Black Ops SB, NB, Mosfet heatpipe removal

[Wotcher]

Preamble.

It gets pretty warm in Queensland Australia during the summer and a few 35°C+ days brought elevated temperatures which became a concern for me.

Originally my purpose in doing this was to try and lower my NB temperatures as they were reaching mid 50°C readings with stock vNB at 1.25v utilising the supplied air cooling fan option.

But as often happens other things become apparent when we delve deeper.

I found that the mating surfaces were not "ideal" in certain areas and the contact area of some mating surfaces less than desirable in others so I resolved to lap/reseat these affected areas to hopefully improve temperature levels.

Originally I didn't take pics of the components after disassembly which I now regret (for reasons I'll explain later) as I never considered making a thread about it but an expression of interest in the heatpipe assembly configuration from a couple of other members resulted in what you are now reading.

Specs:

Black Ops
E8600
Thermalright Ultra 120 Extreme with supplied Foxconn 120mm fan
4* 1GB Kingston HyperX DDR3 1375Mhz
8800GTS 640MB (9800GTX is rma)
2* SATA (no RAID)
1* Optical
2* 120mm intake fans
2* 120mm exhaust fans
Thermaltake Sonata case

On with the show.

The board in the case showing the NB cooling option below the 120mm CPU fan. The cpu fan exhausts down onto the video card at present which appears to work funnily enough, I originally had the TRUE120 rotated 90° so the heatpipes were north/south but decided to try east/west on reassembly, next is to try the fan as a push as opposed to the current pull. Not a ton of room in this case but not cramped either.

http://img205.imageshack.us/img205/5990/firstpickp0.th.png (http://img205.imageshack.us/my.php?image=firstpickp0.png)


The removal is fairly straight forward, eight screws to remove

http://img206.imageshack.us/img206/2168/rearviewia0.th.png (http://img206.imageshack.us/my.php?image=rearviewia0.png)


Naked board as a result, bear in mind this is not the original TIM as I replaced it with MX-2 during the first disassembly/reassembly, I use the spread method. What is not immediately obvious is the poor contact area between the SB cooler and the chip evidenced by the thicker TIM in the top right corner of the chip in the below pic. This was more pronounced with the original TIM as the stock (approx) 10mm square had not spread very much at all. This will be plainer in a later pic.

http://img205.imageshack.us/img205/1626/sanscoolerto9.th.png (http://img205.imageshack.us/my.php?image=sanscoolerto9.png)


Having unscrewed it from the board I was presented with the assembly. I have two areas of concern here, the SB as mentioned before and the NB both appear to be making a less than desirable contact on the mating surfaces.

http://img206.imageshack.us/img206/9685/3in1coolerremovedmp7.th.png (http://img206.imageshack.us/my.php?image=3in1coolerremovedmp7.png)


This surface also has quite noticeable milling marks (as opposed to the SB which is mirrorlike) which are not immediately apparent in this pic (I'll try to get a better one when the TIM is removed) so I will be lapping it.

http://img205.imageshack.us/img205/8392/northbridgelk4.th.png (http://img205.imageshack.us/my.php?image=northbridgelk4.png)



The NB pot is attached to the heatpipe plate with four screws as seen above and when disassembled reveals the heatpipes themselves. Although at first glance these surfaces appear OK they too show milling marks and are slated for lapping. A burr at the bottom of the screw hole presented a minor problem in removing one screw but was soon fixed with a light reaming.

http://img205.imageshack.us/img205/6104/heatpipetonbmatingsurfacf5.th.png (http://img205.imageshack.us/my.php?image=heatpipetonbmatingsurfacf5.png)


The mosfets appear to be making reasonable contact with the thermal tape on the cooler so I will "leave well enough alone" there.

http://img209.imageshack.us/img209/8486/mosfetzp5.th.png (http://img209.imageshack.us/my.php?image=mosfetzp5.png)


Finally a side view of the NB pot suspended above the heatpipe plate.

http://img209.imageshack.us/img209/3264/nbaboveheatpipeplatexn8.th.png (http://img209.imageshack.us/my.php?image=nbaboveheatpipeplatexn8.png)

I'm off to get some supplies then I'll commence lapping

Cheers.

[Wotcher]

The northbridge heatpipe plate to chip mating surface, you can clearly see the milling marks. I used 600, 800, 1000 & 1200 grit paper, some would say overkill but it is what I am comfortable with.

Before.

http://img385.imageshack.us/img385/8936/nbtochipfacesmallqj7.th.png (http://img385.imageshack.us/my.php?image=nbtochipfacesmallqj7.png)


During.

As is quite common in devices using heatpipe technology there is a ridge in the base running parallel to the heatpipes

http://img385.imageshack.us/img385/8167/ridgewa7.th.png (http://img385.imageshack.us/my.php?image=ridgewa7.png)


After.

I didn't have any graph paper, an anti static bag will have to do.

http://img385.imageshack.us/img385/4831/nbheatpipetochipmr8.th.png (http://img385.imageshack.us/my.php?image=nbheatpipetochipmr8.png)


Heatpipe to Pot mating surfaces.

To me this is the area that has the potential to have the worst surface contact of the whole assembly, the grooves between the heatpipes themselves as well as the plate edges are massive non contact areas, I will be fairly liberal with the MX-2 in this area as I reason that some TIM contact and heat transference would be preferable to nothing at all from what in reality would be giant air gaps, I guess I'll find out after reassembly.

http://img81.imageshack.us/img81/4162/dscf0889smallak3.th.jpg (http://img81.imageshack.us/my.php?image=dscf0889smallak3.jpg)


Being unsure of the wall thickness of the heatpipes I stopped lapping as soon as the last dimple in the casting was removed from the most prominent heatpipe, I'm convinced this is not the ideal lapping job but I will err on the side of caution here so I do not sand right through the wall.

http://img81.imageshack.us/img81/7383/dscf0892smallar3.th.jpg (http://img81.imageshack.us/my.php?image=dscf0892smallar3.jpg)


Given the fact that the distance between the "legs" on the pot is 41.5mm I had to get a piece of glass cut to fit, I used 6mm thick glass and the nice man at the glass shop didn't charge me, bonus. Will shop there again :up:

This factor was also restricting in that the directions for movement were limited, there was no 90° rotation between sets of passes.

Before lapping you could see milling marks once again though this time they were in a circular motion, the base is obviously from a seperate piece of material and then (by the looks) it is pressed into the moulded pot and silver soldered(?) in place. It was also concave which became apparent during the lapping as seen below.

http://img508.imageshack.us/img508/1552/pottoheatpipes2sq3.th.png (http://img508.imageshack.us/my.php?image=pottoheatpipes2sq3.png)

I lapped it until the concavity was gone but due to the lack of directional options the surface isn't as "pretty" to look at but I'm sure it is flat nonetheless.

http://img508.imageshack.us/img508/8176/finishedbx1.th.png (http://img508.imageshack.us/my.php?image=finishedbx1.png)

:D

[Wotcher]

Reassembly is complete and initial results at idle are reporting 48°C, not earth shattering but an improvement of 3 - 4°C anyway.

Dinner is next with some gaming afterwards to see how much difference it makes under load.

I hope this helps anyone who has concerns about their SB, NB, Mosfet cooling assembly, if you have any questions I'll do my best to address them.

Cheers.

[RAYTTK]

Thanks Wotcher, Top notch reporting. In the 1st ( elusive) post about this Im pretty sure the guy in question said he used a file on them heat pipes. Next time anything is in for RMA I will be doing this myself so Im very grateful to you for taking the time to share this.

[yiancar]

wow thats a nice try there :) You are the first one i saw breaking down all the heatpipes on this mobo. W8ing for load results :D

[Wotcher]

Thanks Wotcher, Top notch reporting. In the 1st ( elusive) post about this Im pretty sure the guy in question said he used a file on them heat pipes. Next time anything is in for RMA I will be doing this myself so Im very grateful to you for taking the time to share this.No worries mate, happy to oblige.


wow thats a nice try there :) You are the first one i saw breaking down all the heatpipes on this mobo. W8ing for load results :D The decrease in load temps is not what I was hoping for tbh, it is definitely an improvement but only a 3-4 degrees the same as the idle temps.

It is 33°C ambient air temperature here at present and after running prime95 blend test for 10 minutes or so (cpu 3.50GHz, 500*7) with my NB at stock 1.25v it hit 53°C which is an improvement on what I was getting but nothing stunning imo.

Time to turn on the aircon


:rofl:

http://img261.imageshack.us/img261/8528/loadwd2.th.png (http://img261.imageshack.us/my.php?image=loadwd2.png)

[Sailindawg]

Thanks Wotcher. The pics are much appreciated and I got to see what I needed. Can you take a measurement of the mosfet screw holes, from top screw hole to bottom screw hole? I'd like to know that distance. Looking at what you did, I have been thinking of pitching the stock heatpipe assembly and getting a Thermalright mosfet cooler and a copper SB cooler. The NB would go under an aftermarket water block.

Excellent job!!:up: The pics were mucho appreciated!!

Thanks again.

[Wotcher]

Can you take a measurement of the mosfet screw holes, from top screw hole to bottom screw hole? I'd like to know that distance. That would be a little difficult to do accurately as the cooler is on the board and the board is back in the case :)

It wouldn't be more than 3 or 4mm iirc, there are only a couple of threads in it. Given the top side of the cooler flange is quite large I would suggest you would be able to use a set screw and nut instead of utilising the included threaded material if you drilled the hole to a slightly larger diameter.

Cheers

[Sailindawg]

Wotcher, I apologize for not being very clear in my post, I was trying to ask if you knew the distance between the mosfet screw holes from top to bottom, not the size of the screw hole itself. Any how, glad to hear that you have your board back together. Looking at the stock mosfet cooler, I ought to be able to approximate it's length. I need to have an idea of it's length when ordering a Thermalright mosfet cooler. They make them in a 61mm & a 70mm variety.

[Wotcher]

Oh, my bad lol, you meant hole centre to hole centre along the length of the cooler, sorry about that I misinterpreted your question.

It may be possible I can get that measurement with the vernier caliper with the board in place though I think the leg of the vernier will foul on the video card, if that is the case I'll pull the video card out.

I'll power down and have a go and report back shortly.

EDIT:

The leg of the vernier was too long (250mm, 10") to get a reading. I got my 6" steel ruler in there though and measured 95mm centre to centre. While this is not as accurate as the vernier would have been I would suggest it is within +/- 1mm.

Cheers.