PCB cooling, how do you do it?

[Noci]

We can cool our CPU & GPU dies quite well with all kind of different coolers, either liquid, air cooling or even more exotic, but how do you keep the PCB were it is mounted on to cool in general?

In my scenario the heat generated by the VRM's is mostly dissipiated through the cooler on the MB, but a part of heat stays in the PCB, despite a forced airflow over it, or at least it looks that way when I make a thermal image.

Heat is concentrated around the AM4 socket and towards the M2-1, PHC and RAM, remarkably not on the ATX 24-pin and EPS 8-pin (the hotspot on the right are the RAM VRM's without heatsink, on the bottom the heatsink of the PHC, fan is off because of silent mode). I'm pretty sure it goes for other setups like AM5 and LGA1700 likewise.

The system is stable (no worries) so probably within specs, but the Delta-T's on the PCB made me wonder if this has impact on the overall performance and if anybody has more information about this.
All I can do is make sure there is a forced airflow over the front and back of the mainboard to cool as much as possible.

So please share your knowledge/ideas as I'm curious about other opinions.

 

[Deleted member 237269]

Any HWinfo on a 30min stress test run ~
Doesnt seem that hot on your thermal snapshot. How much dBA to loose 2-3°C and no real performance Mhz etc.
Could upgrade for a new AMD 5 setup if you are wealthy

I have the MSI MPG X670E, i hate this stupid looking Wifi antenna.

 

[Noci]

View attachment 329704
Any HWinfo on a 30min stress test run ~
Doesnt seem that hot on your thermal snapshot. How much dBA to loose 2-3°C and no real performance Mhz etc.
Could upgrade for a new AMD 5 setup if you are wealthy
View attachment 329708
I have the MSI MPG X670E, i hate this stupid looking Wifi antenna.

Not wealthy at all but I get by, luckily a normal level of health so better off than some unhealthy billionaire .

At the time I took the thermal images the ambient temps were quite low (18 degrees Celcius and cooling liquid around 22 degrees Celcius), but that is not the point.

I'm sure If I make a thermal image of any X670 board under load, it will also show some Delta-T on spots where you have the PCB copper traces that have to do a lot of transport work. even with that high-end MEG X670 ACE consumer board (feel some comments coming on this remark about that MB already ).
As with any non-destructive testing, you can master the technique(s), but the hardest part is the interpretation of the test results.

It is that we basically cool the (most) heat producing components, but we neglect the interconnecting transport lines which are bound to the same physical laws, more heat means more electrical resistance (at least that is what I remember from school decades ago). So my basic question is: does this have impact? If yes, will it be noticable and can we do something about it?
Buying a high end board is not an option for everyone.

Regarding the antenna (I've pretty much the same), just use the wired 10 Gbps connection of your board, waaay faster if you have the proper setup. Than put the antenna on a place where the sun doesn't shine .
Thanks for the feedback.

 

[Solaris17]

which are bound to the same physical laws

they are, but PCBs can take heavy temps. They are glass and resin mixtures. High temp PCBs like those used in motherboards can do over 150+C

 

[Noci]

I'm very satisfied with my cool board..
I love that VRM cooling which works great too...

View attachment 329709

View attachment 329710
View attachment 329711

Gigabyte Z690 Aorus Master.

Glad to hear you're happy with it, splendid board by the way.
I'm sure the NanoCarbon Aluminium backplate is related to my question and the confirmation that the R&D of Gigabyte gave it some thoughts also.

Additionally, the more phases on the power supply, the less they are individually stressed and thus produce less heat. Not sure about the number of layers of the PCB yet (Yours 8-layer, mine 6-layer) if that will have a mayor impact on heat, correct me if I'm wrong.

Thanks for replying.

they are, but PCBs can take heavy temps. They are glass and resin mixtures. High temp PCBs like those used in motherboards can do over 150+C

Yup, I know, I'm not worried about a smoking MB , but thanks anyway for the confirmation.
And to be honest I don't think PCB cooling is a necessity (yet), just wondering if it makes a difference as mentioned before.

So how far can we stretch the heating of the copper traces on a PCB untill we notice any degradation in performance due to increased electrical resistance? Copper is quite pricy and MB manufacterers will use as little as possible to lower manufactering costs, as they do with components on the PCB's where possible.

Personally I have no idea how thin these traces are, is there an electrical engineer out there who can shed some light on this?

 

[Zach_01]

We can cool our CPU & GPU dies quite well with all kind of different coolers, either liquid, air cooling or even more exotic, but how do you keep the PCB were it is mounted on to cool in general?

In my scenario the heat generated by the VRM's is mostly dissipiated through the cooler on the MB, but a part of heat stays in the PCB, despite a forced airflow over it, or at least it looks that way when I make a thermal image.

View attachment 329695 View attachment 329701

Heat is concentrated around the AM4 socket and towards the M2-1, PHC and RAM, remarkably not on the ATX 24-pin and EPS 8-pin (the hotspot on the right are the RAM VRM's without heatsink, on the bottom the heatsink of the PHC, fan is off because of silent mode). I'm pretty sure it goes for other setups like AM5 and LGA1700 likewise.

The system is stable (no worries) so probably within specs, but the Delta-T's on the PCB made me wonder if this has impact on the overall performance and if anybody has more information about this.
All I can do is make sure there is a forced airflow over the front and back of the mainboard to cool as much as possible.

So please share your knowledge/ideas as I'm curious about other opinions.

The temps you are showing are really nothing for almost any board. Anything below 50C is not worth even mentioning
I'm doing something but its not for the board really, its because PC is caseless and have to have some air flow around components like VRM, DRAM, NVMe drives, chipset and such

and it goes like this
(3x120mm fans)



Ambient temp around PC is usually 21~22C
Upper left fan is bound to VRM temp (custom curve)
Upper right fan to CPU temp (custom curve)
Lower fan to PCI-E_1 (GPU) slot temp (custom curve)









Of course there is a dust issue with such a configuration but there is also a cleaning routine 4-5times/year

 

[Shrek]

they are, but PCBs can take heavy temps. They are glass and resin mixtures. High temp PCBs like those used in motherboards can do over 150+C

I believe it is the heavy copper ground planes in the board that carried a lot of the heat.

 

[Deleted member 237269]

The temps you are showing are really nothing for almost any board. Anything below 50C is not worth even mentioning
I'm doing something but its not for the board really, its because PC is caseless and have to have some air flow around components like VRM, DRAM, NVMe drives, chipset and such

and it goes like this
(3x120mm fans)

View attachment 329725

Ambient temp around PC is usually 21~22C
Upper left fan is bound to VRM temp (custom curve)
Upper right fan to CPU temp (custom curve)
Lower fan to PCI-E_1 (GPU) slot temp (custom curve)

View attachment 329726

View attachment 329727

View attachment 329728

View attachment 329729

Of course there is a dust issue with such a configuration but there is also a cleaning routine 4-5times/year

Holy never paid attention but Aorus bios look nice and easier for fan curve. MSI is crap

Your pc picture is giving me ocd anxiety...

 

[Zach_01]

Holy never paid attention but Aorus bios look nice and easier for fan curve. MSI is crap

That was not BIOS screenshot but Gigabyte software (SIV, System Information Viewer)
This is BIOS (SmartFan5 section)

Your pc picture is giving me ocd anxiety...

Well, I couldn't care less about looks, but I do care about functionality and ...temp

 

[Noci]

@ Zach_01, yeah my Thermaltake core P90 is pretty much an open case too and as mentioned before I do have an forced airflow over the MB, serving the VRM heatsinks, RAM, M2 & PCB heatsinks.

It actually does help with keeping the temps down, thats why I have these nice temps as shown in the thermal image.
I even have a Artic P14 blowing air through on the MB backside (using the hole for the CPU cooler backplate), done that with all my builds ever since back in the days I owned a InWin Dragon Rider case which had a cut out with mesh and a 120mm fan mounting spot in the case side panel on the side of the mainboard's backside.
https://www.techpowerup.com/review/in-win-dragon-rider/3.html

But cleaning your PC only once every 4 to 5 years, hell that is more than a PC lifetime for a lot of forum users here .

The Gigabyte System Information Viewer does look nice, better than MSI Center in my opinion but yeah it's indeed all about functionallity.

@ Shrek, that is an interresting thought, wonder if this is actually a parameter in PCB design, would be logical as we're using more and more power hungry components nowdays.

 

[Calenhad]

The PCB can soak up a serious amount of heat. This is why higher end motherboards are advertised with thicker ground planes than cheaper ones. It does work as a heatsink as well.

Honestly, you are looking for a nonexisting problem. 37C, if I read your IR image correctly, is well within normal operating temperature. When the PCB is hot enough to increase the heat of other components, then you should start considering your cooling solution.

You would also probably be breaking the laws of physics, since those components are what is generating the heat in the first place.

 

[Noci]

The PCB can soak up a serious amount of heat. This is why higher end motherboards are advertised with thicker ground planes than cheaper ones. It does work as a heatsink as well.

Honestly, you are looking for a nonexisting problem. 37C, if I read your IR image correctly, is well within normal operating temperature. When the PCB is hot enough to increase the heat of other components, then you should start considering your cooling solution.

You would also probably be breaking the laws of physics, since those components are what is generating the heat in the first place.

Thanks for the feedback, but I never mentioned it was a problem in this case especially with low temperatures as in the thermal image. My rig has an overkill regarding to cooling for my configuration, so no worries there.

It's just if you look at the thermal image with the delta-T on spots of the PCB, the interpretation of the results is the tricky part and made me thinking.
Thus wondering if I was the only one to apply some sort of cooling to the PCB by surrounding it with a forced airflow. Even if it does not do much, it won't do harm, as you mentioned a PCB can soak up a serious amount of heat, so the faster you get rid of it the better.
Also with the increasing amps going through the traces of a PCB, I wonder when we will hit limitations or even before that notice something in performance drop.

In the end I just hope to learn here what are the facts and not assumptions based on gut feeling. Again thanks for your opinion.

 

[Zach_01]

But cleaning your PC only once every 4 to 5 years, hell that is more than a PC lifetime for a lot of forum users here .

4-5times/year = 4-5 times per year

 

[Noci]

4-5times/year = 4-5 times per year

It was late and I overlooked it , but that makes more sense. I'm guilty of not cleaning it often enough myself, every three months should be a good start.

 

[AsRock]

they are, but PCBs can take heavy temps. They are glass and resin mixtures. High temp PCBs like those used in motherboards can do over 150+C

Or if it such a concern, you could get a fan that covers the back of the cpu some stand off's and put a fan there. I think most newer cases have the CPU cutout already .

I was going to do it but it's just to much hassle for what it's worth haha

 

[kapone32]

Op could buy a new AIO called the Silverstone Ice Mist. You can actually add fans to the AIO header to achieve what you want. Too bad they don't come with the fans. Arctic also has a little fan on it's AIO's to help with what you want.

View attachment 329704
Any HWinfo on a 30min stress test run ~
Doesnt seem that hot on your thermal snapshot. How much dBA to loose 2-3°C and no real performance Mhz etc.
Could upgrade for a new AMD 5 setup if you are wealthy
View attachment 329708
I have the MSI MPG X670E, i hate this stupid looking Wifi antenna.

I actually wanted the Ace but man is that board pricey. I was going to get the Carbon but settled on the Asus Strix E.

 

[Chrispy_]

LOL, this thread is amusing!

The fibreglass that PCBs are made from has a service temperature of 1000F/550C. Cooling it is pointless and a waste of effort.

Outside of the VRMs, the next most heat-sensitive thing on a motherboard are the polymer caps, typically rated to operate from -55C to 125C. That's the service temperature at which the typical 100,000h MTBF rating is tested at. It doesn't mean that they won't still work at 200C, only that their lifespan will be shorter than rated and will probably still work for tens of thousands of hours at those ridiculous temperatures. This is all hypothtetical though - because if your motherboard is over 125C then your CPU or GPU are probably on fire and the system needs to be extinguished.

 

[P4-630]

The fibreglass that PCBs are made from has a service temperature of 1000F/550C. Cooling it is pointless and a waste of effort.

Before reaching those temps all the solder would have melted already...
So all components would have fallen of the board...

 

[Chrispy_]

Before reaching those temps all the solder would have melted already...
So all components would have fallen of the board...

Exactly - just like I said earlier it would already be on fire, most likely.

 

[Shrek]

The fibreglass that PCBs are made from has a service temperature of 1000F/550C. Cooling it is pointless and a waste of effort.

The epoxy can take much less, but one would cool the motherboard, not for the motherboard but for the components using it as a heatsink.

 

[Noci]

LOL, this thread is amusing!

The fibreglass that PCBs are made from has a service temperature of 1000F/550C. Cooling it is pointless and a waste of effort.

Outside of the VRMs, the next most heat-sensitive thing on a motherboard are the polymer caps, typically rated to operate from -55C to 125C. That's the service temperature at which the typical 100,000h MTBF rating is tested at. It doesn't mean that they won't still work at 200C, only that their lifespan will be shorter than rated and will probably still work for tens of thousands of hours at those ridiculous temperatures. This is all hypothtetical though - because if your motherboard is over 125C then your CPU or GPU are probably on fire and the system needs to be extinguished.

Good thing I can still amuse somebody with a techincal question. But where in this thread did you get the idea of somebody being afraid of the PCB catching fire ?

I personally find it amusing how quick a discussion can deviate from its intended purpose without reason.

The original question was to share knowledge/idea's of the fact that the traces on the PCB to the main users show a higher temp than the adjecent parts and if this is or will be noticeable at a certain point in the form of performance loss.
The fact that my gut feeling is that at least I cool a little with just providing a airflow on both sides of the PCB could be wasted energy, but why not it doesn't harm also. I cannot substatiate this by facts (measurements), but prove me wrong

That the question of cooling PCB's is not completely out of the blue is proven by the fact that already two decades ago there was scientific research on cooling PCB's, see the publication in the link below for example:

https://www.researchgate.net/profil...s-Fabricated-With-the-Direct-Write-Method.pdf

My knowledge of printed circuits and electronics is very limited, just the fact that: from Ohm's law, we know that resistance is inversely proportional to current. That is, the lower the resistance, the higher the current and vice versa. Current is also proportional to power according to P = VI. Thus, resistance also becomes a factor when calculating power consumption. It is important to determine the resistance of the trace of a microstrip so that the power dissipated by it can be determined (this text I took from a Trace Resistance Calculator website ).
So with the ever increasing power demand of our hardware this got me wondering how far can we stretch it with the current PCB technology.

So some food for thoughts for the skeptical ones amongst us .

 

[thesmokingman]

This has never been a concern for me, ever.

 

[ShrimpBrime]

The opposite effect.

How do I get my traces warmed up??!!

 

[dgianstefani]

It's not about cooling the PCB for it's own sake, but that the PCB can influence temperatures of RAM, chipset, socket etc. There's reasons people put thermal pads on the backside of mobos, or have direct airflow etc.

 

[Deleted member 237269]

The opposite effect.

How do I get my traces warmed up??!!

View attachment 329846

Not sure the warranty cover this