reducing case volume for better cooling?

Hifriday · Post by **Hifriday** » Mon Feb 05, 2007 10:01 am

I stumbled upon this article by John Cinnamon which applies a duct/housing that is fitted low against the mb and basically covers most of it. The author's principle of maximizing airflow by reducing the volume and increasing cooling efficiency/effectiveness. The CPU fan blows down into the duct and the side of the duct towards front of the case is left open for intake. The duct then rises to the rear case exhaust fans which quickly pulls out most of the heat generated by the mb and CPU, theoretically leaving the rest of the case relatively cool (VGA heat also forced directly out rear of the case).

I know ducting and using seperate chambers are quite common solutions, but I haven't seen such a large mb duct used/mentioned in SPCR. This is a pretty old article and I'm sure it's been discussed/used before but I failed to find them. Are there any problems that make such mods poorly suited for SPCR rigs?

jaganath · Post by **jaganath** » Mon Feb 05, 2007 11:37 am

Wow, a full MB duct, that's the first time I've seen one of those! But you know, this rig exemplifies a point I made in the thread about the BTX form factor; for quiet and effective cooling, you want every heat-producing component to have its own separate airflow- you see in the final layout that everything has its own duct: GFX, PSU and MB/CPU (we can consider MB/CPU as a whole because it is impractical to duct northbridge separately from CPU/VRM etc). One concern I do have is how close the MB duct comes to the intake of the rear case fans, some fans really do not like having obstructions that close to the intake, also you need case fans that work well against restriction (ie not Noctua). It's amazing what a difference all the ducts made to the temps, as much as 30C less! I kind of agree on the "Volume vs Flow" thing but for cooling the more volume you can get in contact with the heatsink the better cooling will be, and maybe that's why this works so well.

An interesting comment at the end:

I don't have a way to measure noise, but it's noticeably quieter. I was able to remove the side and front case fans with no hit on cooling at all.

This echoes what a lot of other SPCRers have found, that intake fans are generally of little utility (maybe only for hard drive cooling and such).

It's kind of funny that an overclocker did this first and not an SPCRer, but I guess the truth is it's become so easy to build a quiet PC these last few years that most people don't need to do such elaborate constructions to get a quiet PC; in that way SPCR is a lot less DIY oriented than it used to be.

jhhoffma · Post by **jhhoffma** » Mon Feb 05, 2007 11:48 am

Because the article uses a (loud) stock heatsink, it leaves a lot of depth in the case unused. Most people with a larger, tower heatsinks would render this style of ducting useless and the HS would either stick out above the mobo duct or the duct would be so deep that it would be almost as big as the case itself.

If a tower HS is not uses usually its a large flower style or horizontal fin/heatpipe heatsink that is much larger and cover much more of the board than a stock CPU HSF. Same reason why it would be not as effective.

regal196 · Post by **regal196** » Mon Feb 05, 2007 4:32 pm

his echoes what a lot of other SPCRers have found, that intake fans are generally of little utility (maybe only for hard drive cooling and such).

I am using a single 140mm fan as an intake fan, the CPU has a Ninga w/ a 120mm (yate loon) and the GPU has a HR-03 on it. I am not using a exhaust fan. I have custom case that inverts the mobo with the I/O port facing up so the air flow is straight up. So in my case, an intake fan is not useless.

As for the idea a whole mobo duct is awesome. On the main problem is the need to move the optical drives to allow a more direct path for air to flow.

ps- sorry if i got off topic

-Pennell

jaganath · Post by **jaganath** » Mon Feb 05, 2007 5:24 pm

I am using a single 140mm fan as an intake fan, the CPU has a Ninga w/ a 120mm (yate loon) and the GPU has a HR-03 on it. I am not using a exhaust fan.

Yeah, what I mean is there is no point in having intake fans if you already have exhaust fans, as the guy who made the mobo duct had. As we all know two fans in series doesn't give you anymore airflow, so either exhaust-only or intake-only is fine, but not both, there will be no benefit.

nici · Post by **nici** » Mon Feb 05, 2007 6:07 pm

It's easier to direct airflow with two fans in series, and if the is any restriction along the way using fans in series does increase airflow.

Hifriday · Post by **Hifriday** » Mon Feb 05, 2007 11:19 pm

jhhoffma wrote:Because the article uses a (loud) stock heatsink, it leaves a lot of depth in the case unused. Most people with a larger, tower heatsinks would render this style of ducting useless and the HS would either stick out above the mobo duct or the duct would be so deep that it would be almost as big as the case itself.

The large surface area of the tower heatsinks allow for effective cooling even with slower airflow, whereas a smaller sink will require higher airflow. However by simply reducing the duct/case volume, the amount of air being blown over the hot components will be increased (assuming the case fans are exhausting the same amount of air) - if the duct effectively reduces the case volume by 90%, then the amount of air drawn through the ducted space would be increased 10x. I assume the amount of heat dissipated from the heatsink is directly proportional to the surface area? in which case a heatsink 1/10th the size could potentially cool just as effectively. Alternatively the fans could be slowed down to 1/10th the original flow for equivalent cooling or a balance somewhere in between a smaller heatsink and slower fan.
Furthermore with the duct you don't have the problem of hot air recirculating around the case which should make the cooling even more effective.

Of course such a tight space and higher airflow could lead to more turbulence/noise, but we would be talking about SPCR components not OC rigs so assume this higher airflow would still be manageable and quiet enough. Well in theory anyways; so anyone have any experience with an actual duct like this?

As for low profile heatsinks with fan blowing downwards onto the mb, there are quiet ones like the XP/SI-120 and the flower sinks should also work (maybe with a slight rise in the duct just around it). Even with a tower heatsink it might work with duct rising up to enclose it, if it is close enough to the rear. However in additon to a duct intake fan just above the tower aimed towards the rear, a second fan might be needed to insure cool air is also circulated through the rest of the mb.

jaganath · Post by **jaganath** » Tue Feb 06, 2007 2:17 am

The main reason this solution works so well is that in a lot of rigs a large percentage of the air simply passes uselessly through the case, without coming into contact with any of the heat-producing components. The tight ducting forces the air to travel over and through the heatsink surfaces, so all the airflow is used as efficiently as possible. It also helps to prevent recirculation and areas of "trapped" flow.

Bluefront · Post by **Bluefront** » Tue Feb 06, 2007 3:20 am

I have extreme reservations just how well this system would work from a noise and cooling perspective. Forcing airflow through a restricted area, either by blowing or sucking, always results in a higher noise level. Blowing through a heatsink with tightly packed fins is an example......sucking through a restrictive filter is another example.

There are limits to just how tight a duct should be for maximum efficiency....I suspect the limit is exceeded with this sort of duct. And as difficult as it would be to calculate the ideal shape of the thing, or even to conduct experimental setups, I think I'll pass on the idea......

jhhoffma · Post by **jhhoffma** » Tue Feb 06, 2007 8:53 am

Hifriday wrote:The large surface area of the tower heatsinks allow for effective cooling even with slower airflow, whereas a smaller sink will require higher airflow. However by simply reducing the duct/case volume, the amount of air being blown over the hot components will be increased (assuming the case fans are exhausting the same amount of air) - if the duct effectively reduces the case volume by 90%, then the amount of air drawn through the ducted space would be increased 10x. I assume the amount of heat dissipated from the heatsink is directly proportional to the surface area? in which case a heatsink 1/10th the size could potentially cool just as effectively. Alternatively the fans could be slowed down to 1/10th the original flow for equivalent cooling or a balance somewhere in between a smaller heatsink and slower fan.
Furthermore with the duct you don't have the problem of hot air recirculating around the case which should make the cooling even more effective.

Err, no? The amount of air wouldn't be any MORE than in the standard setup (if the intake/exhaust fans are the same), in fact with a smaller space (and warm components) you would most likely have increased the pressure inside the duct, decreasing airflow and requiring higher speed intake/exhaust fans to overcome the positive pressure.

Like jaganath said, the amount of airflow in a duct isn't more than in (almost) free air, but more of that airflow is passing over heat-producing components, making it a more efficient airflow.

Don't get me wrong, ducting is all well and good, when done right, but there are limits, like Bluefront suggested.

Hifriday · Post by **Hifriday** » Wed Feb 07, 2007 2:45 am

With a more restrictive duct, agree you would need to increase fan voltage in order to exhaust the same amount of air out of the case. I mean under such a scenario, the amount of air being drawn through the ducted space would increase. However as everyone is saying, the degree of ducting/airflow in the end needs to be balanced for noise and efficiency. For a silent rig, certainly not suggesting an overly restrictive duct where turbulent wind speeds would be required, but simply that a quiet fan would be that much more effective with the reduced volume it needs to cool.

Come to think of it, the only completely inaudible system I own is my Pundit which is a very low-profile (2" height), low volume case that is essentially very similar to one large mb duct. With the PSU seperated out (uses external AC/DC brick), the system is sufficently cooled with a 80mm+60mm intake fans blowing down onto the CPU/chipset sink at very slow speeds and only exhaust holes on the case sides. Yes s754 Semprons are easily cooled, but this gives some merit to this line of thought.

It's a damper that Bluefront with all his ducting skills and experience gives this the thumbs down (I was half-expecting you'd have tried something similar in one of your past projects...).

Bluefront · Post by **Bluefront** » Wed Feb 07, 2007 4:19 am

Here's how I do this testing......I go over each part of the board before installation in a case, then measure everything to determine if any component needs extra cooling. If I find a hot spot (other than the CPU of course), I try to arrange extra airflow in that spot. This pre-testing, avoids problems later on, and certainly avoids building a complicated duct that is probably un-necessary.

ultrachrome · Post by **ultrachrome** » Wed Feb 07, 2007 7:14 am

I agree with Bluefront that it doesn't seem all that necessary. We now have a wide selection of oversized heatsinks that make better use of case volume.

I've seen a fairly recent pic of a what appeared to be an E-ATX server that used a motherboard duct, although I can't rememer how I came across it.

Here's what I found googling "motherboard duct":

jhhoffma · Post by **jhhoffma** » Wed Feb 07, 2007 7:17 am

Well, HiFriday, I guess the key thought is this. Changing the direction of airflow, increases turbulence, and decreases the amount of air the fan(s) are able to move in a given period of time. The same is true for changing the shape and volume you are pushing (or pulling) air through. It's the same reason why those 80mm-to-120mm fan adapters for heatsinks don't really work. They are too restrictive and cause too much backpressure that you have to run your fan faster to make up for it.

I guess the ideal "duct size" for any fan would be one that has the same cross-sectional area as the ID of the fan itself. And tube-shaped would be the most efficient shape, but that's not really practical. It would allow the fan to spin freely without any backpressure and all the airflow would not meet anything that would cause undue turbulence. I think that was one of the key points of the BTX form-factor, the CPU tunnel.

Hifriday wrote:For a silent rig, certainly not suggesting an overly restrictive duct where turbulent wind speeds would be required, but simply that a quiet fan would be that much more effective with the reduced volume it needs to cool.

With respect, I think that is a flawed thought-process, though. Decreasing the volume in which any fan is in, by itself, would do nothing to make it more effective. However, designing a duct to make it airflow more directional, so the air goes where it's needed, would make it more effective.

Think of it this way, you have a certain set of components that put out X amount of heat. No matter what size of case/duct you have, those components always generate that same amount of heat. So you need Y amount of airflow over all of the components to move that X amount of heat out. How you get Y is up to you, but the "whole system aiflow" approach wouldn't really work--why waste airflow cooling the CMOS battery? So you use a duct to move Y though the areas it's most needed. But if you decrease the volume you are working in, with X amount of heat, you increase the pressure of the system and make it more difficult to move Y amount of air. So the fans have to work harder (faster), creating more noise.

I know some would disagree, but IMO, the quietest setup is a negative pressure scenario with the least amount of restriction possible. But there are situations when certain components require a little more air, which is where spot cooling and ducting become a necessity to cool off warmer components without adding unnecessary heat to the rest of the system.

But again, that's just my $0.02.

jaganath · Post by **jaganath** » Wed Feb 07, 2007 10:00 am

I think also the thermal separation aspect of this layout is key; you have a separate duct for each area, as in the P180. It's really surprising how few cases implement this, even though it's a really good technique.

jhhoffma · Post by **jhhoffma** » Wed Feb 07, 2007 3:06 pm

I almost made a sidenote to that effect, but I decided that I had rambled on enough. With the lower volume the heat would be more likely to hang around close to cooler running components, raising their temperature as well. Some components can handle it, but I'm sure most of the secondary components on a mobo (SATA controllers, BIOS, etc.) don't function too well at elevated temps without heatsinks.

ronrem · Post by **ronrem** » Wed Feb 07, 2007 4:51 pm

If you have a lit 100w bulb in a milk carton,it will heat up pretty quick,perhaps start a fire. You may notice that a 100 w bulb won't heat up a small bathroom. Less air volume heats faster. A small case can heat up faster than a big one.

The thing with DUCTS is to suck hot air out ASAP and to supply the coolest air possible to the key components-such as CPU.

Relative temp of the air outside is cooler than what's in the case. Your Ninja is like a car radiator-it does the job on a cool day,but if it's 110F in Barstow,,the car runs hot.

A Duct can bring cool air to the CPU a duct can suck hot air from the CPU/heatsink

Isolating the CPU or the CPU + chipset would tend to make it easier to cool the rest of the stuff,likely the PSU fan is enough. I advocate a base below the compuer,for clearance,so intake is from the bottom. This is easy to make.

ultrachrome · Post by **ultrachrome** » Wed Feb 07, 2007 7:36 pm

ronrem wrote:If you have a lit 100w bulb in a milk carton,it will heat up pretty quick,perhaps start a fire. You may notice that a 100 w bulb won't heat up a small bathroom. Less air volume heats faster. A small case can heat up faster than a big one.

But cases are not sealed.

Mount a 120mm fan on that milk carton and the bulb will run much cooler than it will in a bathroom with the same 120mm fan trying to cool it.

You'll still move the same volume of air through both chambers but more of it will come in contact with the bulb in the milk carton. This is how a duct works.

Hifriday · Post by **Hifriday** » Thu Feb 08, 2007 1:19 pm

Targeted ducting and properly directed airflow are probably more important for quiet and effective cooling, and in a way this mb duct is just that albeit a more crude and generalized implementation. I don't think it will be overly complicated to build and certainly easier to design, of course depening on your rig.

I'll have to admit my knowledge on fans, airflow, pressure, ducts, etc. are really quite poor and still have many questions in this regards, but I think the key question is whether such a solution will be effective in a making a quiet rig cooler and/or quieter.

I thought of trying some light bulbs-box-fan contraptions to help clear things up, but in the end decided a real system would probably give a better idea. I made a quick-and-dirty duct out of some thin cardboard for my Aria system, photos here. I probably won't have time to do any detailed comparisons until next week (and any sound comparisons will be limited to my subjective evaluations), but hopefully it'll give some meaningful data.

jaganath · Post by **jaganath** » Thu Feb 08, 2007 1:52 pm

A good duct can bring down temps massively (as shown with the MB duct). A bad duct can actually make things worse or even cause your computer to overheat; so the implementation is very important as well.