Solder IHS to 3770K?

[Smoblikat]

I am interested ind oing this just for fun, but honestly have no idea where to start.

I saw this thread:
http://forum.hwbot.org/showthread.php?t=75350

and it looks awesome, but I wanted to ask you guys if you knew of an easier way. Any advice on how to go about this would be greatly appreciated.

 

[Idontcare]

Easy way would be to buy some Indigo Xtreme, depackage it, put it on the bare-die and just rest the IHS on top of your little pile of IX shavings (don't clamp it yet).

Fire up the computer, no HSF, get into windows (it will be throttling but no worries), and fire up prime95.

Once the CPU gets to TJmax it will slowly begin to melt the IX. It will take probably a good 20-30 minutes though because the heat of fusion for the solder is large and the thermal output of a throttled 3770K will be low (despite the high operating temp).

Once you see the IHS itself has actually settled down onto the die, meaning the solder had finally melted, use a screwdriver or tweezers (something just so you don't burn your fingers) to position the IHS where you want it and go ahead and latch the CPU socket retention mechanism at that point.

Let the CPU continue throttling for a while longer still, then turn it off and let it cool. Put your HSF on (with TIM of course) and fire it back up. The solder should be set now and won't re-liquefy unless you take your CPU back up above ~95C again.

That is how I'd do it.

 

[Chipfiref]

Idontcare - is it the exact same process you would use for a 4770K?

 

[Idontcare]

Idontcare - is it the exact same process you would use for a 4770K?

Yes.

But I should have noted, not such a big concern with IB but a definite concern with HW, you don't want to put too much solder on the die as there is a risk of the excess solder making electrical contact with the exposed pads that are on the top surface of the PCB (for IB) or the capacitor connections for HW.

A little bit of solder goes a LONG way.

 

[Smoblikat]

Easy way would be to buy some Indigo Xtreme, depackage it, put it on the bare-die and just rest the IHS on top of your little pile of IX shavings (don't clamp it yet).

Fire up the computer, no HSF, get into windows (it will be throttling but no worries), and fire up prime95.

Once the CPU gets to TJmax it will slowly begin to melt the IX. It will take probably a good 20-30 minutes though because the heat of fusion for the solder is large and the thermal output of a throttled 3770K will be low (despite the high operating temp).

Once you see the IHS itself has actually settled down onto the die, meaning the solder had finally melted, use a screwdriver or tweezers (something just so you don't burn your fingers) to position the IHS where you want it and go ahead and latch the CPU socket retention mechanism at that point.

Let the CPU continue throttling for a while longer still, then turn it off and let it cool. Put your HSF on (with TIM of course) and fire it back up. The solder should be set now and won't re-liquefy unless you take your CPU back up above ~95C again.

That is how I'd do it.

Wow, thank you. That is a very detailed answer. Let me try to sum it up:

Get Indigo Extreme
Put some on the bare die of my 3770K
Place the IHS on top, but dont press down (I assume my computer needs to be oriented on its side for this)
I leave the retention bracket open, but will the CPU still function if its not pressed into the socket?
Boot windows like this, no HSF and run a stress test, would IBT speed this process up, or would it create too much heat?
After 30 minutes or so latch the retention bracket on, but is that safe to do while the system is running?
Let it sit for about 5 minutes or so, shut down and let it cool (all of this while the IHS is positioned where I want it, and the retention bracket down).
Install my HSF and TIM and away I go..........

95C sounds like a fairly low melting temperature to me, is there anything that is about 110C, or even 100 - 105C?

Thanks!!

 

[Idontcare]

I leave the retention bracket open, but will the CPU still function if its not pressed into the socket?

Doh! That is the deal killer, totally brain-farted and didn't think about that.

Yeah your CPU will NOT power up with the sort of tentative electrical connections that are going to be made with a processor just held in place by gravity.

 

[Smoblikat]

Doh! That is the deal killer, totally brain-farted and didn't think about that.

Yeah your CPU will NOT power up with the sort of tentative electrical connections that are going to be made with a processor just held in place by gravity.

Shux............any other ideas? What about those indium pads from the thread I linked in my OP?

 

[Idontcare]

Shux............any other ideas? What about those indium pads from the thread I linked in my OP?

Personally I'd just go with CLU at that point.

Another alternative with the same Indigo thought line above is that you could rig a setup to just use gravity to push the CPU into the socket for the initial IX melting phase.

Set computer on side, put some standoffs on the CPU PCB that bridge over the top of the IHS, add a stack of books (pushing down on the CPU, not on the IHS) until it seems reasonable, then fire it up.

 

[Kenmitch]

Shux............any other ideas? What about those indium pads from the thread I linked in my OP?

I want one of these. It looks like it would make bare die doable if one is careful and has a lapped heatsink.

It was rumored to be shipping with a Asus board but didn't. M6E is mentioned.

Sifting thru threads has turned up no source so far.

I'm debating if I want to delid or not currently. My chip doesn't crave vcore but the temps get insane if I crank her up....Even a H100i can't tame the beast.

 

[Smoblikat]

Personally I'd just go with CLU at that point.

Another alternative with the same Indigo thought line above is that you could rig a setup to just use gravity to push the CPU into the socket for the initial IX melting phase.

Set computer on side, put some standoffs on the CPU PCB that bridge over the top of the IHS, add a stack of books (pushing down on the CPU, not on the IHS) until it seems reasonable, then fire it up.

Whats CLU?

I dont feel comfortable putting weight on the CPU, ANY movement would instantly render my 400$ board a 400$ paper weight

I want one of these. It looks like it would make bare die doable if one is careful and has a lapped heatsink.

It was rumored to be shipping with a Asus board but didn't. M6E is mentioned.

Sifting thru threads has turned up no source so far.

I'm debating if I want to delid or not currently. My chip doesn't crave vcore but the temps get insane if I crank her up....Even a H100i can't tame the beast.

I dont have temp issues.......I am not even overclocking my CPU. I just think it would be fun as heck to try this out.

 

[Aikouka]

Whats CLU?

Coollaboratory Liquid Ultra

It's an expensive TIM that is mostly available from the smaller vendors (I never saw it on Amazon or Newegg). I bought it from Performance PCs, but I haven't had a chance to use it yet. Anyway, it's usually recommended for replacing the die-to-IHS TIM as it's full metal; however, it's also electrically conductive (unlike most TIMs), so you need to be careful!

I dont have temp issues.......I am not even overclocking my CPU. I just think it would be fun as heck to try this out.

I'm not sure if I do either, but I find it odd that my CPU runs so much hotter than my GPUs. It wouldn't hurt to replace the TIM and hopefully get some better thermals.

 

[aigomorla]

couldnt u get the near same effect with liquid pro on the die itself?

 

[john3850]

Does anyone know whats in CLU.
That CLU stuff almost looks and acts like the mercury thats been used in relays for years.

 

[Idontcare]

Does anyone know whats in CLU.
That CLU stuff almost looks and acts like the mercury thats been used in relays for years.

It is an Indium/Gallium alloy. Basically solder with a below room-temperature melting point.

 

[SunRe]

Source an electric pan, something with a thermostat would be great.

- Place the IHS upside down in the pan.
- Smear a very thin layer of good quality solder paste / flux inside the IHS, in the center, where it makes contact with the dye. This will stabilize the solder shavings and will allow for better contact with the IHS.
- put some lowtemp solder on it, Indigo Xtreme, whatever. Shavings, as idontcare suggested, it will melt faster.
- put the CPU with the dye upside-down and align it as best as you can with the IHS

Turn on the heat, I doubt that a short burst of over 150 - 180 C will damage the CPU as long as it is not powered, but I can't document this claim in any way.

Surface mount components are soldered this way (inlcluding complicated IC's) at over 200 degrees Celsius.

It would be advisable though to try this with a lesser CPU first, just to get the hang of it.

As for how long this would take, well this is the most common temperature profile I encountered:

 

[BrightCandle]

Can you not just stick the CPU in the oven to melt the IX? You can set a modern oven to 95-100 C and get a very even temperature that would allow you to melt the IX. You would then need a motherboard near by as you removed it with you oven gloves and clamped it down. Given a minute or so I suspect you could happily get the CPU into the board and clamped down with a heatsink and fan on the top (the simple kind anyway).

 

[Chipfiref]

Wow we are getting really mad-scientist with these options Dexter's Laboratory

Not knocking the kitchen method, just laughing when I think of oven mitts! I will be wearing buckskin gloves when I use the razorblade...


I got CLU from Amazon for $15 - it was actually fulfilled by Sidewinder Computers in Indiana, and arrived in 5 days

 

[Smoblikat]

Wow we are getting really mad-scientist with these options Dexter's Laboratory

Not knocking the kitchen method, just laughing when I think of oven mitts! I will be wearing buckskin gloves when I use the razorblade...


I got CLU from Amazon for $15 - it was actually fulfilled by Sidewinder Computers in Indiana, and arrived in 5 days

Haha, welcome to my laboratory

My only concern is that if I do the oven method, when I go to put the CPU in the socket it will melt to the pins.

 

[BrightCandle]

It shouldn't melt the pins, its the same temperature it would reach when run normally with insufficient cooling. It's not like we are making it hotter than it would otherwise be. The pins are metal, it might make the contacts a little softer but its not something they aren't already designed to do.

If I was telling you to bake it to 200C that is a different problem, then its outside of normal spec significantly and the socket likely isn't designed for it.

While some of you may think an oven is a crazy plan it's basically what Intel themselves do to melt their solder. The silicon process contains several hot cycles and the temperature will decrease as it reaches nearer to completion. Its far from crazy, it's what Intel likely does to get the solder to melt.

 

[Idontcare]

Can you not just stick the CPU in the oven to melt the IX? You can set a modern oven to 95-100 C and get a very even temperature that would allow you to melt the IX. You would then need a motherboard near by as you removed it with you oven gloves and clamped it down. Given a minute or so I suspect you could happily get the CPU into the board and clamped down with a heatsink and fan on the top (the simple kind anyway).

Excellent! That's it. Put it upside down (IHS down) with the IHS on a cookie sheet or some such and then set the oven at 100C.

IX melts at 98C, just takes a while of course, and the CPU itself will absolutely be fine at that temperature.

I wouldn't even be too worried about doing the hot-transfer. The PCB+CPU will sandwich the IHS on its own from gravity.

So just let it cool, then put in mobo, latch, install HSF.

Now to get it good and tight go ahead and fire up prime95 and basically take it up to 98C a second time just to ensure a good pressurized seal forms and the IX itself is squeezed nice and thin between the CPU and the IHS.

I think you've come up with the brilliant solution BrightCandle :thumbsup:

 

[Smoblikat]

Excellent! That's it. Put it upside down (IHS down) with the IHS on a cookie sheet or some such and then set the oven at 100C.

IX melts at 98C, just takes a while of course, and the CPU itself will absolutely be fine at that temperature.

I wouldn't even be too worried about doing the hot-transfer. The PCB+CPU will sandwich the IHS on its own from gravity.

So just let it cool, then put in mobo, latch, install HSF.

Now to get it good and tight go ahead and fire up prime95 and basically take it up to 98C a second time just to ensure a good pressurized seal forms and the IX itself is squeezed nice and thin between the CPU and the IHS.

I think you've come up with the brilliant solution BrightCandle :thumbsup:

Thats it!!

We have a winner, im buying some CLU and giving my CPU the biz.......

Ill report results once I have the CLU.

 

[Idontcare]

Thats it!!

We have a winner, im buying some CLU and giving my CPU the biz.......

Ill report results once I have the CLU.

Note - you don't have to do this with CLU as CLU is already a liquid.

The scenario outlined with BrightCandle's idea is for solid-solder like Indigo Xtreme that needs to be reflowed at elevated temps.