Mr. poverty here. Not only do I not have a heater in my house, I don't have an over either. So tomorrow I'm going over to my ex's house for Easter with my kid. I have a GF 5200, an ATI 9800 all in wonder, and my 8800GT not working. I've been reading about this baking video cards thing on the web and I think I'll give it a try in her oven (I hear the process is kind of stinky). I'll let you know how it goes. I wonder...does any body happen to know the difference in melting points for a lead solder card and a non-lead RHOS compliant card with lead free solder?And there are some aluminum heat sinks attached to a couple of the memory chips and voltage regulators that didn't pop off to easily (They were installed with super glue I didn't have money for heat sink epoxy)
While in the oven...if the cards are left in the oven long enough for the card temps to be brought to a steady-state temperature (uniform throughout the video card, e.g. the card is in the oven long enough for temps to "stabilize"), the heat-sinks being present OR not present will make zero difference in HOW hot the chips ultimately get...externally or internally.
ALL component internals, PC layers, and surfaces will ultimately reach the same homogeneous temperature if left in the oven "long enough". The only difference the heat-sinks in such a situation could possibly make is to change the amount of time that is required to produce changes in the temps of the chips in both the upwards AND downwards directions...the RATE at which the temps change on various elements. This could introduce substantial mechanical stress on the bonds as the chips and heatsinks expand and contract at vastly different rates with changing temps. Using pliable heatsink compound and/or tape helps reduce the impact such expansion rate differences introduce but how these function can be seriously and permanently degraded by "too-high" temps. "Epoxy" type thermal compounds rely on brute strength chemistry of the bond to minimize the impact of the expansion rate differences to prevent bond failure. And of course even these can be still eventually brought to the point of failure by excessive temps...but this failure point would be much higher than the failure point of a cyano-acrylate based "superglue" bond, and it would be also more resistant to failure due to the rates of change in the material temps.
Such changes could of course render the heatsinks less effective in later normal use. Leaving the heatsinks on simply adds one more reason why this procedure is a crap-shoot to be used in absolutely "last-ditch" efforts.
The simple use of super-glue to bond the heatsinks to the chips is problematic. If JUST super-glue is used and it's applied between the heatsink and the chip, it can actually act as an thermal insulator to varying degrees. There are a number of factors involved but in general you want much better thermal conductivity between the heatsinks and the chips. Of course, GENERALLY speaking, having heatsinks is better than not having heatsinks even with reduced thermal conductivity. But like most of these things it's conditional. IF the superglue is applied in a way to provide substantial thermal insulation, then even with heatsinks the chips can still be kept from dissipating heat fast-enough to protect the chip(s) during operation and the chips MIGHT be better served by simple air-flow being adjusted to move more quickly over the surface of the chips instead of over heatsinks applied with super glue. You're dealing with radiative v. conductive heat removal from the chip surface to the heatsinks and which has the superior effect in subsequent removal of heat via air-flow over the heatsinks in a specific configuration and environmental conditions. Unfortunately, the only way to know for sure whether you will see improvement is to have a thermal sensor on the chip CASE or IN the case (e.g. on the die)...NOT on the heatsink...and test the thermal difference under both "with heatsink" and "without heatsink" conditions during operation.
If purchasing heatsink thermal bonding agent isn't possible, then it almost certainly would be preferable to scavenge "thermal tape" from some other equipment (e.g. old UPS circuits, mobos, etc) and bond the heatsinks to the case/thermal-tape/heatsink assembly via alternative mechanical methods. (e.g. see my post at http://www.hardwareanalysis.com/content/topic/77318/#588506 ). Hell, I'd probably even re-use old scavenged thermal compound instead of just super-glue (or air-gap) as long as the "paste" had good consistency and wasn't dried-out).
COULD you see improved temp reduction using just super-glue to bond the heatsink(s)? Yes.But it's also possible the super-glue would impair the thermal coupling and aggravate any over-heating problem. The only way to know for sure is to test. It's also possible the degree of impairment would change over time, so even if it's working "OK" now, it could change without warning to an almost complete loss of thermal coupling between the case and the heatsink when the superglue bond fails...which is possible under repeated thermal cycling stresses. Superglue bonding tends to be very strong in axial (compression or extension) directions, but VERY fragile in lateral directions. That's why you can see a guy hanging from his hard-hat glued to a girder with a flat plate, but if you took a hammer and even lightly tapped the plate from the side, the bond would likely instantly fail and you'd likely hear a sudden scream.
Unfortunately, the bond in these instances may see significant lateral forces during thermal cycling as the plastic chip and metal heatsink surfaces physically contract and expand at vastly different rates. This could result in creating even more invisible gaps between the heatsink and the case or even total bond failure.
John I have had the card (the 8800GT) for a while.I was using passive cooling with a Arctic cooling cooler. I've never gone over the 50C even on the hottest summer days. The adhesive that came with the cooler was crap. The heat sinks were falling off and being the impatient type I didn't want to go with the heat sink epoxy if I decided I wanted to go back to the stock fan. The cheap super glue is pretty easy to pop loose. I got them all off except 2 on the memory chips.I'd use the super glue again. The adhesive goes on micro thin.
No i tried with 3 cards at 375F from 7 to 15 minutes. 2 AGP cards. A 5900 an x800 and my 8800GT PCIx. One thing though. everyone commented on the smell. I didn't get that at all so i may not have been high enough...on the temp that is. I didn't have an oven thermometer so I'm not sure if I actually got to the 375F I suppose I will try again at a higher temp since the nothing to lose scenario is in effect here. Any suggestions?
I suggest not to use baking to repair your card..it is very bad idea for there are some chemicals that will evaporate and be absorbed in your oven that is very dangerous for our health...id rather use a hot air equipment to reheat the IC's and chips of the vidoe card at 100 to 200 degrees temp.I was able to recover my 8800 videocard by doing that.