Swapping fans on heatsink with 120+140mm fans

[freeagent]

25 minutes ago, RevGAM said:

Hmmm, I thought the primary carrier of the heat was the vapor inside? Apparently, I misunderstood, yet ...Bah, I should've gone to school for something else. XD

I didn't finish school, so don't feel bad

[doug_locke]

A little caveat, I didn't use the included thermalright paste in any of my testing. I use Gelid GC-extreme. (very thick paste). 

AND, all previous tests were dont on cold days, my home's heat kept the room at 70°F, CPU idled at 39.0°-39.2° today is unseasonably warm room is 73° CPU idle temp up at 40.5-41.5°C.
 

With FS140 3 fans, cinebench dropped from a consistent 21.0k+ to 19.7k-20.1k... The very first Heatsink I triead was the single tower singlefan Cryorig H5 Ultlimate (with Am4 adapter)... It did hit the thermal limit but Cinebench was still >19.3k, I didn't note what the idle temps were. 

 

Apparently cracking the window wouldve been better than all the time I wasted messing around with fans.

20 hours ago, freeagent said:

If I buy one and find FC140 is better I will be pissed.

 

Edit:

 

It doesn't matter though, as Phantom Spirit is rated for higher TDP, and is nice and tiny in comparison And the EVO can take a T30 in the center without moving the fin stacks or stretching the clips.

I can't find the post now, but their was some talk that FC140 mount caused some flexing on AM5 boards and "maybe" thats why it's performing worse, possibley TL's mounting plate would fix the issue. 
I'd bet it would work better on a larger chip where all 5 heat pipes directly contact the chip/plate.

 

 

[freeagent]

You should try out the TF7, you might be surprised. Use the spatula and put a nice thin even layer on. I cannot speak for AM5, but TR had their AM5 contact frame out as soon as the boards were available.

[RevGAM]

9 hours ago, freeagent said:

You should try out the TF7, you might be surprised. Use the spatula and put a nice thin even layer on. I cannot speak for AM5, but TR had their AM5 contact frame out as soon as the boards were available.

TF7 is indeed quite good.

[doug_locke]

21 hours ago, freeagent said:

You should try out the TF7, you might be surprised. Use the spatula and put a nice thin even layer on. I cannot speak for AM5, but TR had their AM5 contact frame out as soon as the boards were available.

TF7 is definitely better on paper.

Very few beat gelid for longevity or when your constantly wiggling the heatsink clipping/unclipping a few dozen fan configs. 

 

 

 

Heatsinks from last decade only show a tenth of a percent for top 10-20 pastes. But, Thermalright heatsinks are cheap chunky things lots of metal between the the heat pipes and processorr 8 vs 12+w/mk paste ad/or lapping might make a bigger difference than they do in pricier brands. 

 

 

 

I'll try TF7 today

 

 

 

[RevGAM]

1 hour ago, doug_locke said:

TF7 is definitely better on paper.

Very few beat gelid for longevity or when your constantly wiggling the heatsink clipping/unclipping a few dozen fan configs. 

 

 

 

Heatsinks from last decade only show a tenth of a percent for top 10-20 pastes. But, Thermalright heatsinks are cheap chunky things lots of metal between the the heat pipes and processorr 8 vs 12+w/mk paste ad/or lapping might make a bigger difference than they do in pricier brands. 

 

 

 

I'll try TF7 today

 

 

 

I've got Gelid GC Extreme, so I hope you'll share how it compares to TF-7. 

 

Alphacool Silver is terrible!

[freeagent]

I use SYY-157 myself, cheap and great. Just go on thinner or else it becomes an insulator.

[RevGAM]

21 minutes ago, freeagent said:

I use SYY-157 myself, cheap and great. Just go on thinner or else it becomes an insulator.

How does it compare to TF-7?

[freeagent]

4 hours ago, RevGAM said:

How does it compare to TF-7?

It’s better

[doug_locke]

6 hours ago, RevGAM said:

I've got Gelid GC Extreme, so I hope you'll share how it compares to TF-7. 

 

Alphacool Silver is terrible!

 

On 3/16/2024 at 6:48 PM, freeagent said:

You should try out the TF7, you might be surprised. 

I am indeed surprised.

 

Idle temp didn't change (my fans stay at 400 rpm til the CPU hits 50° idle stayed between 40 and 41 with all 4 configs.

 

With TF-7 idle temps didn't change at all (my fans are set to 500rpm til 50°)

 

Today's cinebench score and maxtemp

 

TC140 stock fans + gelid extreme 20112, 

91.2° 

TC140 3 fans + gelid extreme 20342, 89.6°

TC140 stock fans + TF7 20538, 85.8°

TC140 3fans + TF7 20644, 85.0°

 

The max temp dropped about 5°,which admittedly is suspicious. 

 

TF7 is about as viscous as gelid extreme, not thin paste. 

 

3 fan setup only hit 85° a few seconds per minute. TC140 with 2 fans pretty much sat above 85° for the full test... Its still not worth buying an extra if you don't already own one, 3 fans "might' smooth out temp. spikes in normal usage.

 

 

[RevGAM]

23 hours ago, doug_locke said:

 

I am indeed surprised.

 

Idle temp didn't change (my fans stay at 400 rpm til the CPU hits 50° idle stayed between 40 and 41 with all 4 configs.

 

With TF-7 idle temps didn't change at all (my fans are set to 500rpm til 50°)

 

Today's cinebench score and maxtemp

 

TC140 stock fans + gelid extreme 20112, 

91.2° 

TC140 3 fans + gelid extreme 20342, 89.6°

TC140 stock fans + TF7 20538, 85.8°

TC140 3fans + TF7 20644, 85.0°

 

The max temp dropped about 5°,which admittedly is suspicious. 

 

TF7 is about as viscous as gelid extreme, not thin paste. 

 

3 fan setup only hit 85° a few seconds per minute. TC140 with 2 fans pretty much sat above 85° for the full test... Its still not worth buying an extra if you don't already own one, 3 fans "might' smooth out temp. spikes in normal usage.

 

 

Thanks for that! I have a long list of projects for my channel, but one is testing all the thermal pastes I have, so it'll be interesting if I get similar results. 

 

Depending on the fans, more can lower noise until it ramps up too high, but placement is important to avoid dead spots and turbulence as I'm sure you know. 

Edited March 19 by RevGAM
Fixed Otto Rong's spelling error.

[doug_locke]

8 hours ago, RevGAM said:

Thanks for that! I have a long list of projects for my channel, but one is testing all the thermal pastes I have, so it'll be inserting if I get similar results. 

 

Depending on the fans, more can lower noise until it ramps up too high, but placement is important to avoid dead spots and turbulence as I'm sure you know. 

 

Elaborating on your point. Most modern cases are able to mount 3x120mm top fans (for AIO). Putting an exhaust fan in the rear and rear most top position helps temps, 2nd and 3rd top exhaust fan with these big air coolers hurt temps. They'd suck air up before it reaches the CPU fan. 

 

 I did try many many case fan setups. 1 rear exhaust fan one top rear exhaust fan (6 intakes). This is where I landed

 

Top intake fan is  set to  20% and spaced away from the top exhaust but still close enough so that airflow is blocked off from the bottom intake by the gpu. It's only there to keep cool air from leaking before reaching the cpu, running it at max speed is worse than not running it at all. 

I also closed off the gap between the two top fans so they wont just cycle, much of, the same air from one another. 

I mounted the top intake at a slight angle, just to limit turbulence. 

 

 

Lastly using the exact same model for the 3rd cpu fan and back exhaust is pretty quiet, if one moves a lot more air, it can get loud unless the back fan is slightly off center.

 

 

 

[RevGAM]

2 hours ago, doug_locke said:

 

Elaborating on your point. Most modern cases are able to mount 3x120mm top fans (for AIO). Putting an exhaust fan in the rear and rear most top position helps temps, 2nd and 3rd top exhaust fan with these big air coolers hurt temps. They'd suck air up before it reaches the CPU fan. 

 

 I did try many many case fan setups. 1 rear exhaust fan one top rear exhaust fan (6 intakes). This is where I landed

 

Top intake fan is  set to  20% and spaced away from the top exhaust but still close enough so that airflow is blocked off from the bottom intake by the gpu. It's only there to keep cool air from leaking before reaching the cpu, running it at max speed is worse than not running it at all. 

I also closed off the gap between the two top fans so they wont just cycle, much of, the same air from one another. 

I mounted the top intake at a slight angle, just to limit turbulence. 

 

Lastly using the exact same model for the 3rd cpu fan and back exhaust is pretty quiet, if one moves a lot more air, it can get loud unless the back fan is slightly off center. FYI, a lot of larger cases have 3 in the front, and yet many cases only have 2 on top. I opted for the Corsair iCUE 5000x RGB, which can hold 10 fans, of which 6 can be 120 and/or 140.

Fantastic! Thanks again!

[TatamiMatt]

12 hours ago, doug_locke said:

 

Elaborating on your point. Most modern cases are able to mount 3x120mm top fans (for AIO). Putting an exhaust fan in the rear and rear most top position helps temps, 2nd and 3rd top exhaust fan with these big air coolers hurt temps. They'd suck air up before it reaches the CPU fan. 

 

 I did try many many case fan setups. 1 rear exhaust fan one top rear exhaust fan (6 intakes). This is where I landed

 

Top intake fan is  set to  20% and spaced away from the top exhaust but still close enough so that airflow is blocked off from the bottom intake by the gpu. It's only there to keep cool air from leaking before reaching the cpu, running it at max speed is worse than not running it at all. 

I also closed off the gap between the two top fans so they wont just cycle, much of, the same air from one another. 

I mounted the top intake at a slight angle, just to limit turbulence. 

 

 

Lastly using the exact same model for the 3rd cpu fan and back exhaust is pretty quiet, if one moves a lot more air, it can get loud unless the back fan is slightly off center.

 

 

 

How did you space the fan? I have a very similar setup and this is intriguing

[doug_locke]

 

21 hours ago, TatamiMatt said:

How did you space the fan? I have a very similar setup and this is intriguing

(this is about to get the opposite of intriguing, and faster to make than to read. Sorry!) 

The spacers are made from dense ¾" PVA foam, drilled and cut to size (anything slightly compressible works). Fan+case+splacer all connected with thin nuts/bolts not fan screws/ I used twist tie wire first, worked just as well.    

 

The for the spacer to be functional, adjustable, and not look ugly I needed to make o-rings, above and bellow the both sides of each of the case's mount holes.    

The 2 higher mount point use standard fan screws as long as those are on tight the spacer side just need to be tight enough to prevent vibration. But putting case screw in at an angle which could jam or dent the your, 'o-rings' on both sides of metal prevents that. 

 

HW store o-rings that fit snugly are too thin and screws kept squeezing through hole tightening, bigger ones were too hard to line up. My fix DIY o-rings worked better. 7 layers of electrical tape, cut to size & a hole poked in the center 
        Making the o-rings took trial and error:
       -anti-vibration pads (from a spare fan), neoprene, insulation foam, packaging silicone mats, felt, was all too mushy.
       -thermo-plastic (like from phone covers), Tupperware lids, wer too rigid
       - composite shims works but not adjustable
Electrical tape compress wonderfully, but using a nail for the hole is necessary it makes the area hugging the scew stronger (drilling, manicure scissors, a screw driver head, exacto knives all tear the tape and make it slide off the scew). 

I'm sure you can but something that does the same thing but junk around the house was readily available, my case is black so color matching was simple. 
 

[doug_locke]

On 3/20/2024 at 3:44 AM, TatamiMatt said:

How did you space the fan? I have a very similar setup and this is intriguing

I'm take back my advice. Top intake hurts hurts GPU temps as much as it helps CPU temps

On 3/18/2024 at 10:56 AM, RevGAM said:

I've got Gelid GC Extreme, so I hope you'll share how it compares to TF-7. ..

 

On 3/16/2024 at 6:48 PM, freeagent said:

You should try out the TF7, you might be surprised...

 

On 3/18/2024 at 5:47 PM, doug_locke said:

...

3 fans + gelid... 89.6°

3fans + TF7... 85.0°

...

It's been 6 days since applying TF7, 

temps are going up a tiny bit more each day. It's still better than gelid, but now just  barley 

 

Running Cinebench max temps now range between 87.4 to 89.°

 

Some pastes perform better after a few hours This is the first that got worse.

[freeagent]

Temps should go down not up, something is amiss. I barely break 60 during an R23 run. But I am not using TF7..

[doug_locke]

4 hours ago, freeagent said:

Temps should go down not up, something is amiss. I barely break 60 during an R23 run. But I am not using TF7..

They should go down and not change but I'm not super concerned TF7 shouldn't have been 5°C less than Gelid. 

 

With PBO limits off/auto & Without CO -30 the CPU goes to 95° on Cinebench/most stress tests. which is normal for air cooling and all but the the largest AIO's.

 

It was 85.0 max day 1, 85-87, 86-88, 87-89, 88-89 hasn't gone past 89 after.

 

R23 consistently between 20.3 and 20.7k.

 

What R23 score are you getting ?

 

[freeagent]

19 minutes ago, doug_locke said:

What R23 score are you getting ?

It depends which CPU I have plugged in. I get about 15,2xx with my 58X3D, and about 23,6xx with my 5900X. I flip flop regularly..

[RevGAM]

5 hours ago, doug_locke said:

I'm take back my advice. Top intake hurts hurts GPU temps as much as it helps CPU temps

 

 

It's been 6 days since applying TF7, 

temps are going up a tiny bit more each day. It's still better than gelid, but now just  barley 

 

Running Cinebench max temps now range between 87.4 to 89.°

 

Some pastes perform better after a few hours This is the first that got worse.

That is fascinating! Thanks!

[doug_locke]

3 hours ago, freeagent said:

It depends which CPU I have plugged in. I get about 15,2xx with my 58X3D, and about 23,6xx with my 5900X. I flip flop regularly..

Without a power limit Zen 4 boosts to 95° (which is why I'm using Cinebench to test cooler and not temps alone). Youre Zen3 having much lower temps make sense. 

(Although 60° is still very very low)

[freeagent]

46 minutes ago, doug_locke said:

Without a power limit Zen 4 boosts to 95° (which is why I'm using Cinebench to test cooler and not temps alone). Youre Zen3 having much lower temps make sense. 

(Although 60° is still very very low)

My 5900X pushes hard up till 90 no problem. Not that I allow it often.. if talking about my 5900X that should be about 73c in R23 at about 215w.

[TatamiMatt]

22 hours ago, doug_locke said:

I'm take back my advice. Top intake hurts hurts GPU temps as much as it helps CPU temps

My GPU has a very large heatsink and very small passthrough, still be an issue do you think? GPU was only hitting like 60C, 85C hotspot with a 2.94GHz OC on Folding@home (7800xt nitro+)

[doug_locke]

9 hours ago, TatamiMatt said:

My GPU has a very large heatsink and very small passthrough, still be an issue do you think? GPU was only hitting like 60C, 85C hotspot with a 2.94GHz OC on Folding@home (7800xt nitro+)

I can only speak of my 6750xt. My CPU won't thermally throttle but Benchmarks don't improve past 50mhz OC. My throttles often at 2.9ghz (top fun push/pull) and my superposition score is lower than 2.85ghz. That extra couple of degrees from 1 top exhaust exhaust fan gives a slightly better score at 2.9ghz (superposition 1440p score 4.5k vs 4.38k)

 

 

If you don't already own 2 or 3 top fans don't get them. It does very little but adds a lot more noise to the room. I have the extra fan for the RGB but leaving it at 0rpm is the best option.

 

If I use  only 1 top exhaust as a baseline temp(+/- 0° GPU $ GPU undervolted w/no OC )

it's roughly 

2 top exhaust=CPU +3°, GPU -2°

3 top exhaust= CPU +4°, GPU -3°

1 top exhaust, 1 top intake= CPU -2°, GPU +2°

1 top exhaust, 2 top intake=CPU -2°, GPU +3°

 

^^this is with top fan #2&3 set to 60% rpm. At 100% temps go crazy.

 

I'd guess if your GPU was small (<220mm) push/pull top fans are fine either way.

 

Big 3 fan 320mm GPU's (like yours and mine) sticks out far past the the CPU heatsink and the GPU fans are going to cool one side of the GPU... The backside doesn't get as hot but does generate some heat.

 

3x120mm bottom fans push radiating heat upwards a top intake feeds cold air to the CPU but also pushes the heat going upwards back down keeping the GPU warmer. Even with baffles and diagonal fans I cant see a way extra top fans to help one thing without hurting the other.

 

...these aren't real world conditions I doubt many games will push a CPU to 100% but many modern games (ie)  Cyberpunk or even a 5y/o game like Borderlands 3  will push my 6750XT or your 7800xt to 100% and still never get a close to 200fps at 1440p (w a decent GPU overclock cyberpunk gets 6750= 60fps 7800=90fps).

 

TLDR: there is no best config even with identical case/CPU/GPU, there will always be some tradeoff even if noise wasn't a factor

 

 

 

 

 

 

 

[RevGAM]

18 minutes ago, doug_locke said:

I can only speak of my 6750xt. My CPU won't thermally throttle but Benchmarks don't improve past 50mhz OC. My throttles often at 2.9ghz (top fun push/pull) and my superposition score is lower than 2.85ghz. That extra couple of degrees from 1 top exhaust exhaust fan gives a slightly better score at 2.9ghz (superposition 1440p score 4.5k vs 4.38k)

 

 

If you don't already own 2 or 3 top fans don't get them. It does very little but adds a lot more noise to the room. I have the extra fan for the RGB but leaving it at 0rpm is the best option.

 

If I use  only 1 top exhaust as a baseline temp(+/- 0° GPU $ GPU undervolted w/no OC )

it's roughly 

2 top exhaust=CPU +3°, GPU -2°

3 top exhaust= CPU +4°, GPU -3°

1 top exhaust, 1 top intake= CPU -2°, GPU +2°

1 top exhaust, 2 top intake=CPU -2°, GPU +3°

 

^^this is with top fan #2&3 set to 60% rpm. At 100% temps go crazy.

 

I'd guess if your GPU was small (<220mm) push/pull top fans are fine either way.

 

Big 3 fan 320mm GPU's (like yours and mine) sticks out far past the the CPU heatsink and the GPU fans are going to cool one side of the GPU... The backside doesn't get as hot but does generate some heat.

 

3x120mm bottom fans push radiating heat upwards a top intake feeds cold air to the CPU but also pushes the heat going upwards back down keeping the GPU warmer. Even with baffles and diagonal fans I cant see a way extra top fans to help one thing without hurting the other.

 

...these aren't real world conditions I doubt many games will push a CPU to 100% but many modern games (ie)  Cyberpunk or even a 5y/o game like Borderlands 3  will push my 6750XT or your 7800xt to 100% and still never get a close to 200fps at 1440p (w a decent GPU overclock cyberpunk gets 6750= 60fps 7800=90fps).

 

TLDR: there is no best config even with identical case/CPU/GPU, there will always be some tradeoff even if noise wasn't a factor

A few months ago, I saw (maybe here?) how someone installed ducting from fan to GC, from fan to CPU, and achieved better temps for both, so I think you should look at that instead of just baffles. Another person did something similar but used different material, and had cold air from their AC directed to the intake. Naturally, that worked well.