Modding the OCZ Hydroflow for Core i7
In my last waterblock roundup (this one), the OCZ Hydroflow HF-MK1 was featured. It's a very affordable block that, according to some tests I saw, performed quite well on Core2 processors.
http://www.abload.de/img/064cv09.jpg
Unfortunately, it's performance on the i7 was pretty bad. While it's nicely unrestrictive, I measured the worst temperatures out of all 16 blocks tested with the Hydroflow. Not so thrilling.
Since it had done well in other tests, I was wondering why this might be. One thing about the Hydroflow was immediately apparent:
http://www.abload.de/img/02tgxl.jpg
The Hydroflow has a stepped base and the stepped area is relatively small. I can imagine that this area is simply too small for an i7, which has a considerably larger die-area than previous CPU generations.
So, I started lapping down the surface to see if it would make a differnece.
Since it's a lot of material that needs to be removed, I used a Dremel for some rough work:
http://www.abload.de/img/05jfko.jpg
After that, I got busy lapping it down using the same method that I also use for lapping CPU-heatspreaders: Sheet of glass and sandpaper of different grits.
http://www.abload.de/img/046dgb.jpg
It took way too long to get the surface flat... one of the problems I encountered is that there was a slight trough around the edges of the stepped part of the base. So it wasn't enough to just lap down that step, I needed to lap down the entire surface of the block further than I had expected.
After a hell of a lot of work, I finally saw a completely even surface and lapped that with continually finer grains until I finished the surface off with 1200 grit sandpaper. This is what the end-result looked like:
http://www.abload.de/img/067kmw.jpg
Now, while I was at it, I tried some more modifications, to try and improve the performance of this block.
http://www.abload.de/thumb/11hgrh.jpg Increased mounting pressure. The screws that come with the Hydroflow can only be tightened to a certain degree before they are stopped on the spacers (or whatever those things are called in english...). So I replaced the spacers with shorter ones. This allowed me to increase mounting pressure significantly.
http://www.abload.de/thumb/07zcwx.jpg Bowing the surface. I placed an O-Ring in the centre of the block and made sure that I tightened the top onto it very firmly. This should produce more bow than usual on the block's surface.
http://www.abload.de/thumb/10yk0q.jpg Polish. I polished the surface to see if it made a difference to the unpolished one.
Here are the results (testing methology same as always, see my wb-roundups for details):
http://www.abload.de/img/graph1isah.jpg
The numbers in front of the labels show in which order the mods were applied. This means that "3) Increased pressure" is also lapped, "4) Bowed base" is also lapped and with increased mounting pressure etc.
As we can see, increasing the mounting pressure is very important. This is obviously the case because after lapping, the base has become thinner (by probably more than 1mm) so the original spacers are now too long.
Combined with the increased mounting pressure, lapping the base drops the temperatures by a respectable 1.9 K. The other modifications fail to further improve the block's performance, though.
Still, 1.9 K is quite good. It still doesn't make the Hydroflow a particularly good block, when compared to others, but that's not mainly what this was about.
My personal conclusion from this is that the Hydroflow has some potential, even on the i7. If OCZ decide to make a version with a flat base and maybe do some minor tweaking on the pin-structure (in their place, I would test what happens if the pins are made slighty smaller), the Hydroflow would still be a block worth recommending to anyone who doesn't want to spend a ton of cash.
That's that. Your thoughts and comments on this are greatly appreciated. :)
Regards,
Shane
