View Full Version : Heat Spreader Removal Failure!!!
I decided a couple of weeks ago to upgrade my BH-6 with the ST-6 and a Tulatin Celly 1.2. I could only get 1400 without voltage tweaking beyond what the SoftSetup allows. (1.60V) I went all out and ordered the ThermalRight Sk-6 with a delta 30 fan. (I don't like loud coolers) When the SK-6 arrived I was ready to try for 1600 (my goal) My chip did not look that good ( only 1400 before and it was Malaysian) I decided to go for it all in one fell swoop!! I tied the pins together to get higher voltage and since I had the cpu out I decided to remove the heat spreader for better thermal conduction to my SK-6. I had gathered info from this forum and Anandtech forums. I began to cut the epoxy around the bottom edge of the heat spreader with a razor blade and soon was under the edge of the spreader. But it appears the epoxy goes pretty far towards the chip from the inside edge of the spreader.( I know this now) In other words this procedure for spreader removal was beyond my patience. I had gotten too aggressive with my cutting and had cut throught the conformal coating and into some of the traces on the chip mount board. Result = DEAD CELLY!! Since I had killed the thing I decided to try a different method for removal. I captured the chip mount board edges (two of them) in a set of adjustable pliers (waterpump pliers) I used a second set of pliers to grab the spreader by 2 edges and twisted. It took a lot of force (and I had cut some of the epoxy loose) to get it to come off. When it came off , it did so cleanly and without damage to the chip or chip mount board. So my advice would be to try all methods to get this chip to overclock before trying to remove the spreader. If you do decide you must remove spreader, be careful and be patient!!! My hurrying cost me $115.00. Not the end of the world, but a definite hit to my Christmas money. HEHE Just wanted to share my experience and try to prevent others from making my mistake. I must say I consider my self to be fairly adept at working with electronics. (I do it for a living) So the cutting method is one to be avoided by all but the most careful and patient (IMHO) My other chip is on the way, (had to downgrade to my son's Celly 566 @877) I am hoping for a Phillipine chip. Needless to say, I won't be removing the spreader unless I totally fail to get to 1600. If I do remove it, I will use the pliers and twist method. Bottom line- Don't nick the conformal coating (green layer on top of chip board) Those traces are small and numerous.
Happy Overclocking!!
Kid
I've all but given up removing mine until a better method comes along... the razor blade thing didn't work for me (thank god I stopped when I felt myself getting frustrated and realized I was stil getting 1500+ with the IHS on :rolleyes: ). It required too much pressure and I wasn't comfortable cutting downward (like you had) to get the proper angle on the epoxy.
Hmmm, and I was ridiculed by a couple of members when I urged caution about removing those heat spreaders. Looks like maybe it doesn't hurt to be a little cautious from time to time. Hey Kid, thanks for posting your experience and sorry for the costly error. Hopefully it might prevent others from making a similar mistake.
Something tells me that all of these things aren't glued equally. I know Ol'Man and a couple of other spoke of how easily theirs came off, but I worked with mine for two hours and finally gave up. Mine had no gap at all, perhaps because I had run it for a week with a tight heatsink attached, but that thin razor blade refused to gap it period. If I had continued, I would have wound up just like you did. The glue on mine goes all the way around, but has a small gap between both ends of the bead, about a 16th of an inch wide, with no glue at all. I can't even get the razor blade under there. I sure wish I could remove mine, but I can't afford to replace it right now.:)
Rick
Originally posted by batboy
Hmmm, and I was ridiculed by a couple of members when I urged caution about removing those heat spreaders. Looks like maybe it doesn't hurt to be a little cautious from time to time. Hey Kid, thanks for posting your experience and sorry for the costly error. Hopefully it might prevent others from making a similar mistake.
I remember that. Rmcky explained it well: not all of these are made the same. I ran into the exact same problems he did... no one should be too cautious they don't want to at least try, but have patience and know that something just might not work as well as it did with another member.
Thanks for sharing the exprience, really weird how much pressure is required to take that off anyways?
Would putting it in the freezer for 15-30mins help?
Most modern epoxies need a very large temp change from ambient to become more brittle. I don't think putting it in the freezer would make much difference. Perhaps others will see things differently and give it a go using your tip. The fact that these things are not epoxied with the same technique suggests that Intel is trying different methods and/or these methods differ between fabs. I will not be removing the spreader on my new chip unless it fails to overclock to ~1600 and I feel that this is related to a poor mating between the actual chip and the spreader. I hope I don't have to remove it. I will know in a few days. I am sorry but I can't be very accurate in quantifying the force needed to twist it off. My first suggestion would be don't try. Beyond that, be sure you are willing to live with the very real possibility you might damage the chip.
Kid
Grande Juan
12-01-01, 05:43 PM
Perhaps if you had a true 0 degree freezer and worked quik.
Mike360000
12-01-01, 05:53 PM
Geez fellas, you all aren't painting a very good picture of this once *supposedly* easy job!
I am getting a P4 by the weekend and up till now, I thought removing the IHS was going to be easier than this.
I'm just wondering here, but couldn't a person, when they get the razor started in under some of the epoxy, couldn't they twist or bend the razor blade up slightly to help try and seperate the epoxy?
Also what about taking an exacto knife with a square front blade, (about 3/16 to 1/4 wide blade at front) and with the blade in the handle, carefully go around (following the expoxy) the IHS along the edges instead of towards the cpu?
Cheers,
Mike Lamb
i saw a chip once where someone had lapped the spreader out of existence and the remains of the spreader were acting like a kinda shim
has anyone tried this?
Here's another thought for those willing to risk it. The heat spreader is much larger than the actual core. What about taking a dremel tool to cut it off around the top edge of the cap? Once the top of the spreader has been cut off it would be easier to remove or file down the part that is left glued to the package. I have not tried this - just brain storming. Would take a steady hand and have the cpu locked into a vise of some sort.
Originally posted by TC
Here's another thought for those willing to risk it. The heat spreader is much larger than the actual core. What about taking a dremel tool to cut it off around the top edge of the cap? Once the top of the spreader has been cut off it would be easier to remove or file down the part that is left glued to the package. I have not tried this - just brain storming. Would take a steady hand and have the cpu locked into a vise of some sort.
I thought the same thing... problem is the material being used for the IHS and it's thickness isn't conducive to dremel work... too thick and steel/aluminum (too tough for most dremels).
The problem is messing with blades and traces, cut traces don't work well :p. It's also a matter of once you get under the epoxy (if you can) the remaining epoxy is too strong to just 'pop' it off... which is good work by intel since the point is to NOT have this fall off easily...
CrystalMethod
12-01-01, 09:16 PM
Hmm... and I was thinking it was going to be easy when I got one, now I'm having second thoughts.
ol' man
12-02-01, 05:00 PM
Originally posted by CrystalMethod
Hmm... and I was thinking it was going to be easy when I got one, now I'm having second thoughts.
Keep those thoughts. It may not be worth it.
----------------------------------------------------------------------------------------------------------------------------------------------
Well I have something to add I guess to this. I have removed two of these and I do agree this is the last desperate action one should take to OC this chip.
The first chip I did this too I used a hacksaw to get it off. Well I hit some TRACES as many of you here have done now. I did this to a $315 tualatin 1.2 PIII though. Well any way I was the one who found out that the IHS is put on with black SILICONE adhesive. Not EPOXY as has been stated above. Well it is dead so lets move on.
Second chip I got was a .13u celery which was a wonderful OC'r. I use the term "was" cause you will find out in a minute why;)
Anyway the lesson I learned from the first chip was that the soft silicone could possible be CUT not SLICED( as in running the razor blade back and forth like a saw) like some here possibly seem to think I meant. I never sliced to get the IHS off. I only forced the very very thin razor blade under the IHS but left the razor blade flat against the organic base. The way I had the razor blade there was no way I could cut a trace and I thought I had made that clear as many others here have also. It does not make sence to have the razor blade at even a slight angle when trying this. The area is so small for the blade to go under that it would be impossible if it were any other way but flat against the base. I MEAN ABSOLUTELY FLAT. No ifs, ands, or buts about it. I couldn't see any sort of a gap under my chip either as many of you are stating. There may not really be one until you get that first corner to move up with the forcing of the flat blade in but not SLICING under that corner as I show in this pic.
http://hhrol.bobsville.com/cpu5.jpg
I simply pushed the razor blade into the glue. Getting that first corner started seems to be the hardest part and once you do that it is a cake walk from there. If you would have followed my instructions you may not have got your IHS off but you also would not have killed your chip. I removed my second IHS without harming the IHS or the chip and core.
Now back to my second celery...... :(
This hurts.......
Well I did a cryo burnin' and in the process I stayed up all night babysitting it. Come morning all the dry ice was used up so I took the board out and the chip and cleaned them up and then put them under a heat lamp for about 1 hour or better. I then inspected them to see if there was any residual water vapor and there was none. Good to go. So I went and installed my chip and it booted fine. Ran if for awhile like this too but when I was messing around in the bios checking core temps I noticed that I could get my temps to go down if I pushed in the middle of the HSF. Obviously the HSF was not touching the core all the way so I decided to use a shim on it. Well I put it on the chip with out testing it to see if indeed the HSF was making contact with the core. Well I tried booting at 1.8v and 1600MHz to no avail. I figured no big deal. So I pulled off the HSF to find the core wasn’t even touching the HSF. :eek: Not good so then I took the shim out and tried rebooting it at default vcore and MHz but to no avail again. I know one reason why intel puts these IHS's on the chip anymore. The IHS does help for the thermal shut down mechanism if the core is not touching the HSF like mine was. This makes the IHSless .13u celery the same as a Athlon as if your HS is not touching the core it has a very good chance of dying like mine did. It was dead and I know it died when I booted it at 1.8v and 1600MHz without the HSF touching the core. It wasn't touching one bit. There where no ASII smudges on it at all it was clean as a whistle and that is the last time it booted. Luckily soon I will should have another celery soon. WHat can you make of this. Well for one I am a dips*** after no sleep and huffing CO2 vapors and two you need to be as careful with the IHSless tualatin as you would a athlon core for HSF/ core contact. That most likely and also giving the HSF a good base to lay against. I found when I set the mobo in my case and the HSF was sticking out the side that it would hang down a little bit and I had to use a rubber band to hold the HSF up otherwise my temps were through the roof. All in all removing th IHS should be the last resort in getting the last 50~100MHz out of your chip. I learned the hard and when I get my new chip I am going to cool it with this
http://hhrol.bobsville.com/cryo.jpg
and leave the IHS on for now.
Thanks for sharing that story Ol'man ;).
I've been doing a burnin all weekend, seen 30mhz of results from the prior 2 days I've done this. With my trip in January to florida, I figure I will be able to leave my living room door to the deck open for at least a few of those days (have a friend close it when they feed my cat), then I can really see some of the results with below 0C temps. Right now I'm running 35C under full load at 1.83V and 800mhz and 29C case temps (those are good numbers for this chip, even at the lower frequency).
I tried the IHS removal on 2 occassions. Each time I put a lot of the correct pressure on one cornet of the chip, ensuring the blade was flat against the board... it just requires either bigger Kahouna's than I have or too much pressure.
And it really does look like Epoxy on the edges of my IHS, not some other funky stuff (it's clear and takes a lot of razor power to remove enough a bit of it).
I'll let all of you know how the burnin goes... I think I can try some tweaking now ;).
Well, 36 hours of burn in as described above and no changes. I think I've reached the max for this CPU without taking more extreme measures (like getting an AMD chip ;) ).
funnyperson1
12-02-01, 09:59 PM
Originally posted by Pinky
Well, 36 hours of burn in as described above and no changes. I think I've reached the max for this CPU without taking more extreme measures (like getting an AMD chip ;) ).
youre right, that is extreme:p.....doesnt matter anyways, because when i ran Celly 1.2@1450 i saw that i benched higher than a 1.4 AthlonMP so i think 1530 is plenty fast.....
Originally posted by funnyperson1
youre right, that is extreme:p.....doesnt matter anyways, because when i ran Celly 1.2@1450 i saw that i benched higher than a 1.4 AthlonMP so i think 1530 is plenty fast.....
In that case, here's an article (http://www.daocplace.com/index.php?review=xpvstual) you may find interesting...
funnyperson1
12-02-01, 10:14 PM
nice article pinky, now when people come in the intel forums preaching about how you must get amd and intel is the devil....i have something to shut them right up....hopefully ill be able to do this with a PIII1100 as well....if my dad ever gets around to ordering it :rolleyes:
Originally posted by funnyperson1
nice article pinky, now when people come in the intel forums preaching about how you must get amd and intel is the devil....i have something to shut them right up....hopefully ill be able to do this with a PIII1100 as well....if my dad ever gets around to ordering it :rolleyes:
The only point I was making is that they are comparable, or at least closer to cost/performance than they used to be... the AMD's memory bandwidth smoke's Intel's, which is reflected in almost all multimedia applications. I think it's sad Intel had to lose to win in this case (it under performs the AMD, costs more, but Intel fans tout it as victory) :confused: ...
Thanks for the kudos... there will be more (I have a new infrared heat sensor I will be using for future tests).
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.