click to expand...
I've seen posts here advising the lapping a Thermalright Ultra-120 Extreme if it is found to have has a convex base.

So which is best ? :confused:
 
Leave it as it is. Lapping a Thermalright wrecks it's performance. They have micro-grooves in the surface to increase TIM contact area as well and lapping removes those as well.

Thermalright are No. 1 in performance air cooling for a reason. It doesn't need improving.
 
The Thermalright designs do use the convex (bowed) base plate which has been shown to increase cooling by 2C to 3C and therefore increase overclocks.
Click to expand...

I spotted that too Sinopa and had to read it twice to make sure I hadn't misunderstood it.

It does tend to make sense since you would like to think the manufacturer knew what they were doing.

It's surprising this hasn't been commented on elsewhere, to my knowledge.
Every other reference I’ve seen recommends lapping to get rid of the 'fault' and many have done so.
Perhaps most simply follow the popular theory and lap.

I'm sure both sides of the argument will get strong support.
 
While not wishing to doubt Thermalright's design skills I'd sure like to read somewhere how a convex base improves cooling.

The only thing I can think of is greater contact pressure at the highest point which corresponds to where the actual cpu core is under a heatspreader.

Is the base convex on one plain (like a barrel) or is it like a magnifying glass?
 
I could see how the increased pressure at the centre of a single die processor might lower temps.

But with quad core processors having two dies the same convex base plate would only lower temps on core 2 and 3 but raising the temps on 1 and 4 where there is little contact with the base plate.

So I guess that if you have a quad you would be better off lapping the heatsink.
 
I recently tested my CPU without a heatsink attached to find out the temp it would start throttling at (to verify CoreTemp was detecting TjunctionMax properly) and the results were interesting.

CoreTemp was correct with it's 85c Tjunction detection of my E2180 M0 stepping CPU but what was interesting was when I lay the heatsink (Tuniq Tower) back on top of the CPU (had already applied the arctic silver) the temps quickly dropped back to the same as it was when I had the heatsink tightly bolted down. So I then properly reattached the heatsink and this time I made sure not to screw it down tightly and the temps are still the same.

So the lesson learned was C2D's don't need the heatsink on tightly and in fact having it too tight can warp the motherboard yet it isn't necessary to subject it to such punishment. :)

Here's a screenshot when it hit the throttle temp:

 
Last edited:
But how tight is too tight?
Many peops have found that by adding a washer to the clamp of the Thermalright 120 Extreme they can increase the pressure on the cpu and also by chance lower the temp of the cpu.
And if pressure is not an issue, why would the convex work better than flat?
 
Well if the motherboard isn't being warped due to the pressure I'd say it's not too tight. The pressure I guess it depends on the CPU and possibly it's fine for mine as it's relatively flat. As for convex, a lot of IHS's are concave so maybe it's a better fit?
 
It has been said that when the heatpipes are attached to the H.S. from the heat makes the base either convex or concave....so what T.R claim there may be a bit of B.S. in there maybe not....as from what i've seen Big Wayne had a concave base others convex and mine although the TRU120A was convex i have also read a review where they said the older TRU120A was supposed to be flat....so firstly you would have to check what kind of base you got to see wether concave/convex...and then if you decide on lapping do both CPU/HS as having both flat surfaces obviously will make a better contact.:)

Oh btw my Q6600 was convex and as my H.S was lapped already so due to BIG uneven temps i had to lap cpu...if my H.S. was still convex i think it would have made things even worse.:( ( something to think about )

After lapping both my H.S. is doing a pretty good job for the cheaper older version.:D
 
AnandTech were very specific so they may have guidance from Thermalright and elsewhere as to the effectiveness of the base design.
They may well have determined from testing that a convex base gives the benefit they refer to – at least with these particular units.

If Thermalright bases are available in convex, flat or concave flavours then it would seem to be a lottery.
If you feel lucky you might get the better option assuming you are confident what the better option is.

In terms of overclocking and load temperatures at stock speed they appear to give marginal improvement compared to some competitors, who don’t seem to have this or similar issues, which are resolved by some resorting to lapping.

There is perhaps a better option.

62j3zx5.png
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom