BadRobot -- You're not entirely wrong -- just a bit sensitive.
There have been some very stunning examples of configurations which suck air from the heatsink bottom and duct it past the heatsink top and out the case rear. One of these examples used the Zalman CNPS-7000 copper-"flower" heatsink. The user took the wrist part of a long rubber glove -- basically an elastic tube -- and fitted it over the flower, leaving narrow apertures where the fins meet the heatsink base. He then fitted a fan and right-angle/L-shaped duct to the top of the flower -- secured to the case rear.
What we're discussing here is whether air becomes ineffective at picking up thermal energy if it has been slightly warmed after passing through heatsink fins. If the CFM of that air is sufficient, the gain in thermal energy per molecule of air will be insignificant to its remaining potential to pick up additional thermal energy before it has been exhausted.
An interesting point made in the "Cheap As Free" ducting article at OverClockers deals with the effective use of fans. There seems to be a prevalent myth that pressurizing cases is not an effective cooling method, or that it leads to stale air-pockets which increase in temperature within the case.
But on the contrary, if the pressurized part of the case is isolated from the warm electronic components by various panels, that air will not become significantly warmer and will serve as a reservoir of cool air which is then ducted from the case and exhausted after picking up heat from the warm components. This is one situation in which fans feeding air to each other in series actually augment airflow -- ordinarily, putting two fans together with one blowing air into the other do not add to airflow significantly, unless the "serving fan" is larger and the "receiving fan" is smaller. But here, we're creating a pressurized reservoir of air within the case, which is then pulled through narrow apertures around the warm components and ducted to the exhaust.
Someone asked for pictures from my Prescott system, and I'm going to oblige in just a minute here. [I had to configure my Ipswitch software to access a "personal web-page" to upload these photos -- haven't worked with HTML in several years, and am currently focused on sheet-metal and other case-mod-related work].
All of these innovations serve to reduce the number of fans to a minimum within a case -- conserving a few watts of electrical power-usage, and reducing noise. Basically, from my experience with the Prescott system, I choose the larger fans for intake, selecting them to provide larger CFMs than the total exhaust CFMS -- by some reasonable margin. Of course, filtering the intake fans will reduce rated CFMs, but if the intake rpm-fan-speeds are adjustable and the exhaust speeds are adjustable, everything can be brought to an optimum.
Someone also asked if a single 120mm exhaust fan is sufficient, and I would say -- if this is a midtower case -- and if the ducts are designed to keep the CFMs through the CPU heatsink at a reasonable level while also drawing air off the motherboard surface -- I cannot see why not. I choose my fans for bigger motors and higher speed/CFM ranges, while paying attention to their dBA noise ratings. For that reason, my Prescott system's CPU fan is a Delta Tri-Blade which has a top-end speed of 3,700 rpms and a maximum CFM throughput of nearly 150 CFM, but I don't run the fan up that high (the dBA level approaches 50). The Delta is marvelously quiet in the range between 1,000 and 2,500 rpm.
Another good fan which I'm using for my Core-2-Duo build-in-progress is a Sanyo-Denki San-Ace, which top-ends at around 2,500 rpm, draws 0.52A at 12V, and seems relatively quiet. But there are many good fans "out there" to fit the various ways in which you wish to deploy them.
The choice of exhaust fans depends on whether they are augmented by the CPU fan through ducting, or otherwise serve for both exhaust and a CPU fan. For the ThermalRight Ultra-120 and Extreme, the opportunity exists to use the exhaust fan to pull air through the U-120 fins if the fins sit close enough to the fan's intake side.
For the other ThermalRight coolers such as the XP-120/90, SI-120 and SI-128, there is the opportunity to work the CPU FAN and exhaust fans in "serial-augmentation" with the ducted-area in-between. The better design of a duct would limit the largest volume within the duct to the area just below the heatsink fins, so that the motherboard duct-panel would sit just a quarter-inch above either the chipset cooler or the memory modules.
My duct was of "cruder" design. I originally used black foam-art-board -- a step up from cardboard -- but good enough for permanent use -- it's not conductive, it doesn't accumulate static charge, it's rigid and can be glued together with anything from "Automotive Goop" to "Shoe-Goo" to Contact Cement or similar substances. You can secure duct-panels (of either the foam-board or clear Lexan) by gluing patches of Velcro to the parts, case-panels etc. Or you can tap 6-32 screw-holes in case metal for screws or standoffs which fit the panels.
So my first photo shows the duct made exclusively of foam board. By the way -- Michael's Arts and Crafts is usually touted for carrying foam board, and I once thought that $6 per 2'x3' panel was reasonable. Not so. Target and other stores carry Elmer's foam board 2'x3' panels for something like $1.90.
Initial foam board duct-box
Note that the upper-left-corner of the construction shows a protruding box -- a separate component -- which is cut to fit two 92mm exhaust fans as precisely as possible. There wasn't any room in this old case for a 120mm exhaust fan.
Replaced main duct panel with Lexan
This second photo shows the replacement of the flat duct-panel with Lexan. The duct-box around the exhaust fans is still shown to be foam-board, but I have since learned to bend Lexan with a heat-gun, and so that duct-component has also been completely replaced with Lexan.
The folding front-side of the duct-system is still foam board. It sits just a quarter-inch above the memory modules so that air is forced past them at a high-velocity. Except that the drive cages cannot be illuminated by the blue-LED lights, it is convenient to leave this flexible foam board piece as is.
Before replacing my 3.2E processor with a 3.4E, the 3.2 was over-clocked to 3.5 Mhz using the ASUS mobo's ability to unlock the multiplier and drop it by a value of 2. Therefore, the front-side bus was boosted to 1,000 Mhz to run the DDR500 memories at their full spec. Temperatures with any processor or over-clock setting in this machine have never exceeded 43C or 110F at full load and room ambients exceeding 75F.
Facebook
Twitter