Do you recommend lapping this waterblock?

petreza

n00b
Joined
Feb 16, 2006
Messages
52
Hi,

I bought these waterblocks:

For CPU GPU SR Water Cooling Heatsink Block Liquid Cooler Aluminium Waterblock

I know they are junk but I needed a lot of them and could not afford the good ones.

I am planning to use it to cool a load up to 130W.

It is too big for the application so I will have to glue using thermal epoxy an aluminum plate just a bit smaller on the bottom of it.
The plate is 2.3mm thick.

The waterblock has been anodized blue.
Will that interfere with the heat transfer and/or the gluing?
Should I lap the wb and what grit paper do you recommend?

Thanks!
 
Last edited:
As an eBay Associate, HardForum may earn from qualifying purchases.
what's your goal with that thing? And if you are thermal epoxying something to it you'd want to lap the base of that, not the water block.
 
Don't lap it, that looks super cheaply made and I'd be worried about it coming apart. Plus the thermal epoxy will spread into any slight gaps.

And there is no way I could put a 130w load into those. They don't even have a cold plate or finned interal area foe dissipating heating into the water...
 
Last edited:
Piece of advice. Make sure the rest of the parts you use are aluminum too or else you're in for some corrosion issues
 
what's your goal with that thing? And if you are thermal epoxying something to it you'd want to lap the base of that, not the water block.

Its for cooling a 8 core CPU.
Well, heat transfer is heat transfer. I will lap the base and the top of the plate that I am epoxying but wouldn't it help if I lap the waterblock too? I know it supposed to be better lapped, I just don't know how much better. Maybe it is not worth the effort. Maybe the epoxy will stick just fine to anodized aluminum.

Don't lap it, that looks super cheaply made and I'd be worried about it coming apart. Plus the thermal epoxy will spread into any slight gaps.
And there is no way I could put a 130w load into those. They don't even have a cold plate or finned interal area foe dissipating heating into the water...

I measured the wall to be 1.5mm - how much would lapping take off? I was hoping no more than a tenth of that.
As a last step of the manufacturing process before anodizing they have shaved off the base end-to-end with circular tool. That has left the surface with a pattern of concentric circle arcs. The surface is quite flat but if I run my finger nail across the arcs I can feel their jaggedness. Wouldn't a mirror-ish surface finish help with heat transfer or is the difference just academic?
I know this is a risky experiment but I have to run it because of budgetary constraints. The waterblock looks small for a CPU rated at 130W but aren't those small GPU wbs supposed to dissipate even more heat under load? In my case, I am not looking for low temp performance, I just need it to quiet the system. If it manages to give me the same temps as air cooling 60-70C I am happy. I plan to use a large radiator to keep feeding it cool water.

Piece of advice. Make sure the rest of the parts you use are aluminum too or else you're in for some corrosion issues

Thanks! I was aware of the issue. Do stainless steel or chrome plated pars present the same problem with aluminum as copper? I see that some components are made like that.
 
$10-20 air coolers will probably cool better than that thing. The GPU blocks are vastly different. They're made of copper which transfer heat much quicker. They have a proper contact surface. And they have an internal structure optimized for heat transfer. This block you're using has none of that.
 
$10-20 air coolers will probably cool better than that thing. The GPU blocks are vastly different. They're made of copper which transfer heat much quicker. They have a proper contact surface. And they have an internal structure optimized for heat transfer. This block you're using has none of that.

Agreed, That thing looks like a complete pile of junk. You'd likely be better off with the stock cooler.
 
$10-20 air coolers will probably cool better than that thing. The GPU blocks are vastly different. They're made of copper which transfer heat much quicker. They have a proper contact surface. And they have an internal structure optimized for heat transfer. This block you're using has none of that.

Agreed, That thing looks like a complete pile of junk. You'd likely be better off with the stock cooler.

It's for this (and I have 2 of those):

Help with quiet air cooling 4 processors in 2U server?

... so 8 good quality wbs gets quite expensive. Plus these are LGA 1576 (Socket LS) which I assume do not have many wbs made for as well as retention brackets, if any.

Stock air cooling is scary laud.
 
Deal with the noise or pay up. Those blocks aren't going to give you the cooling you need.
 
It's for this (and I have 2 of those):

Help with quiet air cooling 4 processors in 2U server?

... so 8 good quality wbs gets quite expensive. Plus these are LGA 1576 (Socket LS) which I assume do not have many wbs made for as well as retention brackets, if any.

Stock air cooling is scary laud.

So are you also planning to get a pump and radiator for each CPU? I suspect not. $4 (shipped!) water blocks aren't going to effectively cool a whole row of those CPUs. I don't know how you came to the conclusion that this was a good idea, however it's not appropriate for what you are doing. Watercooling works when people put in the effort and select the correct components. What you have done is merely found the cheapest possible blocks that aren't in anyway comparable to a real water setup.
 
Last edited:
I lapped one of them - I bought a spare just in case. It took a long time with 150 grit sandpaper to remove all the anodized aluminum and the manufacturing scratches close to the corners. The scratches in the middle of the waterblock were tough to get to so I went down to 120 grit and it got the job done on no time.

Then I went to the store and got some 80, 150, 220, 360, 1000 grit sandpaper and experimented with all of them. The only moderate effort went toward removing the scratches left from the 80 grit with the 150 grit. The rest of the progression went pretty quick. At 1000 grit the waterblock feels very, very smooth. I don't think there is any tangible gain to be expected by continuing further.

Just as an experiment tomorrow I might use a polishing wheel and rouge with my drill to try to get a mirror finish.

Then I went medieval on one of its corners to see how much sanding it will take to get a hole. It took a lot of sanding. Once I got a hole I used pliers and wire cutters to open the back of it completely. See the pictures below - they show the result only of the 150 grit, not the 1000 grit sanding. It is constructed by welding two U shaped parts forming a box. There are 12 parallel fins inside which are welded to the bottom U but are only (sort of) touching the top U. There is a 5mm gap between the fins and the front and the back wall. (see last picture below) One of the middle fins is (supposedly) extending fully to the front wall creating a barrier so that the water does not flow directly from the inlet to the outlet. The wall thickness is 1.6+mm and the fins are 0.8mm thick. With all my lapping I was nowhere near to making a hole.



01.jpg

.
.
02.jpg

.
.
03.jpg

.
.
04.jpg

.
.
05.jpg

.
 
You don't lap to make the surface smooth, you lap to make it flat (and smooth). To lap you attach the same paper to something flat like a pane of glass and move the heat sink across the surface. I've done this with spray adhesive to hold the sandpaper. The sandpaper must remain absolutely flat so you don't cause a convex or concave deflection in the surface. Based on your pics, you accomplished nothing and rounded off the corners from holding the sandpaper in your hand??? :eek:

And why the hell is the inside of those blue anodized?!? None of this makes sense, I hope that server was like $50 at a yard sale.
 
Last edited:
Deal with the noise or pay up. Those blocks aren't going to give you the cooling you need.

So are you also planning to get a pump and radiator for each CPU? I suspect not. $4 (shipped!) water blocks aren't going to effectively cool a whole row of those CPUs. I don't know how you came to the conclusion that this was a good idea, however it's not appropriate for what you are doing. Watercooling works when people put in the effort and select the correct components. What you have done is merely found the cheapest possible blocks that aren't in anyway comparable to a real water setup.


I knew this is somewhat of a long-shot but for the price it was worth a try. I just hope I don't fry the system experimenting.

Yes, I am planning to connect 4 of these waterblocks in series driven by a Koolance PMP-450S pump @ 24V. I tried this setup and it gave me just about 1 Gallon per minute flow. And there was a weak link in the system - I bought 1 / 2 " -to- 1 / 4 " reducers but they turned out to be more like 1 / 2 " -to- 0.6 / 4 " which I am sure made a big impact on the flow rate. Now I am getting proper reducers and that should help.
I also ran the pump with all 8 waterblocks and it gave me 3L/min flow. So now instead of running both systems in parallel, each with its own pump, I am thinking of running the systems in series: pump---system---pump---system----radiator, reservoir etc. This way I have some redundancy in case one of the pumps fails.

I am thinking of getting this radiator:

Radiator, 9x120mm 18-FPI Aluminum

Anyway, I'll try it. If it does not work - no big loss. I'll just have to get proper waterblocks.

Thank you all for your input!
 
You don't lap to make the surface smooth, you lap to make it flat (and smooth). To lap you attach the same paper to something flat like a pane of glass and move the heat sink across the surface. I've done this with spray adhesive to hold the sandpaper. The sandpaper must remain absolutely flat so you don't cause a convex or concave deflection in the surface. Based on your pics, you accomplished nothing and rounded off the corners from holding the sandpaper in your hand??? :eek:

I used one of those sandpaper holders that painters attach to extension poles, not realizing that the base of those is quite soft foamy rubber. Then I switched to a flat aluminum plate and it is all good now. I even found this 3M "No-Slip Grip" sandpaper that is tacky on the back:

http://www.amazon.com/3M-25320P-G-A...014&sr=8-3&keywords=3M+Pro+Grade+No-Slip+Grip

And why the hell is the inside of those blue anodized?!? None of this makes sense, I hope that server was like $50 at a yard sale.

Yea, I know. How hard would it have been to plug the holes before anodizing - you get what you pay for.
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
I knew this is somewhat of a long-shot but for the price it was worth a try. I just hope I don't fry the system experimenting.

Yes, I am planning to connect 4 of these waterblocks in series driven by a Koolance PMP-450S pump @ 24V. I tried this setup and it gave me just about 1 Gallon per minute flow. And there was a weak link in the system - I bought 1 / 2 " -to- 1 / 4 " reducers but they turned out to be more like 1 / 2 " -to- 0.6 / 4 " which I am sure made a big impact on the flow rate. Now I am getting proper reducers and that should help.
I also ran the pump with all 8 waterblocks and it gave me 3L/min flow. So now instead of running both systems in parallel, each with its own pump, I am thinking of running the systems in series: pump---system---pump---system----radiator, reservoir etc. This way I have some redundancy in case one of the pumps fails.

I am thinking of getting this radiator:

Radiator, 9x120mm 18-FPI Aluminum

Anyway, I'll try it. If it does not work - no big loss. I'll just have to get proper waterblocks.

Thank you all for your input!

You're spending a lot of money on a loop that will be completely bottlenecked by those blocks. As far as I can tell, the LGA1567 socket uses the same mount points as the LGA1366 socket, so I'd get some 1366 blocks and throw away those blue pieces of junk.
 
I knew this is somewhat of a long-shot but for the price it was worth a try. I just hope I don't fry the system experimenting.

Yes, I am planning to connect 4 of these waterblocks in series driven by a Koolance PMP-450S pump @ 24V. I tried this setup and it gave me just about 1 Gallon per minute flow. And there was a weak link in the system - I bought 1 / 2 " -to- 1 / 4 " reducers but they turned out to be more like 1 / 2 " -to- 0.6 / 4 " which I am sure made a big impact on the flow rate. Now I am getting proper reducers and that should help.
I also ran the pump with all 8 waterblocks and it gave me 3L/min flow. So now instead of running both systems in parallel, each with its own pump, I am thinking of running the systems in series: pump---system---pump---system----radiator, reservoir etc. This way I have some redundancy in case one of the pumps fails.

I am thinking of getting this radiator:

Radiator, 9x120mm 18-FPI Aluminum

Anyway, I'll try it. If it does not work - no big loss. I'll just have to get proper waterblocks.

Thank you all for your input!

If you already have the money to do it right from the beginning, then freaking do it right from the beginning. You're just wasting money by doing this. Those fins are way too thick, and too few in number, and the chamber way too large for effective heat transfer.

Get real copper blocks. Low end ones can be found in the $20-30 range. Get a real copper radiator, don't get a cheap aluminum one. Toss any aluminum watercooling part into the metal scrap.
 
Get real copper blocks. Low end ones can be found in the $20-30 range. Get a real copper radiator, don't get a cheap aluminum one. Toss any aluminum watercooling part into the metal scrap.
I dunno, man. If everything in the loop is aluminum, it shouldn't corrode. And this experiment is certainly the most interesting thing I've seen in a long time - the results could be both useful, and entertaining.

I mean, I can only look at flow and efficiency graphs at XS for so long before I start to doze off. All the curves are very similar...temps at CPU/GPU are within a few degrees across the entire field...Components are so efficient these days that water isn't even needed anymore and no one's testing anything and MartinM's last update was Feb. 2015 and the link to his latest review is a fucking virus! Skinee last reviewed a pump in 2010! What I wouldn't GIVE for a little bit of excitement!

I salute you, Petreza. God speed, and please plot your curves.
 
I dunno, man. If everything in the loop is aluminum, it shouldn't corrode. And this experiment is certainly the most interesting thing I've seen in a long time - the results could be both useful, and entertaining.

I mean, I can only look at flow and efficiency graphs at XS for so long before I start to doze off. All the curves are very similar...temps at CPU/GPU are within a few degrees across the entire field...Components are so efficient these days that water isn't even needed anymore and no one's testing anything and MartinM's last update was Feb. 2015 and the link to his latest review is a fucking virus! Skinee last reviewed a pump in 2010! What I wouldn't GIVE for a little bit of excitement!

I salute you, Petreza. God speed, and please plot your curves.

If everything is aluminum there won't be corrosion, but copper parts transfer heat much better than aluminum.
 
If everything is aluminum there won't be corrosion, but copper parts transfer heat much better than aluminum.

What he said. Additionally, no one manufactures quality aluminum blocks optimized for heat transfer from CPUs. The micro-fin structure of modern waterblocks is far superior to that flat U-channel.
 
I think op was trolling. post started on feb 24 and his last post was on feb 24. I was kinda looking forward to seeing the disaster of a system he came up with...
and you guys are totally correct, all same metal equals no corrosion. copper being better than alu, silver being better than both. an all silver loop would be pretty bling-bling!
 
It may not be optimal, but I bet those things would work if he figured out a mounting solution.

I once built a working water cooling loop out of an old Socket 370 heatsink that I cut the outer row of fins off of and then constructed a plastic enclosure around the remaining fins. It actually worked reasonably well, using an automobile heater core as a radiator. This was before the days where you could just go and buy ready made water cooling parts for PCs.

I can't imagine trying to water cool a server unless it's some sort of elaborate custom setup, though. Why not just use low profile heatsinks and big meaty delta fans?
 
Back
Top