Hard Drives Make a Difference in Sound

[Paul R]

Hi Coxhaus -

 

Perhaps I missed it, but what software are you using under Windows to playback your music?

 

In my experience the playback software can and will swamp any differences that might be caused by drive hardware differences, unless the drive is putting out so much electrical noise and vibration that those factors overwhelm pretty much everything else.

 

For example, when I use playback software with true memory playback, I don't hear any differences between drives - SSD to ancient old IDE drives. (Again, with the caveat above.) This is one reason I tend to use only playback software that pre-buffers the music into memory, which is the same effect as using a RAM disk. At least so far as the sound of music playback is concerned.

 

Again, caveat - this is using drives that don't shake like a torso tosser or spark like an arc welder. Those kinds of drives will not only ruin the music for you, but are likely to shake apart or melt down your computer too.

 

-Paul

[Paul R]

I see.

 

This is just my opinion, but if you are not using memory playback, what you are hearing that you like is possibly something being added by the playback chain. That would also make why you hear differences in drives more understandable, as the playback is much more vulnerable to alteration when not playing back with memory.

 

That is just a theory, but I would spend some more listening time and try to identify what you hear as a difference that you like. If it turns out to be a coloration of some kind, you might consider buying a component like a DAC or even cables that give you that favored coloration instead of trying to achieve it with a spinning disk.

 

I am not trying to spoil your fun, but a hard drive should NOT be able to affect your sound. I would consider that a fault and try to fix it.

 

One man's fault is another man's perfect sound though, so please do not take that as me cutting at your system. I'm not, and you should enjoy the music however you like the best.

 

-Paul

 

 

I am running Microsoft Server 2012 playing music using unlicensed JRiver 19 . I have paid for the license, I just have not decided on which hard drive.

 

I do not use memory playback set in JRiver as I like the sound better without checking memory playback. I spent some time listening with JRiver 18 with memory playback and not.

[Paul R]

Hi Roch-

 

Can you still hear differences between drives when you use a memory player or load the files up into a RAM drive- from different physical drives?

 

Either of those choices here will make all the differences completely vanish. Bringing in the music over the SAN will also eliminate storage differences, as will using a NAS device in most cases.

 

That is here of course- YMMV but was rather expecting that to be a general case, just based on what I and a few friends here in Austin experience.

 

Yours-

Paul

[Paul R]

Hey Alex-

 

I am not disputing that your files sound different. But- a binary 1 is a binary 1 and a binary 0 is a binary 0. There is no other possibility there.

 

Whatever is making your files sound different, it absolutely is not the data. Or to put it another way, there are only pure bits. If a bit is corrupted, it will either change in value or be unreadable- both cases are easily and quickly detected as changes in the data.

 

This is very different than having a poorly written CD or storing a file on a USB stick or spinning drive. In those cases it is possible the media could make a difference. The actual data that hits the DAC has been copied several times from the data that is on the media. Not only copied, but converted from one format to another, electrically, at least twice. And transmitted too. But through all of that, excepting any DSP processing, the data remains the same.

 

Media differences, playback chain equipment, yes. Those are possible, but not the data itself. I do not know what the answer to your puzzle is my friend, and I agree you have uncovered a puzzle, but I do know we can rule out the data as the cause. None of your experiments have shown the data changing, to the best of my knowledge.

 

Paul

[Paul R]

I agree those frames were definitely different, and that it could easily have been the power supply in that case.

 

However, the same is true of my blurry (er... Bluray) player when it is floated on Hip Joints. The picture is better, the sound is better, etc. When I add in power filtering, the picture and sound gets better yet. Worse? When I put an aftermarket power cord on the darn BRP, it looks and sounds even better. And yes, when I use a "higher end" HDMI cable, it gets incrementally better yet again.

 

All with exactly the same data.

 

I think you might be able to put *any* video on your Corsair, and it would improve the way it looks?

 

-Paul

 

 

 

Hi Paul

Barry said this about the .wav files made using an external USB powered writer and saved to the same CD-R

"as though some random treble energy surrounded the details in the recordings"

That is consistent with PSU/EMI problems. I have never claimed that the binary data is different, however there seems to be some mechanism by which a " snapshot" of the system noise comes along with the data. Some (including Martin Colloms) also saw this in an earlier experiment where the same 1080 video frame stored at 2 different locations was captured as a .bmp image , and the 2 images placed side by side using Photoshop. Those who picked up on the apparent differences with the jewellery on Christina Aguilera's hand, also found that the preferred image changed sides when the position of the screen grabs was reversed using Photoshop and a new composite image made. The checksums of both .bmp images were the same,

yet the still frame saved from the identical .ts stream on the Corsair Voyager with it's linear PSU seemed to reveal more " sparkle" from the jewellery. It was as if there was a slight haze around the other version. This experiment was suggested by a Melbourne DIY Audio member who also hears these differences.

 

Regards

Alex

[Paul R]

Try floating it and see! I'll bet you a six-pack the picture and sound will noticeably improve right away.

 

This is almost funny Alex - you of all people unwilling to try out something because you don't see how it will work.

 

-Paul

 

Paul

Further down the track I intend replacing the bridge rectifier diodes in my Oppo 103's SMPS with fast, slow recovery types.

I will probably need to make a tiny PCB to use individual diodes though. With my old Oppo 981 that also resulted in better audio, as well as a glossier picture via HDMI. IIRC, your Oppo has linear PSUs.

I have quite a few video clips derived from VEVO and Youtube .mp4 music videos. I convert them to BluRay specs with 16/48 Audio instead of AAC 16/44.1 , and use a tiny amount of " sharpness" to help compensate for the slight loss of detail due to .MP4 encoding at lower than optimum bit rates. I then save the best ones to a Corsair that can plug directly into my 40" Samsung LED backlit TV, and export the audio to my DAC. Some of your U.S.A. SNL derived material looks and sounds fabulous. I just wish we could get our local TV stations to transmit it as you guys see and hear it, often in 5.1 surround as well.

As I said in an email, I didn't get my Oppo to mainly play " spinners" although it can do that better than my previous Oppo 981.

Being behind Perspex doors is enough to stop any mechanical noise from annoying me. I don't believe that hip joints will be of any benefit to me, over existing measures, when playing from USB memory, without any mechanism noise.

Regards

Alex

[Paul R]

Alex - you are just wrong here. Yes, I have put the 105 on sorbothane, and yes, even on a carpet square.

 

It is not doing the same thing, and does not have even close to the same result. There is a big difference between isolation, which you get when you float something properly, and absorption, which is also a good thing, but not the *same* thing.

 

Stuff has been floating here for years, it makes an astounding improvement in things. I'd float the speakers if that were not a disaster waiting to happen (1 dog vs. 2 cats.)

-Paul

 

 

Paul

As previously stated, the Oppo 103 is sitting on a piece of car sound absorbent material which is designed to do similar.

Have you tried sitting your Oppo 105 on the carpet to see how that compares ? Perhaps you don't need carpets in Texas though? IF I ever start using the Oppo 103 more with those shiny round things, I will try your suggestion.

Are you sure you haven't been smoking those joints ? (grin)

 

Alex

[Paul R]

Significant. You seem to insist on talking about about the vibrations caused by the device, which is not at all what floating a device does.

 

-Paul

 

 

Paul

What kind of improvement does it make to playback from USB memory vs. playing a disc ?

Perhaps posting a photo of your listening area may make things clearer to me as to why you need to go to such extremes when not playing shiny discs ? I wasn't talking about on a carpet square, I was talking about on the carpet on the floor. That's very different to on a simple carpet square.

 

Alex

[Paul R]

I understand your point. I just find it a little amusing that someone who usually has an open mind is making a like a mule on something like this.

 

The improvement with isolation is in fact, every bit as significant as, or more significant than, upgrading power supply. (grin)

I think your system is more coupled than isolated.

 

-Paul

 

Paul

And I keep insisting that it is not a problem for me with my gear inside a wooden cabinet behind Perspex doors ,sitting on a shelf using a piece of car sound deadening material under it, speakers on a carpeted floor using cones, and non earth shattering listening levels. I am more interested in reducing the internal vibration from the player's mechanism when playing discs, just as I do when ripping discs on my PC, not further protecting it from room resonances, vibration from speakers etc. That's why my previous Oppo had quite a bit of 3M 2552 tape fitted to it as well. To further clarify. My listening area is in a large converted bed room. I simply do not need to use the kind of power levels you may need in a much larger purpose built listening room. You are likely to find that most listening rooms in major cities are much smaller than in urban areas, so power requirements are considerably lower, as is also the risk of room resonances and vibration.

 

Alex

[Paul R]

I have some nice rugs up on the wall my wife picked out at the first reflection points, and of course, our listening couch is positioned correctly for maximum imaging and bass. There is also a bass trap or two around the joint, cleverly disguised as bookshelves. We also use from diffraction and absorption behind the speakers.

 

None of which will isolate your equipment.

 

-Paul

[Paul R]

That's fine. I didn't mean to imply your listening environment was poor (pot calling the kettle black here, Mr. Mac Hater? ) - I am sure it is a pretty fantastic setup.

 

Just that you seem to have missed an entire area which might provide a nice improvement. That might also be a possible factor in some of the differences you can hear. Over and above power supplies that is.

 

I am pretty sure it is a factor in the disk audible differences question, though not so sure I will state it as a fact anywhere but here.

 

Yours,

-Paul

Paul

Let's just choose to disagree here. If my listening environment was as poor as you seem to think it is, I wouldn't even be posting my findings in C.A. or elsewhere.

 

Alex

[Paul R]

I just can't resist - your headphones have spinning disks? I gotta get me a pair of those...

 

Most of my findings are initially based on headphone listening, not via speakers.

Nice try though. (grin)

[Paul R]

Plus 1 on this for me too- I am about to have some extensive electrical work done on my house, and figure it makes sense to run some new circuits as well.

 

Three 30amp circuits seems a bit much for an audio system though, at least to me. What in heaven's name are you running that can suck that much power over the lines?

 

Better yet, where do you live? I might need to move there, as you must have a great deal from your power company!

 

Yours,

-Paul

 

 

 

Coxhaus: Can you provide a link to specifications/design of the big "power isolators" you use in your mains panel? There are a number of different techniques and technologies which attempt "power isolation." I am always interested in what other people have implemented for their dedicated A/V rooms.

 

Thanks,

ALEX

[Paul R]

Oh yeah! I'm up in Goerogetown but I work in Austin, right where 183 and I35 cross. We can certainly meet up if you would like to!

--Paul

 

 

I think we live close as I live in Elgin Texas and drive into Austin 2 or 3 days a week. It sounds like you live in Austin Texas which is only 25 miles away. If you want to here some of what I am talking about, I bet we could get together.

 

lee

[Paul R]

Man, now I'm jealous. Two CA audiophiles living close enough to get together! I live near Yosemite National Park--great vacation destination--maybe someone will come visit with me someday...

 

Hey, we are planning on spending a few days checking out the Yellowstone Caldera next year! I'm always good for a beer or two.

-Paul

[Paul R]

Just a followup on my own little testing here. Please do not take this as anything more than some interesting results. They don't really mean anything but do suggest some avenues to explore. All testing was conduced by me in a "sighted" environment except for the final pass where I asked my partner to A/B the different choices. This was difficult as it sometimes took 5 mins or so between tests to set up the test. Tests were conducted on the main system in my signature, and involved playing "Time" from DSOTM. (Hi Res 23/96 Edition.)

 

As noted, removing the drive from a Mac Mini and using external Thunderbolt connected storage made a surprising difference in sound to me. But being a suspicious old cur, I came up with and ran a small testing schedule over the past few days. Here's what I did and what I found. Basically, it looks to me like it both the type of connection for the external storage that really makes the difference.

 

Anyone care to suggest a test pattern with the minimum number of steps that would reveal different or more accurate data?

 

-Paul

 

 

[TABLE=width: 500, align: left]

[TR]

[TD]Mac Mini[/TD]

[TD]Drive or Drives[/TD]

[TD]Connection Type[/TD]

[TD]Findings[/TD]

[/TR]

[TR]

[TD]No internal drive[/TD]

[TD]Drobo local[/TD]

[TD]Thunderbolt[/TD]

[TD]Music quality seems to improve[/TD]

[/TR]

[TR]

[TD]No internal drive[/TD]

[TD]NAS appearance from SAN[/TD]

[TD]iSCSI[/TD]

[TD]Music quality seems to be equal to or better than Thunderbolt. I cannot tell for sure.[/TD]

[/TR]

[TR]

[TD]No internal drive[/TD]

[TD]External SSD[/TD]

[TD]Thunderbolt[/TD]

[TD]A little better than an internal drive/SSD[/TD]

[/TR]

[TR]

[TD]No internal

drive[/TD]

[TD]External SSD[/TD]

[TD]USB3[/TD]

[TD]A little better than an internal drive/SSD[/TD]

[/TR]

[TR]

[TD]No internal

drive[/TD]

[TD]External SSD[/TD]

[TD]Firewire 800[/TD]

[TD]A little better than an internal drive/SSD[/TD]

[/TR]

[TR]

[TD]Internal

drive[/TD]

[TD]SSD[/TD]

[TD]N/A[/TD]

[TD]Sounded very good, but not as good as the connections above[/TD]

[/TR]

[TR]

[TD]Internal

drive[/TD]

[TD]WD Spinning[/TD]

[TD]N/A[/TD]

[TD]About the same as internal SSD[/TD]

[/TR]

[/TABLE]

[Paul R]

We will definitely do just that! I am beginning to wonder about this again, and that always leads to lots of experimentation and listening. Also a good excuse to get some new headphones I suppose, since I then can do a lot of this at night. I wan to see if it makes any difference using MacOS or Windows, which is pretty easy to do with everything but Thunderbolt.

 

-Paul

 

You know Paul we should get together before I spend more money on my server and compare your MAC mini to my server. I don’t know about MACs but if it sounds better to me I would be willing to buy one and mod it.

[Paul R]

Well, playing around tonight, I figured out I had somehow turned off the memory play option in JRMC. Once that setting was restored, the differences that I was so surprised about almost disappeared. Though surprisingly, Thunderbolt and iSCSI still sound better than the local drives. I think that is because they are cached differently than other connections, but that is only a guess at this point.

 

I cannot believe I missed something so simple, but I did. That pretty much invalidates all the testing I did over the past week. (I mean, it invalidates it to me.) Back to more testing.

 

-Paul

 

 

Nice report Paul. You might enjoy some related tests I made and just posted about at: http://www.computeraudiophile.com/f8-general-forum/uh-oh-i-beat-my-sd-card-trick%3B-bypass-your-ethernet-switch-and-make-your-external-drives-sound-close-ram-disk-using-apple-thunderbolt-ethernet-adaptor-and-second-network-connection-18475/#post277280

 

Have a good weekend,

ALEX

[Paul R]

iSCSI is a little bit more complex than you may be used to with a NAS, and it really works best if you have two ethernet adapters, but it works okay with just one too.

 

Check out globalSAN iSCSI Initiator - SNS (Studio Network Solutions) which I use, and the Atto Product at Xtend SAN | iSCSI Initiator for Mac® | ATTO, which I own but don't generally use.

 

They both work very well. You do need to build a SAN with iSCSI capability (or a NAS with iSCSI capability) first. You can easily use a Linux system as an iSCSI SAN.

 

-Paul

 

Paul: Now you got me very interested in exploring the iSCSI protocol! I had not heard of it until now, but five minutes with Google and Wikipedia has me wondering if there is anyway to do iSCSI targets with OS X and regular hardware. I don't see a NAS in your signature (but I do see you run Windows). There seems to be Mac s/w for iSCSI, but without the time to study it looks like it is just for use with certain vendors' NAS hardware.

Can anyone here tell us more about this?

 

Thanks,

AJC

[Paul R]

Turning memory play back on made the sound better to my ears, but of course, that easily can vary between other ears.

 

It doesn't mean I did not hear what I heard. It just means that what I reported should not be taken as credible (or "valid") because I made a mistake.

 

To be honest, I would not have thought the different connection methods would make a difference in any case, but when you have facts in your face, you don't ignore them because you are uncomfortable with them. At least, I try very hard not to do so.

 

Ram sounding different? Wow - I never thought of that. If RAM does sound different, there are a whole lot of other variables that would have to be examined to account for it. Speed and buss of course, but also DMA, DMA controllers, processor arch (pipelining) and gosh knows how many others.

 

Someone needs to build a matrix with everything that affects audio playback in a computer taken into account, and put in an order of precedence or importance. Perhaps build a testing "checklist" that several people can work on, changing just one factor at a time? OS, Player, and Player Configuration, DAC interface, and Power are at the top of my list, though each one of those has plenty of sub categories.

 

I wonder if it could be made fun enough to interest enough people to get some difficult to dispute results?

 

Just brainstorming...

 

P.S. What configuration are you using with JRMC? Mac or Windows? There are some other settings in Windows that make a lot of difference, WASPI, etc.

 

In General, under Windows, these are the settings I am using...

 

-Paul

 

 

 

I don’t see why not setting memory playback invalidates your testing. If setting memory playback makes them all sound the same this is not necessarily a good thing. So you are saying the bad ones sound better and the good ones sound worst because they all sound the same now.

 

What if my RAM does not sound as good as your RAM? I don’t know whether this is true or not but lately I have been questioning everything on the computer as affecting the quality of sound.

 

I am going to test memory playback with JRiver again. I have not tested version 19 so it might of changed.

[Paul R]

Hi Paul,

 

Are you iSCSI and TBolt powered externally or through the motherboard?

 

External only.

[Paul R]

Just plain old fashioned 1gb copper is all I am using to the Macs Lee. Not really much of any other choice with the Minis or the Powerbooks. We have an older Mac Pro here, but the IBM QLogic cards I have do not work in it and I am not going to spend that kind of money to put an old machine on the SAN.

 

-Paul

 

 

Hey Paul, are you using fiber or copper with your ISCSI? I thought about it many years ago when I was working. I was going to set it up in my home rack of servers and virtuals but I just don’t run enough servers any more to share a set of ISCSI drives.

[Paul R]

I did not know you could run Windows with no paging file. That doesn't foul up the way Windows works? A Miracle!

 

Under Unix/Linux, I always have paging space (i.e. swap space) available, but I try to tune the system so that it is never used.

 

-Paul

 

 

I did not think to mention this before but I always use no paging file which some of you know as no virtual memory. It seems like Windows uses virtual memory even when it does not need it. I don’t want the OS writing to my hard drive when I am trying to read from it. I feel like it interferes with the music stream.

[Paul R]

That depends a bit Lee- if you are running Windows on a Mac Mini, then it absolutely is a true Windows environment. If you are running MacOS, then it is a Unix environment.

 

But in any case, unless one is moving to true high end storage gear, everything but RAM is going to be what is called extremely deterministic. That is to say, its performance will depend greatly upon the exact use to which it is put.

 

It may be far slower writing than reading, or perform much better with sequential operations than with random access operations. In any case, I would suspect that for music, the idea is to get the device that is the most quiet, electrically speaking. That does argue against using a RAM disk. Yet we have people that find using a RAM disk provides the best playback sound.

 

In any case, it should not be difficult to boot a Windows machine over the network, which is pretty much the easiest way to run Windows from a RAM disk. VMWare provides some easy ways to do this as well, using network delivered VMs.

 

Or for those wanting to enter the high end in computer terms, something like this might work. Be warned, it is high end in cost as well as performance...

 

xpress — Kove

 

Paul

 

I think I want to wait for someone to post good results to running Server 2012 using a RAM disk. I am leery of this as I don’t like the play from memory option in JRiver 19. I also think the windows environment is a lot different than the Apple mini environment. I don’t think you will be able to dismount the boot drive in windows server 2012 like you do in the Apple world.

[Paul R]

If you have any page faults at all, you have at some point, exceeded the available/usable/free/os-dependent memory in your system, and should NOT turn off the page file.

 

Indeed, Windows seems to want to use a page file even when it has absolutely no need to.... so I usually just leave some paging space available on the system.

 

-Paul