120mm or 140mm case fans for my build? [UPDATE] Temp Test with both

[ClockHound]

As they say: If it ain't broke...

You haven't drilled out enough pop-rivets...

 

[Alpha0mega]

Oh, I am not planning on removing the HDDs because of their heat. I want to move them away from my system because of their noise. Their constant hum is the noisiest thing in my system by far, which I find a little annoying, especially in the winters, when everything is super-quiet. Any reduction in total energy being dumped into the case is just a secondary advantage.

I only mentioned this at all because of your suggestion of using a smaller HDD cage, to reduce restriction to the air flow, pointing out that my future plans will probably eliminate the HDD cage altogether.

You haven't drilled out enough pop-rivets...

:biggrin::biggrin:

 

[ClockHound]

If you want to reduce hard drive noise, then try the elastic hairband/sewing elastic mounting method. I have a couple WD Blacks suspended in my case and they are the least noisy element. CPU fan is louder.

Here's an older thread on SPCR with some nice pics down the page a bit.

 

[BonzaiDuck]

If you want to reduce hard drive noise, then try the elastic hairband/sewing elastic mounting method. I have a couple WD Blacks suspended in my case and they are the least noisy element. CPU fan is louder.

Here's an older thread on SPCR with some nice pics down the page a bit.

That's a good idea, I'd seen it done with reasonable elegance with brown rubber surgical tubing.

Some cases like my HAF 922's have 3.5" drive trays fitted with rubber noise [vibration] isolators. But it shouldn't be too difficult to rig up something else to deal with the same problem. Of course, SSDs don't make noise. But I insist on having at least one HDD in my builds.

 

[ClockHound]

I've been suspending hard drives for almost a decade. Antec used the elastic mounting for the original Solo. Hasn't seemed to catch on tho. Probably cause it eats precious spec space. It's the one area that almost all case designers have failed. Hard.

I use drugstore hairbands, because they're available cheap in a variety of pretty colors.

The looser the elastic the less mechanical noise is transmitted to the case. Those old hard and most new semi-hard mounts are hideous, turns the case into a resonator. Spindle drives are slow enough without having to actually hear how slow the seeks are.

First thing I look for in a case is removable drive cages. Until we have cheap 2-4TB SSDs I'll be collecting drive cages to display my unsed fans in.

Could be worse, have an ancient Control Data full height 51/4 40Meg drive in the archives. When that thing powered up in the old desktop 386, the desk would shake, the cats would run and I'd think to myself, 40 friggin' Megs! I'll never fill this up.

 

[BonzaiDuck]

I've been suspending hard drives for almost a decade. Antec used the elastic mounting for the original Solo. Hasn't seemed to catch on tho. Probably cause it eats precious spec space. It's the one area that almost all case designers have failed. Hard.

I use drugstore hairbands, because they're available cheap in a variety of pretty colors.

The looser the elastic the less mechanical noise is transmitted to the case. Those old hard and most new semi-hard mounts are hideous, turns the case into a resonator. Spindle drives are slow enough without having to actually hear how slow the seeks are.

First thing I look for in a case is removable drive cages. Until we have cheap 2-4TB SSDs I'll be collecting drive cages to display my unsed fans in.

Could be worse, have an ancient Control Data full height 51/4 40Meg drive in the archives. When that thing powered up in the old desktop 386, the desk would shake, the cats would run and I'd think to myself, 40 friggin' Megs! I'll never fill this up.

Those were the days. Those were the days.

 

[Alpha0mega]

That's a good idea, I'd seen it done with reasonable elegance with brown rubber surgical tubing.

Some cases like my HAF 922's have 3.5" drive trays fitted with rubber noise [vibration] isolators. But it shouldn't be too difficult to rig up something else to deal with the same problem. Of course, SSDs don't make noise. But I insist on having at least one HDD in my builds.

The 530's mounts also have rubber vibration dampeners. Not entirely sure how effective they are, though I can feel next-to-no case vibration when I put my hand on the cage.

The sound I don't like is the HDD's inherent, slightly whiny, hum. While I doubt that will be effected by the rubber tubing method, I will give it a try. What's the harm?

Though, eventually, I still aim to move all mechanical HDDs to a HTPC/NAS. I can't think of a use-case for mechanical drives where they have to be present in the main system itself. The data speed of the GbE connection should be close to the peak sequential performance of even the newer ones. The only thing that I use mechanical for nowadays is media and long-term storage, neither that are particularly sensitive to performance differences.

 

[ClockHound]

Maybe the 530's mounts are lossier than most, but so far haven't heard any of that type that can decouple as well as a floppy elastic hairband.

If you use surgical tubing be careful with tension. Ideally you want the 'suspension' to be tuned as low as possible < 10 hz. Need to see some serious droop from the suspension - at least an inch when you let gravity have its way. The lower the frequency of the suspension the better the decoupling works.

 

[BonzaiDuck]

NAS . .. that's OK, I'd think. I've got a WHS box. It has files which are accessed directly by this or that client system, some specific to user. But I have my "main rig" functioning with the low-level background process of feeding the AVR/HDTV with MC Live TV, and the DVR captures are made on the only HDD in the system. Eventually, I choose to transfer some of those DVR captures to the server, and erase others.

The server duplicates "serious" stuff in the drive pool -- not the movies. This is then backed up (without the duplication) on a folder and file basis to an HDD not part of the pool. The client backups of OS boot/system disks are made nightly, similarly duplicated, and backed up on a discretionary basis to the same type of hot-swap HDD outside the pool.

I think I've shifted from having a RAID5 4xHDD on the "flagship" workstation toward configuring clients now with the fewest number of HDDs, and SSDs with OS/boot/system and as many programs as possible.

The CoolerMaster 3.5" plastic drive racks with the little rubber isolators seem to work fine, and the server has the same HAF 922 as two of the workstations. But the server is the noisiest of all three, and I think it is mostly because of the 5.25" hotswap bays-- even though they're built with "cushions" or shock-absorbers.

Ain't it a b****, though? The worst of it: this or that hot-swap caddy has fans that buzz. I've decided, since the case is mildly pressurized, I'm going to unplug those fans in any caddy that shows this noise. And I'll bet the drives in such caddies will show the same 75F to 80F temperatures as those with the fans connected.

 

[Alpha0mega]

Ok, so I finally received the Corsair SP120 LEDs yesterday. Took longer than I had hoped, but I ran some tests last night. As promised, here are the results.

TEST PROCEDURE
I have tried to control for as many variables as possible, without having access to any specialized equipment.

• The tests were done late at night, so as to minimize possible temperature variations of day progression.
• Each test was a 10 minute simultaneous run of Unigine Heaven and Valley. I have games that push temps somewhat higher than these, but, being games, they aren&#8217;t as consistent.
• After each test, I had a 10 minute idle interval, to allow dissipation of built up heat. During this time the fans were all at max. I also opened the door to my room and turned on the ceiling fan to full, to equalize the ambient temps and rid the room of built up heat.
• In all test the starting temps were the same: 28C for the top GPU, 26C for the bottom. CPU at 31C
• Common to all test: 1 x 200mm front intake, 1 x 140mm rear intake. Both NZXT fans, which came as default in the Phantom 530.
• All tests at 12v
• Ambient temps estimated to be around 18C (I don&#8217;t have an ambient temp sensor).
• GPU temp limit set to 91C in MSI Afterburner (highest possible).
• Temps were monitored and logged using MSI AfterBurner and GPU-Z

Test 0 - an unrecorded run, to ensure that the first test would have had the same conditions as the subsequent ones.

Test 1 &#8211; 1 x AF140 QE, blowing air between the SLI cards lengthwise. *
Top Card Max: 80C
Bottom Card Max: 73C
Top Card Average: 72.9C

Test 2 &#8211; 1 x SP120 LED, blowing air between the SLI cards lengthwise.
Top Card Max: 76C **
Bottom Card Max: 74C
Top Card Average: 69C

Test 3 &#8211; 1 x SP120 LED, blowing air between the SLI cards lengthwise, 1 x SP120 LED bottom intake
Top Card Max: 74C **
Bottom Card Max: 72C
Top Card Average: 67C

Test 4 &#8211; 1 x SP120 LED, blowing air between the SLI cards lengthwise, 1 x SP120 LED bottom intake, 1 x AF140 QE top exhaust
Top Card Max: 74C **
Bottom Card Max: 71C
Top Card Average: 68C


* I did not do a test without a fan blowing between the GPUs this time. I had done it earlier, when I had just got the AF140 QE. Those numbers aren&#8217;t directly comparable to this run, since the weather was warmer. However, during that time, the temp reduction between no-card and AF140 was 3C for both cards.

** In all cases where the GPU cooler was the SP120, the cards touched their peak temps later. E.g. in Test 1, the top card hit 80C six minutes in, and stayed there. In the later tests, the peak temps were hit towards the end of the run, and kept dropping down on 1C below the peak, then rising back, then dropping&#8230;
Additionally, I am unsure why the temps didn&#8217;t rise beyond 80C in the first test. I have seen them go higher while playing games (particularly if I forget to turn up the fan controller).


SOUND
At 7v and below, the SP120 LED sounds indistinguishable from the AF140 QE. At 12v, the sound caused by its air flow is a little higher with the case open and the ear close to the fans, but nothing that&#8217;s really noticeable. With the fan controller at 12v max, now, with two additional SP120 LEDs, my system is of course a bit louder than it was before. Even the min setting of the fan controller is slightly louder.

Mounting the SP120 as bottom intake introduced some really bad case vibrations at 12v (but not at lower voltages). Closing the case properly reduced them, and pulling out the bottom dust filter a millimeter removed their audible component entirely, though touching the bottom-side of the case confirms there is still a slight vibration. The vibrations aren&#8217;t caused by the filter, since totally removing it does nothing. I wish Corsair hadn&#8217;t cheaped out by not including the rubber grommets that it included in the older, non-LED fans.
Mounting the SP120 LED inside for the GPUs caused no vibrations however, audible or tactile.


OTHER THOUGHTS
The SP120 LED lighting (I have the white ones) is bright and clean, and looks quite good, if you are into that sort of thing. I personally don&#8217;t like unnecessary noise, visual noise included, so I will have to snip the LED wires. I wish Corsair had put in a switch like some other brands do, so I wouldn&#8217;t have to mutilate the fans.

The AF140 that I moved to the top exhaust had to be fixed in place using wire ties, since there was no way I could use screws without dismantling the system. Still, it&#8217;s quite firmly in place, without any vibrations.

I might try using wire ties on the bottom fan too, to check if it makes any difference in the vibrations. I will also cut some foam and put it in.


CLOSING
I am quite satisfied with this purchase. I was skeptical if it would make much difference, but replacing the AF140 QE with the SP120 LED dropped the temps by 4-5C, without any noticeable difference in the noise. Adding another as the bottom intake dropped the temps by around 2C, increasing the noise (both max and min) only slightly. I think I may use a LNA to limit the bottom fan to 7v, or perhaps start disconnecting it altogether during the winters.

 

[BonzaiDuck]

Are those the "short" model GTX 970 cards, perhaps with a single fan each?

 

[Alpha0mega]

They are shorter than most, around 8", but have two fans.

 

[BonzaiDuck]

They are shorter than most, around 8", but have two fans.

Ah! But two fans! Two! I guess my point in anticipation of your answer is as follows. IF you're prepared for a day's tedium to get optimal air-cooling for all the motherboard hot-spots (VRMs etc.) and the graphics cards, that's a bigger view. But with the Noctua cooler, you could simply add some duct-plates to the two cards so that the two pairs of fans work in tandem, pulling air up and over the top of each card and down the back sides -- eventually drawn into the intake of the Noctua fins. Then you could eliminate the blower fan behind your drive cage. I'm optimistic that this would be more effective in cooling the cards.

But it's extra trouble. Some Xacto blades, $10 to $15 in foam art-board and glue.

Talk about the disincentive of tedium. I've got plans on the drawing-board to turn my motherboard into a "Sabertooth" for cooling. The stencil is almost ready for printing in 1:1 scale to ATX. But it's been a couple months, and I'm not all that eager to embrace getting finished with it.

Also useful for that type of graphics card ducting project: a 92mm x 15mm Zalman OP-1 fan, if there's room for a "puller" near the Noctua.

Just some thoughts . . . that's all . . .

 

[Alpha0mega]

pulling air up and over the top of each card and down the back sides -- eventually drawn into the intake of the Noctua fins

One problem with this is that the NH-D15's design won't allow it. I think you are going by your D14's design, whose fin assembly was "open".

vs

Another problem, which might just be me misunderstanding what you mean, is that the fans of the cards don't really move the air up or around much. They suck it in to cool the card components, and exhaust it out from around the card body. Which means that the same air keeps getting recirculated, getting warmer and warmer. This is doubly true for the top card in the SLI stack. Not sure what kind of ducting would resolve that.

The whole aim of my putting the fan blowing between the cards was to remove the hot air building up, and if possible, to push it out of the case on its own through the slits in the Phantom 530's PCI-e slot guards (if you look at the picture in my first post you will notice that the guards aren't solid).

The AF140 helped, reducing the card temps by 3-4C. However, even though it was completely free from all restrictions on intake and exhaust, plus being fed in parallel by the front 200mm intake, plus having a higher CFM rating, it just couldn't push the air out of the case through the guard slits. Putting my hand there, even at 12v, I could barely feel a slight breeze. The SP120, on the other hand, is amazing. Even with the fan controller at the minimum, the airflow I feel matches (perhaps even slightly exceeds) that of the 12v AF140. At 7v and 12v, there is just no contest. I feel a strong wind, that keeps ratcheting up as the voltage keeps going up, and the numbers from my temperature test back it up.

A 4-5C improvement from just switching out the AF140 for the SP120 LED is quite impressive, even more so since there is essentially no difference in the noise. Adding another SP120 as the bottom intake dropped the temps by another 2-3C.

At this point I am not even sure what purpose the AF series could possibly serve, where the SP ones wouldn't be better. I see no reason for anyone to pick the former over the latter, even where static pressure isn't really a necessity. The AF fans are supposed to shine where there are no restrictions around them, and there weren't any in the my AF140 deployment. Still the SP120 solidly outperformed it, at no additional noise cost, and lower monitory cost. The AF simply can't keep up with the strength of the SP's airflow.

Keep in mind that this is coming from someone who was, prior to this, a bit of a skeptic on high static pressure where radiators weren't concerned. I always assumed that SP fans would be pointless and/or noisy when not being used on radiators and that AF would be better as case fans. Well, not anymore.

 

[BonzaiDuck]

My mistake. You're correct about the closed sides of the D15, and its size would probably pose an obstruction.

I'm still keeping my mind open to the idea of an SLI duct-box, though. By mid-week, I'll have two Maxwell cards in my system, and double cards will likely mean temperatures above 55C for gaming or above 62C for Heaven or KomBustor.

It's a lot of trouble and tedium -- that's a dead certainty.