Quiet Silverstone SG03 build

Show off your quiet rig.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Quiet Silverstone SG03 build

Post by edh » Tue Apr 15, 2008 1:37 pm

This is my first build that is intended to be quiet from the start. My previous Shuttle SN45G got a lot quieter over it's life but this time I didn't want to end up replacing half of it just to get it quieter.

The SG03 is an unusual Micro ATX case. There are various reviews of it up, here's the first on Google:
http://www.phoronix.com/vr.php?view=9554

Components:
Intel Core 2 Duo E8200
Asus P5E-VM HDMI
2x1Gb PC6400 Kingston (temporary)
Leadtek 9600GT
Samsung HD501J 500Gb HDD
LG 18x DVD-RW
Enermax Modu82+ 425W
Silverstone NT-06 Lite cooler
Arctic Cooling Accelero S2
2x Akasa AK-191SM 120mm fans
PaQ PWM -> DC convertor

Here's the internals sans PSU:
Image

Having a modular PSU really does help in this case just to be able to get the thing in and out. Hooking up the PSU fan sensor cable to the header underneath the back of the CPU heatsink is not so easy however. The case is rather lacking in PSU mounting holes so I had to lightly file one edge to be able to get a screw in up against it. 4 screws is much more stable than three.

The PWM to DC fan convertor can be seen at the top right hand corner of the motherboard. I have both fans running 800-850rpm although I can slow them down or speed them up independently over a range of 250-1300rpm. I know the choice of Akasa fans has been questioned here before but I really do like these. They disappear underneath the sound of the quiet hard drive idling and can not be recognised below around 750rpm, despite being right at the front of the case.

Here are some detailed fan speed/temperature testing results obtained via Speedfan with manual fan control. These are all idle temps, it's to compare fan speeds, not overall cooling efficiency. Although PSU fan speed can not be controlled, I noticed that with the front fans running under 300rpm there wasn't sufficient airflow over the CPU and PSU so the PSU fan felt the need to rev up, peaking as high as 630rpm. It was nice to see that it slowed back down to 480rpm after airflow had increased. Although 450rpm is meant to be standard this one revs a little higher due to the increased air resistance that the CPU cooler introduces. This is aerodynamic, not thermal.

Table 1

Code: Select all

Top fan rpm    Lower fan rpm   CPU   GPU   System   HD0
0              0               42    50    43       31
289            0               42    51    36       33   Slowest starting speed
291            276             41    51    35       32   Slowest starting speed
474            285             39    54    30       32
469            464             38    49    29       30   Similar to PSU fan speed
692            479             35    47    25       29
703            703             34    40    24       27   Highest totally inaudible
811            781             33    38    22       26
879            844             32    37    22       25   Defaults
1172           1278            30    31    19       22   Max
Room temperature is pretty low, probably about 18C. System temp is very low because the sensor is somewhere near the front, right in the way of the fans. All temperatures were left until stable. The GPU temp is slightly confusing as it seems to go up for a while.

Overall the system is so quiet that sometimes I've come into my living room and been surprised to find that it's on, with my old Shuttle my sensitive hearing means I could have heard it down the corridor if the door was open.

The main noise now is from the HDD at idle and it's seeks are quite noticeable when in front of it. The hard drive mounting isn't that bad but doesn't cut out much low vibration. I'm considering converting the HDD tray into a sandwich mount. There's about 7.5mm on either side of the disk and a burly aluminium tray around it to use as the frame.

The DVD made a bit of vibration at high speed so I've damped it down with a bit of adhesive backed felt. The drive is only used for those few Windows games I play that still require a disk in the drive so a bit of nosie doesn't bother me. Resonating case noise is bad though, so I wanted to stop it.

The case came with a rather nasty overbright white power LED. I've disconnected this now which might save a Watt and looks much nicer. power consumption is ~76-80W idle and the highest I've seen under load is 139W.

Here is the system running with a Keytronic KT2001 keyboard, Logitech MX400 and Samsung SM245B:
Image

Edit (2008-04-17):
The S2 has now been replaced with an S1 rev2. This has dropped the GPU idle temp by around 5C and the load temp by around 9C.
Here's the discussion of it:
viewtopic.php?p=407862#407862

Todo (updated 2010-02-07):
- Further improve hard drive mounting.
- Better cable management. Some cables are longer than they need to be.
- 2D/3D clocks on the graphics card. I've now read that this is not possible in the BIOS, I may resort to software.
- Possibly modify the 3.5" 'external' bay. Having the blanking plate attached to the side bracket (read an SG03 review to see what I mean) makes installing and removing the bracket much harder.
- Replace case fans with Noctua S12B
- There must surely be a way of adapting the propreitary fan voltage output from the Modu82+ to enable use of a standard fan.
- Experiment with ducting

Comments?
Last edited by edh on Sun Feb 07, 2010 3:14 am, edited 16 times in total.

bonestonne
Posts: 1839
Joined: Mon Feb 05, 2007 2:10 pm
Location: Northern New Jersey
Contact:

Post by bonestonne » Sat Apr 19, 2008 12:40 pm

:roll:
i swapped my ECS 9600GT's stock cooler with an Accelero S1 Rev 2, and the Idle temps dropped ~8-9C after burning in.

I had also removed all the plastic cr4p from the heatsink, the two plastic pieces on the edges, and the plastic cap over the base.

the installation had gone well, but i really had to be careful with the core because i didn't trust any stray thermal junk from the stock cooler.

i wish there were 3 more VGA heatsinks to stick onto the VRMs, i don't trust them, even though stability is rock solid.

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Sat Apr 19, 2008 1:26 pm

bonestonne wrote:i wish there were 3 more VGA heatsinks to stick onto the VRMs, i don't trust them, even though stability is rock solid.
I'm thinking of taking a hacksaw to the stock 9600GT cooler to cut off the VRM cooler part of it. It's got a pair of screwholes and is made to measure so will fit on the VRM's nicely. Don't think anyone else has done this but it would potential be doable on 8800GT, G92 8800 GTS and 9800GTX with the S1. This idea won't help you if your card came with an S2 originally of course. There are quite a few cards which differ from the reference cooler and don't come with VRM cooling or even memory cooling so I don't think it's too much to be worried about in terms of reliability. It may make a tiny power difference though as the VRM's handle a lot of power and heat may effect their efficiency.

I was intending to retest temperatures soon both to see how the GPU temp has changed with the paste curing and also to see how the S1 performs at different airflow settings as I had with the S2.

Mats
Posts: 3044
Joined: Fri Dec 26, 2003 6:54 am
Location: Sweden

Post by Mats » Sat Apr 19, 2008 1:43 pm

Are you using Real Temp for the CPU?
Other programs seems to be reporting the wrong temp.

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Sat Apr 19, 2008 2:19 pm

Mats wrote:Are you using Real Temp for the CPU?
Other programs seems to be reporting the wrong temp.
I am not using Real Temp but I'm not bothered by that either as it may well give the wrong temperatures.

Initially the BIOS was reporting a CPU temp sometimes below 0C. That was obviously wrong but was because the initial BIOS release did not support Wolfdale. I have updated the BIOS to the latest which specifically adds full support for Wolfdale and this reports CPU temps anywhere from 30C upwards in the BIOS which matches what the ASUS PC Probe Utility says ( a horrid program) and also matches Speedfan and GKrellm under Linux.

One perculiarity with the CPU temp sensor on this motherboard is that it seems unable to give readings below 30C. Even on a really cold boot it'll still come on at 30C and stay at that reading even when other components gain 10C. During cold running in Speedfan it's really obvious by poltting the temperatures and seeing a big flat line at 30C except when under load when there's lots of fluctuations.

Reading core temperatures direct from the CPU is different and not something I trust right now with such a new CPU design. Coretemp reports core temperatures 13C above the motherboard CPU sensor and recent Speedfan betas that add Wolfdale support report core temps 8C above the motherboard sensor. I'm pretty confident that both Coretemp and Speedfan are not reading core temperatures correctly. I have little reason to put any trust in any program that claims to read core temperatures at the moment.

Just by touching parts of the motherboard and CPU cooler (by comparison with other measured components) I concur with the CPU temp being reported by the mobo and believe the other core temp readings to be incorrect.

Edit:
Just rebooted to Windows and tried Realtemp. It gives the core temps to be 3C above the CPU temperature reported by the motherboard. I'm much more inclined to believe this temperature as the TjMax is reported correctly and having read into it the premise seems alright.

The core temperatures can vary by a degree or two from the temperature on the surface so the 3C difference may still mean there is an error in calibration in one component or another. This could be either the CPU core calibration or the motherboard. Doesn't seem much point in calibrating that out as they are not measuring the same point anyway.

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Mon Apr 21, 2008 10:30 am

I have now cut a VRM heatsink from the original stock cooler. This has all of the right points to cool the VRM's. It was running stable anyway and I'd seen no issues but this is partly a personal comfort thing and also to show other people who aren't so sure about using the Accelero's that VRM cooling is possible if you improvise:
Image
I had to completely dismantle the stock cooler to get a hacksaw in where the fan normally sits. It only took a couple of minutes as it's only soft aluminium and with nothing but a hacksaw the edges didn't come out to bad. If you have a file to hand, better still.

This kind of mod should also work on 8800GT's, G92 GTS's and 9800GTX's.

Installation was tricky with the S1 already installed. I would suggest to anyone else doing this that it may be easier to install the heatsink along with the memory heatsinks prior to installing the main cooler. As it was I had to unclip one of the edge clips just to get a few millimetres more room. The VRM sink I then slid underneath having applied some thermal paste. As the original cooler has two screw holes in this part of it, it screwed on nicely with a pair of the original screw. There is no risk of this falling off, unlike the memory sinks.

This is what the card now looks like. The VRM heatsink can be seen just below the right hand clip running across the card:
Image

Having reassembled, the system works as before so no harm done. Maybe the VRM's are healthier, there is no way to measure this. If anything the GPU temp might have changed by -1C but that'll only be because I further tidied the cables around it, maybe also a little bit of movement in checking the screws were tight.

One more thought I had when the card was out. Someone previously asked if an Accelero Sx card could benefit from a backside cooler. S2 not powerful enough for you? S1 not powerful enough for you? Then why not use both:
Image

That's a flipped S2 sitting on top of my card with the S1 attached underneath. I have not actually attached the S2 and don't intend to, it's just that it would fit. It would be a bugger to fit and you'd need some kind of soft thermal interface so as to not damage the back of the card. Thermal improvement would be questionable. You could not do this with dual S1's as then the top one would have to hang out of the back due to the inverted arrangement. Just an idea for anyone with some spare time, a spare S2 and a very hot card... with a big gap above their graphics card... and tiny fingers for assembling the complicated arrangement.

Ash
Posts: 116
Joined: Tue Apr 15, 2008 12:36 pm
Location: London

Post by Ash » Mon Apr 21, 2008 12:23 pm

correct me if im wrong, but your cpu hs cooler is the enzotech extreme-x right?

i am looking into getting a new cpu cooler, and want to get either the thermalright 120 extreme or the encotech extreme, which is the only other one i could find that can better temps than the 120 extreme and i dont really like the scythe hs's.

only problem im in uk, and i dont think i can get it here from what i have seen so far.

my new build is mainly to get rid of as much noise as i can, so i wanted to run the cpu cooler without the fan, as you have done. tests that i saw used the stock fan at 1300 and 2000rpm and the noise was too loud, but i read that the airflow was around 90. so im sure i can replace the fan with a quieter one if needs be but not with as much airflow.

as i mentioned temp results were better but the fan speeds where higher aswell.

how are you finding the cooler without the fan, would you recommend it?
i also read somewhere that when installed it wasnt very secure?

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Mon Apr 21, 2008 2:42 pm

Ash wrote:correct me if im wrong, but your cpu hs cooler is the enzotech extreme-x right?
Well actually it's not. As shown in this thread and in my signature it's a Silverstone NT-06 Lite. It's the one recommended for use in the Sugo series of cases.

hozer2k
Posts: 35
Joined: Fri Nov 24, 2006 2:51 pm

Post by hozer2k » Mon Apr 21, 2008 9:43 pm

edh: I have a very similar setup as you and my hard drive is the loudest part as well, especially the seeks. I don't have the same HD as you, but was actually about that buy that one and try some vibration mounting solutions in the 5.25" bay. Have you found anything to get the seek noise down?

Ash
Posts: 116
Joined: Tue Apr 15, 2008 12:36 pm
Location: London

Post by Ash » Tue Apr 22, 2008 1:22 am

apologies, i was looking through your spec before but i didnt recognize the silverstone as your cpu cooler.

it looks identical to the enzotech extreme, i think the enzo has 4 pipes instead of three.

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Re: Quiet Silverstone SG03 build

Post by edh » Tue Apr 22, 2008 9:50 am

@hozer2k: I've not done anything about the hard disk yet. Initially I was quite dissappointed with it as it seemed louder than I expected. However, having run it against the 160Gb Seagate 7200.7 in my old machine, it's actually very quiet, it's just because the machine is so quiet it's more noticeable. The seek noises are particularly bad not because they are loud but because the spin noises are relatively quiet. There is a bit of a low hum that occasionally comes through when cold. Perhaps it's a thermal behaviour thing.

Do you have an SG03? I don't want to go for a 5.25" enclosure as my DVD-RW goes in it quite nicely and I don't want an external one. The HDD bays in the SG03 are 115mm wide, giving 7.5mm on either side for silencing material. I think my first thing to try will be a set of silicone grommet mounds going into the drive from below which will mean drilling 4 holes in the HDD tray. This will give a mounting like an NSK3480 has which works quite well for a colleague of mine.

I have now repeated the idle measurements in Table 1 but with the Accelero S1 rev 2 replacing the S2, the custom VRM heatsink and minor cable reworking. This time I also recorded the PSU fan speeds, these are not manually set but controlled by the PSU. The difference in GPU temp is also shown:

Table 2

Code: Select all

Top fan RPM    Lower fan RPM   PSU fan RPM  CPU   GPU   GPU diff System   HD0
0              0               608          42    44    -6       44       33
305            0               628          42    46    -5       36       35   Slowest starting speed
293            293             611          42    45    -6       35       35   Slowest starting speed
485            291             554          41    48    -6       30       32
501            481             493          38    42    -7       28       30   Similar to PSU fan speed
715            465             493          36    40    -7       24       29
710            721             495          34    44    -6       24       28   Highest totally inaudible
816            773             495          34    33    -5       23       27
875            829             495          33    32    -5       22       26   Defaults
1220           1298            502          30    26    -5       20       23   Max
The S1 gives 5-7C drop in temperature when idle versus the S2. There is quite a clear trend that the extra heatsink size helps most when the lower front fan runs ~500rpm. The GPU temps also seem to rise when the top fan is on, possibly because it blows some hotter air down.

Load CPU and GPU temps to come when I have a big chunk of time.

meansizzler
Posts: 43
Joined: Thu Jul 12, 2007 3:44 am

Post by meansizzler » Fri Apr 25, 2008 5:08 pm

I have a similar setup toyours, but if you would believe it much quieter.., would like to know load temp of CPU please, as i'm using E6420 and looking otupgrade to it, I got an S2 on my 8800GT, and you can fit the floppy bracket it there with card reader or floppy installed withthe S2, jsut removes the black plastic that clips onto the side of the S2 cooler.....

Let me knwo if you want to know myfull specs

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Sat Apr 26, 2008 12:53 am

meansizzler wrote:I have a similar setup toyours, but if you would believe it much quieter..,
If it's this system here then it's debateable either way. My PSU should be quieter than yours and the fans are going to be pretty similar. Where you do win though would be HDD noise. That's the big noise issue left for me and what I'm planning to do next.
meansizzler wrote:would like to know load temp of CPU please,
Just what I was about to post, all with various fan speeds. This was done with Orthos running small FFT's. The CPU did actually hit 58C at maximum but the increase in PSU fan speed soon caught up with it and brought it down to 54C. The two fans were varied together this time as the GPU isn't doing anything:

Table 3: Orthos 100% CPU

Code: Select all

Top fan RPM    Lower fan RPM   PSU fan RPM  CPU   GPU System   HD0
0              0               804          54    44  43       28
293            293             892          54    45  37       32
480            467             804          53    43  32       30
715            715             648          50    35  25       28
877            834             565          50    35  25       28      Defaults
1223           1282            503          47    29  24       25
So the simple answer to your question is that the CPU load temp under normal fan settings is 50C. You can also see that the PSU fan revs up with lower fan airflows through the system. With the two fans being some distance from the main heat sources there may be a problem at lower fan speeds of the airflow not being able to penetrate far enough into the case. During this test the power consumption of the system was ~110W. This roughly matches with the 27.7W E8200 power consumption that has been mentioned before as there will be additional power loss in the VRM's and the PSU versus the idle power of ~78W.

I've also done some tests with Orthos and ATiTool together to test CPU and GPU at the same time. Running just ATiTool on it's own isn't very useful as it uses significantly more than 0 CPU.

Table 4: Orthos + ATiTool

Code: Select all

Top fan RPM    Lower fan RPM   PSU fan RPM  CPU   GPU System   HD0
0              0               1110         54    70  46       35
318            0               1140         54    71  38       36
317            310             1110         54    74  34       36
495            486             1082         54    71  32       35
727            485             959          51    75  29       35
731            730             827          50    62  28       32
827            787             753          49    60  27       31
890            861             709          48    58  26       30   Defaults
1215           1318            502          45    46  24       27
The increased power load causes the PSU fan to rev up much more. The listed top speed is 1000rpm but the airflow restriction caused by the CPU cooler may be raising it over 1100rpm. It's not that quiet at that speed so not good to run the case fans at those load speeds. It's also easy to see that the PSU fan is regulating the CPU temperature as before as it doesn't rise any higher than 54C.

The GPU temp is seen to go up when the top fan is speeded up (at low speeds at least), this effect is really obvious in Speedfan and I can only guess that it's because some hot air is being blown down onto the graphics card. It still only peaked at 75C which is pretty amazing with no artefacts recorded.

I also monitored the power consumption in this test and at the lower fan speeds it came to 152W. When the fans were speeded up the power consumption was only 144W. This must be because the PSU would not be running so hot (seen by the different PSU fan speed) and therefore running more efficiently. 8W just from temperature.
meansizzler wrote:I got an S2 on my 8800GT, and you can fit the floppy bracket it there with card reader or floppy installed withthe S2, jsut removes the black plastic that clips onto the side of the S2 cooler.....
I think you've not understood my problem with the blanking plate. It does fit past the graphics card OK and always has done, I run it in place as it should help guide the air on to it. I don't want to put a drive in it either. The problenm is that the blanking plate attached to the front makes fitting it much more difficult. It would be much easier if the blanking plate was fixed to the front panel, not the bracket. This could be fixed by getting a basic clip in blanking plate.

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Sun Apr 27, 2008 2:46 am

Having read the methodology discussed here I decided to have a go at some frequency analysis. In the same way a mobile phone is the only digital camera I had to hand, a mobile phone is the only microphone I had to hand so the recordings had to be converted from .amr to .wav for analysis and the quality isn't so good. I recorded 20s of sound at 1m which was analysed in Audacity. I would suggest not using this data for comparison anyway as your equipment will differ.

Image

One of the clearest things that comes out is the peak highlighted (and partly obscured) by the bar at 120Hz. 120Hz x 60s = 7200rpm so this noise is probably from the hard drive. That's the only really obvious thing that comes out of this recording. There also might be an 8Hz tone from the PSU fan and the main fans should produce something at 14Hz, in addition to a broad band of higher frequencies. The signal levels shown are not the same as the sound levels, there's probably about 90dB of amplification required.

Can anyone else see anything in these spectra? This probably isn't something many people have done but it would interesting to know if other people have seen frequencies which can be matched to particular components.

I'll try to get hold of a proper microphone and do some better recordings to analyse. What I can then do is compare the spectra after different modifications to the system, for instance with better hard drive mounting. It would also be interesting to take different measurements during 2D and 3D work by the graphics card to find the high pitch whines some users find with their cards.

D-Three
*Lifetime Patron*
Posts: 6
Joined: Tue May 06, 2008 1:32 pm
Location: Belgium

Re: Quiet Silverstone SG03 build

Post by D-Three » Fri May 09, 2008 1:24 pm

edh wrote:- Redesign hard drive damping, possibly with a sandwich, maybe only grommets. Whatever, it's got to sit in the same tray arrangement.
I'm very interested to see how you will do this. I've replaced my SG02 with a SG03 about six months ago. Although my current setup is much cooler I'm not very satisfied with the noise it produces and I plan to rebuild this setup later this year. I've silenced my HD (T2K250) with a tool from Hitachi; the drive is noticeably more silent now but also noticeably slower.

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Re: Quiet Silverstone SG03 build

Post by edh » Sat May 10, 2008 7:49 am

D-Three wrote:I'm very interested to see how you will do this.
This is my plan:
Image
What this shows is a cross section through one of the hard drive bays as if viewed from the side of the case. Currently the disk is clipped in to the sides of the tray by plastic clips. Just by touching the drive you can tell that there is some vibration going on. What is shown in the picture is that I'll drill a set of 4 holes through the underside of the tray that match up with the holes in the underside of the disk. A set of screws can then go through some soft gromits into the underside of the disk, holding the disk in place like in an Antec NSK3480 which works quite well for noise. Beyond that foam can be packed into the 7-8mm gaps around the sides of the disk.

I just need to find some suitable screws and gromits.

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Mon May 26, 2008 5:01 am

Having run the system for about 3 months, the temperatures inside have been rising over time, I thought this would mostly be due to the warmer temperatures now it's coming through to summer, however, a lot of it was from dust building up on the dust filter. I'd only vaguely cleaned it before but today I took the front grille off to vacuum out the inside. This can be done without having to turn the machine off. CPU and GPU temps both dropped by 3C from doing this. Shows that the dust filters do pick up a lot which does restrict airflow. The inside seems clean which is good. I would hope that the dust build up on the filter does less to negatively effect cooling than if the dust could build up on the heatsinks. It's still much easier to clean though.

maalitehdas
Posts: 166
Joined: Tue May 27, 2008 12:04 am
Location: Finland

Post by maalitehdas » Tue May 27, 2008 1:40 am

About frequencies (you might already know this) is important to notify those double levels (120Hz, 240Hz, 480Hz, 960Hz) which show in your map too and are all caused by HD.

What is important to check that the whole case is not peak-resonating at any of those frequencies. You can check it with a speaker inside the case submitting different frequencies, and hear when the sound is audible louder. It can also be calculated with the case sizes, knowing material strength and density etc...

I've been hoping to see some "freq.values" in case reviews (i mean the case's own freq.lvl) - some cases are probably worse with 7200rpm HD:s, and some others might have difficulties with 10000 rpm HD:s

To those who didn't quite get what i mean: You all have probably tried this yourself - make a noice in a small room and notify that your own voice sounds louder in certain heights - resonating from the walls at the rooms own frequency level :)or one of its doubles

KyonCoraeL
Posts: 45
Joined: Sun Jun 01, 2008 6:40 pm
Location: Vancouver BC, Canada

Post by KyonCoraeL » Sun Jun 01, 2008 8:17 pm

Have you tried flipping your supply upside down so the air gets taken from outside of the case?

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Mon Jun 02, 2008 12:32 am

KyonCoraeL wrote:Have you tried flipping your supply upside down so the air gets taken from outside of the case?
No, I have not tried this. As the PSU fan does cools the CPU, I don't think it would be such a good idea. It would also mean having no clear airflow path through the entire case.

KyonCoraeL
Posts: 45
Joined: Sun Jun 01, 2008 6:40 pm
Location: Vancouver BC, Canada

Post by KyonCoraeL » Mon Jun 02, 2008 6:53 pm

makes sense, afterall the hottest component is your GPU not the CPU.

As for your hard Drive situation, have u considered dremmelling the metal between the two hard drives so you could make a silent suspension setup?

kinda like a hard drive suspension bridge.

I was thinking about this case, how much room do you have approximately for the heatsink?

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Tue Jun 03, 2008 11:49 am

KyonCoraeL wrote:have u considered dremmelling the metal between the two hard drives so you could make a silent suspension setup
I want to keep the hard drive cages more or less as they are so that I have the spare bay for future use. I'm also wary about a suspension system if the system has to be moved.
KyonCoraeL wrote:I was thinking about this case, how much room do you have approximately for the heatsink?
The official clearance from the mobo to the PSU is 78mm although it's actually around 80mm. This limits the number of 3rd party coolers that'll fit. Forget about the big tower coolers unless you can come up with an alternative, non-ATX PSU to put somewhere else. The Silverstone NT-06 is recommended and I've had no problems with it. I have also seen a guy who bent a Thermalright SI-128 slightly so that it got beneath the PSU. Debateable if that is really a better solution. Lateral and longitudinal clearance depends on mobo.

akromatic
Posts: 34
Joined: Wed Jul 30, 2008 11:56 pm
Location: australia

Post by akromatic » Thu Jul 31, 2008 12:00 am

hi, im wondering when you use the accelero in the SG03 does the side brace still fits and can i still use the external 3.5" bay to put a hard drive or a fan controller?

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Sat Aug 02, 2008 7:47 am

akromatic wrote:hi, im wondering when you use the accelero in the SG03 does the side brace still fits
The side brace fits in fine. It helps the cooling of the graphics card just by keeping the air from the front channelled down a bit more. This also shows that air guidance would improve things.

The biggest problem with the side brace is cable space. As the 3.5" drive cover is attached to the side brace, you can't put any cables in the gap immediately behind the 3.5" opening as the brace will then not go in place. You really have to have one of these cases to work out the cabling.
akromatic wrote:and can i still use the external 3.5" bay to put a hard drive or a fan controller?
Depends on the device. The plastic outer cover of the S1 does protude laterally by about 5mm into this space so might prevent fitment of an HDD or FDD but shouldn't stop a fan controller. The plastic cover can be removed which would give you a few more mm and might solve the problem, if not the Accelero could be bent up around the outer edge of the graphics card.

Cable routing would suffer in such a setup and I wouldn't think of that 3.5" as being a quiet place for an HDD.

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Sat Aug 23, 2008 5:46 am

I've made a number of updates to the system now.

Some time ago I updated the BIOS again and the CPU temperatures have changed again. Up by 8C I think as it put the CPU temps inline with the Speedfan core temps. Looks like Wolfdale support wasn't complete in the previous version. Still not sure about the temp reported, feels quite a bit cooler than that. This update has also changed how fans rev down after boot. They stay at there full speed perhaps 10 seconds longer. There's probably some logical reason for doing this. The BIOS update has also stopped the system from hanging on resuming from a sleep state.

I've also dropped vcore to 1.1V. This is the minimum the board allows which is disappointing as MikeC picked up in his review. Other Asus boards can go quite a bit lower and as a uATX board you'd expect low power systems to be a target market. CPU temp may have fallen ~2C, need to check power consumption to make sure the voltage drop is across both idle and load.

I have redone some of the power cabling. The PCI-E power cable used to have spare connectors for other cards. As these weren't required, I've pulled out all of the extra connectors and wires to make a far smaller braided cable which tidies up the cabling a lot. The cable is still far too long but I can easily fold that out of the way and it leaves me more open to future cable rerouting.

The optical drive used to be powered by a string of molexes, one of which had a molex-sata adaptor. I've cut the cables short and wired up a rewirable sata power connector on the end so that there is no longer a bundle of cables behind the optical drive. In total ~400mm of cable was removed.

I also found that the inner plastic sheath on the Accelero S1 was hitting one of the sata cables on the mobo so I removed the inner sheath.

Things are much tidied now. Not too sure about any temperature of noise difference as any temp difference would only have been +/-1C and the HDD is still the loudest component:

Image

The HDD vibration issue I have had an idea on which I'm going to try in the next few days hopefully. I'll try and take some before and after recordings for FFT comparison.

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Sun Aug 24, 2008 3:01 am

For comparison of different HDD mounting methods, I have taken another baseline noise reading for the system with the case closed at idle. The spectrum looks like this:

Image

This was recorded in a different setting to the original readings and was taken using a different microphone that may have cost as much as 79p. The results are therefore incomparable with previous recordings. The desk the system is now on has much worse vibration characteristics, hence the 120Hz HDD hum is worse. There is also a large 50Hz peak which will be from different mains electrical equipment.

My first effort has been to prevent the vibration from the HDD. I had been looking to damp the vibrations internally to the case but due to the tight space and not wanting to do anything definite yet, I ran the system open with the HDD sat on a chunk of foam external to the case. This was noticeably quieter for vibrational noise and was like rediscovering the Barracuda IV. Obviously there is some more airborne noise you wouldn't get internally. Just touching the HDD in this setup showed how much it vibrates and the transmission to the desk was undetectable. Seek noise was slightly softer.

The spectrum for this setup looked like this:

Image

The 120Hz peak has disappeared from the spectrum. There are changes in other parts of the spectrum but they're hard to compare at anything other than the 120Hz vibrational range.

Noise subtraction did not work well on these recordings as the volume was so low. Instead what I've done is a subtraction of the second spectrum from the first spectra. This is not an accurate representation of what the HDD sounds like but gives some idea how much the foam helps:

Image

There is an obvious peak around 120Hz which comes to 14.15dB. Depending upon what you consider a doubling of volume to be, this could mean that the vibrational noise of the HDD is cut to less than a quarter of what it was originally. There is also a very big high frequency area which becomes much quieter. Some of this noise is boarding on inaudible anyway so may not have much effect on overall loudness of the system.

In this setup the overall quietness of the system is so much better that it becomes obvious that HDD mounting needs improving greatly in the SG03.

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Wed Aug 27, 2008 10:34 am

I have now completed the first attempt of a hard disk sound deadening setup. I want to keep the disk trays in one piece so that it's all reversible but drilling a set of holes in them doesn't affect the current clipping arrangement:

Image

The holes are 4.5mm in diameter and are arranged three on each side of the tray at the correct spacings for the screw holes in the sides of the disk. The drive is being reversed so that the connections are at the motherboard side. The trays are much harder than I expected and clearly are more than just plain aluminium so drilling took some time and required pilot holes and WD40.

The drive was then suspended by three lengths of sewing elastic of around 5mm in width. This was thin enough to be fed through the screw holes in the drive to go entirely round the disk. This arrangement requires the disk to have an open underside.

Image

This setup is easy to adjust the balance of by pulling through extra elastic on individual holes. The ends are just tied off at the top.

The drive now sits with the connections at the motherboard side. This requires the drive to be further out in the tray so that there is the space round the back of the disk for the cables to run in to it. This makes cabling in the system much tidier as I have been able to swap the 450mm long SATA cable with a 150mm one. I have also modded the SATA power cabling so that there is only a single connector.

Image

The ends of the elastic still flap around but the drive seems quite secure in place, enough for careful movement of the system without damage.

Sound wise it is definitely better than the original clip mounting but not quite as good as the external foam bed. Drive seeks are certainly softer and the whirring doesn't lead to major vibration. I can no longer feel it through the desk. I took another recording of the system to see if there was any quantitative improvement:

Image

The peak at 120Hz is still there but greatly reduced though not as good as on foam.

Again, I have carried out a spectral subtraction with the original control recording:

Image

The <10Hz difference should probably be ignored but the obvious thing is the peak at 120Hz. This corresponds to an 11.59dB reduction in noise at this frequency. Not quite up with the 14.15dB improvement from the foam but still a definite improvement and the disk is safe inside the case again.

The elastic is not an ideal material for the straps and the tight space does not help. This must be one of the smallest 3.5" suspension setups running as there's only 5mm on either side of the disk. I think the trays themselves may pickup some of the airborne sound which could be improved with some thin foam on the inner surfaces. Something else I am intending to try is to drill out the holes in the disk to the same 4.5mm diameter which would then allow a set of soft fan mounts to be used to hold the drive in place. Silicone rubber probably will work better at isolating vibration than sewing elastic.

The HD501J is quite a popular disk on these forums so I doubt they're actually that loud but I think my particular one might be quite a shaky unit. Just holding it when it was out on the foam showed how much it vibrated. Each disk seek has a real kick to it and the whole think vibrates as it spins. A Caviar Green would have been better.
Last edited by edh on Fri Sep 05, 2008 12:22 am, edited 1 time in total.

KyonCoraeL
Posts: 45
Joined: Sun Jun 01, 2008 6:40 pm
Location: Vancouver BC, Canada

Post by KyonCoraeL » Thu Sep 04, 2008 10:28 pm

Good Job, I just bought the sg03 after the p5q-em has been released... I am really impressed with the cooling and silence of the case without any modification. I feel that the build quality is a tad flimsy, I already stripped one of the screws by accident. Have you tried rubber gromets? or mounting the hard drive into the floppy brace?

edh
Posts: 1621
Joined: Mon Mar 17, 2008 1:49 pm
Location: UK

Post by edh » Fri Sep 05, 2008 12:44 am

KyonCoraeL wrote:I just bought the sg03 after the p5q-em has been released...
The obvious advantage of the P5Q-EM is the X4500HD graphics. As I'm running a 9600GT it doesn't bother me now but the X4500HD is certainly enough for most peoples uses. Seems to perform better than a top of the range card from 5 years ago. I definitely wouldn't get one as an upgrade but for a new build it would seem to make sense.
KyonCoraeL wrote:I feel that the build quality is a tad flimsy, I already stripped one of the screws by accident.
You do have to be quite careful when the case is in pieces as some of the parts aren't too strong on their own. Doesn't seem anything like as bad as the Lian Li cases I've seen through work. None of the systems that are in them seem to have a full complement of planar panels. Half the motherboard trays are blatantly not perpendicular to the back panels and just about all of the USB port flaps fall off.

I would have preferred to have seen more thumb screws as it tends to prevent people making a mess of things.
KyonCoraeL wrote:Have you tried rubber gromets?
I was considering this for a while but it is generally not considered as good at HDD dampening compared to suspension:
viewtopic.php?t=8240

There are also problems with doing this in the SG03. If you put grommeted bolts through the sides of the cage, the outside of the grommets and bolts will interfere with the other drive cage meaning that you can't use both cages. If you drill holes through the bottom of the cage to mount the HDD from the underside like in an Antec and as illustrated in my previous diagram, the 'foot' of the cage gets in the way on one side. Unless you drill massive holes through that, it won't be possible.
KyonCoraeL wrote:or mounting the hard drive into the floppy brace?
That would mean metal on metal contact and just by touching the disk against any metal it gets a lot louder. That won't be quieter and it will make disk fitment much more difficult, putting it into contact with the Accelero S1.

KyonCoraeL
Posts: 45
Joined: Sun Jun 01, 2008 6:40 pm
Location: Vancouver BC, Canada

Post by KyonCoraeL » Sat Sep 06, 2008 2:05 am

yeah that floppy brace is really bothersome. I put in a scythe 3.5" fan controller and the only way I could get it in is sliding it in thru the front.

Alot of the noises from the case seem to be magnify the noise because of the thin side panels, numerous holes and the direct contact from the harddrive to the bottom of the case.

I think I am going to try to replace the rails with pieces of foam eventually.

unfortunately I haven't been able to test the noise levels recently because I am still missing ram, which I will have by tomorrow.

Post Reply