How to: Improve your XPS M1330 Cooling

yeh keep us updated...

 

Thx for the tip. They've even gone as far as "advertizing" that size cut-off as "used for M1330 cooling", lol! I ordered it via their Ebay store.

Just one question, to those that have already "mounted" theirs: the site states that only one side of the copper sheet is polished:

"The front face of the sheet is protected with a plastic film; the back face may have scratches caused by the sheet cutting process."

did you "lap" the unpolished side, or just implemented it as is?

 

When mine came it was smooth on both sides. Did not do anything other than using thermal paste on both sides.

 

In my thread http://forum.notebookreview.com/showthread.php?t=264854 someone suggested bending the copper piping to get the heat pipe to touch the GPU.

I have heard of the thinkpad crowd doing this. Has anyone done this with the 1330?

How much did that cost?

 

Has anyone considered soldering the copper or aluminum sheet onto the heatsink to improve thermal conductivity. I really am worried about this as I just bought a refurb M1330 and was really excited to get it until I just read about this.

 

12.64 for the piece of copper, including shipping.

31.00 for the cooling pad.

Time wasted-- priceless.

 

I wrote an e-mail to them. Basically they stopped responding after they found out I'm outside of the US. Does it matter where their customers come from? It's the same product with the same problem anywhere in the world.

 

Damn I saw this thread and Im scared now...

Anyway its time to get some copper and grab my trusted AS5, maybe I can even overclock the GPU doing this mod.

 

You really don't need to do that. The life of the CPU and GPU starts dropping after the temp starts hitting 80C. If it's below that, I don't think you really need to worry anymore, especially since you won't really notice the on the bottom anymore. The GPU is rated greater than 80, but I'd use 80 as a rule of thumb for long life.

 

I can overclock without seeing any temp increase yet, but that's not to say that the internal temp doesn't become more critical. I'll stick with the rated clocks for now.

 

anyone know how can I get this copper sheet in melbourne, australia?

 

I can look at what it would cost to ship you some from my leftovers from Japan. I ended up paying around $25 US for the copper. I don't think I'll end up cutting it to the exact size...The guy who did mine cut around the "L" and it works just as great.

 

Anyone in the United States willing to send me an extra piece of copper 14mm x 14mm x 1.5mm cheap? I just bought this laptop from Dell Outlet and can't wait to get it, but I have only a 1 year warranty and need it to last well over 4 years for college. Can't afford for anything to go wrong with the laptop.

 

I wouldnt mind getting if its 1 layer of copper with a thickness of 1.5mm.

 

Can't help you there...only 1mm and .5mm.

 

Can be ordered here, they ship internationally. Why not e-mail them for a quote?

http://metaloffcuts.co.uk

 

I emailed them yesterday and got an immediate reply, with a quote for the piece plus shipping at $12.15 USD, that's 50 cents cheaper than shipping from a US company. I cancelled my US order and placed an order with the UK firm because they could deliver KenHT's magic specs exactly. And they shipped out right away. Whole process took under 15 minutes.

 

That would definitely be an issue warranty-wise, but aside from that, the copper heat-pipe is sheathed by a cast-iron (or cast-alu?) brace which supports the heat-pie & fan assembly.

See my 1330 photo "set" to get a better idea of what I mean.

 

Lol, I'm betting it will become their (Metaloffcuts) best-selling sheet-metal cutoff for a while to come... :twitcy:

 

It's Aluminum
that's why i said aluminum sheet will have the same cooling effect

 

Wow, your pictures of the notebook are very well taken

Btw, I don't hold any interest with Metaloffcuts, just thought it works for me, it'll work for others too. Also I had no issues dealing with them, thought I'd do a bit to help them make up from volume, they do price it very reasonable.


Cheers

 

Thx

I actually wanted to order a 2mm version too but I was to lazy to call them up and they only had "your" custom size on ebay available...

I don't know much about the cost of sheet metal & effort required to precision-cut a square, but I've got no problem parting with ~ €7 if it means saving me the sheer aggravation of another mobo replacement!

 

Does the piece of copper cost like nothing but the shipping is killer? We should set up a group buy so people can get their copper for so much cheaper if like 5 people participate in a group buy and each piece costs 50c then it'll come out to be like .. $2.50 a person.

 

Has anyone been able to contact metalcutoffs? I've tried a couple of times already.

 

Yes. They replied in a few minutes to my email. Try sending them another email from another email provider, maybe your email got stuck in their Spam folder.

 

Hmm.. it's actually quite ridiculous reading about your ordering and prices for a piece of cutoff.. $25?!? + shipping? for a sq.cm. of copper/alu?
Just use a flat, thin fin from an old heatsink, go to your local mechanics shop, electrician, roofing company or similar to obtain a free piece of sheet cu/alu.. but first I really would remove all heatpads and check the clearance difference between the GPU, NB and CPU - maybe 1,5-2mm are needed for the GPU when the two others remain?

 

The CPU is not using heatpads. The NB is fine on it's own. The GPU is a 1.5 to 1.8 gap. 1.5 is just right to make it seat at the stock height with enough room left for the spring loaded screw to cope with thermal expansion. 2mm is cutting it too close, i'd be concerned about cracking the chip.

That said yes, USD25 is very expensive. Even buying it off from another site in the UK and shipping overseas will be less than half that price.

 

I know you meant to say "on it's own with original thermal pad" but I just thought I'd point that out.

I think mine ended up at €7 ($12) incl. shipping - which is half the price your quoting (thanks to the minimal shipping costs "within the EU"). Sure, that's still a lot of spare change for a tiny piece of Cu but I don't have the time to go looking for ways to "re-shape" some existing piece of copper from some old HS and then hope the size is close enough.

No way. I admit that if you're earning USD, $25's too much but €7 won't even get me a couple of (good) beers, so it's a no-brainer if you live on this side of the pond. For the North Americans, there must be a similar dealer of sheet metal that can deliver with "local" postal fees

p.s. I ordered mine Friday & got it today...

 

I got quotes from both a US manufacturer and Metalcutoffs. The US was ~12.62 including shipping to the US, the UK was 12.12 including shipping to the US.

I think the person who originally quoted $25.00 is currently in Japan, not known for bargain basement prices. And, even at that rate, I'd pay it. Better than the $350.00 I paid (and cancelled) for extending my warranty.

I will post when I receive my copper chip, ordered last week from UK, to let you all know how long it took to ship.

 

That is one of Dell's moves I believe: letting people buying extended warranties after one or two repairs in the first year. The owner of a 1330 gets scared about the repeated mobo burnouts and orders an extended warranty.

 

That's exactly what happened to me. Until, after a week or so of ruminating, I realized that 1) it would be cheaper to buy and replace the motherboard myself; and 2) if i really NEED to keep replacing motherboards for three years, I will get a better computer. Luckily I caught my error before the option to cancel my purchase expired.

 

Thx once more...
... and I hope you all like the additional "Project chill Out!" pics I added recently after completing the Cu shim mod

As I posted elsewhere, I'm still holding my breath (at least until the AS5 "cures" overnight) because while the GPU temp's better now, the difference is nothing to write home about...

After doing the mod, I realized that if the HS doesn't sit flush on the Cu shim (due to any angle other than parallel), heat dissipation will be anything but optimal...

 

Hi, I am a mechanical engineer and would like to point a few things out.

The property of a material to conduct heat is identified as thermal conductivity.

Thermal conductivity normally is identified as letter k.

In layman terms, the bigger the thermal conductivity k, the better the material to conduct heat. Why I point this out?

Aluminum k = 230 Watts/[meters*kelvin]
Copper k= 400 Watts/[meters*kelvin]
diamond k=1000 Watts/[meters*kelvin] <--for the extreme and rich
This means that on same temperature conditions Copper will be 174% better than aluminum.

You need almost twice the aluminum to dissipate as much heat as copper do. The downside of copper is its weight. In XPS M1330 weight was a very important factor, thats why they chose aluminum instead of copper.

Copper is a better solution for its thermal properties compared to aluminum.

In my opinion, Dell engineers overlooked heat issues which is critical using densely packed electronic systems.

 

I have also discussed this matter with an "expert" and he said that because of the contact of the HSF that is made from Aluminum, Copper and Aluminum foils will get the same temperature decrease.

 

i got my motherboard and heatsink replaced today by a dell technician
temps are still about the same on iidle about 46 degrees where as before it was 51 degrees. I was expecting a bigger improvement.

Iv got the CPUID Hardware monitor, how can i check my GPU temps, only has Core 0 and Core 1 and HDD temps?

 

I got this issue too. It was from the video driver. Sleep state was also missing and came back after I switched to another video driver. Try and change the video driver. 169.09 and 177.41 from laptopvideo2go.com worked for me.

 

Yes, that's the mod-world's understanding too, however there is one factor that interests me that maybe you could (also) shed some light on.

In my older, Pentium 4 based PC, I have a Zalman CNPS7000-CuAL CPU HSF and as the name implies, it's a mix of Cu & AL.

The base which mates with the CPU is of solid Cu and most of the fins are in AL. The reasoning behind this combination was that AL is better at "releasing" heat than Cu and so Zalman made most of the upper-part (fins) out of AL.

Zalman also made the exact same HSF out of pure Cu ( CNPS7000-Cu) and in a comparison test, the CuAL actually was more efficient. The difference was maybe a 3-4C reduction in CPU temp but nevertheless, a reduction!

What's your take on that side of the equation (i.e, retention of heat)?

 

Now that me & my 1330 had a good night's sleep (that lasted roughly 5hrs, lol), I am ready to post the figures and one or two "observations" that one else has experienced / noticed / or taken the time to mention.

I did 4 specific tests to verify not only GPU, but CPU temps and stability. Pre-Cu followed by post-Cu column. CPU & GPU values taken from RivaTuner, all others from 18kfangui. Ambient @24C.

TEST 1: @Idle
(Sidebar + RivaTuner Monitor + 18kfangui + the usual Dell TSRs. Note that temps react linearly to fan speed which is automated by the [A09] BIOS)
Core0 ..... 33-43 .... 38-48
Core1 ..... 35-46 .... 42-50
GPU ....... 58-67 .... 56-64
Memory .... 44-47 .... 44-46
Chipset ... 44-49 .... 45-47
Fan RPM ... 3050-0 ... 3035-0


TEST 2: Intel Thermal Analysis Tool
(@80% load for both cores)
Core0 ..... 79 ..... 73
Core1 ..... 81 ..... 74
GPU ....... 80 ..... 74
Memory .... 55 ..... 56
Chipset ... 58 ..... 59
Fan RPM ... 4352 ... 4276


TEST 3: 3DMark06 - CPU Tests
Core0 ..... 73 ..... 68
Core1 ..... 79 ..... 73
GPU ....... 80 ..... 73
Memory .... 55 ..... 56
Chipset ... 57 ..... 59
Fan RPM ... 3725 ... 4352


TEST 3: 3DMark06 - Freeze (GPU) Test
Core0 ..... 55 ..... 57
Core1 ..... 57 ..... 60
GPU ....... 95 ..... 71
Memory .... 56 ..... 56
Chipset ... 57 ..... 57
Fan RPM ... 4352 ... 3754


TEST 3: Half-Life 2: Lost Coast Tech Demo
(1280x800, textures & effects on high)
Core0 ..... 57...... 60
Core1 ..... 58 ..... 66
GPU ....... 100 .... 75
Memory .... 58 ..... 57
Chipset ... 58 ..... 58
Fan RPM ... 5018 ... 4352


So those are pretty much the results. "Idle" results are below average (read: not as good), from what I've read on NBR. The key result - being the GPU temp @load, however is looking great! Take 3DMark06, for instance: 95C to 71C, a drop of 24C, is as much as I could have hoped for

But of course, there is the question of how the other components are affected by the mod. At first glance, the cores are now idling a bit higher, but - what's this - they are running cooler @load! A byproduct of the mod? I think not, more likely a byproduct of my better application of AS5**. I can't explain the decrease in any other way...!
**I also had the rep use AS5 when he swapped out my mobo, but maybe he was too sloppy, dunno...

Unless... look at how the GPU temp increased when the CPU was @load. Why? This must be due to the common HSF solution, because I cannot come up with any other logical reason... . This would imply that the opposite case is also true: Hotter CPUs result in a hotter GPU, even if the GPU is not in use.

--------- Start of Edit --------
Another interesting observation I made is that during 3DMark06' CPU tests, the GPU temp was hotter than it was during the Freeze (GPU) stress-test...! 73C compared to 71C...

Finally, back to the idle "test": what the above figures don't reveal is the period of time between the min & max temp (fan rpm) values. The whole process can be "pictured" as a simple sinus curve.

Thermal-pad configuration:
GPU Starts out at it's min of 58C & rpm is 0. After "n" mins, GPU has reached it's approx. max value of 67C and the fan kicks in (~3050 rpm). After "m" mins, the GPU temp's back down to 58C & the fan cuts out.

Cu configuration:
Same deal, different values (56-64C, 0-3035rpm). *But*, the time factor has changed! Now it takes quite a while longer, maybe 1.75*n before the GPU hits it's max of 64C and the fan kicks in. Furthermore, the cool-down time (m) is shorter!

Long story short, the less the fan is on, the more battery-charge life you get
--------- End of Edit--------


In any event, I am pleased with the results and I think it was worth the effort in the end :yes:

 

So helpful!! I tried to rep you, and got the message that I have to find somebody else to be grateful to first. Oh well.

I will post my own results as soon as I get my chip and do the mod. I have a couple of questions:

1. I was quite surprised by the amount of AS you used judging from the photo, and the AS oozing out the side-- my understanding has always been that one should use much less than that-- a "skim coat" at best. Could you (or anyone) give a brief description of how much you use (AS website describes the ideal starting amount as 1/2 an uncooked short grain of rice...)?

2. I am not a gamer, but I AM interested in the performance of the GPU at load over time, which I can't get using 3dMark (I assume). Could you point me in the direction of a demo that might allow me to run a game like deep freeze for 20 minutes or so? (I downloaded the demo you referenced above, but it looks to be just an .avi file....)

Thanks again!!

 

Great analysis traveller, you're probably right in assuming a hotter CPU will mean a hotter GPU, that's what others have reported as well. Interesting to see that a hotter GPU does not necessarily result in a hotter CPU though.

I'll get my metal next week, will report back with results too.

 

Yes, I fully agree.

I would have thought it would be quite the opposite and I was worried that the GPU's improved heat dissipation would decrease the CPU's, but the numbers show otherwise...

 

Thx

I figured that we should all see what the mod does in all possible scenarios (idle, GPU@load, CPU@load, mobo "surface" temps, etc.).

We don't want some poor 1330 owner that spends 90% of the time editing video and only 10% doing 3D work to get an overheated CPU because of the mod, etc.

I'm quite thrilled about the side effects (battery-charge life) and can certainly recommend this over the A11 "workaround" any day! Those of you Cu modders who already updated to A11 can relax though, because the temps are anyways lower and thus the fan still shouldn't kick in more often*!

*read: 65 and 70C triggers are now, for all intensive purposes, equal.

 

Did all those who did copper mod also replace the CPU thermal paste? Is it even required to meddle with the CPU and its existing thermal paste when doing the copper mod?

 

You have to clean and reapply new Termal paste each time you remove the HS, or else the cpu will be damaged.

 

Bump....?

 

Dell XPS M1330 - nVidia GeForce 8400M GS - Copper Mod - Step by Step Guide.

 

Thank you so much. Timing is perfect-- my chip just arrived in the mail. From the UK to the US in 3 business days!!! More to follow....

 

how tight should the screws be?

 

The screws should be tightened without applying pressure. If you can no longer turn a screw unless you apply pressure, you have tightened it correctly.

 

traveller:
Thanks for the thermal tests. Now I have something to compare with. My GPU has stayed at or under 75C for every test I have run to include overclocked test which is really nice. The CPU and GPU connection does not really affect the other's temp even over a prolonged period. This is probably due to the A11 and the fan kicking in.
I would recommend staying away from the 80C on your CPU as that is action the point were CPUs start burning out over time. 85C is actually the point were CPUs will typically have less than 5 years of life. 80C only means your chip will not last longer than you do. I ended up using RMClock to kick the CPUs max multiplier from 11 to 10. This runs mine at 79C as opposed to 82C and drops the speed from 2.2 to 2.0GHz. I think I could possibly have done better with the thermal mod, but I'm not going to mess with it for now.