Thermal paste Silliness.

honegod

[H]F Junkie
Joined
Aug 31, 2000
Messages
8,327
Still Air is a pretty good insulator.

Air bubbles in thermal paste are bad.

So how to minimize air bubbles in the CPU/ heatsink interface ?

A Vaccume chamber.

Put the paste CPU and heatsink into a chamber.
Draw as hard a vacuum as possible.
As the pressure drops any bubbles in the paste will increase in size as the air inside them expands to fill the abhorrent vacuum.
Making the bubbles want to float up through the paste (whose density does not change) more energetically.

Hit the paste with a bit of ultrasonic to help the now big bubbles burst.
= reduced air content paste.

(there will still be air pockets but there WILL be less insulation.)

Then apply the paste as normal (in the chamber) & mount the HS to the CPU, as normal.

All of the air bubbles trapped in the space between the contact surfaces will be low pressure bubbles that contain fewer air molecules than bubbles of the same size in ambient pressure.

Vent air into the chamber to bring the air pressure up to ambient and all of the remaining bubbles will shrink in size, drawing the paste in to fill the volume of the dwindling bubbles.
 
Still Air is a pretty good insulator.

Air bubbles in thermal paste are bad.

So how to minimize air bubbles in the CPU/ heatsink interface ?

A Vaccume chamber.

Put the paste CPU and heatsink into a chamber.
Draw as hard a vacuum as possible.
As the pressure drops any bubbles in the paste will increase in size as the air inside them expands to fill the abhorrent vacuum.
Making the bubbles want to float up through the paste (whose density does not change) more energetically.

Hit the paste with a bit of ultrasonic to help the now big bubbles burst.
= reduced air content paste.

(there will still be air pockets but there WILL be less insulation.)

Then apply the paste as normal (in the chamber) & mount the HS to the CPU, as normal.

All of the air bubbles trapped in the space between the contact surfaces will be low pressure bubbles that contain fewer air molecules than bubbles of the same size in ambient pressure.

Vent air into the chamber to bring the air pressure up to ambient and all of the remaining bubbles will shrink in size, drawing the paste in to fill the volume of the dwindling bubbles.

This guy says that air bubbles in thermal paste is nonsense. You can find that part at the 11:40 mark.

 
This is silliness. Assembly line robots applying TIM a millimeter thick in an 'x' pattern on the GPU. Air gaps and minimal contact.
 

Attachments

  • msi.rx480.jpg
    msi.rx480.jpg
    170.6 KB · Views: 0
That's bad cooler contact problem, not TIM application.
Poor flowing paste resulting in the latter causing the former I'd say. In any event, doubtful any of us would goop on TIM at millimeter thickness in this manner.
 
Poor flowing paste resulting in the latter causing the former I'd say. In any event, doubtful any of us would goop on TIM at millimeter thickness in this manner.
Could be wrong viscosity TIM for the target mounting pressure, but there's no such thing as too much TIM.
 
Okay I'll ask, how the f do I create a vacuum chamber and then draw all the air air out of it?
 
I was looking around EBay and saw a bunch of ~$100 degassing chambers available, add a small vacuum pump = Vacuume chamber.

I think Waldos would be easier to setup than a pair of spacesuit gloves to do stuff in vacuum.
 
This guy says that air bubbles in thermal paste is nonsense.

I understood him to say air in thermal paste is bad so manufacturing process pointedly excludes air.

He specifically said that oxygen is very bad mixed with powdered metal and must be excluded for safety.
A component of air is oxygen.

@11min he mocks the notion of air bubbles when applying the paste. reasonable.

What I know is that polishing a surface just reduces the size of the irregularities which will make the possible size of air bubbles trapped in those irregularities smaller.

Molecules however are really tiny. And a bunch of them could fill irregularities too small to see without mechanical augmentation.

The application of "vacuum" would force those air molecules to grow out of the irregularity and be more easily swept away by the tsunami of paste sweeping along the HS surface.
 
Air bubbles do not stay. The TIM is under pressure between the cooler and the CPU. And heated by the CPU. All of the air bubbles are pushed out, once you actually start loading/heating your CPU.
 
I got some baddd paste man...this Generic Tensan stuff runs like 5 degree hotter loads man...The Sycthe paste is a good bit better...and I moved and lost the g1/4 Plug fill for that new pump...the old school laying down ones it's strong as shit too. So I need a plug and I guess just run the cpu's on a 240mm still...you can still put a better sink on gpu's man..they don't need water...
 
Could be wrong viscosity TIM for the target mounting pressure, but there's no such thing as too much TIM.

Depend on the pressure
The not to much tim is a "NEW" thing ( for us oldies)
There has absolutly been times where the pressure has been to week to remove excess TIM and having to much TIM was a bad thing.

Over time TIM has improve on conductivity ( lessens the punishments for to much TIM)
improvet in visqiei...vasqui..... more liquid ( reduce the risc of punishment)
and mounting pressure has increase (reducing the risc of punishment)


But current times with CPU's you are absolutely correct IMHO
 
There is a such thing as too much TIM. You don't want that crap oozing all over your motherboard and squeezing out from under your thermal solution like some burger ruined with mayo.
 
There is a such thing as too much TIM. You don't want that crap oozing all over your motherboard and squeezing out from under your thermal solution like some burger ruined with mayo.
Sure, but in the context of thermals...
 
Last edited:
Yeah, but what would he know, he just develops and sells the thing, and he made a fortune by extremely overclocking stuff.

Nice argument-from-authority.

Good authority for sure, but at no time in the video does he directly address what I am talking about.


There is 0 data on any investigation into micro-bubbles entrapped in micro irregularities in the surfaces to be TIMed.

The TIM is NOT applied with pressure in the range of HS mounting pressure , it is just a drop dropped onto the surface which is spread across the surface at the rate dictated by the viscosity of the NON compressible fluid.

Being NON compressible the goo will spread as the clamp pressure increases.
The internal pressure of the goo remains at ambient until the goo cannot flow enough to escape the applied pressure.

This is where "pump-out" comes into play.
 
i used to work work latex casting a lot. you dont need to do the entire process in a vacuum chamber, just put the paste in there with a big enough cup to expand. run it for 10-15 seconds or until you see the large bubbles stop popping. then take it out and use it as you would. air doesnt go back in when its released from the vaccum, thats not how it works, you now how a more dense material.

you can also make a soda completely flat in the chamber if you like. it works on any liquid.
 
Nice argument-from-authority.

Good authority for sure, but at no time in the video does he directly address what I am talking about.


There is 0 data on any investigation into micro-bubbles entrapped in micro irregularities in the surfaces to be TIMed.

The TIM is NOT applied with pressure in the range of HS mounting pressure , it is just a drop dropped onto the surface which is spread across the surface at the rate dictated by the viscosity of the NON compressible fluid.

Being NON compressible the goo will spread as the clamp pressure increases.
The internal pressure of the goo remains at ambient until the goo cannot flow enough to escape the applied pressure.

This is where "pump-out" comes into play.
I'll take it from someone who deals with these things for a living and has years of experience both using and producing the stuff, over theoretical musings. Not every "fallacy" is a fallacy.

You're talking about something of academic importance at best. Even if it were a factor, the variability in the physical HS mounting would overshadow it.
 
i used to work work latex casting a lot. you dont need to do the entire process in a vacuum chamber, just put the paste in there with a big enough cup to expand. run it for 10-15 seconds or until you see the large bubbles stop popping. then take it out and use it as you would. air doesnt go back in when its released from the vaccum, thats not how it works, you now how a more dense material.

you can also make a soda completely flat in the chamber if you like. it works on any liquid.


Yup, that's what I'm talking about.

The sustained vacuum isn't for the paste, it should have already outgassed.

The reason for mounting the HS in low pressure is to expand the tiny air bubbles so they are protruding out of the surface micro fissures of both mating surfaces.

Allowing the paste flow to push them aside.

If we increase the air temperature in the chamber to say 100c and allow the alloy surfaces to heat up, not only would the air expand more, but the surface scars will ALSO expand, allowing the paste to more easily flow into them.
 
I've dealt with thermal interfaces for 35 years.
I also own a high vacuum pump capable of degassing TIM's.
Degassing TIM's is essentially irrelevant.

Meeho is 100% correct.This is purely academic.
Surface smoothness and surface flatness are FAR more important.

The repeaters they use under the ocean for fiberoptic lasers use multi-stage peltier blocks with oriented lattice diamond substrates.
They don't even degas those.

An optically flat surface lapped to another only requires a TIM layer a few molecules thick.

Entrained air is not an issue.

;)
 
You're talking about something of academic importance at best. Even if it were a factor, the variability in the physical HS mounting would overshadow it.

Two data points you are using are not facts.
1. You have 0 facts to use for determining the relative importance of vacuum application of thermal paste.

2. Having 0 facts about what an improved mounting process might do to influence the heat flow you cannot estimate the relative value of this process by comparing it with any other mounting process.

Depending on how MUCH air is trapped in micro-bubbles, and how effective this process is at reducing the insulative power of the micro-bubbles,
There MIGHT be some serious gains.

Don't know.
 
Two data points you are using are not facts.
1. You have 0 facts to use for determining the relative importance of vacuum application of thermal paste.

2. Having 0 facts about what an improved mounting process might do to influence the heat flow you cannot estimate the relative value of this process by comparing it with any other mounting process.

Depending on how MUCH air is trapped in micro-bubbles, and how effective this process is at reducing the insulative power of the micro-bubbles,
There MIGHT be some serious gains.

Don't know.
It goes both ways, though, and my zero is at least tilted to 0.001 by all the current practical knowledge and experience.
 
An optically flat surface lapped to another only requires a TIM layer a few molecules thick.

Absolutely, no argument.

Are YOUR processor & heatsink polished to optically flat and lapped to each other ?
(If so, do mine ?)
Difficult to accomplish with sandpaper and a mirror.

I looked into lense grinding tech, but copper is no way near as hard as glass.

Thought about using Happich Simichrome, but I fear the abrasive is probably not diamond and so I don't want that substance embedded in the soft copper contact patch.( I WOULD be ok with bits of diamond )

Reduction of surface irregularities is indeed very good.(lapping)

Reducing the size of surface scratches is also very good (polishing)

But I don't see how to do that in my kitchen.

My Threadripper surface is not shiny, I might try lapping /polishing it with the nickel plated silver arrow surface.
 
I can make you a video how to do this.

A 29.5" vacuum pump is 700$.
A vacuum chamber to fit the mobo etc. is a couple hundred bux if you make it.
gain: .001 %
Don't fuckin bother.

Listen.
Degassing is a legitimate thing.It's crucial in 'potted-for-space' stuff obviously, and even for fiberglass/concrete/carbon fiber.

TIM's on PC's (and even servers) ain't one of these times.

Edit 2 :

Lapping is easy.
get a 15" x 10" sheet of one inch thick glass.
Goto NAPA, get 3 sheets of 400,600,800 and 1000 grit wet/dry sandpaper.
Round and round she goes bud.Many hours.

Nice lapped surfaces can use cheese slices for TIM.

:ROFLMAO:
 
Last edited:
I would very much like to watch such a video.
I was joking, the guy made a mockery of DIY PC building. Used factory TIM pad on CPU cooler, then smothered it in paste. Like enough to do 5 cpus. Plus kept the factory pad on. its amazing it didnt start a fire.
 
Wait a sec.
Op's name does not check out.

You would think someone with 'honegod' as a name knew about the importance of lapping (or 'honing', haha)

Just jokes, =)

(y)
 
Wait a sec.
Op's name does not check out.

You would think someone with 'honegod' as a name knew about the importance of lapping (or 'honing', haha)

Just jokes, =)

(y)


When I created the title I was the sole Hone Operator at the spin-foundry/ machine shop I worked at.
The most interesting bit I got to work on was a submarine (US Navy) propeller shaft bushing. Specced at ambient Hawaii temperature, +/- .0015"

Fun because I was working in an unsealed, 'heated' only by the foundry at the other end of the building, in winter, in indiana.
So I had to correct the shape of the freezing huge micrometers to accurately reflect the final finished size and shape of the inside diameter reletive to the OD. (Wall thickness)
Correcting for temperature increase caused by the Honing process.

All corrected to Hawaii standard temperature.

I nailed it ! Took me about 2 weeks.

Roughly (very) 30" dia & 15' long.

Did I mention the deformation caused by gravity ? Measurable. And corrected for.

So yes I DO know.
Valid name. :)
 
The smaller-than-micro-sized air bubbles you would be removing via vacuum would represent a total equivalent surface area of 0.000000001 m^2 (just throwing some tiny number out there) which means f$&k all in terms of thermal heat transfer compared to the entire surface area of the CPU to heat sink thermal interface.
You won't be able to tell or measure the difference and the amount of torque you apply to the heat sink mounting screws will matter more than vacuum applied TIM every single time.
I mean if the air bubbles were 1mm in diameter and there were a lot of them, then sure, but we're talking about bubbles the size of a micrometer or nanometer or maybe even smaller. At that scale, you've got variances in the TIM compound itself that are larger. "boulders in a sea of pebbles"
 
Wikisurfing indicates the actual size of O2 and N are oscillating at a quantum level but generally would use angstroms as being about the right scale.

So, every bit of the sink surface starts with being covered by pressurised gas atoms (N) and molecules (O2) miles thick.
All WE are interested in is using thermal paste liquid to replace the gas coating at the angstrom level.

How many layers of gas particles will the paste displace ?

I'm looking at the paste like a ball of pizza dough on a stainless steel countertop being rolled out by a rolling pin.

To keep the dough from sticking to the surfaces, flour is scattered on the area.
The dough sticks to the flour instead of the rolling pin and table.

So does air stick to thermal paste ?

If the dough did NOT stick to the flour it would roll over the flour and not push it away. ending with a flat layer of flour between dough and table.

The stainless steel countertop tends to collect tiny scratches which fill with tiny amounts of flour. Which requires washing to dislodge the flour from the teensy scratches at the end of day.

(title prior to Honegod was Pizza Surgeon)
 
There is a such thing as too much TIM. You don't want that crap oozing all over your motherboard and squeezing out from under your thermal solution like some burger ruined with mayo.

Bullshit. Optimum application is your entire case is filled up with paste. Like a bucket. Paste gotta be better than air right? o_O:confused::LOL:
 
Back
Top