Page 1 of 2

Puget Serenity SPCR Edition v.2

Posted: Wed Sep 29, 2010 3:34 pm
by MikeC

PSU, intake filtering

Posted: Wed Sep 29, 2010 4:47 pm
by Worker control
Looks like a lovely build.

Isn't the power supply a bit over-specced? For the same cost, you could go with a Seasonic X-Series model with lower max output and better efficiency.

The one angle I was surprised not to see covered in the review is intake filtering. Usually, on my Solo builds, I replace the stock filter with wire mesh filters, such as these from jab-tech:
http://www.jab-tech.com/92mm-Aluminum-m ... -3622.html
http://www.jab-tech.com/120mm-Aluminum- ... -1501.html
(similar filters are available other places -- I can't find any on newegg at the moment, but I swear I've seen them there before).

Maybe the stock filters on the P183 are good enough?

Posted: Wed Sep 29, 2010 5:07 pm
by MikeC
Agreed re the PSU -- it could easily have been a Seasonic X-560, for example, tho I am not sure it's cheaper than the Antec. That might have reduced power a bit at idle and a bit more at full load (maybe 10~15w at full load, a lot less at idle.)

The filters... well, it matters most when the fans are spinning hard. slower fans = less dust. In this case, with the fans so slow, I really don't feel it's an issue. Any system needs periodic dusting anyway.

Posted: Wed Sep 29, 2010 5:21 pm
by Enzo_FX
Maybe I missed this...

but I'm curious as to what Slipstream fans they used, the beloved Mediums? Also, how did they regulate their speed? No variable speed perhaps?

Re: the PSU
Price is prob the main reason, but they could have some kind of deal with Antec =P.

Posted: Wed Sep 29, 2010 5:36 pm
by netmask254
It's an amazing work. Thanks for update and review!

It's difficult to image with many noise sources (3 fans and one disk, considering their individual test dBL), the whole system can still be so silent.

I'm very curious that if we don't apply the AcoustiPack (or doesn't apply it everywhere, e.g. just apply to one side panel), what's the raw noise level? Is this the secret? :roll:

Posted: Wed Sep 29, 2010 6:04 pm
by Vibrator
Slipstreams sound nice, but sleeve bearings + a smaller than normal motor in a hot environment worry me a bit.

Would it be possible to add Gentle Typhoons as an upgrade from the default configuration?

Have to disagree

Posted: Wed Sep 29, 2010 6:14 pm
by Johnsy
Huge rises in several key temperatures compared to the original, making this machine simply the "or" in the eternal either/or of thermal/acoustic performance.

Re: Have to disagree

Posted: Wed Sep 29, 2010 6:37 pm
by MikeC
Johnsy wrote:Huge rises in several key temperatures compared to the original, making this machine simply the "or" in the eternal either/or of thermal/acoustic performance.
Not really. The earlier sample was too conservative, wasting thermal headroom. The new version is closer to the limits... but who runs Prime95+Furmark for an hour? Is that a real app? No. I'm confident whatever 3D game will run fine w/o overload. And that is invariably the most demanding task asked of most computers these days.

Posted: Wed Sep 29, 2010 6:44 pm
by MikeC
netmask254 wrote:I'm very curious that if we don't apply the AcoustiPack (or doesn't apply it everywhere, e.g. just apply to one side panel), what's the raw noise level? Is this the secret? :roll:
No it is not "the secret" -- the real secret is in balancing everything, it's in the details of component choice and execution. The damping helps reduce standing waves, some of the hum of cavity air resonance.

Aside from the nicer sounding fans (especially the CPU hs fan), I think the cherry picked WD Green Power drive is a major difference. We just sorted through half a dozen new samples of those for another project and found they could be separated into 2 groups -- those with lower vibration (rank 8 on SPCR's HDD vibe scale) and those with higher vibration, rank 7. The former will definitely be quieter unless they are all mounted in the best elastic cord suspension.

Posted: Wed Sep 29, 2010 7:34 pm
by twitch
The WD20EARS green drives actually come in two flavors with the newer ones modeled after the f4 with 3 platters of 667 each whereas the earlier model WD20EARS used more. Incidentally they are both called the same model. Perhaps this may partially explain the increase in vibration and noise.

It is plausible that even within the newer gen wd20ears there is some variation in build accounting for some difference but the std dev cannot be that wide.

Re: PSU, intake filtering

Posted: Wed Sep 29, 2010 7:58 pm
by jonbach
Worker control wrote:Isn't the power supply a bit over-specced? For the same cost, you could go with a Seasonic X-Series model with lower max output and better efficiency.
I can address this question! Yep, the Antec CP-1000 is certainly more than is needed for this build. We wanted to certify something that could scale up. But more than that, the Antec CP series is just extremely cheap compared to other units on the market in the same wattage range. The Antec CP-1000 costs us $25 less than the Seasonic X-Series 750W. We wanted to stay away from passive, because we wanted the power supply to draw air through the hard drive area, so we would need no other fans down in the bottom compartment. The size and front-to-back airflow of the Antec CP series is ideal for that purpose. On top of that, and possibly most important to us, these power supplies have AMAZING reliability. We've sold a lot of them, and we're maintaining a 0.27% failure rate right now. Normal for power supplies is in the 5-10% range. Granted, we're broad with our internal definition of "failure." We treat a fan tick, or higher-than-normal electrical noise as a failure. But pretty amazing regardless!

Posted: Wed Sep 29, 2010 11:28 pm
by Das_Saunamies
Cor. But with an astronomical price tag, you only get a 5750? Really? Doesn't seem to follow suit with an unlocked i7, 8GB of RAM and an SSD.

I am duly impressed by the fact that ticking, whine etc. get weeded out at manufacturer level.

Re: PSU, intake filtering

Posted: Thu Sep 30, 2010 1:29 am
by Olle P
jonbach wrote:We wanted to certify something that could scale up. ... the Antec CP series is just extremely cheap...
I still think staying with the CP-850 would allow sufficient up-scaling at a lower cost and marginally higher efficiency.

Cheers
Olle

Re: Have to disagree

Posted: Thu Sep 30, 2010 2:50 am
by Johnsy
MikeC wrote:
Johnsy wrote:Huge rises in several key temperatures compared to the original, making this machine simply the "or" in the eternal either/or of thermal/acoustic performance.
Not really. The earlier sample was too conservative, wasting thermal headroom. The new version is closer to the limits... but who runs Prime95+Furmark for an hour? Is that a real app? No. I'm confident whatever 3D game will run fine w/o overload. And that is invariably the most demanding task asked of most computers these days.
I wasn't suggesting that the machine was dangerously undercooled.

The point I was trying to make was simply that nothing particularly clever or innovative has been done here. It runs quieter at the expense of running hotter: by a substantial 23% in the case of the the hdd, for example, even at idle - ie, most of the time.

Those of us who use our machines for professional music production are often running CPU loads of 70-80% continuously for many hours at a time. Not quite Prime, perhaps, but thermally demanding nonetheless.

Re: Have to disagree

Posted: Thu Sep 30, 2010 6:20 am
by MikeC
Johnsy wrote:I wasn't suggesting that the machine was dangerously undercooled.
That's what it sounded like to me...
The point I was trying to make was simply that nothing particularly clever or innovative has been done here. It runs quieter at the expense of running hotter: by a substantial 23% in the case of the the hdd, for example, even at idle - ie, most of the time.
Maybe... but no other company has ever used component binning for low noise or achieved such low noise levels before, and we've been testing systems for nearly 9 years. Also you can't just say the HDD is running 23% hotter -- it is NOT the same HDD or even the same model, and there are always SAMPLE VARIANCES! There's no way to tell for sure whether these components with the old cooling would run hotter or cooler than the old system components.
Those of us who use our machines for professional music production are often running CPU loads of 70-80% continuously for many hours at a time. Not quite Prime, perhaps, but thermally demanding nonetheless.
That really is NOTHING like Prime95+Furmark. I'd guess about 80~100W less. With that kind of load, this system would not even break a sweat going 24/7 - 365.

Posted: Thu Sep 30, 2010 6:36 am
by frenchie
8) Great !! This build is a nice example of what can be done with the right components ! Imagine what it would sound like with some minor DIY mods !!

Posted: Thu Sep 30, 2010 7:09 am
by PartEleven
MikeC wrote:The damping helps reduce standing waves, some of the hum of cavity air resonance.
Interesting. I never thought about using damping for this reason. I think cavity resonance is stopping my system from achieving absolute silence at my current air flow. What's the reasoning behind this? Less empty volume?

Posted: Thu Sep 30, 2010 7:17 am
by MikeC
PartEleven wrote:
MikeC wrote:The damping helps reduce standing waves, some of the hum of cavity air resonance.
Interesting. I never thought about using damping for this reason. I think cavity resonance is stopping my system from achieving absolute silence at my current air flow. What's the reasoning behind this? Less empty volume?
Standing waves occur mostly when sound is reflected multiple times across parallel surfaces. (Think of mirrors on opposite walls.) Applying sound absorption material reduces this reflection... but generally it is much less effective at lower frequency -- ie, if the hum of cavity resonance is at very low frequency (like the fundamental of HDDs -- 90~120Hz) it is virtually untouched by typical foam damping. There, you just have to reduce the noise at the source -- use a lower vibration/noise HDD.

Posted: Thu Sep 30, 2010 8:53 am
by cloneman
Is there no coil whine on that motherboard?

I've got the same Asus board (except it's not the "-E" version) with a Core i5 750 and I've got noticeable whine, however I "fixed" it by disabling Intel C-State and Intel Speedstep in the bios. Gives me 10C higher on idle, but hey, sanity is well worth it.

Re: PSU, intake filtering

Posted: Thu Sep 30, 2010 8:59 am
by dhanson865
jonbach wrote:
Worker control wrote:Isn't the power supply a bit over-specced? For the same cost, you could go with a Seasonic X-Series model with lower max output and better efficiency.
The Antec CP-1000 costs us $25 less than the Seasonic X-Series 750W. We wanted to stay away from passive, because we wanted the power supply to draw air through the hard drive area, so we would need no other fans down in the bottom compartment. The size and front-to-back airflow of the Antec CP series is ideal for that purpose. On top of that, and possibly most important to us, these power supplies have AMAZING reliability. We've sold a lot of them, and we're maintaining a 0.27% failure rate right now. Normal for power supplies is in the 5-10% range. Granted, we're broad with our internal definition of "failure." We treat a fan tick, or higher-than-normal electrical noise as a failure. But pretty amazing regardless!
That is a huge reason. It's bucking to beat the Intel SSD failure rate which is also insanely low.

Posted: Thu Sep 30, 2010 9:09 am
by dhanson865
Das_Saunamies wrote:Cor. But with an astronomical price tag, you only get a 5750? Really? Doesn't seem to follow suit with an unlocked i7, 8GB of RAM and an SSD.

I am duly impressed by the fact that ticking, whine etc. get weeded out at manufacturer level.
There isn't a stock fanless card made that is faster than the 5750.

They could have gone for a MSI Hawk 5770 but it would have added some (minimal) noise to the setup and it has two fans so if either one has a click/tick/failure you are dealing with a binning/failure rate issue compared to a stock fanless card.

Code: Select all

The TDP is reference design TDP values from AMD.
Different non-reference board designs from vendors may lead to slight variations in actual TDP.

Card Load Idle
5670  64W 15W
5750  86W 16W
5770 108W 18W

6750 114W 20W
6770 146W 23W
You just aren't going to see stock fanless cards that have a TDP above 100W.

To get something better and still be fanless you have to wait for AMD to come out with a 6730 or 6670 or whatever they'll call it that fits under that 100W limit. Preferably it will be closer to the 75W mark than the 100W mark.

Re: Have to disagree

Posted: Thu Sep 30, 2010 10:04 am
by Johnsy
MikeC wrote:
Johnsy wrote:I wasn't suggesting that the machine was dangerously undercooled.
That's what it sounded like to me...

Here's my original post in it's entirety:

Huge rises in several key temperatures compared to the original, making this machine simply the "or" in the eternal either/or of thermal/acoustic performance.

What part of that suggests that I think the machine dangerously undercooled? Surely an increase in mainboard temp. of more than 25% at idle (38C v 30C) and of more than 55% (55C v 35C) under Prime95 is huge? Not dangerous, not even particularly worrying in absolute terms, but very substantial nonetheless. It's an observation - that's all
Quote:
The point I was trying to make was simply that nothing particularly clever or innovative has been done here. It runs quieter at the expense of running hotter: by a substantial 23% in the case of the the hdd, for example, even at idle - ie, most of the time.

Maybe... but no other company has ever used component binning for low noise or achieved such low noise levels before, and we've been testing systems for nearly 9 years. Also you can't just say the HDD is running 23% hotter -- it is NOT the same HDD or even the same model, and there are always SAMPLE VARIANCES! There's no way to tell for sure whether these components with the old cooling would run hotter or cooler than the old system components.
The WD Caviar Green 1.5TB used in the original machine, and the WD Caviar Green 2.0TB drive used in the new feature identical power specifications:

http://www.wdc.com/en/products/products.asp?driveid=772
http://www.wdc.com/en/products/products.asp?driveid=773

Quote:
Those of us who use our machines for professional music production are often running CPU loads of 70-80% continuously for many hours at a time. Not quite Prime, perhaps, but thermally demanding nonetheless.

That really is NOTHING like Prime95+Furmark. I'd guess about 80~100W less. With that kind of load, this system would not even break a sweat going 24/7 - 365.

Firstly, I have not referred in any of my posts to Furmark. I referred only to Prime95, and even then I conceded that it was not a flat comparison. Secondly, perhaps you could tell me which VST/VSTi's and host enviroments you're familiar with?

Re: Have to disagree

Posted: Thu Sep 30, 2010 10:20 am
by nutball
Johnsy wrote:Surely an increase in mainboard temp. of more than 25% at idle (38C v 30C) and of more than 55% (55C v 35C) under Prime95 is huge?
Erm... you do realise that the Centigrade/Celsius temperature scale has a quasi-arbitrary zero-point? You can't sensibly make arguments about percentage rises in temperature on that scale. Convert the temperaures to Kelvin (clue: Kelvin = Celsius + 273), redo the calcs and you'll have a very different picture about the fractional increase.

Re: Have to disagree

Posted: Thu Sep 30, 2010 10:35 am
by Johnsy
nutball wrote:
Johnsy wrote:Surely an increase in mainboard temp. of more than 25% at idle (38C v 30C) and of more than 55% (55C v 35C) under Prime95 is huge?
Erm... you do realise that the Centigrade/Celsius temperature scale has a quasi-arbitrary zero-point? You can't sensibly make arguments about percentage rises in temperature on that scale. Convert the temperaures to Kelvin (clue: Kelvin = Celsius + 273), redo the calcs and you'll have a very different picture about the fractional increase.
In this case, that's a red herring. 'Zero' here is ambient temp.

Notwithstanding that, may I point out that the 'max operating temp' given by ASUS in their enviromental spec for this board is (in common with the majority of consumer-grade boards) 55C. So even with an ambient (room) temperature of only 21C, this machine is right at the limit.

Posted: Thu Sep 30, 2010 11:53 am
by MikeC
Johnsy --

That 55C temp you cite is surely for the operating environment -- not for specific chip temps on the board. The embedded sensors on mobos are all over the place; I never know where they are exactly. But in any case, all kinds of chips are perfectly safe to run near water boiling temps -- not that they should be run there -- and 55C seems absolutely tame to me. I assure you the average air temperature inside the case w/ the system at full tilt is NOT 55C.

I agree with you that quiet PCs generally tend to run a bit hotter than noisy ones that have much higher airflow. It's a basic tenet of quiet computing to use some of the cooling headroom of advanced heatsinks to run the fans slower -- which is what your first post alluded to. But what's too hot or cool enough if the system isn't showing any signs of stress (slowdown, instability, crash, etc) is mostly a matter of personal preference and opinion.

IMO, this one runs perfectly cool at moderate loads, marginally hot at full CPU load, and fairly hot with both CPU and GPU at full tilt. If I was running in the latter mode (CPU+GPU @100%) for hours on end and often, then for added safety (especially in hot weather), I'd tweak the BIOS fan control to give a bit better cooling at the expense of a dB or 2. But I really can't imagine why/how anyone would have such loads other than for pure lab testing like we do.

Posted: Thu Sep 30, 2010 2:46 pm
by CA_Steve
Nice build.

Posted: Thu Sep 30, 2010 2:48 pm
by Johnsy
[quote="MikeC"]Johnsy --

That 55C temp you cite is surely for the operating environment -- not for specific chip temps on the board.

You're right, of course. (Interestingly, every server/workstation board I'm aware of - single and dual CPU alike, whether Intel, Supermicro or Tyan - calls for a rather more onerous operating enviroment of between 0-35C, rather than the 0-55C typical for consumer boards). To be honest, I was just trying to avoid a long, digressionary argument about physics with Nutball there! (no offence, Nutball).

On the Intel and Gigabyte LGA1156 boards I'm familiar with, the 'motherboard' temperature is read from the thermal diode in the PCH (ICH10/10R). I believe the 945/946 chipsets (2005-7) were the last to employ 'remote' (in Intel parlance) diodes - usually having one located near the CPU, and (often) another between the north and southbridges - which qualified more as a 'enviromental' sensors, albeit very localized ones.

Posted: Thu Sep 30, 2010 10:50 pm
by Vetotat
dhanson865 wrote:There isn't a stock fanless card made that is faster than the 5750.
But isn't the 5770 available as fanless.

Posted: Fri Oct 01, 2010 12:27 am
by flemeister

Posted: Fri Oct 01, 2010 3:54 am
by Parappaman
I was about to post the same exact thing. :mrgreen:
A CP-850 and this card will make this computer a better balanced machine. I also think that using the 1200 rpm Slipstream variant @5v on the internal slot or even at the rear exhaust will greatly improve thermal performance without affecting base noise, considering how well the case is packed and sealed and the mere 200 RPM difference in fan speed.