Vostro 3560 cooling mod

spleenharvester

spleenharvester

spleenharvester

Associate
Joined
4 Dec 2008
Posts
1,812
Hi all,

I've just upgraded my trusty old Vostro 3560 from an i3 to an i5, and it's running quite a bit toastier (75*C under load with the i3, but now peaking at 90*C with the i5), despite the two processors both having a 35W TDP.

This model comes in GPU and non-GPU flavours (mine is the latter) - both of which have a different heatsink:



7ZTqKVC.jpeg.png





I've done some digging and it seems the chassis for each model is the same, and the only difference is that the ATI chip is physically missing from the UMA motherboard (even the solder points are all there). My understanding of how heatpipes work is a tad limited, but am I right in thinking that if I replace the heatsink with the GPU version, the heat should be distributed to an extent across both sections (+that extra aluminium bit surrounding the fan), and thus this should lower the temperatures a fair bit?

Thanks :-)
 
Apart from the risk of shorting out where there is no gfx chip only solder points i do not think it will help
The heat should flow from the heatsinks to the fins,not bleed back into the gfx heatsink
Maybe if the gfx heatsink fan was faster it might help but if not i do not think it would help.Might even make temps worse
 
Just had a closer look at your pics and the pipe length is different from gfx to non gfx version so i would agree it won`t fit/work
 
I`m not sure whether thats correct. The heatpipe length is different because the exhaust end comes into the outlet at different locations if you look closely. If you can get the GPU version cheap enough say sub £10 might be worth a punt. Providing you can isolate the bottom heatsink so its not touching the motherboard, use some kind of foam pad. I would be tempted to dismantle the whole lot and see if you can remove the bottom heatpipe/sink altogether if its seperate to the top one. In the interest of one reason for modding (ie because you can) might be worth examining.

What TIM did you use ?
 
Thanks all - I'm 99% sure the GPU version will fit as the chassis is exactly the same, and the GPU/UMA boards are virtually identical excepting that the radeon chip is physically missing. I think the heatpipe just looks shorter as the back of the fan has metal on the GPU version, and connects up with the CPU area. (that, and the different insertion point - good catch)

Have purchased one and going to try soon - I'm going to place a spacer under it to prevent it shorting out, and might attach a spare heatpipe with thermal adhesive. Will report back with the results - could be an interesting experiment :)

EDIT: The TIM I'm currently using is new Arctic MX-2.
 
Last edited:
varkanoid said:
I`m not sure whether thats correct. The heatpipe length is different because the exhaust end comes into the outlet at different locations if you look closely.
Click to expand...

That's what I meant in post #3. GPU heatpipe attached to the heatsink nearer the fan, the CPU along side that.

The OP's original point about more distribution area is false. Doubly so as heat pipes don't work like that. ;)

Can't find a Dell one, but this is from a ThinkPad :
s506875824555543720_p175_i2_w800.jpg
 
MagicBoy said:
The OP's original point about more distribution area is false. Doubly so as heat pipes don't work like that. ;)
Click to expand...

How so - surely a certain degree of heat will be drawn back into the GPU end down a temperature gradient? My understanding of heatpipes is limited, but they aren't unidirectional are they?

A few people have told me it will just take longer to reach max temp but not actually reducing the temp, and I see where that argument comes from. But the GPU version has to take a total load of ~55W TDP, but the CPU version only 35W, despite it looking like the fin array and fan are the same - is the extra metal surrounding the fan casing perhaps the main factor?
 
They work on phase change. The heatpipe is a vacuum with a liquid in it. Liquid evaporates at the hot CPU/GPU end.The vapour is drawn towards the cooler heatsink end where it gets cooled and condenses back into a liquid then flows back to the CPU/GPU end and the cycle repeats. Given the non attached GPU end isn't a heatsink it'll have virtually no cooling effect.
 
MagicBoy said:
They work on phase change. The heatpipe is a vacuum with a liquid in it. Liquid gets evaporates at the CPU/GPU end.The vapour is drawn towards the cooler heatsink end where it gets cooled and condenses back into a liquid then flows back to the CPU/GPU end and the cycle repeats. Given the non attached GPU end isn't a heatsink it'll have virtually no cooling effect.
Click to expand...

Interesting - without knowing the substance in the heatpipe, surely there must still be a dynamic range here though? ie the heat at the fins is still going to be greater than the heat at the GPU end, which will be at room temp ish, so there will still be some movement from the fins to the GPU end?
 
You pick the substance depending on the temperature range you want. GPU end doesn't have a heatsink or active cooling, so you'll get a tiny bit of heat soak into the pipe and die contact area then nothing I expect. You'd get better results running the existing fan faster...
 
I might have a go at epoxying a heatpipe between the CPU-GPU plates after all then - would have been great to compare them unmodified to the UMA heatsink for the sake of discussion, but the extra bits of metal surrounding the CPU side would make the data noncomparable.

Anyone know the best way to remove aluminium fins from an old heatsink?
 
Last edited:
So I did the mod - went fine, and all fitted OK. Temps at idle are unchanged (with the fan tripping on/off @ 50*C this is perhaps to be expected), but load temps have been reduced by about 12~15*C. Haven't yet tried bridging the CPU and GPU plates. Might give that a go, and if there's no advantage, I might just chop the GPU bit of the heatsink off (since I believe the metal casing around the fan is probably most responsible for the temp drop).

The Sunon-branded fan is a fair bit quieter than the Forcecon fan though at the same RPM, so I'm definitely keeping that either way :-)
 
So odd problem - I've actually noticed that the CPU cores are down to ~35*C at idle, and it's the HD 4000 core (and hence the package temperature) that is idling at ~50*C. Any idea why there would be such a big disparity here?

Also in reply to above - both fans are 2.4CFM and are used interchangeably on both types of heatsink so I don't think it's attributable to the fan alone, and the old TIM was fairly new arctic MX-2 :)
 
I recently was given another dead system of the same model, that had the GPU version with heatsink and backplate. Although I repaired the dead board so that it would boot, it still won't POST, so I decided to use it to do a slightly neater version of this mod on my working system ;).

I used the new heatsink (as there was a small dent in one of the heatpipes on the other one I bought), cut down a rubber spacer to size and placed it in the unpopulated GPU section, then peeled off the backplate from the dead board and stuck it on my working board and put screws through:

oBPqULO.jpg


fsQrUc5.jpg


Much more robust than the first attempt (which held the GPU section in place by kapton tape and hope!), and temps have now dropped even further - the UMA heatsink was peaking at CPU 92*C~ usually with the fan running full tilt. Under the same usage conditions, the CPU package now peaks at 70*C, with the fan not running at full tilt. I consider it to be a great success!

I might have a go at adding a fan grill to the system, as with the UMA version temps would drop about 15*C with the back cover removed. I'm not sure if I'd see the same improvement now that the mod has been done but it's worth a shot I think.
 
You must log in or register to reply here.
Back
Top Bottom