Thinking I may need to rearrange fan setup

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
I've got a Fractal Design Node 804 case that is a 2 chamber setup (MB side and PSU side) that is configured for water cooling. With its setup, I am wondering if I need to change the layout of my fans and/or buy some higher output fans. Below is the current setup:

MB Side:
  • AMD Ryzen 9 3900XT OC'd (4375, 4375, 4275, 4250 at 1.256mV) watercooled with Optimus Foundation AMD water block
  • RTX 3090 FE using stock fan profile and stock clocks
  • 1 Front intake fan (1x BeQuiet! Silent Wings 3 120mm, 1500RPM max with custom fan curve)
  • 1 Rear exhaust fan (1x BeQuiet! Silent Wings 3 120mm,1500RPM max with custom fan curve)
  • 2 Top exhaust fans (2x BeQuiet! Silent Wings 3 High Speed 120mm PWM, 2200RPM max with custom fan curve)
PSU Side:
  • EKWB Quantum Kinetic TBE 200 D5 Water Pump
  • Corsair XR7 240mm x 54mm Radiator mounted to top
  • 1 Front intake fan (1x BeQuiet! Silent Wings 3 120mm, 1500RPM max with custom fan curve)
  • 1 Rear exhaust fan (1x BeQuiet! Silent Wings 3 120mm,1500RPM max with custom fan curve)
  • 2 Top exhaust fans mounted on radiator (2x BeQuiet! Silent Wings 3 High Speed 120mm PWM, 2200RPM max with custom fan curve)
With essentially a single intake fan but 3 exhaust fans on each side, I am thinking I may be creating too much negative pressure, causing the exhaust fans to work harder and not perform as efficiently. Take my coolant temps, for example. With the case closed up and at idle (ambient temp around 25c), my coolant temps sit around 32-33c with my custom fan profile (quieter at lower temps) and around 31-32c if radiator fans are set to max speed. When I remove the side panel on the PSU side, my coolant temps at idle drop to around 30.5-31c with my custom fan profile and down to 29c when fans are set to max speed. The difference in CPU idle temps based on Ryzen Master and Corsair Link shows a difference of 1-2c lower temps with the side panel removed (39-40c vs 38-39c). Under full load such at CineBench or Furmark CPU Burner, peak CPU and coolant temps are a few degrees lower too between the two scenarios.

As far the the MB side, I haven't removed that side yet to see if a difference is seen. So far, the GPU stays relatively cool. I never see it go above 69c. Now, whether I can get it to go lower or get it to clock higher at the same temp remains to be seen.

If it wasn't for concern of dust or curious kittens, I would just leave the side panel off. So I am wondering if I should try and replace the side panel with maybe a filter screen of some sort, change my fan layouts, replace some fans, or a combination of the three.

Since I am exhausting out the top with fans that turn at a higher speed than my front and rear fans, should I reverse my rear exhaust fans and make them intakes? If I do, I would want to put a filter screen on them to minimize dust and hair getting in from the back. Do you think my existing fans are fine for the front and rear intake/exhaust or should I swap them out for higher speed fans?

For the GPU, the bottom of the case is vented and filtered. It can handle up to 2 80mm fans on the bottom force feed cooler air to the GPU and may also help create some positive pressure in the case. I'm thinking of maybe using Noctua NF-A8 PWM Chromax for this if I do. Any thoughts on this?
 

mnewxcv

Supreme [H]ardness
Joined
Mar 4, 2007
Messages
7,675
You need more intake. The 3090 exhausts as well, so you are looking at 3-4x as much exhaust as you are intake. Def not ideal for airflow.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
You need more intake. The 3090 exhausts as well, so you are looking at 3-4x as much exhaust as you are intake. Def not ideal for airflow.
Yeah, I’m figuring that. Initially, I was to have two intake fans in the front for each chamber (so 4 fans total), but the length of the 3090 prevented me from adding a fan in the MB chamber. As for the PSU chamber, the stock power supply cables are so long, to cleanly organize them so as to not impede airflow in other places, I removed the bottom intake fan to give more room for the cables.

While the top exhaust fans are a higher speed than the other fans, all 4 were initially going to be installed on radiators, which would restrict airflow and most of the time will never be run at top speed. Now I have two of them not on a radiator due to some slightly incorrect measurements in my original design, bringing down from a planned two radiators to one radiator. So these two high speed fans in the MB chamber do have the capability to exhaust much more air than planned due to no radiator restricting them.

TL;DR
My original build design had a pretty equal intake to exhaust airflow. But due to going with a 3090 vs a 3080 (3090 became available first for me) and some slight speed bumps in the build process, my original airflow plans got changed and now I am trying to determine the best path to fix this.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
I went ahead and ordered 2 Noctua NF-A8 PWM Chromax fans last night to use as intakes under the 3090 and some foam fan filters for the rear fans so that I can reverse them.

Also, I remembered that I removed the foam filter off the top panel when building the rig since I felt it wasn’t needed using the tops as exhaust. Turns out that it is almost the exact same size as the side panel. So I ordered some magnetic tape and plan to sandwich the foam filter between the magnetic tape and stick it to the PSU chamber side. That way, if reversing the exhaust fan still doesn’t allow enough flow through the radiator, this will certainly help seeing that I got a decent drop in temp by removing the panel all the while reducing the amount of dust and hair from entering the case. At least it should make do until I decide to replace the front and rear fans with higher flow rates. That side of the case faces a wall, so it cannot be seen by anyone.
 

mnewxcv

Supreme [H]ardness
Joined
Mar 4, 2007
Messages
7,675
adding the 2 80s will help a lot. I would set them to a fixed RPM personally for noise reasons. I would try not reversing the rear fans yet and see how it goes.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
adding the 2 80s will help a lot. I would set them to a fixed RPM personally for noise reasons. I would try not reversing the rear fans yet and see how it goes.
Since the 2 80s are primarily for just feeding the 3090 and offsetting the the airflow displacement when the 3090 is in use and its fans are running, I may configure them to not run when the 3090 fans are not spinning (I believe under 40c) and then have them crank up to a fixed RPM when they do spin up. That way, the case will be quieter when doing day to day activities outside of gaming where the added airflow (and noise) isn’t needed.

But...if the fans are quiet enough when running, I may just leave them on. We will see.
 

Nobu

Supreme [H]ardness
Joined
Jun 7, 2007
Messages
4,952
If they're like my slim 92 noctua fans, you may end up finding a certain rpm is just quieter than the rest, and still performs well enough.

There are spots in the curve where my fans rattle, or just have too much wind noise, but thankfully they don't have to spin up too often in the computer they're in. I set mine above where they rattle, and set a slow rising curve so they don't get loud unless I really need them.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,290
You have a case setup blowing air everywhere instead of flowing air.
Radiator side of case will probably work very well with no intakes and just the radiator fans as exhaust.​
Motherboard side will probably perform very well with either just top intakes or just front intakes.​
be quiet! Silent Wings 3 are very good fans. I don't think you need more fans, but just need to setup some of what you already have so fans are flowing air through case instead of sucking / blowing it everywhere.

I've found good hi-pressure rated front intakes (3x in your case) with all back PCIe slot covers removed to increase rear vent area around GPU, block all opening in front half of case not covered by intake fans so air they are flowing into case has to move on through case to go out, (not leak around in front of fans and go in circles) and no exhaust fans are needed. Using both intake and exhaust fans is like using push/pull on radiator or cooler. It might improve temps by a degree or two, but only at full speed because more fans make more noise. Slow them down to same noise level as case with only intake fans and temps are the same, maybe a little warmer. Only time intake and exhaust improve cooling is at full speed and then only by a couple degrees. If you are like me (and most others) our fans rarely if ever run full speed. ;)

Below link is basic guide to how airflow works and how to optimize case airflow:
https://hardforum.com/threads/basic-guide-to-improving-case-airflow.1987938/
 

Nobu

Supreme [H]ardness
Joined
Jun 7, 2007
Messages
4,952
Having exhaust fans helps when all your exhaust ports have restricted flow, because air that might have gone through will instead turn and create a current of air inside the case which pulls air away from the exit.

In my antec isk 300-150 for example, even though it seems like air should flow easily from one end to the other and straight out, with only intake fans (3 on the same side) the cooling performance is awful. The air hits the cpu cooler, and is circulated by the cpu fan. Hot spots are created, and the CPU just keeps heating up.

In a larger case with fewer obstructions you may fare better with intake only, but it's not a perfect solution for every case. You'll have to see what works for you.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,290
If your case is actually flowing air toward a vent then the air will not be turning. It's simple displacement in action. For it to turn it has to push the air that is there out of it's way, and if case is actually a little higher pressure than room the air hitting vent grill will flow on through vent to lower pressure area of room.

The few times I've used exhaust fans over the years has been to create a stronger flow path, but the problem with that is PCIe slot covers open / vented or other venting below GPU the GPU fan nearest back of case can (and does) pull air in and draw dust in with it.

Good reason for no exhaust fan is there is no hassle trying to balance intake to be pushing air a little harder into case than exhaust is pulling air out. All intake fans with no exhaust means all air entering has to flow out, and with even just a little more exhaust vent area than intake vent area there is little restrictions from exhaust grills so almost no difference in amount of air flowing through case.

Years ago Silverstonetek and PugetSystems tested vents noise and airflow with the results being rather interesting. I've done several builds with not back grills, or replaced punched metal grills with wire grills:
https://www.silverstonetek.com/techtalk_cont.php?tid=wh_chessis&area=usa
https://www.pugetsystems.com/labs/articles/Effects-of-Grill-Patterns-on-Fan-Performance-Noise-107/
 

Nobu

Supreme [H]ardness
Joined
Jun 7, 2007
Messages
4,952
If your case is actually flowing air toward a vent then the air will not be turning. It's simple displacement in action. For it to turn it has to push the air that is there out of it's way, and if case is actually a little higher pressure than room the air hitting vent grill will flow on through vent to lower pressure area of room.

The few times I've used exhaust fans over the years has been to create a stronger flow path, but the problem with that is PCIe slot covers open / vented or other venting below GPU the GPU fan nearest back of case can (and does) pull air in and draw dust in with it.

Good reason for no exhaust fan is there is no hassle trying to balance intake to be pushing air a little harder into case than exhaust is pulling air out. All intake fans with no exhaust means all air entering has to flow out, and with even just a little more exhaust vent area than intake vent area there is little restrictions from exhaust grills so almost no difference in amount of air flowing through case.

Years ago Silverstonetek and PugetSystems tested vents noise and airflow with the results being rather interesting. I've done several builds with not back grills, or replaced punched metal grills with wire grills:
https://www.silverstonetek.com/techtalk_cont.php?tid=wh_chessis&area=usa
https://www.pugetsystems.com/labs/articles/Effects-of-Grill-Patterns-on-Fan-Performance-Noise-107/
The problem is there are more places for air to escape than just where you want it to. You might have 1mmH2O at the inlet where the fan is, but only 0.8 at the outlet. Some air will go out, but not as much as what came in through the fan (yes, everything that comes in leaves, but not all through the exhaust vent). This difference between intake and exhaust is what causes turbulence and hotspots in a case.

Removing the grill may make up for the difference, but it will depend on your case and it's layout whether that will be enough to keep things cool.

Fwiw, the reason it doesn't happen with all exhaust fans, despite it being basically the same scenario, is because air is coming in from all the gaps instead of going out. This fresh air disrupts eddies that might have formed in corners, and cools hotspots, and the exhaust fans remove hot air from the case actively instead of pushing it around and hoping the hot air finds an exit.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,290
The problem is there are more places for air to escape than just where you want it to. You might have 1mmH2O at the inlet where the fan is, but only 0.8 at the outlet. Some air will go out, but not as much as what came in through the fan (yes, everything that comes in leaves, but not all through the exhaust vent). This difference between intake and exhaust is what causes turbulence and hotspots in a case.

Removing the grill may make up for the difference, but it will depend on your case and it's layout whether that will be enough to keep things cool.

Fwiw, the reason it doesn't happen with all exhaust fans, despite it being basically the same scenario, is because air is coming in from all the gaps instead of going out. This fresh air disrupts eddies that might have formed in corners, and cools hotspots, and the exhaust fans remove hot air from the case actively instead of pushing it around and hoping the hot air finds an exit.
Not as many as you might think .. assuming we block openings in front fan mounting panel so air those intakes push into case isn't circling in front and going round and round. Sometimes I block top and bottom venting in front half of tower case.

If fan has a 1mm H2O rating it will never be pushing 1mm of differential into case. The way they measure static pressure is seal a fan's exhaust to an air tight container and measure how much pressure it can push into that sealed box at full speed .. not really a functional measurement, but only one we are given.

True case design is key, but most tower designs (and others) work well with only intakes pushing air into them.

Actually fresh air leaking can disrupt smooth airflow from intake to exhaust vents just as much or more. We (at least I) don't want more turbulence in my case. The more turbulence there is inside of case the more heated air mixes with cool air we want to get to components. The smoother air flows through our cases the better it flows over components cooling them and the better it can push heated air on back and out.

Few cases have components in corners. Good example of only intake fans is our power supplies, and they have components closer to sides and corners than anywhere in our cases. ;)
 

Nobu

Supreme [H]ardness
Joined
Jun 7, 2007
Messages
4,952
Not as many as you might think .. assuming we block openings in front fan mounting panel so air those intakes push into case isn't circling in front and going round and round. Sometimes I block top and bottom venting in front half of tower case.

If fan has a 1mm H2O rating it will never be pushing 1mm of differential into case. The way they measure static pressure is seal a fan's exhaust to an air tight container and measure how much pressure it can push into that sealed box at full speed .. not really a functional measurement, but only one we are given.

True case design is key, but most tower designs (and others) work well with only intakes pushing air into them.

Actually fresh air leaking can disrupt smooth airflow from intake to exhaust vents just as much or more. We (at least I) don't want more turbulence in my case. The more turbulence there is inside of case the more heated air mixes with cool air we want to get to components. The smoother air flows through our cases the better it flows over components cooling them and the better it can push heated air on back and out.

Few cases have components in corners. Good example of only intake fans is our power supplies, and they have components closer to sides and corners than anywhere in our cases. ;)
They also have very optimized airflow designs, whereas you'll have to optimize your airflow yourself— my silverstone psu, for example, blocks airflow from the fan in front of the exhaust, to force air to go through the heatsinks and out the exhaust instead of just hitting the top and then doing whatever.

1mmH2O was just an arbitrary number, to help explain my point, not based on any real measurements or specs. ;)
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
I don't know if I would say I have air blowing everywhere instead of flowing. I would think that directional flow in the case about right but it's the balance of intake to exhaust that is off, more exhaust than intake. This is causing some fans to need to work harder to reach a particular flow rate. Here are some pics with a quick and dirty layout of expected airflow:
MB_Chamber.jpgPSU_Chamber.jpg

The first pic is the MB chamber (sorry for the window panel being in the way). The second pic is the back side, the PSU chamber. Blue arrows are representing intake and internal flow direction. Red arrows represent exhaust. Green arrows on the PSU side represents the possibility of me flipping that fan as an intake. The green rectangles are areas I am considering blocking off to minimize the chance of intake fans recycling air. I forgot to label the flow of the PSU's fan, which is pulling air in from under the case and exhausting out the back, so no internal case flow is effecting it.

I don't think flipping the top fans in the MB chamber around as intakes would be a good idea. For one, I do eventually plan on putting another radiator in there in that spot. Another reason is I would think that would risk pulling in warm air being exhausted out the top next to it from the PSU chamber. Lastly, I think it would fight against the air being exhausted from the video card, creating added turbulence in the case instead of encouraging a path of flow.

The areas of green rectangles are ventilated but have no fans, either because you can't fit a fan there or it would be a really tight fit (only on PSU chamber side). The bottom of the case is ventilated as well but not marked. Unless I decide to try and squeeze another fan in on the PSU side, it could be possible, I was thinking the other night it might be best to block that area off to minimize any recycling going on from the front intake fan. As far as the MB side, I am thinking those areas plus the bottom is where the video card is primarily feeding its fans from even though there are no fans. The video card is within millimeters of the front vents and with the size of the card in comparison to the case, I would think it isn't receiving much recycled air above it, getting its intakes mostly from the bottom, front and back ventilation areas next to it. When my 2 80mm Noctua's show up and go in the bottom of the MB chamber under the video card, I might block those other ventilation areas off to reduce any recycling.

For the PSU chamber, I am probably going to test with the rear exhaust changed to an intake and the bottom front blocked and then compare it to replacing the side panel with the foam filter sheet I mentioned earlier.

The other night I did do a rough flow rate test on the PSU chamber with the side panel on and off using a dryer sheet placed over the top exhausts and making note of how high it blows up. With the panel on and fans set to their custom profile, I made note of the height and behavior of the dryer sheet movement. It wasn't consistent in one spot and would fluctuate. Turning just the front intake to max (custom is around 1,100 RPMs and max around 1,500 RPMS) steadied the waving/fluctuation of the sheet but the height didn't change much. Changing the fan profile of the rear exhaust fan from custom (which sits around 700 RPMs) to max (around 1500 RPMs) the dryer sheet dropped in height by a noticeable amount. It took raising the RPMs of the radiator fans a good few hundred RPMs to get it back to the height it started at. Removing the side panel the dryer sheet height was quite a bit higher with no change to the fan profiles. Front intake fan at custom or max didn't make any difference. At the height the sheet goes, to meet an equivalent height with the panel installed I would have to raise fan RPMs by quite a bit. And with the case open, the custom fan profile wants to run the fans at an even lower speed and still manager to keep the radiator cooler than with the case closed and fans running at a higher RPM. So, clearly, the radiator fans are fighting for air.
 
  • Like
Reactions: Nobu
like this

Nobu

Supreme [H]ardness
Joined
Jun 7, 2007
Messages
4,952
Can air move freely between the two chambers, or are they more isolated (only paths are for cables, etc)?

If they're isolated, I'd set All the fans on the sides of the case to intake. The GPU is in it's own little world down there, so just pretend everything above it is a separate chamber.

Or, reverse the fans on the radiators and set all the side fans to exhaust. Air temp will increase slightly (1-2°), but the exhaust fans will immediately be exhausting that air so it shouldn't build up.

In either case, I think blocking that vent would be a good idea if you have an active fan next to it. Worst case it doesn't help and you just unblock it.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
Can air move freely between the two chambers, or are they more isolated (only paths are for cables, etc)?

If they're isolated, I'd set All the fans on the sides of the case to intake. The GPU is in it's own little world down there, so just pretend everything above it is a separate chamber.

Or, reverse the fans on the radiators and set all the side fans to exhaust. Air temp will increase slightly (1-2°), but the exhaust fans will immediately be exhausting that air so it shouldn't build up.

In either case, I think blocking that vent would be a good idea if you have an active fan next to it. Worst case it doesn't help and you just unblock it.
The openings between the two chambers that do exist are primarily for cabling and tubing and not so much for air movement.
Openings.jpg

The largest opening towards the front is maybe 2"x 7" at best and a good amount of it is occupied by cables and tubing. Plus the front intake fans sit 25mm further in from that, so the air they push gets pushed past the openings. The video card covers around an inch or so of the bottom of the largest opening. The other two small openings are mostly occupied by cables. The largest opening behind where the motherboard sits is pretty much blocked off by the motherboard and its thick backplate (save for the little sliver marked by the thin, vertical rectangle and is occupied by cables), leaving little room for air flow to transfer from one chamber to the other. I can't even slip a fan connector behind it.
 

Nobu

Supreme [H]ardness
Joined
Jun 7, 2007
Messages
4,952
The openings between the two chambers that do exist are primarily for cabling and tubing and not so much for air movement.
View attachment 293107

The largest opening towards the front is maybe 2"x 7" at best and a good amount of it is occupied by cables and tubing. Plus the front intake fans sit 25mm further in from that, so the air they push gets pushed past the openings. The video card covers around an inch or so of the bottom of the largest opening. The other two small openings are mostly occupied by cables. The largest opening behind where the motherboard sits is pretty much blocked off by the motherboard and its thick backplate (save for the little sliver marked by the thin, vertical rectangle and is occupied by cables), leaving little room for air flow to transfer from one chamber to the other. I can't even slip a fan connector behind it.
Yeah, so I would try with all the side fans as intake. If you could put a fan on that vent (with a grill to keep cables out) that'd probably help, but blocking it off would be the next best imho. That should help, and would be the easiest changes to test.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,290
I don't know if I would say I have air blowing everywhere instead of flowing. I would think that directional flow in the case about right but it's the balance of intake to exhaust that is off, more exhaust than intake. This is causing some fans to need to work harder to reach a particular flow rate. Here are some pics with a quick and dirty layout of expected airflow:
View attachment 293073View attachment 293074

The first pic is the MB chamber (sorry for the window panel being in the way). The second pic is the back side, the PSU chamber. Blue arrows are representing intake and internal flow direction. Red arrows represent exhaust. Green arrows on the PSU side represents the possibility of me flipping that fan as an intake. The green rectangles are areas I am considering blocking off to minimize the chance of intake fans recycling air. I forgot to label the flow of the PSU's fan, which is pulling air in from under the case and exhausting out the back, so no internal case flow is effecting it.

I don't think flipping the top fans in the MB chamber around as intakes would be a good idea. For one, I do eventually plan on putting another radiator in there in that spot. Another reason is I would think that would risk pulling in warm air being exhausted out the top next to it from the PSU chamber. Lastly, I think it would fight against the air being exhausted from the video card, creating added turbulence in the case instead of encouraging a path of flow.

The areas of green rectangles are ventilated but have no fans, either because you can't fit a fan there or it would be a really tight fit (only on PSU chamber side). The bottom of the case is ventilated as well but not marked. Unless I decide to try and squeeze another fan in on the PSU side, it could be possible, I was thinking the other night it might be best to block that area off to minimize any recycling going on from the front intake fan. As far as the MB side, I am thinking those areas plus the bottom is where the video card is primarily feeding its fans from even though there are no fans. The video card is within millimeters of the front vents and with the size of the card in comparison to the case, I would think it isn't receiving much recycled air above it, getting its intakes mostly from the bottom, front and back ventilation areas next to it. When my 2 80mm Noctua's show up and go in the bottom of the MB chamber under the video card, I might block those other ventilation areas off to reduce any recycling.

For the PSU chamber, I am probably going to test with the rear exhaust changed to an intake and the bottom front blocked and then compare it to replacing the side panel with the foam filter sheet I mentioned earlier.

The other night I did do a rough flow rate test on the PSU chamber with the side panel on and off using a dryer sheet placed over the top exhausts and making note of how high it blows up. With the panel on and fans set to their custom profile, I made note of the height and behavior of the dryer sheet movement. It wasn't consistent in one spot and would fluctuate. Turning just the front intake to max (custom is around 1,100 RPMs and max around 1,500 RPMS) steadied the waving/fluctuation of the sheet but the height didn't change much. Changing the fan profile of the rear exhaust fan from custom (which sits around 700 RPMs) to max (around 1500 RPMs) the dryer sheet dropped in height by a noticeable amount. It took raising the RPMs of the radiator fans a good few hundred RPMs to get it back to the height it started at. Removing the side panel the dryer sheet height was quite a bit higher with no change to the fan profiles. Front intake fan at custom or max didn't make any difference. At the height the sheet goes, to meet an equivalent height with the panel installed I would have to raise fan RPMs by quite a bit. And with the case open, the custom fan profile wants to run the fans at an even lower speed and still manager to keep the radiator cooler than with the case closed and fans running at a higher RPM. So, clearly, the radiator fans are fighting for air.
Have your tried just unplugging all fans in motherboard compartment except top exhaust fans? Doing this shouldn't be too hard and might lower noise without effecting cooling much if even at all.

You need to have matching intake and exhaust potential. This means similar vent area and design so similar restriction. If in/out vent is more restrictive than it need to be bigger area to flow same as opposite vent. Which ever is most limited will determine maximum case flow.

So if top vent as exhaust is 120x240mm you need at least a 120x240mm of intake vent area with similar grill and whatever else is in it. Obviously same applies to back compartment venting. Hope that all makes sense.

You could try same top exhaust fans only in back compartment / PSU chamber. Easy to do could be solution with very little effort. ;)

Post #8 has link to simple guide to airflow and how to optimize it. If you haven't read it I suggest you do. Most find it helpful.




They also have very optimized airflow designs, whereas you'll have to optimize your airflow yourself— my silverstone psu, for example, blocks airflow from the fan in front of the exhaust, to force air to go through the heatsinks and out the exhaust instead of just hitting the top and then doing whatever.

1mmH2O was just an arbitrary number, to help explain my point, not based on any real measurements or specs. ;)
1mmH2O isn't what I was addressing. I was trying to clarify what a "1mmH2O" static pressure specification / measurement is, and how it's very misleading if you don't understand what it actually is.

Some have optimized design, most do not. Silverstone applies airflow much better than most do.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
Have your tried just unplugging all fans in motherboard compartment except top exhaust fans? Doing this shouldn't be too hard and might lower noise without effecting cooling much if even at all.

You need to have matching intake and exhaust potential. This means similar vent area and design so similar restriction. If in/out vent is more restrictive than it need to be bigger area to flow same as opposite vent. Which ever is most limited will determine maximum case flow.

So if top vent as exhaust is 120x240mm you need at least a 120x240mm of intake vent area with similar grill and whatever else is in it. Obviously same applies to back compartment venting. Hope that all makes sense.

You could try same top exhaust fans only in back compartment / PSU chamber. Easy to do could be solution with very little effort. ;)

Post #8 has link to simple guide to airflow and how to optimize it. If you haven't read it I suggest you do. Most find it helpful.
The front and rear fans in the MB chamber are hardly noticeable when it comes to noise as is, so turning them off to quiet the case is going to make little difference there. Plus, while a majority of the air is being exhausted out the top, I am trying to encourage at least some crossflow from front to back to help keep my 2 M.2's cool that sit on a daughterboard right above the video card (can be seen with the glowing "Republic of Gamers" logo).

And while it may make little difference, from a mental comfort point of view, I like having the front intake fan blowing directly onto my tubes since they sit right above the exhaust fan of the GPU. I need to plug in a few more temp sensors (which I can do) to really see what kind of heat that fan is exhausting, but if the front intake fan helps keep cool air flowing over the tubes to keep heating them up via exhaust at a minimum, then I am happy with that.

I am not really that concerned with the MB chamber when it comes to heat. The primary heat source in there is mainly the video card with the M.2's, RAM, the MB chipset, and VRMs following behind. The chipset and VRMs do have their own fans built into the MB and they also exhaust air out the back of the case. I almost forgot about them since I don't even hear them. The video card is doing a pretty decent job not exceeding 69c in its current state and is pretty quiet. Even when the video card fans are not spinning when doing desktop work, the video card is usually sitting between 32-34c. I imagine adding the 2 80mm Noctua's below it feeding it cooler air and blocking off the other parts mentioned should help even more and I may only run those when the video card's fans are running. Now when it comes time to add a radiator in there, then it may be more of a concern when it comes to heat levels, especially since some of that hot exhaust coming off the video card will then be sucked through the radiator. But at that point, the video card may be getting water cooled as well, so then that concern will be moot.

My main concern at the moment is the PSU chamber and ensuring that the radiator is fed cool air and the radiator fans are not having to work harder to achieve that. With the two options I mentioned earlier, I should be able to achieve that. If I had room for a push/pull setup on the radiator, I would do that and run the fans at an even lower speed, but I don't have enough space to do that with this 54m thick radiator. Maybe if it was a 30mm thick radiator, I would.

Yes, I did read the guide in post #8 as well as reviewed the articles about grill types. I have no intentions yet to cut into my case, so the stamped honeycomb grill types will have to stay for now. Filters are restrictive and I could remove those as a quick way to improve airflow into the case. But then I would be letting dust and hair into my computer. And while preventing all dust and hair intrusion is nearly impossible, I would at least like to minimize it since this computer resides in a room with kittens, a temporary covered litter box, and a litter robot. The computer does sit on the corner of my desk instead of on the floor, so that helps, and I do have a BlueAir Blue Pure 211+ air purifier (20" x 13" x 13" in size) sitting within a foot of my case that runs 24/7 as well as a smaller Holmes air purifier that sits near the temporary covered litter box that only gets turned on during cleanings to minimize any spread of litter dust.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,290
Nice detailed reply, thanks!

Key to good cooling is not more fans blowing air around mixing heated with cool raising air temp we are trying to cool components with. Good cooling is airflow from intake to exhaust with as little disruption / turbulence as possible so air reaching components is as close as possible to room ambient. Temp sensors make it easy to monitor air temp near components, then compare air temp to component temp.

What are M.2 temps? I'm guessing not a problem.
What are CPU temps idle and heavy / max load?
32-34c nice and cool. What are GPU temps at max load?

I really think extra fans in PSU chamber are not helping much if any. Don't know for sure unless you monitor air temp entering radiator with and without other fans in there running.

While removing grills does help it isn't essential to good airflow and cooling. No reason to remove grills.

Kittens are so much fun! Hard to really control them, but sounds like you have things setup as good as can be.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
Room temp is around 25c.

Coolant temp, taken from an Alphacool coolant temp sensor located in my pump reservoir, with case cover on is around 32-33c at idle and will get up to around 40c under extended heavy load using custom fan profile. Temps drop by around 1c in both idle and max load at max fan speed. An example of a extended heavy load is running Furmark CPU Burner for an hour. Typical daily use never reach those max temps.

Coolant temp with case cover removed is around 30c at idle and will get up to around 38c under extended heavy load using custom fan profile. Temps drop by around 1c in both idle and max load at max fan speed.

With case cover closed, CPU temp (Ryzen 9 3900XT all core overclocked 4375/4375/4275/4250 at 1.256v and RAM at 3600 at 1.35v) at idle is between 39-41c (it bounces around a lot) and will get up to around 74c under extended heavy load using custom fan profile and water pump set to 50% (2500 RPMs). Temps drop by around 1c in both idle and max load at max fan speed. Day to day use max CPU temps tend to stay below 70c.

With case cover opened, CPU temp at idle is between 38-40c (it bounces around a lot) and will get up to around 72c under extended heavy load using custom fan profile and water pump set to 50% (2500 RPMs). Temps drop by around 1c in both idle and max load at max fan speed. Day to day use max CPU temps tend to stay below 69c.

I believe 85-88c is the temp threshold where the CPU will start to down clock and 95c is when the CPU will start implementing safety measures to protect itself. So I am staying clear under that. The measurements given above I get from Corsair Link 4 software. It matches what Ryzen Master shows and when comparing to HWInfo, the measurements appear to be taken from the Tctl/Tdie which has some kind of offset in it. Measurements at the individual Tdies (and individual CCXs) can come in around a few degrees c cooler. But since it is the Tctl/Tdie that AMD seems to be the most focused on, I tend to use that.

The RTX 3090 FE sits around 32-34c at idle and tops out around 69c under extended heavy load.

There are two M.2s on the daughterboard (2x 2TB Seagate Firecuda 520 PCIE Gen4x4), one on top and one on underside both covered with an aluminum heat sink. This daughterboard rests just a few mm above the graphics card so it is bound to receive heat soak from that. I haven’t taken good measurements yet of them in heavy use, but using Samsung Magician, at idle they sit in the upper 30s c and have been seen to climb to mid to upper 50s c if the video card has been running under heavy load for a while.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,290
All sound good to me. 1c open vs closed is good. I often get lower component temps with case closed vs open .. because case airflow paths are not as good open as closed. I don't check every build for open / closed temps, just go for temps well within same range with no / low noise levels.

My understanding of 3900XT is 84c, so best to stay below that. From what I've seen throttling doesn't hurt CPU, but I prefer staying below throttle point anyway. My 3600 runs upper 30's to mid 40's 5-8% load most of the time, but I have lots of windows / web-pages open and fans are 400-600rpm. M.2 temps are fine. I think as long as they are below 60c you are golden. Specs I've seen show 70c as max safe operating temp.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
All sound good to me. 1c open vs closed is good. I often get lower component temps with case closed vs open .. because case airflow paths are not as good open as closed. I don't check every build for open / closed temps, just go for temps well within same range with no / low noise levels.

My understanding of 3900XT is 84c, so best to stay below that. From what I've seen throttling doesn't hurt CPU, but I prefer staying below throttle point anyway. My 3600 runs upper 30's to mid 40's 5-8% load most of the time, but I have lots of windows / web-pages open and fans are 400-600rpm. M.2 temps are fine. I think as long as they are below 60c you are golden. Specs I've seen show 70c as max safe operating temp.
Yeah, temps are not too bad between being open and close. Peak temps it seems to span a few degrees more vs idle where the difference is smaller. What is noticeable and you can’t really extrapolate from just highs and lows is that heat soak of the radiator takes longer with the case open. Also, these are mostly based on my custom fan profile. With that fan profile, the fans want to run at a lower speed by a few hundred RPMs with the case open and still manage to achieve slightly lower temps. If I adjusted the curve to raise the speed or set them up at a fixed speed, the temp difference becomes a little larger between case open and closed, but not by much.

I believe you are right about 84c being the where the down clocking threshold starts for the CPU at default settings and doing single core PBO boost. With my all core overclocks settings, I don’t get that single core speed boost. I am using CTR to set my overclocks and in the settings there is an option for max temp threshold where if reached it will revert to stock clocks and voltage automatically. In the first release of the software, they had it set to 85c. With the latest (beta 7) it’s been set to 88c as their default setting. Ryzen Master lists max temp threshold as 95c, but that isn’t for where down clocks begin, just where safety measures start to cut in if I understand correctly.

The M.2’s, luckily, have their own safety measures in place if it gets too hot and will start to throttle after a certain temp is reached. According to the manual, 10% throttle-back occurs at 82c and 20% throttle-back occurs at 85c, both based on SMART measurements. Thermal protection kicks in at 95c based on SMART measurements. Controller protection kicks in at 110c based from on die sensor and hardware failure occurs at 120c. Now providing there is at least some airflow and I am not slamming both M.2’s at once while the GPU is at full load, I don’t think temps are too much of a concern. I just want to stay below throttle thresholds.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,290
Sounds like you've got it pretty well sorted.
Didn't even try OC'ing my 3600. Did get fastest RAM I could get ad managed to bump it a little.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
Sounds like you've got it pretty well sorted.
Didn't even try OC'ing my 3600. Did get fastest RAM I could get ad managed to bump it a little.
I didn't want to go for an aggressive overclock, just something that was stable and consistent. Plus, while I know it is in the design, I didn't care for watching how much the CPU voltage fluctuates doing these single core boosts and increased CPU temps since it was bouncing all over the place. I vaguely recall seeing it frequently go over 1.3v and even above 1.4v for blips. But when I would see that, I would see these quick spikes in temps. Setting a fixed voltage and overclock lowered these temp spikes, calmed some things down, and allows me to get a lower idle temp. While peak single thread performance may suffer a little for things that only need a quick boost, multi-thread tasks and single thread tasks that last a while perform better.

Right now, I am still using the XMP timings (or whatever AMD calls it) for my RAM, which is rated at DDR 3600 CL16. Once I get everything else settled, I may try and tinker with the timings using the Ryzen DRAM Calculator. I've heard some good things about it and might be able to reduce my latency and timings a little more. But I don't want to do it at the expense of too much added heat and risking stability. So we will see.
 
Last edited:

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,290
I didn't want to go for an aggressive overclock, just something that was stable and consistent. Plus, while I know it is in the design, I didn't care for watching how much the CPU voltage fluctuates doing these single core boosts and increased CPU temps since it was bouncing all over the place. I vaguely recall seeing it frequently go over 1.3v and even above 1.4v for blips. But when I would see that, I would see these quick spikes in temps. Setting a fixed voltage and overclock lowered these temp spikes, calmed some things down, and allows me to get a lower idle temp. While peak single thread performance may suffer a little for things that only need a quick boost, multi-thread tasks and single thread tasks that last a while perform better.

Right now, I am still using the XMP timings (or whatever AMD calls it) for my RAM, which is rated at DDR 3600 CL16. Once I get everything else settled, I may try and tinker with the timings using the Ryzen DRAM Calculator. I've heard some good things about it and might be able to reduce my latency and timings a little more. But I don't want to do it at the expense of too much added heat and risking stability. So we will see.
RAM can take quite a bit of pushing without getting too hot. We used to be able to get a lot more performance overclocking, but now it's much harder. Now it's more the fun of it.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
RAM can take quite a bit of pushing without getting too hot. We used to be able to get a lot more performance overclocking, but now it's much harder. Now it's more the fun of it.
But with the memory controller being on the CPU, wouldn't pushing the RAM add heat to the overall CPU? Or is that only if I bump up the MHz on the RAM (since the Infinity Fabric speed is tied to this)? If I am only fiddling with tighter timings and bumping the RAM voltage by .01-.02v, will that heat up the CPU?

It's not an area I have dabbled much in.
 

doyll

[H]ard|Gawd
Joined
Feb 4, 2012
Messages
1,290
I haven't either. Have a friend who loves OC'ing things who helps (really does) most of mine.
 

VirtualMirage

Limp Gawd
Joined
Nov 29, 2011
Messages
384
Just an update:

Initially, I went ahead and reversed the rear case fan (120mm BeQuiet! SilentWings3, 1500rpm max) and placed a foam PC air filter in front of it. It helped a little but I felt more could still be done.

So I ordered a Noctua NF-A14 PWM Chromax 140mm fan to replace it. While that fan location is designed to fit up to 140mm, it was a tight squeeze behind the radiator. I had to remove all the screws mounting the radiator and while holding it in place with one hand, had to slide the 140mm in place with the other. Then found out that the silicone boots that fit on the corners of the Noctua make it just a tad too thick to fit there with the radiator, so I had to pull those off on the top portion that was bumping into the radiator to get it to fit. I guess there is exactly 25-26mm of space behind the radiator and the silicone boots were just a 1mm too much.

I installed the 140mm as an intake and took the 120mm that was there and managed to fit it into the bottom front in the chamber after rearranging some of the PSU cables. So in the PSU/Radiator chamber I have two 120mm intake fans (1500rpm max) in the front, 140mm intake fan (1500rpm max) in the rear, then two 120mm exhaust fans (2200rpm max) on the radiator. I did try to put the foam air filter over the 140mm but I didn't like the restriction that it was making, causing the fan to run at a higher RPM and make a humming noise I found to be annoying. So, for the moment, there is no filter on the rear. I am going to find something that is a little finer to put in front of it so as to not be as restrictive but help catch most of the dust and keep larger debris out (like hair). Maybe some window screen material? It running without a filter on the rear for a few days shouldn't cause too much mess, especially since I just dusted out the case and filters today and the computer sits near an air purifier.

With the fans configured the way they are and using a custom fan profile that keeps them relatively quiet most of the time, my coolant temp now sits around 2C cooler than before at idle, 30-31C vs 32-33C. CPU idling will drop down to as low as 36C but seems to sit more frequently around 38C. Running FurMark CPU Burner on 24 threads for over an hour my max CPU temps would fluctuate around the 71-73C mark and coolant temp stayed below 36C, an improvement of around 4-5C from before. And this is with the fans never reaching max speed either, so they remain quieter than they did before. My airflow coming out of the top of the case through the radiators is noticeably higher than it was before. I have a used dryer sheet pinned to the top of the case that I use to gauge airflow based on how high or low it floats. I think I might be able to get the coolant temp to drop even more if/when I add a second radiator to the loop. As it stands, coolant temp at idle is around 5-6C above ambient, assuming the coolant temp sensor is accurate. If I max out all the fans in the PSU/Radiator chamber, I can get my coolant temp to drop to a little above 29C (and CPU idle dropped to as low as 35C), about 1-2C cooler than my custom fan profile that is much, much quieter.

Also, I installed two Noctua NF-A8 PWM Chromax 80mm fans directly underneath the RTX 3090 FE earlier this month and have them on a custom fan profile too. They sit at about 1,500 RPMs on the low end since they are so quiet and will ramp up to 2,000 RPMs when the GPU starts to get under load. They will top out at 2,200 RPMs, but I don't have them reach that unless the GPU exceeds 75C since that is when they become noticeable over the GPU fans. Otherwise, at 2,000 RPMs and below you barely notice them. Now, when doing daily activities or idling, the GPU will get as low as 27C and frequently sit at or below 30C. This is a nice improvement over the 32-34C it would idle at or run at doing desktop activities. Under load, GPU temps still never exceed 69C but the core clock boost is now higher without even applying an overclock.

All in all, I think it is a step in the right direction. But when I stop to think about it, it seems pretty crazy having 11 fans in this "small" case (not including the PSU fan and the two fans on the video card). At least it is pretty quiet when doing daily activities. Not silent, but certainly quieter than the air purifier in the room. It does get louder under heavy load, but it is still tolerable and quieter than some of my older builds.
 
  • Like
Reactions: Nobu
like this
Top