- Jun 30, 2004
- 16,154
- 1,757
- 126
[THE PICS AND NARRATIVE I PROMISED DURING THE GROWTH OF THIS THREAD ARE PRESENTED BELOW THIS ORIGINAL POST, WITH THE TITLE "DUCTING FOR DECIBELS"]
Some folks are still interested in the general Nidec Servo "Gentle Typhoon" fan line. I would suspect that most would pick those with lower CFM, RPM and therefore -- lower dBA.
I had been making a search for a 120x38 (or 120x25mm) PWM fan with output between 110 and 150CFM. Obviously to get 150CFM (the RATED airflow), a Delta 1.0A 50+dBA fan will make a lot of noise. I'd considered several options, and in the interim use a ducted 140mm Akasa to provide 113CFM to a 120mm case-exhaust port.
I don't know how I missed these GT 120x25mm fans:
http://www.performance-pcs.com/cata...oduct_info&cPath=36_407_696&products_id=36549
It tops the Delta fan I'd looked at for RPM by 500RPM or more. but the noise level at the top end is only ~44dBA.
The Akasa generates half as much noise, but only has 3.5 mm H2O static pressure. But someone independently tested the Gentle Typhoon 4250 (see #2 of the "120mm High Speed Series" list:
http://dvtests.com/?p=13497
As opposed to the rated ~117CFM output, the test measured the GT4250 @ ~245 cu-m/hr or 144CFM!
I wager that I can both control the noise thermally (a given), and I can control it or attenuate it "acoustically." This is a "ducting" experiment, and you can use Spire acoustic foam to line a duct made to compensate overall inner diameter for the foam's thickness.
Anyway, people were complaining about the availability or perception that these GT fans weren't anymore in manufacture. So I thought I would post this, since this and other resellers still have them in stock to offer.
ADDENDUM -- A FORGOTTEN QUESTION:
Does anyone have experience with this particular 4,250 RPM Gentle Typhoon model? If the measured CFM output exceeds the rated spec by ~28 CFM, what would be the possibility that the rated top-end noise level is either more or less than the actual measured value, since the reviewer doesn't show any decibel measurements? The promotional "overview" of the fan suggests it was meant to be as quiet as possible, or that the "tone" of the fan and some design feature reduced the annoying "frequency" of the fan for human ears.
===========================
DUCTING FOR DECIBELS
===========================
As a veteran fan-collector, the Nidec-Servo Gentle Typhoon PWM AP-30 seems pretty incredible.
At speeds up to just over 2,000 RPM, the fan's characteristic hum seems difficult to hear if you could hear it at all.
The fan has variously been called a "beast" for its airflow and even resellers call it "loud." Acknowledging this, Nidec-Servo promotes the unit with the features of built-in, carefully-engineered acoustics. It is only my suspicion, but the ring of vent-holes around the rotating fan-hub doesn't seem intended for ventilation, and I could only guess that it has some sort of noise-cancellation purpose.
It does get better.
It is possible to attenuate further any motor noise as you explore three objectives: reduce temperature, maximize CFM at certain temperature thresholds, and take successful measures to bury the noise.
Of course, many in the overclocking community won't experience stress-test wattage peaks during their computer's ordinary usage. So this suggests creating two temperature profiles with thermal fan control: one for regular use and likely guaranteed to provide better than average cooling with less noise at known, proven overclock settings; and another for benchmarking or overclock stressing. There will be exceptions: enthusiasts who use their computers 24/7 for CPU-intensive tasks, "folding@home," etc.
You make the choice to use air-cooling or water. Either way, you need a good, efficient exhaust fan or fans of one type or another. If the throughput at working peak load sounds like an AC vent in a small room on a hot day, then the fan is mostly generating white-noise from air turbulence. Comparing the turbulence/white-noise and noise coming from the motor, the white noise is much more tolerable.
As I said, the Gentle Typhoon 4,250RPM PWM fan -- known as the "AP-30" -- has very little of any noise up to about 2,000 RPM without noise abatement. That is, "silence" seems fairly guaranteed between lowest possible idle (around 1,200) and 2000. Moving from there toward the 100% duty-cycle and ~4,200 RPM, both the white noise and the characteristic tone are more obvious. The tone is subtle at first, rising to its highest level above 3,700 RPM.
The AP-30 easily idles at 1,400 RPM at 25% duty-cycle. It is difficult to avoid assumptions or presuppositions that could prove inaccurate with real scientific measurement. We might want to think that CFM airflow increases linearly with both speed and noise, but we really don't know that for sure. What we do know: the fan is rated at ~117 CFM by the manufacturer, but independent tests showed something more promising:
http://dvtests.com/?p=13497
That review -- if you can believe it -- shows airflow results for the fan powered at 12V to be 244 cu-m/hr, or 144 CFM. Either way, it is "a beast."
On first testing the fan, I thought its top-end might have promise for a noise-attentuation project, and I could see how the high-end Typhoon hum could really seem annoying. People would complain that they're "loud" at high speed.
There are several ways to categorize fans. Some fans are promoted as case fans but may have cpu-fan usage; other fans are bundled with heatsinks and supposedly chosen to serve a wider range of personal tolerance. There are 3-pin fans and PWM fans. There are "silent" fans and "high-output" fans, and there are "Low Speed," "Medium," "High" and Ultra-High" fans. Those in the high-output or "high-speed" categories will usually have dBA noise ratings above 40 dBA.
The Gentle Typhoon D1225C12B9AP-30 can fit all of these profiles. It is quiet up to 2,100 RPM, where the most barely perceptible hum seems to arise. The hum is audible but hardly annoying to 2,400 RPM, but rises in both pitch and volume toward the top end. With the case open or otherwise no provision for noise abatement with the case closed, the high-end speeds exhibit noticeable sounds, but those sounds are at least matched by the white-noise of full-throttle output. The white noise doesn't have so much of an annoyance edge; the characteristic whirr of the Typhoon's top-end, is something else -- rising in pitch and volume as RPM increases.
It is unclear to me whether this whirring sound comes from the motor, or from a tone created by the fins. The white-noise has nothing of an easily identifiable tone. The high-end noise of the Typhoon other than the white-noise has a tonal quality and its pitch or frequency rises with RPM. We may attenuate the tone's loudness or intensity, but it will still be present more or less, even if we can attenuate it to a point where one must strain or even put on a hearing-aid to notice it. We want to make it less noticeable.
This is as much to say that we want the tone's volume so attenuated that the RPM at which it seems annoying his shifted upward and as near as possible to the top-end 4,200 RPM speed. "Near-as-possible" may just mean that we can attenuate the noise acceptably to run the fan as fast as we need to.
THE DUCTING PROJECT
There are two benefits for ducting your heatpipe cooler to an exhaust fan. The first of these has always obviously been more efficient cooling. The idea is to make all pressurized air in the case pass through the CPU cooler fins on its way to a quick exhaust.
There are a number of approaches. One could build a duct-box that would fit over an entire single or double-tower cooler, assuring that the only air that gets into it comes from the intake side, and all the exhaust air goes immediately through the case exhaust-fan without further mingling with case-interior air. Since this duct must be closed to the case-interior air except through an intake, it must be designed so that it has a "bottom" that mates up with the lowest cooler fin, with a hole in the bottom of the duct that matches the cooler-fin width and length. For a double-tower cooler, there would be a bottom with two such holes, with sides and a top that would seal off the space between the towers.
There is also a "minimalist" approach. If the duct mates up with the exhaust side of a single-tower cooler, you can assume that intake air will come from the other three sides and even the rear area of fins not covered by the duct -- as long as the duct is very close to the fins. With a double-tower cooler, one might put a "pusher-puller" fan like an AKASA 140R in the center of the cooler, which then adds a sort of duct of its own. It will draw air from three sides of the front tower. An exhaust-fan duct will then draw air from the three sides of the rear tower, including the exhaust air from the center fan.
THE FIRST AND SIMPLE DUCT -- CIRCA 2011: YEAR OF THE SANDY-BRIDGE BUILD
The nice thing about accumulating old fans that you don't want to use anymore: You can cut out the motors and glue together the shrouds for a ready-made duct:
120x38mm fan, one whole 120x25mm shroud, and a second 120mm shroud cut down and fitting the remaining gap to the cooler fins. The last two items were bonded with Super Glue; rubber rivets secured the fan to the case; nylon screws and nuts secured the middle fan shroud to the forward part of the fan.
WHY I DECIDED TO MAKE IMPROVEMENTS
This trouble all began two months ago, when I thought I had a problem with a 3-pin fan-port on my motherboard. I wanted to use more PWM fans with a Swiftech 8W-PWM-SPL splitter, available at
www.sidewindercomputers.com .
This would allow me to power the fans directly from the PSU but controlling them from the mobo's CPU_FAN port. First on the list of my attentions was a Panaflo 120x38 exhaust fan shown in the last photo, which ran quietly to its top-end of 2,300 RPM, rated output of 103 CFM. At that speed, it would still offer some slight degree of white-noise under CPU load.
So I considered replacing it with a Delta PWM fan of the same size rated at ~150 CFM. I was avoiding the 120x25mm fans, just because I had a ready-made duct for a 120x38mm unit.
My case has a 120mm exhaust vent. I'd replaced the Noctua fan of my cooler with an Akasa 140R, and giving up the idea of the Delta, I got the idea to rig a duct which accommodated a square 140mm version of the Akasa -- offering rated output at 110 CFM. This worked fine, and here are the pictures of the ducting-mod I made which I might well have just left alone: it was virtually noiseless.
Unfortunately, I won't go through the trouble of re-installing this configuration in my machine, just to snap you a picture. Installation and removal is one of the primary challenges to ducting, and I've known myself to sacrifice convenience for in-place effectiveness.
The Akasa 140x25mm ducting mod
The idea with the Akasa fan was to put it in the middle of a duct -- to bury it deeper in the case. And of course, it wouldn't otherwise fit the 120mm vent.
But the problem with this and similar mods -- even the initial ducting of the 120x38mm Panaflo -- was the difficulty in assembly and removal. Using fan shrouds or a fan-adapter, it was tedious business getting all the pieces installed or removing them. I don't have the E-trade baby's hands and fingers. I was inclined to use rubber fan-rivets wherever I could in the assembly. Where I couldn't do so, I used nylon screws and nuts. When you want to clean kruft from the system, you'd want to be able to remove the ducting easily, and these concoctions I'd made didn't fill the bill.
Later, as forum members worried at a rumored prospect that the Gentle Typhoons weren't available or had gone out of production, I found a reseller who had them and discovered this AP-30 model. If my colleagues here revered some of the Typhoons for noiselessness, the exception was probably the AP-30, but the AP-30 had both a rated and tested output that I favored.
At first I thought to bury the AP-30 in the middle of the duct as I had done with the Akasa fan, thinking that pulling it a bit away from the metal case vent might promote airflow and reduce noise. I concocted a duct such that the last piece installed at the rear cooler fins would just slide into rails on the other part of the assembly:
But the nature of the mounting holes in the AP-30, my desire to use rubber rivets, and an error in measurement made me see the complexity of all this. I found that the complicated construction was 1mm too big to clear the cooler fins and I saw the trouble of mounting the fan midway between case and cooler. The AP-30 was also an ounce heavier than other fans, and I didn't like to see it hanging from another 120mm fan shroud secured with rubber rivets to the case. I threw up my hands and said "Simple is best!" But I nevertheless save this construction; it covers a wider area of the rear fins of my NH-D14 cooler. Perhaps I can trim it a bit with my Dremel and try again.
In the meantime, I retreated to the idea of mounting the AP-30 with rubber rivets directly to the case vent -- as most people do. All of these ideas of burying the fan in the case interior in the middle of the duct -- away from the vent -- are feasible, if you have the E-trade baby's hands and fingers. But you don't.
With the immediate concern that I needed to finish a simple duct for the fan and get my computer up and running again, I got out my pencil and paper, drew a simple duct-box for construction with foam art-board, and crossed my fingers that it would just slide in or out of its mount, or that it wouldn't somehow reorient itself. It had to mate up with the fan or just barely overlap one edge of it. So this is what I came up with:
Somehow, once inserted, the duct didn't need any hooks or tabs to keep it in place. The L-shaped frame on one side of the duct held it secure and resting on the cooler itself. And I began to imagine that the art-board duct would attenuate noise. Maybe it did -- a little. But the hard surface of the duct interior also reflected noise.
ACOUSTIC ENHANCEMENT
The second benefit never mentioned in early ducting "literature" of engineering students trying to obtain better cooling with any heatsink or heatpipe cooler: since the duct covers the fan on the intake side, the duct material can be chosen to attentuate sound.
Foam art board also has decent acoustic properties for noise-deadening, but the paper-backing on its surface is so hard that it will reflect sound before it absorbs it. The reflected sound then bounces around the cooler fins.
Once the duct is completed, patches of Spire noise-deadening foam rubber can be applied to its exterior. Better yet, you can carefully add the Spire in rectangular pieces to the interior, perhaps securing them with a little bit of Super Glue if you worry that they might come loose and foul the fan propeller.
I might have made the duct-box bigger just to accommodate the interior Spire pieces, but they are 1/4" thick, and don't significantly reduce the airflow, when you consider that the box or chamber is larger than the circular fan aperture by dimensions of the four plastic corners of the fan shroud.
I applied Spire foam in stages, testing each application to see how the fan's tone had been attenuated at various speeds. For instance, I decided the fan needed an acoustic "horseshoe" or collar.
Since the horseshoe's open side faces the thin metal of the I/O plate devices on the motherboard, I modified it again and added a fourth side.
I further added a second layer of Spire rubber on the duct exterior, but I couldn't put another piece of it on the side that faced the case side-panel. One piece just fit; two pieces would rub against the side-panel when you removed or replaced it.
Without anything in the way of measured acoustic tests, I can only say that a noise-dampened duct box has a significant positive effect in reducing noise volume -- the decibel level. Perhaps it has a greater effect on noise that seems motor-generated, or on the higher-pitched hums and whistles we've come to expect from fans that are not noiseless.
Under this reduced noise volume, certain aspects of noise kick in at the same time whether or not the duct-box and noise-abatements are installed. You can attenuate the higher-frequency tones of the fan, but you cannot get rid of the white-noise from air turbulence. So at higher RPMs and output, your computer may sound like a nearby air-conditioning vent on a hot day.
You might consider cutting out all or part of the perforated case exhaust vent metal to reduce turbulence and improve airflow, but that's too extreme a modification to your case when the white noise is a low-annoyance factor. So you can choose to live with it. With carefully chosen thermal fan control, you won't hear it for the most part unless you're stress-testing, and as I said, it's an "air-conditioning vent" for the noise.
ADDITIONAL FOOTNOTES ON DUCTING
You can't design a standard duct that will fit all cases and all coolers. You can't design a single duct that will fit all cases for a single cooler. It is a customization: there are as many duct designs according to how much you want to cover the cooler, and for either the "minimalist" or total approach. The number of different duct designs will be a cartesian product of all available cases on the market and all relevant coolers on the market.
My duct is better than the original 2011 design because the interior of the box has a greater volume than the fan shroud pieces of the original design. But it doesn't cover the entirety of the cooler fin's rear edges. A 140mm fan mounted symmetrically to my 120mm case exhaust vent exactly matches the lowest fin of the NH-D14. These things are peculiar and specific to my choice of a case (the HAF 922) and the NH-D14 cooler.
The duct based on the dimensions of a 120mm fan misses a few cooler fins at the bottom. It also doesn't cover the entire width of the cooler fins, and it overhangs the cooler on two sides by either 5 mm or 1 cm -- the reason the duct needed the two L-shaped frame pieces. Those were not initially intended to hold the duct in place -- but they do. The intention was to prevent air from being exhausted before entering the cooler fins.
So the square size of the duct could be replaced by a rectangular duct to fit the entirety of the rear cooler fins. It would require some precise measurement, and an additional rectangular piece covering the rear side of the duct and leaving a 120x120mm hole for the fan. This would enhance airflow and efficiency, and allow for at least the same attention to acoustic padding. How much it would further reduce temperatures, I cannot say. I'll have to build one and try it.
But I have improved the duct chamber size and acoustically deadened the characteristic hum of the Gentle Typhoon AP-30. By doing so, the volume of the tone generated at about 2,400 RPM is now about equal to what one hears at 3,400 RPM. And there is an another peculiar result of this experience.
People refer to a certain noisiness of PWM fans as a direct result of the PWM control. That's just what I'd read or heard. I discovered when running my AI Suite's "Fan Xpert" CPU fan test that the software begins at the fan's lowest possible RPM at about 20% duty cycle, and measures the speeds above that in 10% increments -- completing the test at the 4,200 RPM top-end and 100% duty cycle. It makes these changes as quickly as it can -- just so the software can record the speed from the tach wire. The tone and ascending volume created in this test process seem louder at every level where it becomes apparent, and I took note of the sound at 3,200 and 3,600 RPM. Running a LinX or IBT stress-test that brings the fan-speed to these levels doesn't seem to create the same decibel volume of the tone. I cannot explain this, because in both cases the fans are under PWM control. But the difference seems noticeable.
This would only mean that under normal usage or stress-testing, the fan is likely to exhibit its characteristic tone at lower decibel volume than under a fan-testing program like AI Suite's "Fan Xpert." And as I said, I cannot explain why that would be the case. With all things that might otherwise benefit from objective testing of decibel levels with instruments, people might suggest my judgments have a dimension of imagination, and I might wonder the same thing myself.
This computer may be my final choice for air-cooling, and the next one is likely to include carefully-chosen water-cooling parts. I want to use a custom water cooling kit -- not an all-in-one unit like the Corsairs or the CoolerMaster Nepton 280L. I'd prefer that choice for the same reason I've been fanatical enough to do the ducting: I want the ability to improve both the cooling and the acoustics. Eventually, I think I'd like to build a bong cooler extension to an internal water-cooling configuration.
But we also want to keep our eye on developments in air-cooling heatpipes with vapor-chambers and other innovations. For instance, the advent of Noctua's successor for the NH-D14 which we expect some time after April may offer another ducting opportunity for both cooling and noise. By itself, the "D15" promises to increase heatsink-base thermal conductivity by 25%, so comparison benchmark tests are likely to be quite favorable.
Adding another 5C to those results with a custom fan duct may offer some stark thermal improvements over my experience with the NH-D14 -- or for that matter, any other heatpipe cooler on the market. The D15 with ducting mods might just trump all-in-one or CLC coolers like the current-model CM Nepton. And the manufacturers of install-it-and-fahget-it water coolers will need to come up with better models -- even at slightly higher prices.
For all that, this demonstrates that a $10 package of Spire acoustic padding needn't be applied to permanently gum up your case interior. Nor do you need to stick it on your fans, when you can simply build a horseshoe or collar reinforced by foam art-board. It is instead most useful in focused application directly to the source of the noise.
IF YOU'RE EASY BEING CRAZY ENOUGH . . .
Here I provide some guidance about tools and materials.
Tools:
There are a number of tools I always keep in my "computer tool-box." Some folks familiar with "joints" and "doobies" and "roaches" may know about the usefulness of certain surgical implements: the little clamps shaped like scissors which work like pliers and have a locking catch built-in to the handles. These things come in handy for retrieving parts you dropped into your motherboard, or pulling on fan-rivet nipples when your hands won't fit in confined spaces. I've had occasion to use them when attempting to place a 3-pin fan plug on an otherwise inaccessible motherboard fan-port. For the other thing, the legal usefulness now depends on the state in which you reside -- Colorado, Washington -- to a limited degree, states with "medical dispensaries."
Xacto knife -- I wish they made a better tool for cutting foam board, or perhaps a saw-blade for it. But you need this for scoring the paper-backing and cutting through the material in several patiently-executed passes.
Dremel with saw-wheel -- great for trimming foam board, when accessibility allows for holding the saw-wheel perpendicular to the surface. Also great for cutting the guts out of old fans.
Other things to keep handy: metal straight-edges, metric or English rulers, and whatever you think you might need.
Glue:
I had grown fond over the years for a particular formulation by Permatex found at auto-parts stores, known as "Pit Crew's Choice" automotive adhesive. It comes in a red tube, and is absolutely marvelous stuff. It is water-based, and solidifies into a translucent and slightly rubbery solid. You can use it for insulating wires, sticking wires and other parts to case-metal. And unlike the adhesive on Spire foam pads, it is easily removed without using a putty knife.
But I can no longer find it at auto-parts stores. The last tubes I'd purchased were found online, and those sources no longer have it. I finally located a motorcycle shop in Massachusetts several weeks ago, ordered three tubes and received two with a reduced invoice and apology stating that they had no more left and their supplier no longer offered it. Somebody! Please! Find me a substitute for Pit Crew's!
You might be able to use a glue like Pit-Crews for certain aspects of foam-board construction, but it isn't the first choice. Nor do you want to use a silicone-based adhesive sealant. It might work for this or that application, but it's a mess, really . . .
Super-Glue -- You can use Super-Glue to bond the paper backing on art-board and Spire rubber, but it will melt the foam if applied directly to it. You can also use it to build ducts from plastic fan-shrouds.
Poly-Zap Foam-Safe (or similar) -- can be had at a hobby-shop, and is expensive stuff. You get a couple ounces for at least $10. Like Super-Glue, it contains cyano-acrylates, will bond to your fingers. But it will bond just about anything including styro-foam and the art-board foam. There is, however, a cheaper alternative.
Hold-The-Foam -- a non-volatile formulation, it has the consistency of Pit-Crew's, seems to be water-based, and glues foam art-board very well. It just doesn't set as fast as Poly-Zap. But it's cheap! A $5 tube containing maybe three times more in volume than a bottle of Poly-Zap -- can be had at Michael's Arts & Crafts. Very promising stuff, and good for your wallet.
Speaking of Michael's.
Foam Art-board -- It comes in various sizes and colors. A 2'x3' panel can be had at Target for around $5. That's enough to complete a few ducting projects. At Michael's, you can get a 4'x3.5' folded display panel for about $16.
Lexan -- Great for building ducts if you have a heat-gun to shape it. More commonly used for case side-panel windows. A bit pricey. It will also reflect sound in ducting. But it sure looks nice with some LED lighting. You need Lexan glue to bond it. Again, the hobby shop offers "Poly-Zap Lexan glue." You'll also need a $3 plexi-glass knife to score and cut the stuff.
Spire Acoustic Foam-Rubber -- A $10 package is ample for several computer noise-reduction projects. It keeps well in storage: The package I'm still using is about 7 years old. Once you stick it to a side-panel and come back three months later to remove it, you are in for a frustrating exercise with a putty-knife and a can of lighter-fluid or some other solvent. Best applied in a way that doesn't require sticking it to fans and side-panels.
If you want to noise-deaden your case side-panel, consider applying it first to a piece of foam art-board, then applying the art-board to the metal case-panel with dabs of Pit Crew's or perhaps silicone adhesive-sealant. This increases the noise-deadening effect, but makes it easy to remove and then re-apply. Otherwise, the Spire you have to remove with a putty knife cannot be used for much of anything except to line your trash can.
Now. You can air-cool or water-cool. There may be some application for aspects of this project to water-cooling, either with ducting or for noise-reduction. If you choose air-cooling, you can simply drop in a heatpipe cooler and live with the stock fans. Or you can replace the fans. And if you replace the fans with a high-output model, you'll again be at a crossroads. If you want to address the noise and increase cooling, you might buy the supplies and use time with the tedium.
With a heatpipe cooler and ducting, you will still not reach the cooling effectiveness of a custom water-cooled rig. Look at it another way. All the improvements in air-cooled effectiveness are incremental. Lapping the IHS and the heatsink-base may get you 5 to 10C improvement; the choice of a superior TIM might be worth 5C; ducting is worth at least 5C, because I know I can build a better duct than the one shown here. Figure for the uncertainties that this earns you as much as a 17C improvement in temperatures. Maybe I'm wrong for the total, but I think that's a good ballpark.
But here, our focus has been less on the thermal improvements and more about taming a beast of a fan like the Gentle Typhoon AP-30. To some degree, I feel like there's been success all the way around.
It's all a matter of personal choice. If you want to say I'm nuts, I'll tell you I'm . . . "retired," so it doesn't matter to me!
Some folks are still interested in the general Nidec Servo "Gentle Typhoon" fan line. I would suspect that most would pick those with lower CFM, RPM and therefore -- lower dBA.
I had been making a search for a 120x38 (or 120x25mm) PWM fan with output between 110 and 150CFM. Obviously to get 150CFM (the RATED airflow), a Delta 1.0A 50+dBA fan will make a lot of noise. I'd considered several options, and in the interim use a ducted 140mm Akasa to provide 113CFM to a 120mm case-exhaust port.
I don't know how I missed these GT 120x25mm fans:
http://www.performance-pcs.com/cata...oduct_info&cPath=36_407_696&products_id=36549
It tops the Delta fan I'd looked at for RPM by 500RPM or more. but the noise level at the top end is only ~44dBA.
The Akasa generates half as much noise, but only has 3.5 mm H2O static pressure. But someone independently tested the Gentle Typhoon 4250 (see #2 of the "120mm High Speed Series" list:
http://dvtests.com/?p=13497
As opposed to the rated ~117CFM output, the test measured the GT4250 @ ~245 cu-m/hr or 144CFM!
I wager that I can both control the noise thermally (a given), and I can control it or attenuate it "acoustically." This is a "ducting" experiment, and you can use Spire acoustic foam to line a duct made to compensate overall inner diameter for the foam's thickness.
Anyway, people were complaining about the availability or perception that these GT fans weren't anymore in manufacture. So I thought I would post this, since this and other resellers still have them in stock to offer.
ADDENDUM -- A FORGOTTEN QUESTION:
Does anyone have experience with this particular 4,250 RPM Gentle Typhoon model? If the measured CFM output exceeds the rated spec by ~28 CFM, what would be the possibility that the rated top-end noise level is either more or less than the actual measured value, since the reviewer doesn't show any decibel measurements? The promotional "overview" of the fan suggests it was meant to be as quiet as possible, or that the "tone" of the fan and some design feature reduced the annoying "frequency" of the fan for human ears.
===========================
DUCTING FOR DECIBELS
===========================
As a veteran fan-collector, the Nidec-Servo Gentle Typhoon PWM AP-30 seems pretty incredible.
At speeds up to just over 2,000 RPM, the fan's characteristic hum seems difficult to hear if you could hear it at all.
The fan has variously been called a "beast" for its airflow and even resellers call it "loud." Acknowledging this, Nidec-Servo promotes the unit with the features of built-in, carefully-engineered acoustics. It is only my suspicion, but the ring of vent-holes around the rotating fan-hub doesn't seem intended for ventilation, and I could only guess that it has some sort of noise-cancellation purpose.
It does get better.
It is possible to attenuate further any motor noise as you explore three objectives: reduce temperature, maximize CFM at certain temperature thresholds, and take successful measures to bury the noise.
Of course, many in the overclocking community won't experience stress-test wattage peaks during their computer's ordinary usage. So this suggests creating two temperature profiles with thermal fan control: one for regular use and likely guaranteed to provide better than average cooling with less noise at known, proven overclock settings; and another for benchmarking or overclock stressing. There will be exceptions: enthusiasts who use their computers 24/7 for CPU-intensive tasks, "folding@home," etc.
You make the choice to use air-cooling or water. Either way, you need a good, efficient exhaust fan or fans of one type or another. If the throughput at working peak load sounds like an AC vent in a small room on a hot day, then the fan is mostly generating white-noise from air turbulence. Comparing the turbulence/white-noise and noise coming from the motor, the white noise is much more tolerable.
As I said, the Gentle Typhoon 4,250RPM PWM fan -- known as the "AP-30" -- has very little of any noise up to about 2,000 RPM without noise abatement. That is, "silence" seems fairly guaranteed between lowest possible idle (around 1,200) and 2000. Moving from there toward the 100% duty-cycle and ~4,200 RPM, both the white noise and the characteristic tone are more obvious. The tone is subtle at first, rising to its highest level above 3,700 RPM.
The AP-30 easily idles at 1,400 RPM at 25% duty-cycle. It is difficult to avoid assumptions or presuppositions that could prove inaccurate with real scientific measurement. We might want to think that CFM airflow increases linearly with both speed and noise, but we really don't know that for sure. What we do know: the fan is rated at ~117 CFM by the manufacturer, but independent tests showed something more promising:
http://dvtests.com/?p=13497
That review -- if you can believe it -- shows airflow results for the fan powered at 12V to be 244 cu-m/hr, or 144 CFM. Either way, it is "a beast."
On first testing the fan, I thought its top-end might have promise for a noise-attentuation project, and I could see how the high-end Typhoon hum could really seem annoying. People would complain that they're "loud" at high speed.
There are several ways to categorize fans. Some fans are promoted as case fans but may have cpu-fan usage; other fans are bundled with heatsinks and supposedly chosen to serve a wider range of personal tolerance. There are 3-pin fans and PWM fans. There are "silent" fans and "high-output" fans, and there are "Low Speed," "Medium," "High" and Ultra-High" fans. Those in the high-output or "high-speed" categories will usually have dBA noise ratings above 40 dBA.
The Gentle Typhoon D1225C12B9AP-30 can fit all of these profiles. It is quiet up to 2,100 RPM, where the most barely perceptible hum seems to arise. The hum is audible but hardly annoying to 2,400 RPM, but rises in both pitch and volume toward the top end. With the case open or otherwise no provision for noise abatement with the case closed, the high-end speeds exhibit noticeable sounds, but those sounds are at least matched by the white-noise of full-throttle output. The white noise doesn't have so much of an annoyance edge; the characteristic whirr of the Typhoon's top-end, is something else -- rising in pitch and volume as RPM increases.
It is unclear to me whether this whirring sound comes from the motor, or from a tone created by the fins. The white-noise has nothing of an easily identifiable tone. The high-end noise of the Typhoon other than the white-noise has a tonal quality and its pitch or frequency rises with RPM. We may attenuate the tone's loudness or intensity, but it will still be present more or less, even if we can attenuate it to a point where one must strain or even put on a hearing-aid to notice it. We want to make it less noticeable.
This is as much to say that we want the tone's volume so attenuated that the RPM at which it seems annoying his shifted upward and as near as possible to the top-end 4,200 RPM speed. "Near-as-possible" may just mean that we can attenuate the noise acceptably to run the fan as fast as we need to.
THE DUCTING PROJECT
There are two benefits for ducting your heatpipe cooler to an exhaust fan. The first of these has always obviously been more efficient cooling. The idea is to make all pressurized air in the case pass through the CPU cooler fins on its way to a quick exhaust.
There are a number of approaches. One could build a duct-box that would fit over an entire single or double-tower cooler, assuring that the only air that gets into it comes from the intake side, and all the exhaust air goes immediately through the case exhaust-fan without further mingling with case-interior air. Since this duct must be closed to the case-interior air except through an intake, it must be designed so that it has a "bottom" that mates up with the lowest cooler fin, with a hole in the bottom of the duct that matches the cooler-fin width and length. For a double-tower cooler, there would be a bottom with two such holes, with sides and a top that would seal off the space between the towers.
There is also a "minimalist" approach. If the duct mates up with the exhaust side of a single-tower cooler, you can assume that intake air will come from the other three sides and even the rear area of fins not covered by the duct -- as long as the duct is very close to the fins. With a double-tower cooler, one might put a "pusher-puller" fan like an AKASA 140R in the center of the cooler, which then adds a sort of duct of its own. It will draw air from three sides of the front tower. An exhaust-fan duct will then draw air from the three sides of the rear tower, including the exhaust air from the center fan.
THE FIRST AND SIMPLE DUCT -- CIRCA 2011: YEAR OF THE SANDY-BRIDGE BUILD
The nice thing about accumulating old fans that you don't want to use anymore: You can cut out the motors and glue together the shrouds for a ready-made duct:
120x38mm fan, one whole 120x25mm shroud, and a second 120mm shroud cut down and fitting the remaining gap to the cooler fins. The last two items were bonded with Super Glue; rubber rivets secured the fan to the case; nylon screws and nuts secured the middle fan shroud to the forward part of the fan.
WHY I DECIDED TO MAKE IMPROVEMENTS
This trouble all began two months ago, when I thought I had a problem with a 3-pin fan-port on my motherboard. I wanted to use more PWM fans with a Swiftech 8W-PWM-SPL splitter, available at
www.sidewindercomputers.com .
This would allow me to power the fans directly from the PSU but controlling them from the mobo's CPU_FAN port. First on the list of my attentions was a Panaflo 120x38 exhaust fan shown in the last photo, which ran quietly to its top-end of 2,300 RPM, rated output of 103 CFM. At that speed, it would still offer some slight degree of white-noise under CPU load.
So I considered replacing it with a Delta PWM fan of the same size rated at ~150 CFM. I was avoiding the 120x25mm fans, just because I had a ready-made duct for a 120x38mm unit.
My case has a 120mm exhaust vent. I'd replaced the Noctua fan of my cooler with an Akasa 140R, and giving up the idea of the Delta, I got the idea to rig a duct which accommodated a square 140mm version of the Akasa -- offering rated output at 110 CFM. This worked fine, and here are the pictures of the ducting-mod I made which I might well have just left alone: it was virtually noiseless.
Unfortunately, I won't go through the trouble of re-installing this configuration in my machine, just to snap you a picture. Installation and removal is one of the primary challenges to ducting, and I've known myself to sacrifice convenience for in-place effectiveness.
The Akasa 140x25mm ducting mod
The idea with the Akasa fan was to put it in the middle of a duct -- to bury it deeper in the case. And of course, it wouldn't otherwise fit the 120mm vent.
But the problem with this and similar mods -- even the initial ducting of the 120x38mm Panaflo -- was the difficulty in assembly and removal. Using fan shrouds or a fan-adapter, it was tedious business getting all the pieces installed or removing them. I don't have the E-trade baby's hands and fingers. I was inclined to use rubber fan-rivets wherever I could in the assembly. Where I couldn't do so, I used nylon screws and nuts. When you want to clean kruft from the system, you'd want to be able to remove the ducting easily, and these concoctions I'd made didn't fill the bill.
Later, as forum members worried at a rumored prospect that the Gentle Typhoons weren't available or had gone out of production, I found a reseller who had them and discovered this AP-30 model. If my colleagues here revered some of the Typhoons for noiselessness, the exception was probably the AP-30, but the AP-30 had both a rated and tested output that I favored.
At first I thought to bury the AP-30 in the middle of the duct as I had done with the Akasa fan, thinking that pulling it a bit away from the metal case vent might promote airflow and reduce noise. I concocted a duct such that the last piece installed at the rear cooler fins would just slide into rails on the other part of the assembly:
But the nature of the mounting holes in the AP-30, my desire to use rubber rivets, and an error in measurement made me see the complexity of all this. I found that the complicated construction was 1mm too big to clear the cooler fins and I saw the trouble of mounting the fan midway between case and cooler. The AP-30 was also an ounce heavier than other fans, and I didn't like to see it hanging from another 120mm fan shroud secured with rubber rivets to the case. I threw up my hands and said "Simple is best!" But I nevertheless save this construction; it covers a wider area of the rear fins of my NH-D14 cooler. Perhaps I can trim it a bit with my Dremel and try again.
In the meantime, I retreated to the idea of mounting the AP-30 with rubber rivets directly to the case vent -- as most people do. All of these ideas of burying the fan in the case interior in the middle of the duct -- away from the vent -- are feasible, if you have the E-trade baby's hands and fingers. But you don't.
With the immediate concern that I needed to finish a simple duct for the fan and get my computer up and running again, I got out my pencil and paper, drew a simple duct-box for construction with foam art-board, and crossed my fingers that it would just slide in or out of its mount, or that it wouldn't somehow reorient itself. It had to mate up with the fan or just barely overlap one edge of it. So this is what I came up with:
Somehow, once inserted, the duct didn't need any hooks or tabs to keep it in place. The L-shaped frame on one side of the duct held it secure and resting on the cooler itself. And I began to imagine that the art-board duct would attenuate noise. Maybe it did -- a little. But the hard surface of the duct interior also reflected noise.
ACOUSTIC ENHANCEMENT
The second benefit never mentioned in early ducting "literature" of engineering students trying to obtain better cooling with any heatsink or heatpipe cooler: since the duct covers the fan on the intake side, the duct material can be chosen to attentuate sound.
Foam art board also has decent acoustic properties for noise-deadening, but the paper-backing on its surface is so hard that it will reflect sound before it absorbs it. The reflected sound then bounces around the cooler fins.
Once the duct is completed, patches of Spire noise-deadening foam rubber can be applied to its exterior. Better yet, you can carefully add the Spire in rectangular pieces to the interior, perhaps securing them with a little bit of Super Glue if you worry that they might come loose and foul the fan propeller.
I might have made the duct-box bigger just to accommodate the interior Spire pieces, but they are 1/4" thick, and don't significantly reduce the airflow, when you consider that the box or chamber is larger than the circular fan aperture by dimensions of the four plastic corners of the fan shroud.
I applied Spire foam in stages, testing each application to see how the fan's tone had been attenuated at various speeds. For instance, I decided the fan needed an acoustic "horseshoe" or collar.
Since the horseshoe's open side faces the thin metal of the I/O plate devices on the motherboard, I modified it again and added a fourth side.
I further added a second layer of Spire rubber on the duct exterior, but I couldn't put another piece of it on the side that faced the case side-panel. One piece just fit; two pieces would rub against the side-panel when you removed or replaced it.
Without anything in the way of measured acoustic tests, I can only say that a noise-dampened duct box has a significant positive effect in reducing noise volume -- the decibel level. Perhaps it has a greater effect on noise that seems motor-generated, or on the higher-pitched hums and whistles we've come to expect from fans that are not noiseless.
Under this reduced noise volume, certain aspects of noise kick in at the same time whether or not the duct-box and noise-abatements are installed. You can attenuate the higher-frequency tones of the fan, but you cannot get rid of the white-noise from air turbulence. So at higher RPMs and output, your computer may sound like a nearby air-conditioning vent on a hot day.
You might consider cutting out all or part of the perforated case exhaust vent metal to reduce turbulence and improve airflow, but that's too extreme a modification to your case when the white noise is a low-annoyance factor. So you can choose to live with it. With carefully chosen thermal fan control, you won't hear it for the most part unless you're stress-testing, and as I said, it's an "air-conditioning vent" for the noise.
ADDITIONAL FOOTNOTES ON DUCTING
You can't design a standard duct that will fit all cases and all coolers. You can't design a single duct that will fit all cases for a single cooler. It is a customization: there are as many duct designs according to how much you want to cover the cooler, and for either the "minimalist" or total approach. The number of different duct designs will be a cartesian product of all available cases on the market and all relevant coolers on the market.
My duct is better than the original 2011 design because the interior of the box has a greater volume than the fan shroud pieces of the original design. But it doesn't cover the entirety of the cooler fin's rear edges. A 140mm fan mounted symmetrically to my 120mm case exhaust vent exactly matches the lowest fin of the NH-D14. These things are peculiar and specific to my choice of a case (the HAF 922) and the NH-D14 cooler.
The duct based on the dimensions of a 120mm fan misses a few cooler fins at the bottom. It also doesn't cover the entire width of the cooler fins, and it overhangs the cooler on two sides by either 5 mm or 1 cm -- the reason the duct needed the two L-shaped frame pieces. Those were not initially intended to hold the duct in place -- but they do. The intention was to prevent air from being exhausted before entering the cooler fins.
So the square size of the duct could be replaced by a rectangular duct to fit the entirety of the rear cooler fins. It would require some precise measurement, and an additional rectangular piece covering the rear side of the duct and leaving a 120x120mm hole for the fan. This would enhance airflow and efficiency, and allow for at least the same attention to acoustic padding. How much it would further reduce temperatures, I cannot say. I'll have to build one and try it.
But I have improved the duct chamber size and acoustically deadened the characteristic hum of the Gentle Typhoon AP-30. By doing so, the volume of the tone generated at about 2,400 RPM is now about equal to what one hears at 3,400 RPM. And there is an another peculiar result of this experience.
People refer to a certain noisiness of PWM fans as a direct result of the PWM control. That's just what I'd read or heard. I discovered when running my AI Suite's "Fan Xpert" CPU fan test that the software begins at the fan's lowest possible RPM at about 20% duty cycle, and measures the speeds above that in 10% increments -- completing the test at the 4,200 RPM top-end and 100% duty cycle. It makes these changes as quickly as it can -- just so the software can record the speed from the tach wire. The tone and ascending volume created in this test process seem louder at every level where it becomes apparent, and I took note of the sound at 3,200 and 3,600 RPM. Running a LinX or IBT stress-test that brings the fan-speed to these levels doesn't seem to create the same decibel volume of the tone. I cannot explain this, because in both cases the fans are under PWM control. But the difference seems noticeable.
This would only mean that under normal usage or stress-testing, the fan is likely to exhibit its characteristic tone at lower decibel volume than under a fan-testing program like AI Suite's "Fan Xpert." And as I said, I cannot explain why that would be the case. With all things that might otherwise benefit from objective testing of decibel levels with instruments, people might suggest my judgments have a dimension of imagination, and I might wonder the same thing myself.
This computer may be my final choice for air-cooling, and the next one is likely to include carefully-chosen water-cooling parts. I want to use a custom water cooling kit -- not an all-in-one unit like the Corsairs or the CoolerMaster Nepton 280L. I'd prefer that choice for the same reason I've been fanatical enough to do the ducting: I want the ability to improve both the cooling and the acoustics. Eventually, I think I'd like to build a bong cooler extension to an internal water-cooling configuration.
But we also want to keep our eye on developments in air-cooling heatpipes with vapor-chambers and other innovations. For instance, the advent of Noctua's successor for the NH-D14 which we expect some time after April may offer another ducting opportunity for both cooling and noise. By itself, the "D15" promises to increase heatsink-base thermal conductivity by 25%, so comparison benchmark tests are likely to be quite favorable.
Adding another 5C to those results with a custom fan duct may offer some stark thermal improvements over my experience with the NH-D14 -- or for that matter, any other heatpipe cooler on the market. The D15 with ducting mods might just trump all-in-one or CLC coolers like the current-model CM Nepton. And the manufacturers of install-it-and-fahget-it water coolers will need to come up with better models -- even at slightly higher prices.
For all that, this demonstrates that a $10 package of Spire acoustic padding needn't be applied to permanently gum up your case interior. Nor do you need to stick it on your fans, when you can simply build a horseshoe or collar reinforced by foam art-board. It is instead most useful in focused application directly to the source of the noise.
IF YOU'RE EASY BEING CRAZY ENOUGH . . .
Here I provide some guidance about tools and materials.
Tools:
There are a number of tools I always keep in my "computer tool-box." Some folks familiar with "joints" and "doobies" and "roaches" may know about the usefulness of certain surgical implements: the little clamps shaped like scissors which work like pliers and have a locking catch built-in to the handles. These things come in handy for retrieving parts you dropped into your motherboard, or pulling on fan-rivet nipples when your hands won't fit in confined spaces. I've had occasion to use them when attempting to place a 3-pin fan plug on an otherwise inaccessible motherboard fan-port. For the other thing, the legal usefulness now depends on the state in which you reside -- Colorado, Washington -- to a limited degree, states with "medical dispensaries."
Xacto knife -- I wish they made a better tool for cutting foam board, or perhaps a saw-blade for it. But you need this for scoring the paper-backing and cutting through the material in several patiently-executed passes.
Dremel with saw-wheel -- great for trimming foam board, when accessibility allows for holding the saw-wheel perpendicular to the surface. Also great for cutting the guts out of old fans.
Other things to keep handy: metal straight-edges, metric or English rulers, and whatever you think you might need.
Glue:
I had grown fond over the years for a particular formulation by Permatex found at auto-parts stores, known as "Pit Crew's Choice" automotive adhesive. It comes in a red tube, and is absolutely marvelous stuff. It is water-based, and solidifies into a translucent and slightly rubbery solid. You can use it for insulating wires, sticking wires and other parts to case-metal. And unlike the adhesive on Spire foam pads, it is easily removed without using a putty knife.
But I can no longer find it at auto-parts stores. The last tubes I'd purchased were found online, and those sources no longer have it. I finally located a motorcycle shop in Massachusetts several weeks ago, ordered three tubes and received two with a reduced invoice and apology stating that they had no more left and their supplier no longer offered it. Somebody! Please! Find me a substitute for Pit Crew's!
You might be able to use a glue like Pit-Crews for certain aspects of foam-board construction, but it isn't the first choice. Nor do you want to use a silicone-based adhesive sealant. It might work for this or that application, but it's a mess, really . . .
Super-Glue -- You can use Super-Glue to bond the paper backing on art-board and Spire rubber, but it will melt the foam if applied directly to it. You can also use it to build ducts from plastic fan-shrouds.
Poly-Zap Foam-Safe (or similar) -- can be had at a hobby-shop, and is expensive stuff. You get a couple ounces for at least $10. Like Super-Glue, it contains cyano-acrylates, will bond to your fingers. But it will bond just about anything including styro-foam and the art-board foam. There is, however, a cheaper alternative.
Hold-The-Foam -- a non-volatile formulation, it has the consistency of Pit-Crew's, seems to be water-based, and glues foam art-board very well. It just doesn't set as fast as Poly-Zap. But it's cheap! A $5 tube containing maybe three times more in volume than a bottle of Poly-Zap -- can be had at Michael's Arts & Crafts. Very promising stuff, and good for your wallet.
Speaking of Michael's.
Foam Art-board -- It comes in various sizes and colors. A 2'x3' panel can be had at Target for around $5. That's enough to complete a few ducting projects. At Michael's, you can get a 4'x3.5' folded display panel for about $16.
Lexan -- Great for building ducts if you have a heat-gun to shape it. More commonly used for case side-panel windows. A bit pricey. It will also reflect sound in ducting. But it sure looks nice with some LED lighting. You need Lexan glue to bond it. Again, the hobby shop offers "Poly-Zap Lexan glue." You'll also need a $3 plexi-glass knife to score and cut the stuff.
Spire Acoustic Foam-Rubber -- A $10 package is ample for several computer noise-reduction projects. It keeps well in storage: The package I'm still using is about 7 years old. Once you stick it to a side-panel and come back three months later to remove it, you are in for a frustrating exercise with a putty-knife and a can of lighter-fluid or some other solvent. Best applied in a way that doesn't require sticking it to fans and side-panels.
If you want to noise-deaden your case side-panel, consider applying it first to a piece of foam art-board, then applying the art-board to the metal case-panel with dabs of Pit Crew's or perhaps silicone adhesive-sealant. This increases the noise-deadening effect, but makes it easy to remove and then re-apply. Otherwise, the Spire you have to remove with a putty knife cannot be used for much of anything except to line your trash can.
Now. You can air-cool or water-cool. There may be some application for aspects of this project to water-cooling, either with ducting or for noise-reduction. If you choose air-cooling, you can simply drop in a heatpipe cooler and live with the stock fans. Or you can replace the fans. And if you replace the fans with a high-output model, you'll again be at a crossroads. If you want to address the noise and increase cooling, you might buy the supplies and use time with the tedium.
With a heatpipe cooler and ducting, you will still not reach the cooling effectiveness of a custom water-cooled rig. Look at it another way. All the improvements in air-cooled effectiveness are incremental. Lapping the IHS and the heatsink-base may get you 5 to 10C improvement; the choice of a superior TIM might be worth 5C; ducting is worth at least 5C, because I know I can build a better duct than the one shown here. Figure for the uncertainties that this earns you as much as a 17C improvement in temperatures. Maybe I'm wrong for the total, but I think that's a good ballpark.
But here, our focus has been less on the thermal improvements and more about taming a beast of a fan like the Gentle Typhoon AP-30. To some degree, I feel like there's been success all the way around.
It's all a matter of personal choice. If you want to say I'm nuts, I'll tell you I'm . . . "retired," so it doesn't matter to me!
Last edited:
Need to get around to finding my low RPM fans so they can run full bore and not care about the noise.
Facebook
Twitter