Care and feeding of Scythe low-pressure (SlipStream) fans

[Felger Carbon]

I'm bringing up a uATX case-based computer using the Abit AN-M2 mobo. This mobo has two controlled fan headers, one 3-pin PWM and one 4-pin PWM, and BIOS fan control utilities for both, with selectable thermal sensors for both. I'm using a new Ever Case E3505B mATX tower case, which I'm extremely familiar with (I heavily modded 4 of them, culminating in mounting a 220mm fan on one's left side panel, before throwing them all away! Now, I'm baaack!)

I've read with interest postings on the high-pitch Noctua fans, and their problems with back-pressure. Problems that also involved increased noise.

The exhaust fan:

My new case will use a 500RPM Slipstream as the exhaust fan, well-ducted to an original Ninja. Yesterday, I dremeled out the "grill" on the exhaust port. Now I need to provide a low-backpressure air path from the outside to the input of the Ninja/duct, and that one Slipstream fan should "see" satisfactorally low total backpressure.

This would mean I could hear my exhaust fan only if my ear was within 3 inches of the fan, and that's at full load - the fan will be running more slowly at idle (browsing). The 500RPM Slipstream fan is incredibly quiet (even at 563RPM).

The PSU fan:

The PSU will use another 120mm fan, controlled by the mobo, not the PSU. Because of the chassis layout, there is no way to provide a low-backpressure path for this fan. My option is to use a SlipStream fan anyway, or to instead use a GW NCB (at very low voltage) or even a Scythe Kama PWM 120mm modified for use as a 3-wire fan. If you ground the PWM wire on the Scythe fan (while plugged into a 3-pin fan extension; we don't want to short out a mobo), the Scythe runs 175 to 496 RPM at 7.5 to 12V (from an analog fan controller). The fan starts fine from <7.5V to 12V when power is switched on. (Copied from here.)

So, the PSU fan is up in the air, with 3 possible solutions to be chosen on the basis of lowest noise at a given exhaust temperature from the PSU.

Testing for low backpressure

I strongly suspect (99.44%) that the existing CAGs will not be sufficient. Here's how I'll test: first, measure the exhaust temp of the PSU and exhaust fan (two measurements). Then, simply remove the left side panel - that'll provide a low backpressure path! A retest will provide a handle on how serious the problem is.

Next, replace the left side panel with a cardboard replica. Carve holes in the cardboard, patching with duct tape as necessary, to provide the final air-inlet configuration. Then take the dremel to the real left side panel...

How bad a problem is backpressure?

On the SlipStream fans, that is? I just now ran an 800RPM version (so I could hear it, the 500RPM model is downright hard to hear even inches away from my ear) at 12V. Then I moved a 5" square piece of cardboard up close to the fan exhaust, almost blocking it. Result: the very low motor/bearing noise got even lower but the whoosh turbulence increased somewhat. The net result may have been an increase in noise, but nothing major; nothing like what I've read about the Noctua fans. This means the first fan I'll try in the PSU will be a duplicate of the exhaust fan.

I've written this to explain how I approach this kind of new situation. Sometimes my guesses are wrong and sometimes something pops up out of left field, completely unexpectedly. I have all the parts on hand, so only watching football will interfere with this project.

Meanwhile, some of you get to explain what I'm doing wrong, perhaps before I do it!

[Felger Carbon]

This is ready to turn on, less the hard drive. I first just run bios, then boot to floppy to FDISK and FORMAT the HDD. A quick 3D look-see showed there's really only one logical place to put an extra CAG: A 3" wide strip just between the Ninja and the 3.5" floppy, 6 to 9 inches high. The green "X"s outline this area:



For the longer (higher) extra CAG, you can see the HDD enclosure partly gets in the way. Only partly, and besides, that's with a 3.5" HDD. What about a similar enclosure using a 2.5" HDD and a 40mm Scythe fan?



That's a new Hitachi 7K200 100GB 7200RPM, that transfers data to/from the disk media (not cache) slightly faster than the 3.5" drives I've been using. I shouldn't see any "2.5 inch performance penalty" at all.

[Felger Carbon]

Well, the big experiment is a flop. What I wanted to do was find out how much more intake area I had to use to avoid backpressure problems (low air flow, high noise) using the high-pitch Scythe Slipstream fans. I had a carefully thought-out plan, which I shared with you, to attack those problems. The result of the experiment came as a complete surprise:

There were no problems. Honest. No temperature problems, no noise problems.
------------------------

There's big advantages in firing up a computer without HDDs. You can move the case around, add and subtract side panels and bezels, tilt the case if needed to look at something, all with no danger to the missing HDD.

I used the Abit mobo (AN-M2) bios hardware monitor to read the exhaust fan RPM and CPU temperature. I used several Oregon Scientific indoor/outdoor thermometers to read the chassis exhaust temp, the PSU exhaust temp, and the average of the two standard CAGs' intake temp (the higher CAG is at a higher temp, natch). The thermometers had been calibrated against each other; and I just wanted the relative temperature offsets and not the absolute value(s).
-----------------------

I used the closed-off storage bedroom for these tests because it's cool (in the event of temperature-rise problems) and very quiet in the back. For the record, here's what I got:

Kill-A-Watt sez wall-power drain is 56W (no HDD) with an AM2 Athlon 64 3000+ 90nm Orleans. With the side cover on, average CAG intake temp is 15.6C; exhaust RPM 527, CPU temp (bios) 47C, PSU exhaust 25.8C. The regular chassis exhaust is much lower than the PSU's, so I just measured the PSU.

Removed the side cover and waited 30-45 minutes. CAG intake went up 0.5C (CRT {yes, CRT} and computer were heating room). CPU temp now 42C, PSU exhaust now 23.8C. RPM and wall power constant.

This particular case, with a 120mm PSU fan and a passive Ninja results in PSU exhaust temperatures well higher than the main exhaust. Fact of life. But a full-load (100% CPU duty cycle in bios) exhaust rise of 10.2C (side panel on) is fine with me, expecially since I never see full load when my computer is running (as opposed to under test).
---------------------------------

My guess is, 500RPM on the Slipstream fan is like ~650-700RPM on a regular 120mm fan... except the motor and blade noise is far less (same whoosh). Holding the 500RPM SlipStream in my hand, I can't hear it if it isn't within 3 inches of my ear. I'm not especially flexible these days, so I couldn't get my ear closer than about 6 or 7 inches to the rear computer exhausts, so it's not surprising I couldn't hear anything with the side cover either on or off.

Those of you now running YL, SFlex, or NCB exhaust fans at 1000RPM might give a try with an 800RPM SlipStream. Your computer's exhaust might get a lot quieter. If you do try this, let me - and the rest of us - know. None of my computers draw enough power to need that much exhaust airflow.

edit: fixed two keyboard typos

[sea2stars]

I've been looking for Globalwin NCBs and haven't been able to find any at my usual internet shops, but I came across the SlipStreams. I bought two 800RPMs last night to toss into a NSK2400 system to replace the current NCB intakes.

I'm really looking forward to see..um.. hear.. um, not hear.. what these puppies can do. I'll post impressions, temps, etc when I get them installed.

[slimeballzz]

Would you think these fans are better than the S-Flex for radiator use (Thermochill PA series)?

[NeilBlanchard]

Greetings,

The 9-blade Scythe fans are now available on NewEgg. Here's the 500RPM model:

http://www.newegg.com/Product/Product.a ... 6835185056

And here's the 800RPM model:

http://www.newegg.com/Product/Product.a ... 6835185057

I'm getting two of each...

[djkest]

From the unofficial tests that members have been doing, it looks as if these fans are slightly superior to even the Noctua, S-Flex, and Nexus fans! I can't remember who tested that they produced the same airflow at a lower RPM, but anywho... interesting.

Interesting to me because the design is similar to the SilenX fans with the many small blades and the smaller hub, but the silenX fans weren't very good.

In any case, I ordered an 800 rpm model on wednesday, should be here monday. I'm sure it will be the new quietest thing in my system, lol.

[Tzupy]

@Felger Carbon: from the pictures you posted I can't tell if the case has the original exhaust grill, or you cut it out and maybe replaced with a wire grill.
I never mod the cases I use, only get a new case when buiding a new system and that's it, right now I have a P150 (with controlled Nexus fans).
I'd like to know how the Slipstream fans behave when mounted in a case with standard exhaust grill (not cut). Thanks in advance.
From my experience, some fans are noisier when case mounted than in free air, the Antec Tricools surprisingly are not.
PS. are there any 92 mm Slipstreams, or only 120 mm ?

[Felger Carbon]

I'd like to know how the Slipstream fans behave when mounted in a case with standard exhaust grill (not cut). Thanks in advance.
PS. are there any 92 mm Slipstreams, or only 120 mm ?

I guess if you want to know about std. exh grilles you'll have to ask someone who has them. I don't. No 92mm SlipStreams have been announced that I know of.

[NeilBlanchard]

Hello,

I got the two 800RPM models today (the 500RPM units are coming from a different location -- go figure), and they are awesome! Whisper quiet at 12 volts, and only a very mild bearing noise only in certain positions -- I wouldn't be surprised if it went away with a little use. I will try them some more after I move a client's (slightly noisy) computer out of my office.

They also seem to blow a decent amount of air -- subjectively more than a Nexus? Nine blades (of somewhat steeper pitch than typical), small hub, small(ish) outer gap -- in free air, it blows more air than I would have expected from an 800RPM fan. And again, the dominant noise is a "whoosh" with a very mild bearing noise audible only from just inches away from my ear...

[MikeC]

Interesting that this thread should pop up now...

I just got an email from Ron, who seems knowledgeable about fan design. He says...

You may also wish to comment on the strut designs in the fans in the power supplies that you review. I know from the article on fan design on your web site that you understand strut design.

I note that both the Corsair HX520 and HX620 have Adda fans with struts that are not oriented for minimum audible noise. I have alerted Corsair to this problem, but I continue to be amazed that more than half of the fans that I see on the market still have their struts in the wrong orientation. I attach an image of two Scythe fans. The one on the left has excellent strut design: (1) struts emerge from hub in a direction OPPOSITE to rotation, which enables maximum orthogonality with the nearest edges of the passing fan blades; (2) in addition, the struts are slightly curved to improve orthogonality further. Thus, these struts will spread out in time the interaction with the nearest edges of the passing fan blades and thus will minimize the generation of tones. But the struts in the larger fan on the right emerge from the hub in the SAME direction as rotation, and thus tend to interact with the nearest edges of the passing fan blades all at one time, producing tones which are more audible to human beings.

The image on the left he mentions shows one of Scythe's new fans, which have very unique struts:

When the closest blade edges pass by the struts, they meet at nearly right angles, and not the entire blade edge at the same time; rather, the "contact" area is a point that moves. Ron's comments suggest that this means the turbulence noise that's created is spread over time and over a wider range of frequencies, thus causing less tonality, and more broadband sound.

The more conventional design looks like this:

Each blade meets each strut across its entire length virtually in parallel. This means a more concentrated or condensed "contact time".

My comments are just conjectures on Ron's notes. I need to find out more about all this... but does this make sense and correlate to what owners of these new Scythe fans hear?

[Sooty]

*cough* What I said *cough*

[NeilBlanchard]

Hello,

I've mentioned this before, too:

Yate Loons/Nexus do a decent job at this:

but the curved Scythe struts are better still.

Also, the corners of the Scythe fan are open:

[MikeC]

OK, I guess I'm late to the party...

[sea2stars]

So far I'm really impressed with these fans.

However I've only been able to compare how the SlipStreams compare to an Globalwin in the open air; the CPU in my HTPC died and I'm looking at my options for a decent, ie cheap, Core 2 Duo.

[Tzupy]

@MikeC: I agree that the curved struts are reducing the turbulence noise, but they are not 'unique'.
Here you can see them on the Scythe Kama PWM 92 mm fan:
http://www.overclockers.co.uk/showprodu ... subcat=820

[AuraAllan]

The KAZE JYU fans has the same thing going on.
http://www.scythe-usa.com/product/acc/0 ... etail.html

[MikeC]

Tzupy & AuraAllan -- I am aware of those fans... they have something in common in that they are all Scythes. I'm not sure I've seen such designs on other fans. The thing that's totally new to me is the idea that this strut design reduces not turbulence noise, per se, but the tonal aspect of the turbulence noise.

[NeilBlanchard]

Hello,

If the fan with 7 blades and 4 struts that are parallel to the trailing edge produces a "negative" pulse each time a blade passes a strut, this totals 28 pulses per revolution. If it is turning at 1500RPM -- that's 42,000 pulses per minute, or 700Hz. Bearing in mind the concept of "Fletcher and Munson Curves", where we have a much lower threshold of audibility in the midrange:


With the struts at an angle to the trailing edge -- there is almost a constant state of the air flow, since each blades is always passing by one (or two) struts at any given moment. In other words, it would produce a very low frequency -- or almost no noise (from this phenomena) at all...

[MikeC]

If the fan with 7 blades and 4 struts that are parallel to the trailing edge produces a "negative" pulse each time a blade passes a strut, this totals 28 pulses per revolution. If it is turning at 1500RPM -- that's 42,000 pulses per minute, or 700Hz. Bearing in mind the concept of "Fletcher and Munson Curves", where we have a much lower threshold of audibility in the midrange....

With the struts at an angle to the trailing edge -- there is almost a constant state of the air flow, since each blades is always passing by one (or two) struts at any given moment. In other words, it would produce a very low frequency -- or almost no noise (from this phenomena) at all...

Yeah, Neil, you nailed it. That neatly explains Ron's comment about reduced tonality.

[Felger Carbon]

I have some real data on true CFM in my PC case for these SlipStream fans. This is continuing data on the same case I was using at the start of this thread; I'm still using it... with two very minor mods, so far.

I replaced the 500RPM (nominal) fan with an 800RPM fan, and connected the fan to a Spire SP-205 analog fan controller. I don't count this as a change. I'm still monitoring some temperatures and the fan RPM in bios (no hard disk, no XP OS). Wall power is 56W when first turned on, dropping to 55W when the CPU warms up.

Change #1: I added a 3.6" x 1.25" scrap of cardboard to the bottom of the Ninja duct, blocking more air from sneaking into the exhaust fan "under Ninja's skirts". This simple mod improved the CPU temperature a lot.

Change #2: I unplugged the fan in the PSU. The PSU "exhaust" is now an additional air inlet for the exhaust fan, resulting in less backpressure on the exhaust fan. Also significant is the fact that all 55W is being exhausted thru the dremeled-out exhaust fan hole. This makes it possible to monitor the temperature of "the (only) exhaust air" and therefore to calculate the true exhaust CFM, which I have done.

The raw numbers: 578RPM. 1/8" from fan exhaust hub, 38.7dBA; 5" (1/8 meter) 33.7dBA room ambient 26.9dBA. 578RPM chosen to provide same sea-level CFMs as 500RPM would at sea-level. (KF is at 4100' and has about 86.4% of sea level air pressure.)

Raw temps: 18.2C amb, 23.4C exh, 40C CPU, 28C System, 37C PWM; 55W wall power. CFM = K*P/C where K = 1.738 at sea level, P is watts, and C is degrees C. The exhaust temp is 5.2C higher than the CAG intakes, so C = 5.2.

sea level CFM = 1.738*55/5.2 = 18.38CFM at 578RPM in KF, which would be 18.38CFM at sea level at 500RPM. The rated CFM of the SlipStream fans is 24.5 at 500RPM. 18.38/24.5= 75.0%. With two CAGs on my case left panel, the PSU as an additional CAG, and the slight (and I mean slight) air from the front bezel, my high-pitch SlipStream fan is pumping 75% of its free-air CFM. Not bad!
--------------------------------------

I'm going to continue to tune my Ninja duct for lowest CPU temp. Then I'll drop my fan's RPM; 40C CPU temp and 5.2C exhaust rise are both ridiculously low.

Anybody who needs formulas derived, please contact Jaganath, who is my derivation specialist. Also, he's been through that CFM formula before.

[Cerb]

Looks good. I guess when I upgrade, I'll get a little assortment of fans, and see how well they do. If these will report tach around 5-7v, that'd be wonderful.

About the only bad thing appears to be the bearings. It seems to me that fluid bearings would be ideal. But, they would have been more expensive (these fans are cheap!), and might not have been available from the true manufacturer.

[Tzupy]

Madshrimps has a comparison table (airflow and noise) of Slipstream fans versus other good fans:
http://www.madshrimps.be/vbulletin/f22/ ... ans-39458/
At the same rpm, they are a bit noisier than their 7-bladed (I assume) competitors, but they push a lot more air.
A Slipstream at 1,200 rpm pushes a bit more air than S-Flex at 1,600 rpm (really that good?) and is 8 dbA quieter.
I'd like to know how they perform on a Thermalright heatsink (SI-128, Ultra-120, etc).

[Felger Carbon]

Madshrimps has a comparison table (airflow and noise) of Slipstream fans versus other good fans:

I'd like to know how they perform on a Thermalright heatsink (SI-128, Ultra-120, etc).

I believe that's a compilation of manufacturers' published data.

I guess some incredibly brave person will have to spring for $10 or so and give it a try. Somebody who has an SI-128, U-120, etc.

[NeilBlanchard]

Greetings,

[Edit: you're right -- this is probably all manufacturer's data...]

Thanks for the link! The 800RPM model is both quieter than one of the other 800RPM fans, but it pushes the same amount of air as the 1100RPM fan! Seems like a winner to me.

The S-Flex 800RPM and the Scythe "Slip Stream" 500RPM are the other winners, for sure. The faster Slip Streams get a lot noisier -- maybe there is an optimum speed for this design? It would seem the 800RPM model is the best CFM/SPL ratio, while the 500RPM is the outright quietest. The 800RPM S-Flex splits the two Slip Streams...

[Felger Carbon]

The 800RPM S-Flex splits the two Slip Streams...

Minor point: the 800RPM S-Flex requires a pretty high starting voltage. The 500RPM SlipStream doesn't have that problem. This is only a problem if you use fan controllers to drop RPM, though.

[Tzupy]

@Felger Carbon: I'd get one today, but they aren't yet available in my country. I'll probably have to wait a month to replace my Nexus on SI-128 with the 1,200 rpm Slipstream. Since the S-Flex are about 13 euros and NMB 15 euros, I suppose the Slipstream price shouldn't go over 15 euros. I belive the 1,200 rpm one is optimal for me, since right now my Nexus runs at 440 rpm and CPU temp is 24C. But in 3 months I might upgrade to a Q9450 and will need better cooling. Are you sure the Slipstreams undervolt well? I am also considering swapping the fan in my S12E+650 with a Slipstream, if it starts at 4V, even if it's sleeve bearing.

[NeilBlanchard]

Hello,

I think the 800RPM Slip Stream would be a better choice to replace the 1,000RPM Nexus. It blows as much or more at 12v, I think, and it certainly is very quiet.

[Luminair]

Hello,

I think the 800RPM Slip Stream would be a better choice to replace the 1,000RPM Nexus. It blows as much or more at 12v, I think, and it certainly is very quiet.

We're right at the cusp of needing a modern fan roundup

I see a few different use cases for case fans, and I'm not sure which fans are best for which situation. It is a topic worthy of an article.

Obstructed input (case input fan), obstructed output (case output fan or typical heatsink configuration), obstructed input and output (rare heatsink configuration or case input with hard drives blocking output), and finally unobstructed input and output (like the optional middle P180 fan).... which fan is best where? All these new fans from the different manufacturers demands a roundup!

[Cerb]

How do they undervolt, and how do they sound when undervolted? IoW, how well does the 1200 RPM replace the 500 and 800, and how well does the 800 replace the 500?

This series of fans really throws off my controller plans (again). I might be able to use fans at >7v, again...