Well, cold bug is destroyed, no doubt.. but I'm faced with a new bug.. the hot bug
My temps seem to have increased significantly since I first booted up with the reattached spreader. I didn't think anything of it until I noticed F@H WUs ending early
I used to be 110% prime95 stable at 3.2GHz 1.52V. This is my everyday gaming/folding config. Strangely, I can't even keep it prime stable for longer than 1-2 hours now. Temps while folding are in the positives once again (jumps around between 0 and 5'C). I thought I'd just go ahead and bump up my vcore a notch to take care of that problem, but now the chip responds to vcore very negatively. The temps skyrocket once I surpass 1.55V. At 1.56V, I was actually loading at about +10'C, and stability is not improved as a result.
I have no problems with cold-bug freezeups at 3.3GHz, but It's now nearly impossible to maintain stability at 3.3GHz. Running Prime95 small FFT's, the load temp gets up to +19'C at 1.65V, which is ridiculous. The evaporator temp barely breaks -39'C, so it is clearly not pumping the heat away from the core efficiently. I get prime failures within 1-5 minutes now at 3.3.
Out of frustration, I've backed it down to 3.1GHz. All seems ok, but I was not expecting this
I haven't had too much of a chance to tweak, but I'll update again once I make some more progress.
I don't know, but maybe when you reattached the IHS, it raised up enough off the core so that it isn't making a good clean contact bwtween the core and IHS now and is depending upon the TIM for contact too much. I'm thinking that you might have to get all the original silicone off from around the edges of the proc pcb, then reattach the IHS or maybe even taking the IHS and lightly sanding down the lip (the part that contacts the proc pcb) just a little bit so that you make a better contact between the IHS and core. What do you think, lemonlime?
I don't know, but maybe when you reattached the IHS, it raised up enough off the core so that it isn't making a good clean contact bwtween the core and IHS now and is depending upon the TIM for contact too much. I'm thinking that you might have to get all the original silicone off from around the edges of the proc pcb, then reattach the IHS or maybe even taking the IHS and lightly sanding down the lip (the part that contacts the proc pcb) just a little bit so that you make a better contact between the IHS and core. What do you think, lemonlime?
I think you are right about that mudd.. When I did my test fittings of the IHS, the contact appeared good, but I doubt it has the same amount of pressure than the factory seal had. I purposly left the original silicone there too, becase I didn't want to do too good of a reattachment job I think that I'm going to have no choice but to pull it off again and correct this. Its strange that the temps are getting worse and worse, but thats probably due to the TIM changing/warping slightly with use. It started out with better temps than the original IHS, now it's actually significantly worse. I've moved back to stock clockspeeds, because the temps really are getting out of hand.
Just for perspective, I brought it back to default (2.2GHz). I'm running prime95 right now, with small FFT's. It's full load temp is -9'C. Without the IHS, I had the same load temp at 3.2GHz, with 1.52V. I remember full loading at stock clocks at -19'C before.
I'm actually considering to run the chip without the IHS once again I know, whats wrong with me.. but perhaps I can simply raise the temps a degree or two by using a non-phase friendly TIM, along with weaker mounting pressure.
Or.... I could snap and pick up an opteron 165 although too many of those are bugged as well Where are these conroe chips??
I'll keep you guys posted.. this adventure is far from over
I couldn't believe that 3.1GHz wouldn't remain prime stable either. This chip reacts so poorly to vcore increases now, that I pretty much have to stick to default vcore. Although the temp is higher than it should be, I was surprised to see 3.0 at default vcore This will have to do until I can get some time to tear everything down again .
I'm very tempted to just pick up one of the new LDBHE 06xx stepping manchester X2's. They have rock-solid memory controllers that are good for very cold operation (up to about -50'C). Many are hitting 2.6 undervolted.. I do have to do something about this Opty 148 though, I can't leave it this way
Pure silicone can handle extreme temps without noticeable movement we send it through ovens at 240-260 C without shrinkage or expansion. Though high grade silicone can go as low as -60C and lower without being affected, the molecules do slow down i.e. the product gets stiffer when cold or speed up when hot etc. When your using low grade silicone sealant this could be greater. We use e-43 silicone to stick metal and other silicone products together this is just as acidic as bath sealant, but once cured any adverse effect from the acid is minimal your IHS should be fine. Sealant can take up to 6 days to cure properly and sufficient weight should have been applied for the duration
Pure silicone can handle extreme temps without noticeable movement we send it through ovens at 240-260 C without shrinkage or expansion. Though high grade silicone can go as low as -60C and lower without being affected, the molecules do slow down i.e. the product gets stiffer when cold or speed up when hot etc. When your using low grade silicone sealant this could be greater. We use e-43 silicone to stick metal and other silicone products together this is just as acidic as bath sealant, but once cured any adverse effect from the acid is minimal your IHS should be fine. Sealant can take up to 6 days to cure properly and sufficient weight should have been applied for the duration
Thanks for the info calous3.. I let the silicone cure for about 36 hours, with some weight on the chip. I basically centered a spindle of 50 CD-R's on the chip and let it sit there. It seemed to do the trick nicely. When I inspected it, it seemed to be fairly hard, and the IHS didn't budge. I didn't bother doing any more than just the four corners, because I did plan to remove it again once this chip gets demoted back to air cooling.
I'm starting to wonder if condensation got under the IHS and degraded the contact and TIM. I thought I had corrected the insulation problems, but perhaps not. I hope the CPU is not floating in a pool of water again.
I suppose I have the following options:
1) Leave it as-is, use default vcore, and stay limited to 3.0GHz (yeah right.. you guys know me well enough that I'd never settle for this )
2) Attempt another reattachment job, which may get me back to 3.2GHz at reasonable temps. It is possible that it will degrade once again
3) Remove the IHS again, and try to restrict temperature by reducing the evaporator mount pressure, cranking heaters to 100%, reducing vapo fan speeds and using 'too much' TIM.
Right now, I'm really leaning towards option #3. Worst case scenario, I'll just have to settle for 3.2GHz with my ram on a divider, which seems to work just fine. I think I've come to the conclusion that this chip simply won't do 3.3GHz 24/7 prime95 stable. It failed after about 4-8 hours, before I even touched the IHS.. With it removed, it gets coldbugged hard, although primes no problem 24-hours (just cant take load off the CPU). With the reattachment job, it was impossible and failures happened in less than one minute.
Now I just have to put down Oblivion for longer than a few minutes to get this done!
0
LeonardoWake up and smell the glaciersEagle River, AlaskaIcrontian
edited March 2006
calous3, welcome to Short-Media. Thanks for your technical input. We always enjoy that here very much. I hope that you stay active here and find us a fun, learning community.
no real expertise in phase change or any other exotic cooling, however i can't help but wondering if the the fact that you only put silicon around the 4 corners of the IHS is the problem. i think chip makers use a complete seal around the IHS for a reason---probably thermal.
i would remove the IHS and inspect the core (hopefully it isn't covered in condensation), clean everything, and then re-attach the IHS and make a complete seal this time.
I am willing to bet things will be better...
if this doesn't work, i then think TIM material between core and IHS needs to be examined. maybe the cheapo white paste from the factory acts better under extreme cold than AS Ceramique/AS5?
no real expertise in phase change or any other exotic cooling, however i can't help but wondering if the the fact that you only put silicon around the 4 corners of the IHS is the problem. i think chip makers use a complete seal around the IHS for a reason---probably thermal.
i would remove the IHS and inspect the core (hopefully it isn't covered in condensation), clean everything, and then re-attach the IHS and make a complete seal this time.
I am willing to bet things will be better...
if this doesn't work, i then think TIM material between core and IHS needs to be examined. maybe the cheapo white paste from the factory acts better under extreme cold than AS Ceramique/AS5?
Hey Ryko, thanks for the tips.. When AMD attaches the IHS, there is a 3-4mm openining in the adhesive, so even out of the box, condensation could get under the IHS. You can see that opening in some of my earlier pics. I didn't worry about that too much, as I covered the sensitive bits in dielectric.
Even still, it may be worthwhile sealing it with some other material (something very easy to remove). I'm not sure if there are cons to preventing it from breathing under there though. Hmmm... can't win eh
I wouldn't chance 100% seal ...especially with those extreme temps.
I really liked the idea of only adhering 4 corners personally ...wish you could get that to work.
Update: I really don't want to rip apart my PC now.. Oblivion is too good
3GHz holding OK, temps do not appear to be getting any worse, although they are way higher than they should be. Might break everything down in a few days.
I've pretty much made up my mind to simply remove the IHS again, and do my best to combat the cold bug. Putting my ram on a divider will do the trick. 3.3GHz is just not obtainable for 24/7 use, I need to accept that fact and move on
Well, I finally broke everything apart and decided to fix what I broke, after I broke it a second time.. err.. well, you get the idea. It was clear that the best bet was to simply run the CPU without the IHS once again, and leave 3.3GHz for another day. Anyhow, first task: Remove the evaporator.
What the bloo'e'hell? I could have sworn that I reattached the IHS. Where did it go?
Oh there it is! WTF! After looking at it closely, there was a bit of condensation that made its way under the IHS. It was not nearly as bad as the last time, but it definitely got a little damp. The TIM peeled right off the core. The IHS was held down tightly with perssure from the mount, but the TIM appeared to get eaten away by condensation as I had originally feared. The silicone adhered well to the PCB, but came loose at the IHS. Clearly the four corners are not enough, and it needed to cure longer.
Okay. There is a lesson here.. I'm certain of it..
#1: Don't remove your IHS on phase-change unless you are damn sure your chip is cold bug free.
#2: If you were crazy enough to remove it anyway and need to reattach, use silicone around the entire IHS, and ensure that you remove all of the AMD adhesive. The four corners are clearly not enough.
#3: Create a seal to prevent condensation from slowly eating away the TIM.
#4: Let the silicone cure for longer than 24 hours.
#5: Don't remove your IHS on phase-change.
#6: Don't remove your IHS on phase-change.
So there you have it. After all of that work, I'm right back where I started. Although my temps are fantastic now to say the least. I'm at 3.2GHz, 1.52V and I'm full load prime95 small FFT at -8'C. Memory is on a divider (228MHz currently with no sign of cold bug). I used arctic ceramique this time around.
I'd follow calous3 tips here and seal all around after a good cleaning. Do it with the chip in the socket and then just put the heaviest heatsink you have or just clamp the standard amd heatsink on with a couple of sheets of paper in between the heatsink and the ihs for added pressure.
I'd follow calous3 tips here and seal all around after a good cleaning. Do it with the chip in the socket and then just put the heaviest heatsink you have or just clamp the standard amd heatsink on with a couple of sheets of paper in between the heatsink and the ihs for added pressure.
Yep, I'd agree that is definitely the best way to do it. I've decided just to run it without the IHS. I had to modify the evaporator to protrude about 2mm from the clamshell, but all is well now. It is difficult to get a consistent mount like this-- when I put the case window on, it fudges up the contact. I think I may need to put another thin spacer in for the evaporator head.
Since my computer will be down for a week while I wait for a RAM replacment (and a new HDD... don't ask). I started to think about popping the IHS on my Opti 165 to get the temps nice and low with 1.5v on the core.
My questions:
My waterblock (DangerDen TDX) uses spring-loaded bolts to hold it down. Is there any possibility of crushing the core with this meathod (the springs should keep regular pressure on the core no matter how much I tighten the bolts.. I believe)?
How much space is there between the top of the CPU core and the top of the surrounding caps?
Would I be better off buying a small, thin peice of copper with just a tad larger diameter than the CPU core, polishing both sides, placing one side on the core with AS5, and using AS5 on the other side to interface with the waterblock?
I would imagine this would help a lot with both protecting against chipping the core, and keeping the bottom of the waterblock off of the surface mount capacitors surrounding the core, all without effecting the temp much at all.
good thread - I use an artic cooling Freezer 64 Pro - with my Opty 170, if I popped the IHS would it be possible to still use this CPU cooler without crushing the core?
anyone tried something similar??
Am only considering this when I get a new updated rig April - Aug time next year, as I am thinking of going for a dual CPU clovertown build.
Since my computer will be down for a week while I wait for a RAM replacment (and a new HDD... don't ask). I started to think about popping the IHS on my Opti 165 to get the temps nice and low with 1.5v on the core.
My questions:
My waterblock (DangerDen TDX) uses spring-loaded bolts to hold it down. Is there any possibility of crushing the core with this meathod (the springs should keep regular pressure on the core no matter how much I tighten the bolts.. I believe)?
How much space is there between the top of the CPU core and the top of the surrounding caps?
Would I be better off buying a small, thin peice of copper with just a tad larger diameter than the CPU core, polishing both sides, placing one side on the core with AS5, and using AS5 on the other side to interface with the waterblock?
I would imagine this would help a lot with both protecting against chipping the core, and keeping the bottom of the waterblock off of the surface mount capacitors surrounding the core, all without effecting the temp much at all.
Thoughts? Ideas? Am I just crazy?
I put a lot of pressure on the bare core using the vapo evaporator head. It mounts similarly to the TDX (two spring and post based mounts). I'm pretty sure it can handle a lot of mounting pressure.. so long as it is even pressure. Just alternate from one post to the other tighten the same amount each time. Most people have crushing issues around the corners. I wouldn't bother with the piece of copper.. you should be fine on the bare core so long as you are careful.
As far as the capacitors are concerned.. The A64 cores are much shorter than the AXP ones. There is literally a fraction of a millimeter between the capacitor height and the core height. I had dielectric all over mine, so shorting wasn't a problem if the evaporator head shifted at all. I'd recommend 'taping' off the area surrounding the core with standard electrical tape. Should give you some extra short protection just in case.
Good luck.. be sure to post pics!
Just a couple tips: Be very careful when cutting the epoxy. Always angle the blade just a tad 'upwards'.. make sure you use a blade from a shaving razor, it will save you a lot of frustration as well.
I don't think its a crazy idea.. crazy for phase-change.. maybe, but for water/air, you can see some pretty significant decreases in temperature.
Well I found a treo of "Double bladed" CVS brand disposable razors for only $1, so I bought them and got to work.
I noticed it's a whole lot easier to get the blade under the IHS by first starting in the area without any glue, and in short sawing motions while pulling twords you against the IHS.
Here are a few pictures. The first couple are with the IHS removed, the IHS itself, and the blade w/ cotton cloth I used to cut the glue.
The blades in the razor are easily the thinnest razors I have ever seen or used. Excellent idea LL!
And yes, that is some of my blood on the cotton cloth.
Here is the CPU with the core cleaned off and black electrical tape applied over the surface mount caps (ala LemonLime's suggestion).
In the end all it took to remove my IHS was $1 in disposable razors, a cotton cloth and about a tablespoon of blood.
The topless CPU still works! At 9x300, 1.5v it tops out at 43*C under load where I normally would get 50*C.
The temp could in fact be lower, but the install of Windows I used to test was pre-SP1, and if I remember correctly SP2 w/ the latest A64 platform drivers report the temps a bit lower.
As for the IHS, I think I'll drill a hole in it and make it a new keychain.
EDIT:
Take a look at this article about checking to make sure the CAM isn't pushing your waterblock out of alignment. I'll check mine as soon as I get home.
The topless CPU still works! At 9x300, 1.5v it tops out at 43*C under load where I normally would get 50*C.
The temp could in fact be lower, but the install of Windows I used to test was pre-SP1, and if I remember correctly SP2 w/ the latest A64 platform drivers report the temps a bit lower.
As for the IHS, I think I'll drill a hole in it and make it a new keychain.
EDIT:
Take a look at this article about checking to make sure the CAM isn't pushing your waterblock out of alignment. I'll check mine as soon as I get home.
The CAM clearance was fine. Great news for me because I really didn't want to take it off the system again and sand it down.
The problem with those razors is they're so sharp they can cut pretty deep without realizing it until you're already bleeding all over the place. Luckily the cut isn't much worse than a really bad papercut.
this in response to someones question about mounting pressure... be very careful about it. ive run my fx-60 on watercooling without the IHS now for about 2.5 months. the first round, i had some decent temps with the IHS just sandwiched via mounting pressure. now, when i finally decided i didn't want the IHS, i pulled it off, and reattached the waterblock (apogee). however, this time, i did something i shouldnt have. i tightened down the block waay too much. i would be getting frequent lockups/glitches and even prime95 would fail the torture test after about 20 seconds. both when doing something intensive and at idle. at first i thought i had damaged the core, but then i remembered that while researching what i should do for watercooling, i had read an article about someone who had tightened down their block way too much and it showed the exact symptomes as mine. sure enough, i loosened the nuts, and the problems went away. crazy eh?
this in response to someones question about mounting pressure... be very careful about it. ive run my fx-60 on watercooling without the IHS now for about 2.5 months. the first round, i was using straight distilled water, and had some decent temps (note that the IHS was on the chip, but had been removed, so it was just sandwiched via pressure. now, when i finally decided i didn't want the IHS, i pulled it off, and reattached the waterblock (apogee). however, this time, i did something i shouldnt have. i tightened down the block waay too much. i would be getting frequent lockups. both when doing somehting intensive and at idle. at first i thought i had damaged the core, but then i remembered that while researching waht i should do for watercooling, i had read an article about someone who had tightened down their block way too much and it showed the exact symptomes as mine. sure enough, i loosened the nuts, and the problems went away. crazy eh?
That happened to me too but with the IHS on. You tell it's too tight when the mobo warps (bends) at the socket.
That happened to me too but with the IHS on. You tell it's too tight when the mobo warps (bends) at the socket.
hadn't thought about it like that... although it wouldnt have helped me too much in that case. i was working on it while the board was still in the case. hell, i hadn't even taken my cooling setup apart becasue i didnt want to drain it (too much hassle, the only downside to watercooling)
Comments
Well, cold bug is destroyed, no doubt.. but I'm faced with a new bug.. the hot bug
My temps seem to have increased significantly since I first booted up with the reattached spreader. I didn't think anything of it until I noticed F@H WUs ending early
I used to be 110% prime95 stable at 3.2GHz 1.52V. This is my everyday gaming/folding config. Strangely, I can't even keep it prime stable for longer than 1-2 hours now. Temps while folding are in the positives once again (jumps around between 0 and 5'C). I thought I'd just go ahead and bump up my vcore a notch to take care of that problem, but now the chip responds to vcore very negatively. The temps skyrocket once I surpass 1.55V. At 1.56V, I was actually loading at about +10'C, and stability is not improved as a result.
I have no problems with cold-bug freezeups at 3.3GHz, but It's now nearly impossible to maintain stability at 3.3GHz. Running Prime95 small FFT's, the load temp gets up to +19'C at 1.65V, which is ridiculous. The evaporator temp barely breaks -39'C, so it is clearly not pumping the heat away from the core efficiently. I get prime failures within 1-5 minutes now at 3.3.
Out of frustration, I've backed it down to 3.1GHz. All seems ok, but I was not expecting this
I haven't had too much of a chance to tweak, but I'll update again once I make some more progress.
I think you are right about that mudd.. When I did my test fittings of the IHS, the contact appeared good, but I doubt it has the same amount of pressure than the factory seal had. I purposly left the original silicone there too, becase I didn't want to do too good of a reattachment job
Just for perspective, I brought it back to default (2.2GHz). I'm running prime95 right now, with small FFT's. It's full load temp is -9'C. Without the IHS, I had the same load temp at 3.2GHz, with 1.52V. I remember full loading at stock clocks at -19'C before.
I'm actually considering to run the chip without the IHS once again
Or.... I could snap and pick up an opteron 165 although too many of those are bugged as well
I'll keep you guys posted.. this adventure is far from over
I'm very tempted to just pick up one of the new LDBHE 06xx stepping manchester X2's. They have rock-solid memory controllers that are good for very cold operation (up to about -50'C). Many are hitting 2.6 undervolted.. I do have to do something about this Opty 148 though, I can't leave it this way
Thanks for the info calous3.. I let the silicone cure for about 36 hours, with some weight on the chip. I basically centered a spindle of 50 CD-R's on the chip and let it sit there. It seemed to do the trick nicely. When I inspected it, it seemed to be fairly hard, and the IHS didn't budge. I didn't bother doing any more than just the four corners, because I did plan to remove it again once this chip gets demoted back to air cooling.
I'm starting to wonder if condensation got under the IHS and degraded the contact and TIM. I thought I had corrected the insulation problems, but perhaps not. I hope the CPU is not floating in a pool of water again.
I suppose I have the following options:
1) Leave it as-is, use default vcore, and stay limited to 3.0GHz (yeah right.. you guys know me well enough that I'd never settle for this
2) Attempt another reattachment job, which may get me back to 3.2GHz at reasonable temps. It is possible that it will degrade once again
3) Remove the IHS again, and try to restrict temperature by reducing the evaporator mount pressure, cranking heaters to 100%, reducing vapo fan speeds and using 'too much' TIM.
Right now, I'm really leaning towards option #3. Worst case scenario, I'll just have to settle for 3.2GHz with my ram on a divider, which seems to work just fine. I think I've come to the conclusion that this chip simply won't do 3.3GHz 24/7 prime95 stable. It failed after about 4-8 hours, before I even touched the IHS.. With it removed, it gets coldbugged hard, although primes no problem 24-hours (just cant take load off the CPU). With the reattachment job, it was impossible and failures happened in less than one minute.
Now I just have to put down Oblivion for longer than a few minutes to get this done!
Leonardo
i would remove the IHS and inspect the core (hopefully it isn't covered in condensation), clean everything, and then re-attach the IHS and make a complete seal this time.
I am willing to bet things will be better...
if this doesn't work, i then think TIM material between core and IHS needs to be examined. maybe the cheapo white paste from the factory acts better under extreme cold than AS Ceramique/AS5?
Hey Ryko, thanks for the tips.. When AMD attaches the IHS, there is a 3-4mm openining in the adhesive, so even out of the box, condensation could get under the IHS. You can see that opening in some of my earlier pics. I didn't worry about that too much, as I covered the sensitive bits in dielectric.
Even still, it may be worthwhile sealing it with some other material (something very easy to remove). I'm not sure if there are cons to preventing it from breathing under there though. Hmmm...
I really liked the idea of only adhering 4 corners personally ...wish you could get that to work.
3GHz holding OK, temps do not appear to be getting any worse, although they are way higher than they should be. Might break everything down in a few days.
I've pretty much made up my mind to simply remove the IHS again, and do my best to combat the cold bug. Putting my ram on a divider will do the trick. 3.3GHz is just not obtainable for 24/7 use, I need to accept that fact and move on
What the bloo'e'hell? I could have sworn that I reattached the IHS. Where did it go?
Oh there it is!
Okay. There is a lesson here.. I'm certain of it..
#1: Don't remove your IHS on phase-change unless you are damn sure your chip is cold bug free.
#2: If you were crazy enough to remove it anyway and need to reattach, use silicone around the entire IHS, and ensure that you remove all of the AMD adhesive. The four corners are clearly not enough.
#3: Create a seal to prevent condensation from slowly eating away the TIM.
#4: Let the silicone cure for longer than 24 hours.
#5: Don't remove your IHS on phase-change.
#6: Don't remove your IHS on phase-change.
So there you have it. After all of that work, I'm right back where I started. Although my temps are fantastic now to say the least. I'm at 3.2GHz, 1.52V and I'm full load prime95 small FFT at -8'C. Memory is on a divider (228MHz currently with no sign of cold bug). I used arctic ceramique this time around.
Yep, I'd agree that is definitely the best way to do it. I've decided just to run it without the IHS. I had to modify the evaporator to protrude about 2mm from the clamshell, but all is well now. It is difficult to get a consistent mount like this-- when I put the case window on, it fudges up the contact. I think I may need to put another thin spacer in for the evaporator head.
My questions:
My waterblock (DangerDen TDX) uses spring-loaded bolts to hold it down. Is there any possibility of crushing the core with this meathod (the springs should keep regular pressure on the core no matter how much I tighten the bolts.. I believe)?
How much space is there between the top of the CPU core and the top of the surrounding caps?
Would I be better off buying a small, thin peice of copper with just a tad larger diameter than the CPU core, polishing both sides, placing one side on the core with AS5, and using AS5 on the other side to interface with the waterblock?
I would imagine this would help a lot with both protecting against chipping the core, and keeping the bottom of the waterblock off of the surface mount capacitors surrounding the core, all without effecting the temp much at all.
Thoughts? Ideas? Am I just crazy?
anyone tried something similar??
Am only considering this when I get a new updated rig April - Aug time next year, as I am thinking of going for a dual CPU clovertown build.
I put a lot of pressure on the bare core using the vapo evaporator head. It mounts similarly to the TDX (two spring and post based mounts). I'm pretty sure it can handle a lot of mounting pressure.. so long as it is even pressure. Just alternate from one post to the other tighten the same amount each time. Most people have crushing issues around the corners. I wouldn't bother with the piece of copper.. you should be fine on the bare core so long as you are careful.
As far as the capacitors are concerned.. The A64 cores are much shorter than the AXP ones. There is literally a fraction of a millimeter between the capacitor height and the core height. I had dielectric all over mine, so shorting wasn't a problem if the evaporator head shifted at all. I'd recommend 'taping' off the area surrounding the core with standard electrical tape. Should give you some extra short protection just in case.
Good luck.. be sure to post pics!
Just a couple tips: Be very careful when cutting the epoxy. Always angle the blade just a tad 'upwards'.. make sure you use a blade from a shaving razor, it will save you a lot of frustration as well.
I don't think its a crazy idea.. crazy for phase-change.. maybe, but for water/air, you can see some pretty significant decreases in temperature.
I noticed it's a whole lot easier to get the blade under the IHS by first starting in the area without any glue, and in short sawing motions while pulling twords you against the IHS.
Here are a few pictures. The first couple are with the IHS removed, the IHS itself, and the blade w/ cotton cloth I used to cut the glue.
The blades in the razor are easily the thinnest razors I have ever seen or used. Excellent idea LL!
And yes, that is some of my blood on the cotton cloth.
Here is the CPU with the core cleaned off and black electrical tape applied over the surface mount caps (ala LemonLime's suggestion).
In the end all it took to remove my IHS was $1 in disposable razors, a cotton cloth and about a tablespoon of blood.
The temp could in fact be lower, but the install of Windows I used to test was pre-SP1, and if I remember correctly SP2 w/ the latest A64 platform drivers report the temps a bit lower.
As for the IHS, I think I'll drill a hole in it and make it a new keychain.
EDIT:
Take a look at this article about checking to make sure the CAM isn't pushing your waterblock out of alignment. I'll check mine as soon as I get home.
http://www.overclockers.com/tips1197/
That's an old article that still comes in so handy!
Based on those temp decreases you saw, I'm pretty sure your CAM clearance is fine--good to check none the less. 7'C delta is pretty damn good.
Hope you didn't lop any fingers off
The problem with those razors is they're so sharp they can cut pretty deep without realizing it until you're already bleeding all over the place. Luckily the cut isn't much worse than a really bad papercut.
http://www.angrygames.com/vids/street_tuts/IHS_removal.zip
j/k smj ...nice job!
hadn't thought about it like that... although it wouldnt have helped me too much in that case. i was working on it while the board was still in the case. hell, i hadn't even taken my cooling setup apart becasue i didnt want to drain it (too much hassle, the only downside to watercooling)