Jump to content


Photo

paste: always better than air?


  • Please log in to reply
22 replies to this topic

#1 Master_Scythe

Master_Scythe

    Titan

  • Hero
  • 20,106 posts
  • Location:QLD

Posted 27 September 2017 - 04:56 PM

I've had something bugging me for a while.

Take these two scenarios, which would cool better?

 

Lets say I have an IC (or in my case, a Transmitter amplifier)

 

Scenario 1:

- The chip is free to air

- Has forced air-flow over the top of the IC.

 

Scenario 2:

- The chip is covered in thermal paste, or ceramic (so it sets).

- The thermal paste has forced air flow over it.

 

I'm thinking in scenario 2, there's more thermal mass, so better cooling.... however.....

Where is the changeover?

 

Lets say, I had an IC, and I dunked it into a bucket of non-flowing thermal paste.

And I had a second identical IC, which I put a fan on, and it's exposed.

Where is the changeover point where 'Thick thermal paste' is more insulating, than it is aiding with cooling?

 

Or is this likely to never be met, because the thermal conductivity of Thermal paste is always better than air?

So as long as the paste had air flow, it's always better?

Risk of getting a 'hot spot' in the paste, with no flow? Or is thermal conductivity just too good in brand name paste.

 

 

 

 

Real world application?

 

The reason I ask, is because an RC hobby of mine is using some 5.8ghz video transmitters.

These things get hot.

Now, most of them are hiding behind anti-interference shields (metal plates).

 

I know an Air-Gap is the worst possible thing, as air itself insulates;

So in the case of VTX modules that are layered  PCB > Transmitter IC > AirGap > EMF Shield

I'm best to pop that shield off, fill the entire gap with non conductive thermal paste, and re-apply the shield.

But the question is, should I be shimming that gap, or can I just use 1mm thick of paste, and call it a day?

 


Wherever you go in life, watch out for Scythe, the tackling IT support guy.

"I don't care what race you are, not one f*cking bit, if you want to be seen as a good people, you go in there and you f*ck up the people who (unofficially) represent you in a negative light!"


#2 Rybags

Rybags

    Immortal

  • Super Hero
  • 35,024 posts

Posted 27 September 2017 - 05:11 PM

There was an article/YT video about air vs water and it mentions that most test scenarios are all wrong thanks to thermal mass.

ie - to get a fair test you need to do more than the simple "3 minutes flogging the crap out of it".

 

For stuff like voltage regs and components that get hot the general philosophy seems to be to have them in contact with RFI shield, case, whatever, to conduct the heat away.  The things are even designed as such in some cases to be screwed to something as well as soldered to the board.

 

The other thing - hot is a relative thing.  You don't want capacitors enduring 60 degree heat all day but with plenty of modern ICs and other components it won't matter.

 

 

For your scenario... I reckon heat conductive paste should help.  A problem with cooling is the transition between materials in different phase of matter, like solid to liquid then through solid to air for most cars.  The air of course being the weak link but ultimately carrying the unwanted energy away, the weakness being lessened by forcing lots of it through, and by the second last medium being a good conductor like copper or aluminium.

 

That should translate to hot components.  If the second last medium is something like ceramic or plastic then the pasthrough of heat to air will likely be poor.


Edited by Rybags, 27 September 2017 - 05:15 PM.


#3 Master_Scythe

Master_Scythe

    Titan

  • Hero
  • 20,106 posts
  • Location:QLD

Posted 28 September 2017 - 08:58 AM

Thanks, makes sense.

 

So, when we talk about thermal paste "Insulating a heatsink", when building computers, that's purely because no paste is as good as metal-to-metal contact, right?

So I guess by that logic, a metal shim  inside the paste is always the answer..... Otherwise we wouldn't lap heatsinks, we'd just use more goop.

 

It's purely the air-forced-over-a-surface vs air-forced-over-a-surface-ON-a-surface scenario that had me stumped.

 

Because without any education on the subject, something says in my mind that 'blowing air on something' cools it down.

But, why can't that air cool the thermal paste that's ON that something?

 

I'm thinking perhaps it greatly raises the thermal mass, but also greatly raises the thermal resistance to convection (of the whole package)?

 

I really don't know enough.

 

Where's Redhatter or Goth when you need them?


Wherever you go in life, watch out for Scythe, the tackling IT support guy.

"I don't care what race you are, not one f*cking bit, if you want to be seen as a good people, you go in there and you f*ck up the people who (unofficially) represent you in a negative light!"


#4 Dasa

Dasa

    Guru

  • Super Hero
  • 14,848 posts

Posted 28 September 2017 - 10:20 AM

a small amount of thermal past over a chip without a heatsink could insulate it not to mention that it would attract dust

the amount of heat created and the level of airflow would all affect which would be cooler

that said if you took it to the point of submerging the entire thing in mineral oil well that's a lot of extra surface area

 

https://www.gamersne...tivity-and-more

Material Thermal Conductivity (W/mK) at 25C Air, atmospheric 0.024 Water 0.058 Thermalpaste (AVG) ~5.3 - 8.5 Aluminum 205 Copper 401

Silverstone TJ08-EW 6700k@4.7 1.375v - Z170-GENE - 2x8g 3866 16-16-16 - 1070@ 2100\9600MHz  -Samsung 830 64G, Sandisk Ultra II 960G, WD Green 3tb - Seasonic XP1050 - Dell U2713 - Pioneer Todoroki 5.1 Apogee Drive II - EK VGA-HF Supreme  - Phobia 200mm Rad - Silverstone AP181 Project Darkling
3770k vs 6700k RAM Scaling, HT vs RAM, Arma III CPU vs RAM, Thief CPU vs RAM
Parts Guide


#5 Rybags

Rybags

    Immortal

  • Super Hero
  • 35,024 posts

Posted 28 September 2017 - 10:38 AM

Yeah, thinking again maybe an interface of paste to air would be worse than paste to metal.  Because the "TIM" aspect of the goop is what's important, we're using an OK conductor instead of the air gap which is a poor conductor.



#6 Dasa

Dasa

    Guru

  • Super Hero
  • 14,848 posts

Posted 28 September 2017 - 11:02 AM

out of the op options a alu\cop shim with two layers of paste would provide by far the lowest temp assuming the shield is a decent enough heatsink

 

that said maybe just get some thermal pads and call it a day https://www.pccasege...ng/thermal-pads

they work best when squished tightly between the chip and the heatsink

those ones are better than most that come on out gpu vrm


Edited by Dasa, 28 September 2017 - 11:04 AM.

Silverstone TJ08-EW 6700k@4.7 1.375v - Z170-GENE - 2x8g 3866 16-16-16 - 1070@ 2100\9600MHz  -Samsung 830 64G, Sandisk Ultra II 960G, WD Green 3tb - Seasonic XP1050 - Dell U2713 - Pioneer Todoroki 5.1 Apogee Drive II - EK VGA-HF Supreme  - Phobia 200mm Rad - Silverstone AP181 Project Darkling
3770k vs 6700k RAM Scaling, HT vs RAM, Arma III CPU vs RAM, Thief CPU vs RAM
Parts Guide


#7 stadl

stadl

    Guru

  • Super Hero
  • 9,042 posts

Posted 28 September 2017 - 11:35 AM

More thermal mass should not equal better cooling in the long term - it's like a capacitor - once saturated, it would not lower the steady state/average temperature, It would slow the rate of climb, and also slow the rate at which it drops.

Theoretically, dampening temperature fluctuations could increase service life, as most electro-mechanical logistics engineers will tell you changing temperatures kill faster than higher temperatures

 

Surface area increases opportunity for dissipation which increases cooling.

 

When it comes to thermal transfer: metal > paste > air. 

 

I'd expect a metal-paste-air interface is poorer than the metal-air of a bare component - unless your paste can significantly increase surface area, while not attracting non-thermal-conductive debris.

 

Thermal paste is intended to improve a metal-metal interface by excluding air. metal-paste-metal being better that metal-air-metal.

A lapped/polished chip and heatsink sans paste can be superior to an interface with paste, but the manufacturing tolerances required make it impractical in most cases.

Lapped with paste allows less paste to be used, and is a good balance.

 

Disclaimer - I have not studied thermodynamics.


A device so simple that it took a mind as brilliant as mine to create it! So brilliant, in fact, that simply by harnessing the power of one live frog, it... it... uhh... <poke> <poke> ... World domination has encountered a momentary setback. Talk amongst yourselves.


#8 Nich...

Nich...

    Professional Tart

  • Mod
  • 43,214 posts
  • Location:Mexico

Posted 28 September 2017 - 11:36 AM

The heat being emitted, how much would  be conductive and how much radiative?


"I think it is a sad reflection on our civilization that while we can and do measure the temperature in the atmosphere of Venus we do not know what goes on inside our soufflés" -- Nicholas Kurti

#9 stadl

stadl

    Guru

  • Super Hero
  • 9,042 posts

Posted 28 September 2017 - 11:44 AM

Not sure.

 

I'm familiar with the differences between conducted and radiated for electro-magnetic energy, but in the thermal domain, I would have thought they are basically the same, just through different mediums - unless by radiated you mean conversion of heat to IR.


A device so simple that it took a mind as brilliant as mine to create it! So brilliant, in fact, that simply by harnessing the power of one live frog, it... it... uhh... <poke> <poke> ... World domination has encountered a momentary setback. Talk amongst yourselves.


#10 Master_Scythe

Master_Scythe

    Titan

  • Hero
  • 20,106 posts
  • Location:QLD

Posted 28 September 2017 - 12:56 PM


I'd expect a metal-paste-air interface is poorer than the metal-air of a bare component - unless your paste can significantly increase surface area, while not attracting non-thermal-conductive debris.

 

That's what I'm trying to figure out.

Why.

 

At a thought, is it safe to assume the plastic chip is better at convection than thermal paste itself would be?

So although convection can still happen, into the paste, once the paste heats up, then the PASTE is bad at convection?

 

I think I'll still fill the RF shield with non conductive paste, because I know from working on car exhaust systems, NOTHING insulates better than an air-gap, and the shield is airgapped from the IC.

 

So once the air pocket saturates with heat, I'd imagine THAT's why it skyrockets.

 

Where as if there was paste, it'll get to the RF shield more efficiently than air would carry it (as there's no flow).

 

Yes?


Wherever you go in life, watch out for Scythe, the tackling IT support guy.

"I don't care what race you are, not one f*cking bit, if you want to be seen as a good people, you go in there and you f*ck up the people who (unofficially) represent you in a negative light!"


#11 stadl

stadl

    Guru

  • Super Hero
  • 9,042 posts

Posted 28 September 2017 - 02:31 PM

Chips in plastic cases are usually not good for any thermal conduction. If you have plastic in the mix rather than ceramic, then all bets are off I would expect, but airflow will always help

 

Many plastic cased ICs dissipate more heat through their legs than the package, so you PCB design - pads and tracks - might be more of a factor than any paste or airflow across the package.

 

And in a solid RF shield, then your options are:

1, conduct to the shield with metal and/or paste.

2.fanforce to equally warm all air in the shielded enclosure and try on conduction through the shield to the outer environment - could benefit from heatsink fins on the inside and outside of the RF shield.

inside the shield - in which case the volume of air in the shielded enclosure, 

 

Vented RF shields are much easier.


A device so simple that it took a mind as brilliant as mine to create it! So brilliant, in fact, that simply by harnessing the power of one live frog, it... it... uhh... <poke> <poke> ... World domination has encountered a momentary setback. Talk amongst yourselves.


#12 Master_Scythe

Master_Scythe

    Titan

  • Hero
  • 20,106 posts
  • Location:QLD

Posted 28 September 2017 - 04:04 PM

Chips in plastic cases are usually not good for any thermal conduction. If you have plastic in the mix rather than ceramic, then all bets are off I would expect, but airflow will always help

 

Many plastic cased ICs dissipate more heat through their legs than the package, so you PCB design - pads and tracks - might be more of a factor than any paste or airflow across the package.

 

And in a solid RF shield, then your options are:

1, conduct to the shield with metal and/or paste.

2.fanforce to equally warm all air in the shielded enclosure and try on conduction through the shield to the outer environment - could benefit from heatsink fins on the inside and outside of the RF shield.

inside the shield - in which case the volume of air in the shielded enclosure, 

 

Vented RF shields are much easier.

 

These are cheap RC transmitters

https://www.banggood...-p-1147691.html

So getting 'good vented ones' or such just isnt going to happen, lol.

 

OK, so it sounds like I'm best trying to shim the space, but since I'm worried about shorting anything out, and since it's a solid shield, I'll fill it with Gelid GCExtreme thermal paste and call it a day, I think.

Might Ram-Sink the shield too.


Wherever you go in life, watch out for Scythe, the tackling IT support guy.

"I don't care what race you are, not one f*cking bit, if you want to be seen as a good people, you go in there and you f*ck up the people who (unofficially) represent you in a negative light!"


#13 stadl

stadl

    Guru

  • Super Hero
  • 9,042 posts

Posted 28 September 2017 - 05:10 PM

In the case of a receiver or transmitter like that I'd generally avoid messing with anything inside the enclosure.

Anything conductive may alter the RF performance and cause interference problems, well before you actually short the transmitter to the case.


A device so simple that it took a mind as brilliant as mine to create it! So brilliant, in fact, that simply by harnessing the power of one live frog, it... it... uhh... <poke> <poke> ... World domination has encountered a momentary setback. Talk amongst yourselves.


#14 @~thehung

@~thehung

    Guru

  • Hero
  • 8,611 posts

Posted 28 September 2017 - 11:38 PM

I'd expect a metal-paste-air interface is poorer than the metal-air of a bare component - unless your paste can significantly increase surface area


great post, but i am having some trouble with this.

seems to me that if metal-air is better, then it is a less restrictive bottleneck to heat transfer than metal-paste. in which case, no matter how good things get on the outgoing side of the metal-to-paste interface, transfer is hard limited by this narrower bottleneck. 

 

by this way of thinking, the advantage of metal-paste, then, is to mitigate against variability in the quality of air (temperature/speed) otherwise in direct contact with the bare metal, which might cause its average conductivity to drop below that of the metal-paste bottleneck. 

 

what bothers me about this, is the existence of circuits in cases without fans or venting that nevertheless contain components with massive pasted on heatsinks.  assuming enough of these components are bare metal, this suggests a very low threshold for how slow moving and/or warm air can be before metal-air is eclipsed by metal-paste.  namely, if this threshold wasnt very low, it would not explain why slapping a chunk of very large surface area aluminium over thermal paste would do much at all in 'stationary' air.   my theory: the heatsink is there to catch as many low temperature air currents as it possibly can, in order to compensate for inherently poor metal-air transference — and the reason it cools the component "faster" is because it retards saturation of the metal-paste bottleneck.


Edited by @~thehung, 29 September 2017 - 12:44 AM.

no pung intended

#15 stadl

stadl

    Guru

  • Super Hero
  • 9,042 posts

Posted 29 September 2017 - 07:52 AM

You can look at a single interface or the whole system.

 

At a single interface, for a given surface area, Metal-Paste is better than Metal-Air.

 
At a system level, the intent of a Metal-Paste interface is usually followed by a Paste-Metal interface with the intent of allowing heat to be transferred to something that can increase the surface area.

So the typical design comparison is Metal(small)->Paste->metal(small)...Metal(large)->Air vs Metal(small)->Air

 

Adding a dob of paste to a chip with no plan to pass it on to a larger surface area piece of metal really just means Metal(small)->paste(small)->air

Without getting any surface area increase, you're not getting heat transfer to the environment improved. You are just adding thermal mass, which will smooth out spikes, but maintain similar average temps.

 

The same thing would probably occur if you simply pasted a piece of 1mm think metal shim to the top of a chip - it would not make any significant change to surface area, the extra effort would offer little benefit until you make the metal on top have a larger surface area.

 

If you got a small amount of paste, and make it spiky, that would add surface area and in theory improve over bare metal - but as soon as it clogged with dust/dirt, you're probably lose any surface area benefit.


A device so simple that it took a mind as brilliant as mine to create it! So brilliant, in fact, that simply by harnessing the power of one live frog, it... it... uhh... <poke> <poke> ... World domination has encountered a momentary setback. Talk amongst yourselves.


#16 Master_Scythe

Master_Scythe

    Titan

  • Hero
  • 20,106 posts
  • Location:QLD

Posted 29 September 2017 - 08:45 AM

In the case of a receiver or transmitter like that I'd generally avoid messing with anything inside the enclosure.

Anything conductive may alter the RF performance and cause interference problems, well before you actually short the transmitter to the case.

 

Thats why I was going to use a cheap silicone heat paste, or an expensive "Non conductive" PC paste.

Because that way I'm adding no material that should mess with RF.

 

Just they're ALL air-gapped, and it's not on purpose, at 400mW+ they burn themselves out.


Wherever you go in life, watch out for Scythe, the tackling IT support guy.

"I don't care what race you are, not one f*cking bit, if you want to be seen as a good people, you go in there and you f*ck up the people who (unofficially) represent you in a negative light!"


#17 stadl

stadl

    Guru

  • Super Hero
  • 9,042 posts

Posted 29 September 2017 - 09:17 AM

 

Thats why I was going to use a cheap silicone heat paste, or an expensive "Non conductive" PC paste.

Because that way I'm adding no material that should mess with RF.

 

Just they're ALL air-gapped, and it's not on purpose, at 400mW+ they burn themselves out.

 

 

I take it they are being fitted to mini UAVs/quadcopters - is there somewhere you can place them to get good external airflow or environmental cooling?

 

There is also the option of removing the RF shield if you're willing to ignore EMI regulations - which are probably non existent with a part like that from banggood - the assumption is probably the system integrator will meet EMI regs for the commercial product they install it into.

 

For the price they are cheap enough for some trial and error until you find a setup that can survive without burning out.

If you can manage to avoid the dust/dirt getting stuck on it, I can't see chip-paste-air being any worse than chip-air, so not much to lose.


Edited by stadl, 29 September 2017 - 09:24 AM.

A device so simple that it took a mind as brilliant as mine to create it! So brilliant, in fact, that simply by harnessing the power of one live frog, it... it... uhh... <poke> <poke> ... World domination has encountered a momentary setback. Talk amongst yourselves.


#18 Master_Scythe

Master_Scythe

    Titan

  • Hero
  • 20,106 posts
  • Location:QLD

Posted 29 September 2017 - 09:27 AM

 

 

Thats why I was going to use a cheap silicone heat paste, or an expensive "Non conductive" PC paste.

Because that way I'm adding no material that should mess with RF.

 

Just they're ALL air-gapped, and it's not on purpose, at 400mW+ they burn themselves out.

 

 

I take it they are being fitted to mini UAVs/quadcopters - is there somewhere you can place them to get good external airflow or environmental cooling?

 

There is also the option of removing the RF shield if you're willing to ignore EMI regulations - which are probably non existent with a part like that from banggood - the assumption is probably the system integrator will meet EMI regs for the commercial product they install it into.

 

Yeah it'll have plenty of airflow, but chip>air>shield means even with flow, it doesn't help.

 

Removing the EMI shield would actually degrade wouldn't it? Because anything near the 5.8ghz band will be able to "get in" to the chip?


Wherever you go in life, watch out for Scythe, the tackling IT support guy.

"I don't care what race you are, not one f*cking bit, if you want to be seen as a good people, you go in there and you f*ck up the people who (unofficially) represent you in a negative light!"


#19 stadl

stadl

    Guru

  • Super Hero
  • 9,042 posts

Posted 29 September 2017 - 10:34 AM

 

 

Removing the EMI shield would actually degrade wouldn't it? Because anything near the 5.8ghz band will be able to "get in" to the chip?

 

 

 

As a transmitter 'getting in' can be less of an issue than getting out.

 

The risks are usually that something mixes before the final stage amplification - ie you mix and repeat other signals, or they get in at one of the lower frequency inputs, so the quality of your output could be degraded, but trial and error will determine if that degradation is significant/detrimental. If you have multiples of them running around near each other that is when you might have problems

 

 

At those frequencies and power levels, it's not great danger, but also not trivial - I don't know how much RF Safety is involved - but you are hardly strapping it to your brain - but with it removed, I would avoid touching it too much when operating - a few cm are probably all you need. 

 

An alternative option would be to remove the shield and replace it with a larger one with more air. If you want to make your own, some metal sheet can be cut, folded and even seam soldered. With the right material, you could increase the volume significantly, while adding very little weight.

 

An alternative would be to try perforating the current shield one to improve airflow.

 

At 5.7GHz, you're around 5cm wavelength, so shielding with holes of a couple of mm are probably fine. It works off maximum dimension, so long slots are not good for shielding, even if their area is small - which when you look at that shield, it's only soldered in 4 corners so it's effectively got a couple of slots about 1.5cm long.

The smaller the holes the better the shielding, but I would expect you could balance the shielding with the cooling if you are a bit casual with meeting the standards :)

 

Yes, changing the metal box around the components may change the characteristics, but changing the box will have less effect that metal directly on components which was my original concern.


A device so simple that it took a mind as brilliant as mine to create it! So brilliant, in fact, that simply by harnessing the power of one live frog, it... it... uhh... <poke> <poke> ... World domination has encountered a momentary setback. Talk amongst yourselves.


#20 Master_Scythe

Master_Scythe

    Titan

  • Hero
  • 20,106 posts
  • Location:QLD

Posted 29 September 2017 - 10:52 AM

 

 

 

Removing the EMI shield would actually degrade wouldn't it? Because anything near the 5.8ghz band will be able to "get in" to the chip?

 

 

 

At 5.7GHz, you're around 5cm wavelength, so shielding with holes of a couple of mm are probably fine.

 

 

How about folding some metal flyscreen?

 

EDIT: and would I be best to connect said flyscreen to the antenna GROUND, to act as a faraday cage for the chip?


Edited by Master_Scythe, 29 September 2017 - 10:55 AM.

Wherever you go in life, watch out for Scythe, the tackling IT support guy.

"I don't care what race you are, not one f*cking bit, if you want to be seen as a good people, you go in there and you f*ck up the people who (unofficially) represent you in a negative light!"





0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users