RAM heatsinks

[eidairaman1]

No. Press down/insert the ram stick by pushing down on the edge of the PCB (red marking) if exposed. Some heatsink covers this so you may have no choice but use the heatsink to insert in socket.



It's real glue. Large collection of DDR2 to repair/restore..

Dude i just explained it.

 

[joemama]

It's always good to have something to cool down the modules, I don't know much about the effect of temperature on the chips but the heat cycle due to the computer turning on and off is definitely bad for any kind of component

 

[delshay]

Here's the result after freezing the memory overnight (G.Skill PI DDR2). I like the bit where not a single chip remained on the PCB. This is a very poor result but it has no effect what will happen to the dram chips.

Second photo you can see reballed chips awaiting install on the PCB (G.Skill Trident DDR2)


EDIT UPDATE:

Just found the "perfect solution", I was wrong. Heat the memory module up to a very hot temperature. Keep it very hot at all times. Not sure of the temperature but I kept it on the preheater. Now lift/separate the heatsink away from the PCB, it should be easy to lift off when extremely hot.
I may upload the tool used to separate heatsink away from the PCB.

Don't worry about the first photo where it went wrong. It will be reballed & find a new home on a new PCB.


EDIT UPDATE2:

I have added a third photo with 100% perfect removal of the second G.Skill PI heatsink. I used that tool to separate the heatsink away from the PCB by placing it in between the gap & opening the pliers while the memory module is extremely hot.

 

[Mussels]

Well, freezing certainly went well

Thanks for the feedback!

 

[delshay]

Well, freezing certainly went well

Thanks for the feedback!

Yeah, completely wrong. I have lots of spare parts which allows me to carry out lots of experiments,. Plus I have the skill to restore., so no big deal
You need to get it very hot but if you go a little over the top it will start to melt the label. I have no idea what the temperature requirements are but it must be over 80c+ & remain at this high temperature until heatsink is removed/separated. Funny enough it looks like if you put the heatsink back & let it cool down, it goes back super sticky & your back to where you started.

 

[Mussels]

I'd test between 60 and 80C if you had a controlled heatbed, not hot enough to damage anything but hotter than expected running temperatures of the device (and they'd never use an adhesive that'd melt right off, close to expected running temps)

 

[delshay]

I'd test between 60 and 80C if you had a controlled heatbed, not hot enough to damage anything but hotter than expected running temperatures of the device (and they'd never use an adhesive that'd melt right off, close to expected running temps)

It is controlled. Preheater is set to 300c but I have to remove the memory module before the label melt. I don't know what the real temperature of the memory module is, then again it's not really required. Just use your judgment when the heatsink parts away from the PCB. Just put the pliers in between the gap with a little opening pressure & as the memory module heats up & hits a certain temperature the heatsink parts away from the PCB with little to no effort. If I really want to find the temperature I could put a thermal probe on the heatsink.

 

[Mussels]

Regardless, nicely done and thanks for letting us know freezing was dumb

 

[Braegnok]

Can never get a clear answer on Bdie temp sensitivity. Is it 45 or 50? Or is 45 just a sub-140ns tRFC or 1.55v+ thing? I have my single rank and dual rank B-die down to 133ns and 140ns without destabilizing at 1.45-1.53v during long repeated TM5/LinX/HCI sessions, as long it stays below 50 - hence why I treated 50 as the magic number.



Sweet DJR setup. Have you noticed any temp sensitivity on Hynix? Heard somewhere that DJR is, but could never confirm and CJR doesn't seem to be

In general I have noticed running tighter timings = more work being done by the memory = more heat output. Temperature impacts stability fairly significantly by increasing the leakage current of the memory cells storing your data. tRFC controls how long the cells are allowed to be recharged and tREFI controls how long a cell can operate before being recharged; so both of these directly effect the temperature induced error threshold. This threshold varies with high binned ICs but can be much lower than you would think when running extremely tight profiles with high voltages, thus increasing tRFC is probably the easiest way to raise your temperature induced error threshold, especially with a high tREFI.

 

[Shrek]

I have no idea what the temperature requirements are but it must be over 80c+ & remain at this high temperature until heatsink is removed/separated.

I wonder if boiling in water would be enough.

 

[delshay]

I wonder if boiling in water would be enough.

Welldone! ..That's a experiment I would love to try. It should work.

..Err but you would lose the sticker/labels plus damage the thermal pads..

 

[Mussels]

For whatever reason that had me thinking of sticking old, unused RAM sticks inside the reservoir in my watercooling loop

I mean yes, it'd corrode and do horrible things, but it'd also look like i was cloning RAM sticks

 

[Keullo-e]

I remember that "despreading" DDR1 with TSOP chips went okay back in the day, but I also experienced that a BGA DDR3 chip came with the heatspreader when I tried the same some years later.

Wonder how my DDR3s would work without heatspreaders on my X58 rig. They get pretty hot even with moderate 1680MHz 9-9-9-27-2T 1.65V settings.

 

[Shrek]

Welldone! ..That's a experiment I would love to try. It should work.

..Err but you would lose the sticker/labels plus damage the thermal pads..

I'd take that over damaging the chips

 

[ir_cow]

Yeah, completely wrong. I have lots of spare parts which allows me to carry out lots of experiments,. Plus I have the skill to restore., so no big deal
You need to get it very hot but if you go a little over the top it will start to melt the label. I have no idea what the temperature requirements are but it must be over 80c+ & remain at this high temperature until heatsink is removed/separated. Funny enough it looks like if you put the heatsink back & let it cool down, it goes back super sticky & your back to where you started.

Yeah heat is always the answer. It also looks like those weren't glued from the photos, just sticky tape.

I just go back and forth with a hairdryer until it starts to come apart. Sometimes the sticker will start to come off. Just need to push it back down and keep going. As long as the solder doesn't melt, you are golden.

 

[damric]

Some heat spreaders are glued on, and glued on such that I have ripped ICs in half despite preheating. Some are just fastened together with clips and use thermal pads and are much easier to deal with.

I have 4 sticks of DJR. I've never run them higher than 1.45v yet, for 3800/16. I might try 1.65v now and see if I can get timings down.

 

[ir_cow]

I have 4 sticks of DJR. I've never run them higher than 1.45v yet, for 3800/16. I might try 1.65v now and see if I can get timings down.

Hynix does not scale down with voltage like Samsung-B. You will be lucky to get CAS14.

 

[AusWolf]

Someone correct me if I'm wrong, but I don't think consumer-grade RAM needs massive heat spreaders (Corsair Dom Plat style and such). A small one (or even nothing) with proper airflow is fine.

 

[AlwaysHope]

Would putting some old DDR2 sticks in the microwave for about a minute warm them up enough to remove the HS?
I'm only asking cause' unsure if that kind of environment could damage the circuitry?

 

[joemama]

Would putting some old DDR2 sticks in the microwave for about a minute warm them up enough to remove the HS?
I'm only asking cause' unsure if that kind of environment could damage the circuitry?

Don't put things with metal components inside a microwave

 

[Solid State Soul ( SSS )]

They are mostly for aesthetics, a way to attract gamers with who has the most gamery designed "heatsink"

i mean DDR2 and DDR3 ran fine at 1.5v and 1.35v without heatsinks since forever, so yeah

 

[AlwaysHope]

Don't put things with metal components inside a microwave

That's what I use to think until I saw a college teacher in cooking put a metal bowl of some food into a Microwave.... & nothing happened but the contents in the bowl cooked! In fact they do this regularly during service times.

 

[maxfly]

Would putting some old DDR2 sticks in the microwave for about a minute warm them up enough to remove the HS?
I'm only asking cause' unsure if that kind of environment could damage the circuitry?

There was a dude at hardforum that thought microwaving his gpu in an attempt to resurrect it was the equivalent to baking it in an oven...more than once. He couldn't understand why his microwave stopped working OR why parts of his gpu flashed and melted.
Snap! Crackle! Pop!
True story.

 

[delshay]

I'd take that over damaging the chips

If you put the memory in a pot of boiling water I would imagine you don't need to do anything, the heatsink would just fall off (not tested).

Yeah heat is always the answer. It also looks like those weren't glued from the photos, just sticky tape.

I just go back and forth with a hairdryer until it starts to come apart. Sometimes the sticker will start to come off. Just need to push it back down and keep going. As long as the solder doesn't melt, you are golden.

Yes those have very strong sticky thermal pads. The true test will come when I take apart "Corsair Dominator 9136" DDR2 ..You can see each chip is glued direct to the heatsink.

 

[Shrek]

If you put the memory in a pot of boiling water I would imagine you don't need to do anything, the heatsink would just fall off (not tested).

You not going to test the idea?