If it's quieter than a 80mm rear wall fan at the same airflow.

Just because it can be used to "evacuate the heat" from the case doesn't mean that it should.

I guess there are several topics in this thread:

1. Is there any advantage to using a 120mm fanned PSU vs. 80mm?
2. Is a PSU duct for fresh air intake better than no duct from a overall noise perspective? (It should be pretty obvious from a PSU noise perspective alone that it is better. But what about overall?)
3. Should the PSU be used for case exhaust?
4. If yes to (3), should it be the only exhaust?

Oh my poor head... coffee...

To summary comments from Ralf, Powergyoza, Bluefront and several others...

For quiet, effective cooling,

• A bottom-intake fan PSU should only be used if the overall heat in the system is low (let's just put a number here, shall we? say max 150W? maybe lower?)
• But with judicious case airflow management (including ducting), a 120mm fan in the PSU can perform very well as the ONLY fan in a lower heat PC.
• A quiet 80mm fan PSU + additional quiet rear exhaust case fans may still be the quieter solution, and cannot be beat for higher power systems.

Any other broad guidelines I've missed?

powergyoza asked the right question. Can air heated by the hard drives, by the Northbridge, the ram, the video card, the CPU, etc. still be useful.....useful enough to cool the PSU? And the answer is Yes, if the temperature of that air is low enough. On my own system described above, the output temp of my modded Fortron, the only output, maxes at about 36c. And that is after cooling the PSU. I find that temp perfectly acceptable. For you non-believers...well you'll just have to try it.

This airflow thing is difficult to explain, as there are so many variables. Basically for maximum quietness, the airflow through your computer should be tweaked for maximum usefulness. To me that tweaking process differs from the norm. Take a look at this picture. This is your standard case setup....a CPU fan blowing down, a rear case fan blowing out, and a PSU sucking air......all trying to get this air from one small space. An ideal setup? Not to me.

Try this experiment....Take two fans, blowing them in different directions. Then move the rear of these fans closer and closer together. What happens? Well the output of each fan diminishes, and the noise level goes up. This is what is happening to the airflow from the fans in the picture, of course not as dramatic as the experiment. But the effects of restricted airflow are still there.

That is why I think a one fan setup could be superior. My setup may not be perfect yet, but I'm still tweaking it....and right now it works just fine.

That's a really useful analogy. I always suspected that fans starve each other of air when you place their intakes too close together. That's the situation that I have right now.... Even though my temps are fine, maybe this'll be a good opportunity to reroute the air...

Either the PSU evacuates the CPU heat and the case fan the rest or the case fan evacuates the CPU heat and the PSU the rest. Arctic Cooling could evacuate the Graphics card on both. I don't think a front duct would work in this sitiation, what other "practical" alternatives are there?

EDIT: removed everything else because it filled space and, heh, a fan drawing heat upwards from the CPU heatsink didn't work like I expected...

Me too. Unfortunately it's a limitation of the ATX design, which wasn't designed with air flow in mind. I guess if the BTX form factor takes off that will be a big step in the right direction.

Meanwhile one possible solution (that I have discussed in another topic ) is to move the CPU fan to the side, so it is going with the airflow, rather than against it. (FLD has a photo or two in this topic showing an experiment he did). It would work better with a HS designed for a fan on the side (I know there are a few).

Another possible solution is to use ducting to force the air the way you want it. I've thought about it but believe it would create too much restriction due to the various sharp angles the air would have to go through.

For what it's worth, I'm with Ralf as he mentions in another post:

I'm finding the same with my system.

Well sure....if you are satisfied with your current setup, stop right there. But I'm trying to invent a new mouse-trap...so-to-speak. The old one works just fine, but there is a better one right around the corner.

And you'll never find it if you stop where you are, and say good enough. I like to invent things, and try new directions.....this computer/case/fan/quiet thing is a field with many opportunities, many roads. Who knows what's to come, what the setup of tomorrow will be? One thing for certain.....it'll be different from now.

I'm of the same mind as Bluefront.

I've been talking about my FSP300 mod in this thread.

In my experience, so far, the FSP300, modded for 5V (really quiet) is adequate cooling for many setups. I've gone so far as to use it as the only case fan in a P4 3.0C system. It worked fantastically. Of course, all my systems I've used this with all have CPU fans, but the CPU fan and single 120mm fan on the PSU are the only fans, in several of the systems. (I've built several cheap single AMD systems with this configuration, which I didn't cover in the other thread.)

The real question I'm forced to think about is this... How hot is too hot, for a PSU?

I've measured the intake and exhaust air temperature on my modded FSP300's. The difference is usuallly about 10 degrees (30C intake temp, usually produces 40C exhaust temp). So, how hot is too hot? I've had different results in different systems, with different variables, but I'm forced to wonder this... at what point will my PSU be damaged or decrease its life?

In short I agree with Mike's proposition & summaries, and agree with Ralf's (et al) repeated advice not to put pre-heated air into the psu.

Except I get to the same conclusion from a different angle. The biggest heat source in the new (>150W) PCs is the (>80W) CPU. To have the best chance of silence you need to keep that CPU heat away from everything else. We know an 80% efficient psu can cool itself quietly. We know our hdd can be cooled quietly with minimal airflow, so the challenges are the CPU & GPU. The ideal way to cope with these (using air cooling) is to get the coolest possible ambient air to them. And because air tends to go wherever it wants, to get the biggest bang for the buck you need to exhaust the heated air ASAP else it will recycle around the CPU itself (plus GPU, maybe hdd ...). Dorothy/jafb2000 posted a while back a value for the typical recycled air on a CPU fan and it is scary. So the CPU needs to be cooled in its own duct. And maybe the GPU. At which stage the resulting consequence is the PSU can either have its own duct, or IMHO a 120mm psu fan can quietly cope with its own cooling as well as the mobo, hdd etc.

The 'practical' part of Mike's question has to cause us all to fail: most of us have close to 20C ambients and notice every winter how we can progress on the silence battle, and every spring how we have gone too far. For those with 30C or more ambients our low airflow (low noise) solutions are not practical and never can be.

how would a CPU duct work when the best heatsinks are designed for a fan to blow downwards? An elaborate two part duct may work... or maybe the DP-102 that Silvervarg is currently looking at would work?

Greetings all.

I'm a newbie to the quest for a quieter PC. and I have really enjoyed this topic (and others) of how best to cool and PC and the role of the PSU in that process (or lack there of.)

One thing I keep thinking about is how Dell has managed to create such a quiet PC in their Optiplex line. At work (I'm a network admin), I popped open on of the Dell Optiplex GX270's to see just how they accomplished creating such a quiet PC. I was fairly shocked with what I saw.

The had a 120mm (or near in size) rear case fan that had a green ducted hood that pretty much completely covered a Pentium 4 3.0Ghz chip with a heat sink only!!! The case fan in the back draws air from the lower part of the case, draws it up past the heat sink on the CPU, and out. THATS IT! The PSU is slotted in the rear and the fan on the PSU seems to handle any air that does not make it into the ducted hood of the case fan from the CPU.

This setup is basically exactly what MikeC is describing here. Why use the PSU as a primary exhaust solution? This also addresses what someone else pointed out about having a down blowing CPU fan competing for air with the rear case fan. Dell decided to not fight for the air on the CPU heatsink, but instead draw the air directly over it and then exit immediatly.

I'm in the process of bulding a new PC, and I think I'm going to try something similiar as much as possible. I'm going to pick up a PSU without a rear fan (SilenX looks interesting) and then I'm going to try and find some ducted hoods from some old Dell or Gateway PC's and see if I can't force all my air over my CPU and then directly out my case without the hot air from the CPU travelling up to the top of the case in the first place.

Interested to hear anyone else's thoughts on Dell's VERY QUIET solution.

Welcome to SPCR

Dell seems to make very quiet computers, as does ARM: SPCR review

One problem with Dell's design is that many high performance heatsinks like Thermalright's line, work much better with a fan blowing air down the heatsink. In this manner the edges of the heatsink become the coolest and a gradiate develops towards the base of the heatsink, where it is the warmest. A duct running inside and bringing air down the heatsink with an outer duct running on the outside of the heatsink and pulling air from the bottom of the heatsink and out the case might be ideal, if a minimal amount of fresh case air entered the outer duct to be exhausted out with the hot CPU's air.

Thermalright makes a little plastic duct that I used for awhile to exhaust air out the back (I had to use paper to connect it to the fan exhaust in a Sonata) but the temp.s weren't as good as when the fan was blowing down onto the heatsink. Of course Dell heatsinks are probably optimised for a different type of airflow, but I doubt they are as efficient as the thermalrights and a system like Ralf's(link at bottom) would probably be much quieter than one of Dell's. It'll be interesting to see just how quiet they get one of the new Prescotts

VGA cooler removes most VGA heat from case and here's an SPCR review of a SilentX PSU.

SilentX recommends 80mm PSUs now

There's actually a Dell computer out now that uses a fully seperated PSU at the bottom of the case but I want to say it's cooled by a 60mm fan...

If you're planning to build a computer, be sure to post a question in the general or newbie section, read what others have done, or just bug Ralf.

Trip,

Thanks for the reply. I agree with you about the advantage that a fan provides to a CPU heat sink. In fact, I think that this is pretty much a given for any heat sink, not just Thermalright's line. The one thing I can say is that the Dell Optiplex line I am refering to here has been our MOST stable PC within my company. This then begs the question, why add a CPU fan if the PC is stable and stays stable over time? I have a home PC with an AMD 1400 in which my CPU/Motherboard temps always measure much warmer than that of any review. However, my box is stable even while running games that generate high temps.

The real reason I bring up the Dell is that I believe there is something to be learned here from a company that has a lot on the line if it fails and probably put a lot of research into cooling as cheap as possible. The original question in this thread is asking if it is better to channel heat from the CPU through the PSU, or have it vented in a way so that the PSU is really only cooling itself. Dell provides a real world example of PC in which the PSU is cooling just itself, and the rear case fan is handling everything else. Not only that, but they were able to do it WITHOUT a CPU fan via some ducting, and provide a completley reliable and stable PC. So what they have done is this... They reduced the cost of the PSU as it only requires one fan to cool just itself. They reduced the cost of the CPU cooling because it does not have a fan. Then, they reduced the sound of the entire box because they only have TWO fans in the entire box, of which one is ~120mm that spins slowly and only when needed. If they can do all that and still keep the box reliable, then I don't really care if their CPU temps are higher then what many extremist desire to see.

I realize that this is only one example, but I do believe that it helps to prove Mike's original point, that you can design a quiter and cheaper PC by not venting hot CPU air through the PSU.

The CPU fan can be the loudest or most annoying fan in the box because of they are required to be high speed. Most of the annoying high frequency noise comes from my CPU fan. Although the PSU fan makes more general lower frequency white noise. So I think it can be helpful to replace it with the ducted case fan, which potentially can run at a lower speed.

The main reason I see for not dumping extra heat into the PSU or having an intake directly to the outside air is that it prevents thermally controlled fans from ramping up or helps avoiding needing a high speed or dual fan in the PSU. Otherwise, there is nothing wrong with dumping extra heat out the PSU as long as it can "take the heat."

My Enermax could run fine at a less annoying 1800 or less RPM if my case was not designed to dump CPU and case heat through the PSU. There is no room on the back panel for any kind of fin (unless you make a few 30mm's vertically stacked). It is designed for the PSU to vent the CPU directly. It's impossible to rotate the PSU to give it direct access to the air. A PSU channel from the front bays would be difficult to, requiring a 180* bend in the channel and fighting through wire bundles and around drives (there is no lower drive cage in front of the intake). So ducting or isolating the PSU would help greatly because it would keep the thermally controlled fan speed down. I've made a crude duct that keeps CPU heat out of the PSU and it does work for a while until heat builds up (used the small vent and a old 486 fan to vent the enclosed CPU area out the back top vent). However, without room for a rear case fan that would leave the CPU with no place to dump its heat. Of course, I'm looking for a new case.

So I agree that the evacuate the heat role is a generally bad one for the PSU. It has drawbacks

* Puts extra strain on PSU (although most can handle it) requiring it to dump CPU and case heat (which can be high with graphics cards if not vented through a cooler)
* Raises RPM of thermally controled fans or requires running fixed RPM fans at higher speeds

If you don't have thermally controlled fans or have a quiet fan sufficient to cool the PSU plus some, then it might do fine in this role if you don't mind the noise. But its likely to be quieter by isolating the PSU.

Ideally, I think a case would isolate the PSU such that it intakes fresh air directly from the outside like the Antec Overture and dumps the heated air directly outside (such as if you cut a hole in the back of the Overture); it would isolate the case air mass from the CPU area and PSU area exhausting air heated by cards out the back, perhaps using a rear or front case fan and front intakes for fresh air; each part of the case would have its own individual intake and exhaust for CPU, PSU and case with minimum airflow for that item. Or in the direction the BTX standard goes, perhaps this could be handled by a single channel supplying each module.

I like that Dell. It is very rational from their viewpoint. It reduces cost and customer service tickets by jettisoning the CPU fan; satisfies consumers who want their computer to be seen and not heard; all while keeing things stable. The large makers have the advantage of precisely calculating power ussage for their chosen components and picking an optimum PSU drawing as little as possible. Also, if you look at some of the HP's and Aptiva's starting a few years ago, they have a wide open intake on the front. This is arranged so that a panel comes down the front and the intake is below to get maximum intake with least noise. Some of the newer machines do not seem to follow this, so they must have other means of keeping cool without large front ducts.

I like the idea of getting rid of the CPU fan. I wonder how a Fortron 120mm PSU running slowly along with a 120mm case fan ducting CPU heat out would do in a SuperLanBoy?

About that silenx article. I agree with the relationship that a large fan may mean lesser room for heatsinks and large components, which are usually a sign of quality. However, I find the statement "120mm fans cannot be as quiet as 80mm fans, no matter what." a bit much. The idea behind the larger fan is that it can run more slowly with less noise than the smaller fan. That idea is still true whether it always works in practice. I am unsure how important "bearing noise" is and seems that would depend on the quality of the fan.

It also depends on the role of the PSU, whether it must dump CPU and case heat.

robb, I was just using the Thermalright as an example but I thought some heatsinks were optimised for exhausting out the top. Searching around I don't see any though...

Dell's PSUs cool the rest of the case as well (HDD, GPU, etc.)

You're right though that it supports Mike's theory and you make a good point.

I just thought I'd add my experience with my PSUs. I have an older model Seasonic 300w modded with an 80mm L1A and a 350w Fortron with the 120mm fan. I've used both in both cases that I've had a Lian-Li PC60 with 1 L1A as ehaust and a 3700AMB with a OEM Panaflo as exhaust. Both had all the same gear inside (suspended 'cuda 4's, zalman on the GPU, etc so in all other respects they were very very quiet systems). All fans were mounted on Ear fan isolators.

Lian Li- With the seasonic, there was insufficent airflow and the CPU would go to unacceptable temps for low noise. The PSU stayed quiet, but I had to increase the exhaust fan too much to have a quiet system. With the Fortron, I was able to keep the exhaust fan and CPU fan low, but at minimal load (or a mildly warm day) the fan would ramp up. There was just too low airflow in the case (I had modded the exhaust for a 92mm fan). I think I would have been fine had I modded the rear for 2 80 mm L1A's.

3700 AMB- Rear exhaust at 5v. I installed the Fortron first in this one, and the fan kept ramping up to much the same levels as in the Lian-Li. Temps in the system dropped by about 2c overall with a better airflow case. I recently put the seasonic back in, and temps dropped by another CPU temps dropped by another 1C.

I think the only way that a 120mm fanned PSU would be able to be run in a quiet system is with a cold air duct for the PSU or a shroud/duct to exhaust the CPU heat. I've found that what seems to work best in my system is exhausting the heat where it is produced (VGA silencer for the vid card and duct for the cpu).

As I said earlier in this thread.....a modded Fortron PSU is the only exhaust fan in my P4 system. When I first tried the Fortron in it's stock form, the fan speeds seemed much too high. And the rpms would ramp up for no particular reason.....computer idling with nothing going on. Measuring the output temps when this was happening revealed very low temps, none over 30c.

IMHO....running this 300w Fortron works well as the only exhaust fan, but relying on the internal PSU sensors to regulate the fan speeds yields rpms too high, and the PSU too noisy.

Solution for me....Mod the PSU with an Evercool 120mm fan, regulate it's speed with an external controller, monitor the output temps under various conditions, and adjust the fan speed to suit the operating conditions. I found I can run this Evercool-modded Fortron most of the time, with an 1100 rpm (using a Zen fanless heatsink). This is with an ambient temp of 22c. Come summer temps, that speed will no doubt need to be higher. But right now, the output temp of the PSU never goes over 34c, perfectly ok to me. Also the Fortron blows into a small muffler setup which is somewhat restrictive.

And contrary to some of the posts in this thread, this setup is very quiet...

It would require some heavy modding but having the PSU at the bottom of the case, so that the "cpu-facing" fan would in fact face the case bottom, where would be an opening for the fan, so the PSU would get fresh air from floor level, and it would actually not be any part of the air circulation around other parts of the system. Because the intake air would be so cold, probably 10C or more, and it would breathe very freely because the psu intake would be in free air, the psu fan could run very slow while providing good cooling to the psu. It might be possible to build a compartment for the psu under a normal case, just making a hole in the bottom of the case for the wires.

Instead of heavily modding an existing case, why not just build one from scratch? That way, you can acheive the ideal layout. You can even make it out of relatively sound absorbative material. I like cheap vinyl floor tiles because they look good and can easily be cut (scissors work). Hitting them gives a dull "thump" rather than the "rap" of cardboard or the "ring" of metal.

I've thought of that also. Using a PSU with 120 mm fan it would cool the GPU too, provided that one doesn't have any other cards.

My other idea is/was to use a "standard" quiet PSU with 80 mm fan, put it on the bottom, lay a duct/channel from front to back, and maybe put the HD in the channel as well. Maybe even modding the PSU so that the fan is inside the case (maybe leave a wire grill in place...).

So, Dremel-owners, what are You waiting for?

Cheers,

Jan

Yeah, but then you'd be evacuating heat from the PSU... which is what we wanted to avoid in the first place.

Um... The fan would blow in, not out.

Jan

This is a very interesting topic, and it looks to me like there is no single right answer.

Based on what I've been able to find out, and based on threads like this one, I think there clearly ARE times when the PSU can fill a role in evacuating heat. I'm sure a lot of it depends on the quality of ventilation the case offers, too.

The bottom line is, I think an 80mm fan will be quieter than a 120mm **for a given rotational speed**. However, this only serves to illustrate that the reason to go with a bigger fan is not simply because it is quieter, but because the 120mm fan has a better CFM-to-noise ratio. In other words, if you were to spin a 120mm fan fast enough to match an 80mm in terms of noise, it would still be flowing more air.

Since 120mm-fanned PSUs also typically have smaller heatsinks to make room for the larger fan, it does also mean that greater airflow is required to keep the PSU cool than would be required otherwise. However, which is worse: requiring more airflow to keep the PSU cool (which is kind of a small point anyway, given that you're going to have more airflow with a 120mm fan, regardless), or requiring a faster case fan to evacuate the heat that the PSU otherwise could?

Anyone else have any thoughts on this?

Well all, I finally got my quiet PC built, and on a VERY tight budget. I ended up going with an Enermax PSU because they are everywhere and I found a DEAL on one that had a speed control for the rear fan.

After putting all the parts in my new Antec case, I must say it is very quiet. I purchased the case off of ebay with 2 120mm Panaflo (louder version) fans in it and a fan speed control module in one of the 3.5" bays. I also purchased the awsome Zalman AlCu heatsink. When I got the entire thing together, it was pretty quiet. However, the Panaflo's were much to loud for me. I took out the front 120mm fan and replaced the rear fan with the Antec fan that the eBay seller included. Once I plugged the Antec fan into the speed control and dropped it as low as it would go, it was almost silent.

So now I have a PC with the CPU running at 32c and the MB at 29C (Asus A7N8X-X mobo.) With the temps this low (compared to my other PC with a CPU temp of 60c) I figure I could slow some fans down even more. I found a spare wire that had a voltage reducer in it so I added that to my already slow spinning 120mm Antec fan. I'm not sure how slow the Antec fan is spinning now, but it is real quiet.

The one thing that strikes me as really odd, and continues to bring me back to the Dell design is this... I now have a 120mm rear fan blowing out with a CPU fan that blows down directly below a 120mm PSU fan that blows up. It sure looks to me like the air would have a hard time deciding what direction it would take, as all of them are opposite from each other. I'm thinking of turning the PSU up-side-down so that the incoming fan draws air from the TOP of the case (not outside air, inside air,) completely away from the CPU. Then, I bet I could turn off the fan on my CPU entirely and just have the rear 120mm exhaust fan cool the CPU as well. That would drop an entire fan. I realize the temps on my CPU would go up, but when I'm at 32c now, I figure I would have another 40c to go before I need to get really worried (AMD says 80c is tops for CPU.) Add some duct work, and I bet my temps would be real manageable.

More from the company that I love to hate... Dell...

Don't know if anyone has seen the inside of the most recent Dell Precision 450 Workstation or not, but yet another interesting design. This time they put the PSU at the bottom and made it the entire length of the case. It vents directly from the front and out the back without any chance to grab any of the air inside the case.

Very interesting, and I bet it is quiet.

60mm PSU fans?

Or 50mm or 40mm. Either way, I don't see how it could be particularly quiet.

The only thing I would add as a very rough rule of thumb is NOT to use a bottom-intake fan PSU in a cooling configuration that utilizes addtional exhaust fans.