![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Let's talk delidding
- Thread starter Bluesun311
- Start date
Do you guys have any suggestions for how to remove the black glue residue after it's been cleaned off? I've gotten the die completely clean from glue and it's flat when I run my fingers across it but there still seems to be an imprint burned in from where the glue was and I haven't been able to get it off with professional grade goo-off. Should I even worry about it as long as it's flat?
I just left the imprint on there. It seemed like a stain that didn't do any harm.
I personally tried the vise method and I failed at it the first time around. The IHS finally popped off, but some of it's edges were a little damaged. Since it was a crappy overclocker (1.340V for 4.4GHz), I went and got a new one from the local MC and was able to get a much better batch (1.265@4.5GHz). The second CPU was delidded using a vise + razor and it went much more smoothly. The vise was used to hold the CPU while I worked with the razor.
CLP between the die and IHS and PK3 between IHS and EK block. It took me 2 attempts to get a good CLP application, but I can do better still.
Stock IBT Very High:
Core0:84C
Core1:84C
Core2:79C
Core3:74C
Delid IBT High:1st Attempt at CLP between die and IHS.
Core0:74
Core1:62
Core2:59
Core3:55
Delid IBT Very High: - 2nd Attempt at CLP between die and IHS.
Core0:49C
Core1:54C
Core2:53C
Core3:51C
Ambient 20-23C for all tests.
I chalk that one up as a W.
- Joined
- Sep 9, 2002
- Messages
- 5,657
The outline of the glue on the pcb might still be visible from the videos I've seen. I'd be more concerned with overdoing it on the harsh chemicals so long as it's flat.
I'll be doing this for the first time too. I pulled the trigger on Coolaboratory Liquid Ultra.... hope it's a solid choice. I was spending too long to decide between the ultra and pro. I was mostly concerned with having to reapply down the road than a degree or so. I ended up from reading people saying that about either one after having read long enough anyway. :/
I'll be doing this for the first time too. I pulled the trigger on Coolaboratory Liquid Ultra.... hope it's a solid choice. I was spending too long to decide between the ultra and pro. I was mostly concerned with having to reapply down the road than a degree or so. I ended up from reading people saying that about either one after having read long enough anyway. :/
urbanman2004
Limp Gawd
- Joined
- Apr 15, 2014
- Messages
- 186
It's all a/b taking your time. If you looked at the video I linked, the poster explains the way to cut to avoid touching the line of transistors.
urbanman2004
Limp Gawd
- Joined
- Apr 15, 2014
- Messages
- 186
Acetone (aka finger nail polish remover, get for less than $2 at walmart), q-tips to apply acetone and rub off weld, toothpick for scraping off weld, alcohol (91%).
urbanman2004
Limp Gawd
- Joined
- Apr 15, 2014
- Messages
- 186
Stick with Ultra, Pro is overkill
Am I the only one who literally didn't see any difference in temps after delidding?
Granted, my chip is a crappy OC'er anyway but I figured I'd see better temps at 1.25 to 1.3v.
My chip runs 4.5 Ghz @ 1.3v, temps hit around 81 running intel burn test at maximum with liquid ultra applied to the die and mx4 applied to the IHS for the heatsink. Nothing seems to be off. Cpu cooler is the NH-D14.
Granted, my chip is a crappy OC'er anyway but I figured I'd see better temps at 1.25 to 1.3v.
My chip runs 4.5 Ghz @ 1.3v, temps hit around 81 running intel burn test at maximum with liquid ultra applied to the die and mx4 applied to the IHS for the heatsink. Nothing seems to be off. Cpu cooler is the NH-D14.
wow, 73c at 1.4vcore? damn that's insane. i don't know what else to try. when i'm putting the ihs on the die before securing the latch, i line up the outline of where the ihs was so that it sits "factory". when i go to close the latch i put reverse pressure against it so it doesn't move and as such the socket cover sits down properly over the two ledges on the side of the IHS. it secures normally and looks like a normal chip that wasn't delidded. and then the cpu hsf i'm sure is making contact. i don't understand what i could even troubleshoot. i use the pea size drop in the middle of the cpu for the TIM for the hsf, every time i've removed a hsf doing that method whether it be for maintenance or to swap chips or something i always find even contact under the plate.
my chip's bus speed hasn't been stable either since the delid. when the bus speed is set to just auto or 100, it goes back and forth between 98.xx and 99.xx constantly. i can only get 100.xx if i set the bus speed to 102 in the bios. i know that you could turn off spread spectrum for sandy bridge to prevent that fluctuation but didn't see it in my bios. is there a haswell equivalent?
so, fluke of a chip?
my chip's bus speed hasn't been stable either since the delid. when the bus speed is set to just auto or 100, it goes back and forth between 98.xx and 99.xx constantly. i can only get 100.xx if i set the bus speed to 102 in the bios. i know that you could turn off spread spectrum for sandy bridge to prevent that fluctuation but didn't see it in my bios. is there a haswell equivalent?
so, fluke of a chip?
Have you set everything to default in the bios and observe temps? Just to make sure nothing else weird is going on. If you aren't seeing better temps when everything is at factory settings then I don't know.
I have a crap chip as well as evidence by the insane voltage to get a measly 4.5Ghz OC stable.
I have a crap chip as well as evidence by the insane voltage to get a measly 4.5Ghz OC stable.
Bluesun311
2[H]4U
- Joined
- Sep 21, 2013
- Messages
- 2,523
Did you remove all the glue completely? The IHS certainly doesn't help, might try direct mount.
The TIM for the IHS spread fine, and I pre-spread the liquid ultra on the die itself with the method I've seen others do with the brush. If it had spread onto the processor itself I'm sure it would have conducted. And yeah I removed all the glue completely.
How does one mount the hsf to the die without the IHS there to keep it in the socket? I'm assuming that you'd need to perform the work of securing the HSF with a tower on its side and the pressure of the mounting hardware would keep it in place. I don't think it would even be physically possible for the base of the D14 to make contact with the die of the cpu.. am I wrong?
How does one mount the hsf to the die without the IHS there to keep it in the socket? I'm assuming that you'd need to perform the work of securing the HSF with a tower on its side and the pressure of the mounting hardware would keep it in place. I don't think it would even be physically possible for the base of the D14 to make contact with the die of the cpu.. am I wrong?
If I'm not mistaken you would a special heatsink/block for a direct die mount as the die lays flush with the CPU bracket.
I used a heat killer water block on an X2 3800+ a long time ago and chipped the die. Still worked you just have to be very careful.
I used a heat killer water block on an X2 3800+ a long time ago and chipped the die. Still worked you just have to be very careful.
PliotronX
2[H]4U
- Joined
- Aug 8, 2000
- Messages
- 2,070
The bracket of LGA115x comes off very easily with a torx driver. Easier than having to jimmy the locking retention mechanism off of LGA775. A Ninja 3 mounted quite nicely on an i3-4330 for completely passive action
The bracket of LGA115x comes off very easily with a torx driver. Easier than having to jimmy the locking retention mechanism off of LGA775. A Ninja 3 mounted quite nicely on an i3-4330 for completely passive action
What kind of temps do you see with that set up? Does it see heavy workloads?I have the same i34330 chip, currently on stock Intel HSF.
PliotronX
2[H]4U
- Joined
- Aug 8, 2000
- Messages
- 2,070
Under Prime it hovers around 50 depending on room temps
Idles at 1-2 above ambient. Barely any airflow in the case, it was built to be a silent NAS w/Plex transcoding and pulls it off wonderfully. Spanks the crap out of my old Q6600 and sips the energy, very happy with it.![Zarathustra[H]](/data/avatars/m/9/9298.jpg?1723407683){kind=avatar}
- Joined
- Oct 29, 2000
- Messages
- 42,428
I'm glad the current preferred method is to replace the lid after, not to simply mate the HSF to the top of the silicon.
I killed two Athlon 1800 XP's and a 1.2 Ghz Thunderbird Athlon back in the day with a Vapochill before I got fed up with that impossible to mount garbage (the hoses were too stiff and short making the install nearly impossible to pull off without wacking the chip and crushing it) and sold it on eBay.
Under Prime it hovers around 50 depending on room temps
Really? That's crazy... I've seen my 4330 hit 60-65c under load (typical gaming) although I know for a fact that one corner of the post screw for the stock hsf is not latched properly.
Guess I should pull it off and re-do. Think I need to pull the entire board to do so.. which is why I haven't already
PliotronX
2[H]4U
- Joined
- Aug 8, 2000
- Messages
- 2,070
It's pretty sweet I wasn't expecting anything special but the i3 is super fast clock-for-clock and efficient. Delidding is scary if you've never done it but there is a popular way of doing it mentioned with a vice and smacking a block of wood on the lid. Then with direct access to the die the cooling efficiency is increased but you have to be careful selecting heatsinks because a lot of them have HDT (Heatpipe Direct Touch) which will not work well with direct die because most of the heatpipes will miss the tiny silicon die:Really? That's crazy... I've seen my 4330 hit 60-65c under load (typical gaming) although I know for a fact that one corner of the post screw for the stock hsf is not latched properly.
Guess I should pull it off and re-do. Think I need to pull the entire board to do so.. which is why I haven't already
Which means models like the popular Hyper 212 won't work so well. I would look towards the Thermalright Macho, I was lucky to find a Ninja 3 for cheap but the Macho would be better though more pricey.
jackofalltrades
2[H]4U
- Joined
- Feb 7, 2006
- Messages
- 3,742
