How to make a Silent PC ?

[hydra]

...
Now for an great USB/eSata external enclosure.
...

This might do the job: Transcend Desktop HDD Case USB 2.0 + eSATA - www.deltapage.com

 

[gobble]

This might do the job: Transcend Desktop HDD Case USB 2.0 + eSATA - www.deltapage.com

I like that one!! Thanks! hyeah:

--G

 

[hydra]

I was pottering about, window shopping on the internet, when I came across WD AV-GP hard drives: WD AV-GP

Apparently these are HDDs optimised for AV use. What attracted me is:

Whisper quiet.
Noise levels have been minimized to less than one sone virtually below the threshold of human hearing.

And to a lesser extent:

24x7 reliability.
These drives are designed to last in always-on, streaming digital audio/video environments such as PVRs, DVRs and surveillance video recorders.

I have no idea what a "sone" is, so I asked wolframalpha to convert 1 sone to dB. Apparently such a conversion is not accurately possible.

Wikipedia says that a "sone":

a loudness of 1 sone is equivalent to the loudness of a signal at 40 phons, the loudness level of a 1 kHz tone at 40 dB SPL

And lower down, they say:
Normal talking, 1 m away = 40 to 60 dB = ~ 1 to 4 sone.

That doesn't sound exactly "whisper quiet", but maybe in a real application it will be better than a regular HDD. theitdepot lists a 1TB (3Gb/s, 32MB cache) WD10EURS at Rs.3480.00, which means it must be just above Rs.3000.00 on the street.

Does anybody have experience with these drives? They sound perfect for HTPC/Music PC applications.

 

[Thad E Ginathom]

It sounds to me like the hard-disk industry has realised what fertile ground the audio market is for selling!

I have spent a lot of money on my quest for a silent PC --- and fellow members have, quite rightly, asked me why, in this country where we are seldom without the ambient noises of at least fan and/or AC, I should be trying to further reduce the sound levels from my computer, when they are barely audible except in the dead of night when fan and AC are turned off. Well, my answer is, like getting Firewire Audio working effectively under Ubuntu Linux ... it's a hobby, a challenge, a pass-time :lol:

The hard disks (three), in my machine, though, I have never considered a problem. Even given that reduced-speed case fans and a whopping 140mm fan (err, two, actually!) on a huge heat-sink have reduced the noise of my computer to little more than the hiss of airflow, I don't think I ever hear hard-disk noise. What does concern me about the disks is their temperature. I would be free to even further reduce the fan speeds, were it not for the rising temperature of the disks, especially the main 1Tb disk. For this reason, it is generally preferable not to buy the fastest RPM disks for an audio or htpc application, just as it is preferable not to buy the fastest hottest processor. If and when that 1-Tb gets replaced, I'll pay serious attention to this, sticking to the "green" ranges, or perhaps even the one you have spotted here.

Personally, I would expect all disks to function 24/7. In fact, I'd suggest that powering them up and down probable causes more wear than several long hours of use.

Otherwise, audiophile-world cynicism aside, if they they really have optimised the disc and cache to run cool and give good results on continuous reads, then it sounds well worth trying!

Just ... don't forget to use an audio-grade SATA cable to connect it :lol:

 

[hydra]

Yup, I'm kind of taken up with the silent PC thing too

My office PC's main disk seems to be on its last legs (read/write of large files is agonizingly slow), so I was thinking of moving my Music PC's 1TB 7200 rpm HDD over to the office PC, and getting something more silent for the Music PC. Right now, the loudest thing in the Music PC is the HDD. I should have gone for a 5400rpm "green" disk when I was building it

 

[Thad E Ginathom]

That sounds like a good plan.

(Of course, you'd be keeping your office machine backed up, even if you didn't suspect the disk might be about to fail, so there's no need to even mention that )

I don't think disk speed is important for audio --- after all, we can get uninterrupted sound from a CD, which is hugely slower than any HDD --- but video is probably more demanding, especially if you are streaming HD. I have no HTPC experience (I don't watch movies) but I'd guess a generous cache on the disk would help. The link to the WD page is not working just now (perhaps they had a disk failure!): I wanted to see if there are more detailed specs there.

I just googled for reviews. Here is an interesting one, explaining why this drive should not be purchased for a general or multi-purpose machine. The link jumps to p4 which discusses this. This is very important information about the AV-GP: If I hadn't read it I might have gone out and bought one. Now I know I shouldn't!

 

[gobble]

@Thad

What do you think will be the implications for an audiophile PC for a drive that skips error correction? Also the features are optimised for writes which we rarely do for playback. Given an audio PC that does nothing else but playback music, a desktop Caviar drive should not introduce too much jitter. Besides reading FLAC/WAV is not the same as redbook and data can still be readahead/buffered.

--G0bble

 

[gobble]

Here is a good review for a fanless board available in India :Asus E35M1-M Pro: AMD Fusion Motherboard | silentpcreview.com

--G0bble

 

[hydra]

@thad, thanks for posting that link! What is said there makes a lot of sense, and it clearly means these drives will not at all be suitable for HTPC/Music PC use.

@gobble, I'm picking up that very board this week. I had originally planned to use that board for my build, but it was not available anywhere then (and I was in too much of a tearing hurry to finish the build). I bought an MSI E350 board instead, which did the job very well, but unfortunately had to be returned for overheating issues.

The MSI E350 board could not be replaced and a refund is on the way. As soon as I get it, I'm picking up the E35M1-M Pro. It does come with a CPU fan, but using it is just optional. I'm going to try using it without the fan. If you don't want USB-3 and that CPU fan, the Asus E350M1-M (non-Pro version) is almost 2K cheaper.

 

[gobble]

I bought an MSI E350 board instead, which did the job very well, but unfortunately had to be returned for overheating issues.

The MSI E350 board could not be replaced and a refund is on the way. As soon as I get it, I'm picking up the E35M1-M Pro. It does come with a CPU fan, but using it is just optional. I'm going to try using it without the fan. If you don't want USB-3 and that CPU fan, the Asus E350M1-M (non-Pro version) is almost 2K cheaper.

If a board with a nice fanny (pun intended) had overheating problems what will be the state of the board without one? Unless the onboard heatsink is really good ...

Looking forwards to your experience. I will search for this board locally on SP road this Saturday.

--G0bble

 

[hydra]

^^

It was nothing to do with the fan. MSI told me the board was defective. In any case the Asus board does have a better heat sink (at least from appearances), and they must be confident enough that the boards will not have heating issues, or they wouldn't be selling the non-pro version without a fan.

I think (not completely sure as I read it 2 months back) that the board's manual also touts its capability of functioning at 0dB. The silentpcreview review also mentions that it can most likely be used fanless even in a MicroATX case with a single fan blowing across it, if we don't stress the GPU too much. They don't exactly go out on a limb and say that it can be used fanless, but they say that they "wager" so.

Also, my cabinet (Silverstone GD-04B) is pretty airy, with three 120mm fans (one of them just an inch or so away from where the CPU will be), so I think going fanless is definitely worth a try.

In any case I don't think I'll be taking delivery of the board before next week wednesday at least -- the refund is taking a bit of time, and I don't want to spend any extra money right now

I'll definitely post impressions once I have it up and running.

 

[Thad E Ginathom]

@Thad

What do you think will be the implications for an audiophile PC for a drive that skips error correction?

I don't know. In practice it might mean almost nothing, but, in theory at least, we want to know that the bits that were on our CD have been written to our disks, and when we read them from the disk, we can feed them to our sound cards or DACs without alteration. This is why people go to the trouble of using drivers that keep the sticky microsoft fingers off the data as much as possible, because they do not deliver our musical bytes as written and read. Some of us spend so much time chirping away at "audiophiles" about error correction in the digital domain, and here's a product that seems to consider it un-important. I don't think I'd buy one for a music machine, and I am probably not an audiophile!

Also the features are optimised for writes which we rarely do for playback.

That strikes me as very odd to the extent of wondering if it is a misprint! Surely any media-playing machine is dependent on consistent disk reads, and disk writes make no odds whatsoever --- perhaps a very small percentage on CD/DVD ripping times. In fact, albeit only on a cursory thought or two, I can't see much point in highly optimising write performance anyway: buffering and cache takes care of that.

Given an audio PC that does nothing else but playback music, a desktop Caviar drive should not introduce too much jitter. Besides reading FLAC/WAV is not the same as redbook and data can still be readahead/buffered.

Jitter is a timing error in the reading of bits when converting to analogue, and is only going to matter when they finally arrive at the DAC, be it soundcard, usb, or whatever. I don't think it can be caused by a disk. However, it is something I have to look up every time I need to think about it!

 

[manoj.p]

I was pottering about, window shopping on the internet, when I came across WD AV-GP hard drives: WD AV-GP

Apparently these are HDDs optimised for AV use. What attracted me is:



And to a lesser extent:



I have no idea what a "sone" is, so I asked wolframalpha to convert 1 sone to dB. Apparently such a conversion is not accurately possible.

Wikipedia says that a "sone":



And lower down, they say:
Normal talking, 1 m away = 40 to 60 dB = ~ 1 to 4 sone.

That doesn't sound exactly "whisper quiet", but maybe in a real application it will be better than a regular HDD. theitdepot lists a 1TB (3Gb/s, 32MB cache) WD10EURS at Rs.3480.00, which means it must be just above Rs.3000.00 on the street.

Does anybody have experience with these drives? They sound perfect for HTPC/Music PC applications.

I have one of this drive. These drives are designed to work in DVR boxes specifically. DVR boxes write continuously and need low noise. These drives are optimized for that. I have one in my TiVo.

as for htpc any normal, less noisy drive will be fine too. Because the need is not to constantly write data. For jitter, no need to worry. Current dac chips are fast enough to correct timing errors.

 

[Thad E Ginathom]

These drives are designed to work in DVR boxes specifically. DVR boxes write continuously

Ahh, yes, good point

 

[gobble]

I can't see much point in highly optimising write performance anyway: buffering and cache takes care of that.

On a desktop, writes and reads have to happen "concurrently" over a larger time slice (measured in 100s of msecs?) hence HDDs have advanced algorithms that optimize buffering and read-ahead, and selection of sectors and cylinders for writes, in such a way that head movement is minimized on subsequent reads thus reducing latency. Moreover there may be assumptions made about the statistical distribution of the size of concurrent read/write requests which may fall into a certain KB to MB range per second (OS paging, office apps). So the algorithm on this DVR specifc HDD may be tweaked to sequentially read/write larger MBs in a single operation with a larger time slice now dedicated per operation.

Ok to tell the truth, I am imagining things here as I read about these things a decade or more back hyeah:

Jitter is a timing error in the reading of bits when converting to analogue, and is only going to matter when they finally arrive at the DAC, be it soundcard, usb, or whatever. I don't think it can be caused by a disk. However, it is something I have to look up every time I need to think about it!

Excessive delay in reading , system bus bottleneck, cpu scheduling, poorly designed or implemented usb/spdif/firewire chip on mobo, all can add to jitter IMHO

--G0bble

 

[Thad E Ginathom]

The last time I looked up jitter, I came away with the information that, whilst in the digital domain, it simply doesn't matter. The writer gave a rather extreme example of copying from digital data to digital data (DAT tape in his experiment) but, even after multiple generations of digital copy, with jitter, could still get perfect data at the DAC when finally listening to the music. I suspect that the word jitter gets generally applied to a lot of the stuff that can go wrong with digital sound, like buffering problems rather than just its actual definition. It sounds onomatopoeic (like wow and flutter, wasn't it? that we had with speed variations on TT and tape) but is not a j-j-j-jitter at all. I struggle to understand the technology behind digital sound (largely because I'm a maths dunce) but, apparently, jitter produces hiss. I guess I never heard it!

On the other hand, problems that you mention have been very much a part of my ubuntu-firewire-audio struggle.

Ok to tell the truth, I am imagining things here

Your imagination on this is better informed than mine!

 

[gobble]

The last time I looked up jitter, I came away with the information that, whilst in the digital domain, it simply doesn't matter. The writer gave a rather extreme example of copying from digital data to digital data (DAT tape in his experiment) but, even after multiple generations of digital copy, with jitter, could still get perfect data at the DAC when finally listening to the music.

Thad
We need to remember that reading a redbook CD is not like reading packet'ised data that can be buffered. With a flac from hdd it is not redbook anymore so it can be bufferred just before the DAC and sent out with few bottle necks, if implemented that way.

It has been claimed that for 44.1khz sampling of a 16bit data stream, 173picosconds is the threshold for timing errors before SQ is affected audibly. 173ps momentary latency/jitter is easy to run into in any system even with a real-time kernel.

--G0bble

 

[manoj.p]

On a desktop, writes and reads have to happen "concurrently" over a larger time slice (measured in 100s of msecs?) hence HDDs have advanced algorithms that optimize buffering and read-ahead, and selection of sectors and cylinders for writes, in such a way that head movement is minimized on subsequent reads thus reducing latency. Moreover there may be assumptions made about the statistical distribution of the size of concurrent read/write requests which may fall into a certain KB to MB range per second (OS paging, office apps). So the algorithm on this DVR specifc HDD may be tweaked to sequentially read/write larger MBs in a single operation with a larger time slice now dedicated per operation.

Ok to tell the truth, I am imagining things here as I read about these things a decade or more back hyeah:

Your are correct. HDD's access the OS vitals, drivers and most importantly the swap space very often. Every time you access any data, HDD is reading it and at the same time, storing it in the swap space too. That's why the recommended practice in the serious workstations and servers is to use one dedicated driver for all the OS operations and programs. Then they use another larger drive just for the data. In Linux, they even go one step ahead and create a dedicated partition on OS drive for swap space. This way, the HDD reading head does not have to move back and forth for swap space/OS operations/programs and data.

Excessive delay in reading , system bus bottleneck, cpu scheduling, poorly designed or implemented usb/spdif/firewire chip on mobo, all can add to jitter IMHO

--G0bble

The jitter again is nothing but timing error. Every device like CD-ROM, DVD drive, HDD drive, mother board, sound card, video card has the clock. They all run by their own clocks. In PC, mother board controls most of the clocks and uses it for PC operations. For music, if set properly, the sound card takes over and bypasses the motherboard. In this process it also bypasses the MB clock and uses its own. If you connect the external DAC, then it takes over and uses its own clock. Then in that case, its just reading the data as its input to it and adjusting the clock. What happens before that, does not matter. The clock is broken when you change the track or stop the track as the data stream is stopped or broken. When you play the next track, the stream is started again, the controlling clock adjusts it and starts playing. today's chips are so fast, they can read/buffer data very quickly and adjust the clock.

If you think about it, its all data. There are so many crucial industries/technologies where timing and clocks matter so much. Yet they never worry about it because its minor. But it's made out to be a big deal in music industry. Because people want to believe in it and are willing to pay for it. Whether one can hear the difference, its beyond human ears. Surprisingly, none talks about jitter once it has passed the digital domain. Once its out of the DAC, then goes to amp and then through interconnects - long cables to speakers. Then through cross-over to each drivers. What happens to jitter or something like that there? None will talk about it. Because there is nothing to talk about it. But still good money can be made on it in the digital domain.

 

[Thad E Ginathom]

So far as my understanding goes, I agree with a lot of what you say.

However, we don't talk about jitter in the analogue domain, because it doesn't exist. It is a specific problem that happens at A/D and/or D/A conversion.

 

[manoj.p]

I know that jitter does not exist in analogue domain, that's why i used jitter or something like that. What I want to bring out is - we freak out so much about jitter or timing errors which are specified or measured in picoseconds (yes, picoseconds) when there are lots of things that can go wrong on the analog domain.

Coming back to jitter - its specified in pico seconds. Meaning, if there is any timing error, the device has to align its clock by few ps. 15 to 20 years back, it was a big deal because chips were not so fast then. In today's time, this problem is almost non-existent because chips and clocks have so much progressed to take care of it altogether. The interfaces alone are designed to have minimum timing errors. I said non-existent because there could be some devices where jitter exists and it is there purely because of poor implementations of interfaces. That is why just the DAC chip does not matter, but its the circuitry that goes along with it adjusts the clock etc. If the design is right, then the jitter problem is already taken care of.

I would suggest reading this topic - The Well-Tempered Computer Since you are into computer audio, that whole site is dedicated to it, so can be a good reading.