Multiple Fan Sandwiched for Improve Cooling

[UaVaj]

know anyone or any website where they have tested cpu cooling using multiple fans sandwiched together for improved cooling.

obviously this video is way way way over kill and outright unpractical.
http://www.youtube.com/watch?v=VUbpb23yTK8

was think more in terms of say 2-3 push + 2-3 pull on an evo 212.

bottom line is what are the pro/con of sandwiching fans.

 

[dma0991]

I found this. Fan stacking does work as it has similar principal to a jet engine where you have a series of progressively smaller and denser blades that compresses the air. This will not be the case with stacked computer fans because it isn't optimally designed (like a jet engine) to do so nor is it spinning fast enough.

The pros, it does improve cooling by a bit but not enough to be noticeable. The cons, more fans = more noise and adds more vibration. You're better off just using a single push pull fan configuration than stacking it.

 

[Charlie98]

I started a thread some time ago giving my opinion on the 212+ cooler... that I thought it was poorly designed (as most finned air coolers are) because there is no shroud to close the sides and prevent the cooling air from just escaping out the sides of the cooler before it can pass all the way through the cooler and complete the heat transfer, even in a push/pull configuration. I wound up wrapping the sides of the cooler in duct tape and the terrible gaps between the fan shroud and exchanger and, indeed, it dropped my CPU temps a degree or two in informal LinX testing. The Noctua D14, for example, I feel is a better design with it's push/push design but still suffers from the loss of efficiency by allowing air loss from the gaps.

I agree with dma that the slower (quiet) fans in a CPU cooler wouldn't lend themselves to a 'compressor' style cooler (although the idea is intriguing! ) I think the real key is closing up the fan gaps and closing the edges of the heat exchanger to force all the air through it.

 

[BonzaiDuck]

I started a thread some time ago giving my opinion on the 212+ cooler... that I thought it was poorly designed (as most finned air coolers are) because there is no shroud to close the sides and prevent the cooling air from just escaping out the sides of the cooler before it can pass all the way through the cooler and complete the heat transfer, even in a push/pull configuration. I wound up wrapping the sides of the cooler in duct tape and the terrible gaps between the fan shroud and exchanger and, indeed, it dropped my CPU temps a degree or two in informal LinX testing. The Noctua D14, for example, I feel is a better design with it's push/push design but still suffers from the loss of efficiency by allowing air loss from the gaps.

I agree with dma that the slower (quiet) fans in a CPU cooler wouldn't lend themselves to a 'compressor' style cooler (although the idea is intriguing! ) I think the real key is closing up the fan gaps and closing the edges of the heat exchanger to force all the air through it.

Well, that's fairly consistent with brief experiments I'd done in '07 with the TRUE or similar coolers. My repetitive assertion may be overplayed in recent weeks, but the dual Noctua fans on the D14 actually hinder the cooling that is possible.

If anyone were to apply a ducting solution to the D14, it could've been me, but I hadn't really done it thoroughly to cover the issues Charlie raised: I'd merely ported my case exhaust fan to the back side of the D14. I'm thinking I could now eliminate the Akasa fan I have in the middle of the cooler, by replacing the exhaust fan with a slightly beefier unit that provides PWM control. But it would need the ducting solution Charlie mentions -- assuring airflow from the front pipes and fins to the rear set.

Something tells me that on a cooler like that, you'd want some degree of pressure on both sets of fins, so the improvement might otherwise be smaller than it could be.

 

[IEC]

Use of a proper fan with good static pressure would be the biggest improvement. Example would be a Scythe Gentle Typhoon AP-15 (1850 RPM) or similar.

You'll get mediocre results by using more than 2 fans in push/pull. Very much diminishing returns.

 

[BonzaiDuck]

Use of a proper fan with good static pressure would be the biggest improvement. Example would be a Scythe Gentle Typhoon AP-15 (1850 RPM) or similar.

You'll get mediocre results by using more than 2 fans in push/pull. Very much diminishing returns.

Incidentally. People recently remarked that either Gentle Typhoons were "out of production," or no longer available. Anyone interested should check an outfit named "Aerocooler."

 

[UaVaj]

correct. with the open sides and all the cracks in between the cooler n fan. air simply escape through those opening and does not fully flow through the fins.

will have to play with it when there is more free time.


on a simpler scientific note: when two regular 120mm fan are sandwiched together. what is the air flow difference +/-? what is the static pressure difference +/-?

 

[dma0991]

on a simpler scientific note: when two regular 120mm fan are sandwiched together. what is the air flow difference +/-? what is the static pressure difference +/-?

Increase in air flow/static pressure is a given but does not increase proportional to the number of stacked fans and there's a diminishing return.

I have personally tried this a long time ago with my old AMD system with stock heatsink. You do get a stronger gust of wind but as I've said, it also increases noise and vibration. There are better ways to improve on a push pull CM Hyper212 Evo. One of them as mentioned in this thread, is to seal the open sides. There was a tower heatsink (I can't recall its name) similar to CM Hyper212 but has its sides crimped down to create this seal.

 

[BonzaiDuck]

Increase in air flow/static pressure is a given but does not increase proportional to the number of stacked fans and there's a diminishing return.

I have personally tried this a long time ago with my old AMD system with stock heatsink. You do get a stronger gust of wind but as I've said, it also increases noise and vibration. There are better ways to improve on a push pull CM Hyper212 Evo. One of them as mentioned in this thread, is to seal the open sides. There was a tower heatsink (I can't recall its name) similar to CM Hyper212 but has its sides crimped down to create this seal.

I maybe said it before -- maybe in this thread. My system had six fans for 2+ years until last month. I reduced the number by two, but my CPU load temperatures are improved 5C.

Whether or not it actually improves the temperatures, I think I can get the total down to three with a beefier exhaust fan -- higher static pressure, higher RPMs, [higher dBAs] -- higher CFMs at full throttle. But I will have to duct my D14 for the effects already cited, per the remark about "electrical tape" on the side of the fins.

I need a 120x38mm PWM fan that's a little less noisy or less "fast" than this:

http://www.performance-pcs.com/cata...oduct_info&cPath=36_407_696&products_id=26058

If I can run it at low therms at 1,000 RPM with "good" dBA's, what would it be like at 2,700 . . . or 3,000? The motherboard fan-control features would allow me to limit the top-end and also the noise -- even at the load temperatures.

I can appreciate this discussion about "fan sandwiches." Some WON'T appreciate my interest in these high-output 120x38's. And the entire purpose: to make it possible for a 200mm side-panel fan to run slower at idle from CHA_FAN1 while I put the CPU and exhaust fans on the PWM plug and draw power for it/them directly from the PSU.

It's almost a nit-picky, obsessive enterprise. Even at full bore, the 200mm is next to noiseless . .

 

[dma0991]

Whether or not it actually improves the temperatures, I think I can get the total down to three with a beefier exhaust fan -- higher static pressure, higher RPMs, [higher dBAs] -- higher CFMs at full throttle. But I will have to duct my D14 for the effects already cited, per the remark about "electrical tape" on the side of the fins.

I currently only consider air cooling to be an easy and nearly maintenance free solution. Its not something that I would take the trouble of ducting to get the most out of it because I doubt a duct would be an interesting centerpiece of your rig. I saw a video (removed, but there's a picture in this thread) a while back, an AIO ducted with high speed fans (3-4kRPM). As I recalled, improvements were marginal but it ended up sounding like a lawnmower.

I'd rather go with custom watercooling if I'm that obsessed with getting the lowest temp delta and lowest noise output, not to mention the tremendous benefits that GPUs get as well from watercooling.

 

[BonzaiDuck]

I currently only consider air cooling to be an easy and nearly maintenance free solution. Its not something that I would take the trouble of ducting to get the most out of it because I doubt a duct would be an interesting centerpiece of your rig. I saw a video (removed, but there's a picture in this thread) a while back, an AIO ducted with high speed fans (3-4kRPM). As I recalled, improvements were marginal but it ended up sounding like a lawnmower.

I'd rather go with custom watercooling if I'm that obsessed with getting the lowest temp delta and lowest noise output, not to mention the tremendous benefits that GPUs get as well from watercooling.

You definitely have a point there. I'm approaching a crossroads about this. On the ducting, it depends on what you duct and how you duct it. But more mobo components are solid-state these days, and mainstream boards are equipped with some decent heatsinks or even heatpipes.

It's short-term tedium -- trouble of short duration. If it's a matter of appearances, there's "Lexan," some "shapes" like oatmeal barrels or small pieces of lumber, and a heat-gun.

I think I might pay more attention to it if I were using H2O and wanted to cool motherboard parts, but then again -- things have changed on that angle . . . too . . .

There may be 'things to do" with high-output fans, but there's also diminishing returns to CFM. So . . . .