|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
87.114.76.91
In Reply to: RE: I play audio files off a RAM disk posted by Tony Lauck on July 10, 2014 at 12:04:50
So, the cable is still there; you are introducing layers of complexity and there is no good reason to believe that there is zero interface issue for RAM playback. It's just that this is 'faster' but so what.
Follow Ups:
The discussion in this thread has been about cables. There are no cables associated with RAM disk playback. There appears to be lower CPU utilization as well, consistent with higher speed.
The SATA cable is not carrying any signal. It seems unlikely that it is creating any noise just sitting there and doing nothing. Possible, but unlikely.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
This is a weak headed comment. To get the data into RAM, the cable is involved, so does a whole new bunch of circuit board traces and sockets when dat is released from it instead of from the hard disk.
"This is a weak headed comment. To get the data into RAM, the cable is involved, so does a whole new bunch of circuit board traces and sockets when dat is released from it instead of from the hard disk."
Huh?
The SATA cable is involved only when the data is loaded into RAM. That could have happened the night before, but usually it as a minute or two before I start playing the music.
As to those motherboard traces, these involve the processor and I/O controller connections to the DRAM modules plugged into the motherboard. These traces are used for nearly all operations in the computer.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
... from the point of view of interaction with the rest of the system. The fact that no music data is being read or written, doesn't mean there's no activity - at the very least, there's OS monitoring the hardware's health.In other words, I'm sure SATA cable will affect the SQ, even if file is fully loaded into memory before playback - but mechanism is NOT that of data being "screwed up", passing through the cable before loading.
Edits: 07/11/14
Do you have any information as to how often and/or how much "health" or "keep alive" interaction goes over SATA cables in Windows operating systems? I suspect this may depend on how the BIOS and OS have been set up.
There isn't a lot of traffic to/from the disks. The disk I/O light on the PC flashes all the time when I play music off the hard drive, but it rarely flashes while playing music out of the RAM disk. The O/S seldom goes to disk as most of its activity seems to come out of cached RAM.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
"The O/S seldom goes to disk as most of its activity seems to come out of cached RAM."
Besides, if your OS disk is separate from your external disk holding your music library, any non-cached OS interactions will be to your internal disk which should not have a long exposed SATA cable outside the shielded computer enclosure.
As an aside, with the Mac Mini or the PC, if I use an external disk for my music library, I can unplug it while music is playing out of RAM. There is no 'activity' to the disk at this point. How could there be, it's unplugged!
Not sure about Windows but you'll find very little OS activity 'chattering' to disk in UNIX based OS's like Mac OS. That's why, in the old days anyway, one would run the UNIX 'sync' command to flush (RAM) cache to HDD before invoking the 'init 0' forced shut-down.
From the Terminal screen on my Mac
If I run the OS off the Mac Mini internal SSD (as I normally do), and the music library off the Mac Mini internal HDD, there are no SATA cables involved. Both disks are direct attached via very short flex circuit header, and tightly shielded within the machined aluminum casing.
I don't have any powered up external drives. All my active drives are inside my PC case. I think it is a mistake to do otherwise. I do have an E-SATA header and cable that goes to a Blac-X box. The Blac-X provides a slot that accepts SATA data and power connectors and allows me to use "naked" SATA hard drives for external backup. This device is used once a month or so for off-site backups. I suppose that the idle SATA cable from my PC case to the Blac-X box is somewhat degrading my sound, but I doubt the sonic benefits are sufficient to justify grovelling in the dust on the floor while doing the requisite extensive listening tests. (Life is too short to spend time trying all possible tweaks. So I haven't tried testing the effect of tying bowlines, grannies and square knots in the window blind shades.)
Windows drives can be configured to flush or not flush the cache. I normally set them to do lazy writes, which means they may take some number of seconds before data gets written to the drive. For my E-SATA drive I have an add-on utility program that allows me to safely remove this drive, even though Windows thinks this drive is purely internal. Plus, the Blac-X box has some rather garish lights to indicate on-going I/O activity. Of course this paranoia is completely inapplicable for drives that are "read-only".
I looked at the SATA spec. The drivers are floating when no frames are being transmitted. I suspect that an idle SATA data cable causes most of its interference through the ground wires, so I am not sure how shielding is going to help. Also, there aren't going to be any data errors passed on, just dropped frames due to bad (32 bit) CRC's. I do have no doubt that an active SATA cable could radiate all kinds of noise that might affect other equipment (that shouldn't be sensitive to the interference).
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
The SATA cable is involved only when the data is loaded into RAM.
On the face of it, that would seem almost self-evident but it's not, in my experience, the whole story.
My audio PC uses a good embedded mobo (Fit-PC2) with a John Swenson designed (i.e. good) C-L-C PSU and two TPR regulators, one at 12v for the CPU and another at 5v for the SSD. The Win_XP OS is "slimmed" to within a inch of its life - data are read off the SSD at boot time only. I use RBCD files exclusively without upsampling. There is no KVM - music and control data are passed via LAN. (Though the system hangs at the end of the current track, I have in the past switched off the SSD during play without ill-effect. That persuades me that the SSD is not accessed during play.)
The drive would normally be powered via the mobo's SATA socket but I provided a dedicated 5v line as much, to be honest, because it was cheap and easy to implement as the result of any systematic testing.
Not long ago, wanting to pinch the 5v regulator circuit to use elsewhere, I popped the SSD back into the Fit-PC2. There was an unexpected but dramatic drop in SQ. The SATA interface might well not be passing data but it was certainly doing something deleterious - the drive was back on its 5v reg PDQ. I cannot say how "busy" SSDs are when nominally idle but my guess is pretty busy.
Bits-is-bits wallahs can ignore the issue as they choose but I was certainly reminded of it on this occasion.
"There was an unexpected but dramatic drop in SQ. The SATA interface might well not be passing data but it was certainly doing something deleterious - the drive was back on its 5v reg PDQ. I cannot say how "busy" SSDs are when nominally idle but my guess is pretty busy."
It doesn't take activity on the SATA bus for the SATA cables to act as antennas to re-radiate the common mode energy that could be on them from some source other than the sata bus in a typical PC type computer. Often the loudest noise radiated from a cable has nothing to do with the signaling intended to be on that cable, especially in the case of low current low voltage signals such as signalling on the sata bus. Though the data rates require pretty good edges on those signals so the radiation wont be zero.
Whatever the source, shielding can reduce the intensity of the noise if done correctly. However, addressing the common mode problem at it's source will most always be the more effective and efficient path by which to beat it down.
I'd be willing to guess the most likely source of the loudest radiation from most sata bus cables is the flybacks in the smps or the bucks on the mother board.
Of course, sufficiently hardening the the circuits likely to receive unwanted ambient energy and also work to prevent the reception in the first place for any equipment likely to mix the energy back into the audio stream might be the best approach. Can't keep the whole world quiet these days.
"Bits-is-bits wallahs can ignore the issue as they choose but I was certainly reminded of it on this occasion."
Who's ignoring these issue? Some here seem more inclined to think logically about solving these types of problems than others but no one seems to be ignoring it that I've seen.
It doesn't take activity on the SATA bus for the SATA cables to act as antennas to re-radiate the common mode energy
Thanks for your comments. The one above more or less echoes the point I was trying to make. I say "more or less" mindful that I should have made it clear that the HDD/SSD in a Fit-PC2 connects direct to the mobo - see pic.
My powering the drive separately involved moving it out of the case and connecting it with a modified Slimline adapter (data cable to the mobo, power cable to an external PSU). In this case, SQ improved despite adding a SATA lead. IOW, though I accept your general point, it's not altogether relevant in this case.
I'm not disputing the benefits of shielding cables and am baffled as to why anyone should see fmak's point as controversial. What I was disputing was TL's suggestion that "[With data in RAM,] the SATA cable is not carrying any signal. It seems unlikely that it is creating any noise just sitting there and doing nothing. ", "The SATA cable is involved only when the data is loaded into RAM. That could have happened the night before, but usually it as a minute or two before I start playing the music" and more besides. [Emphasis added]
For reasons including those you outline, I'd have thought it inevitable that SATA cables radiate noise all the time and that HDDs/SSDs continuously generate it regardless of whether they are performing RW operations.
Wow, that internal mounting is pathological. It's bound to shuffle the EMI environment around even if unplugged. True, the compulab boards tend to have pretty tight current loops but still... No matter what you do there are gotcha's, packing stuff in is usually good for radiation but not so hot for crosstalk.
Watching the Scottish open, but it's in unreal time.
Regards, Rick
Cool little computer there.For me the controversial part is spending money on something which may or may not help. Details about what was done seem important but hard to get in this case.
For example an improperly applied shield might not help a common mode problem at all. Common mode problems can be tricky that way, ie is your shield ground actually ground or is it some resonating pseudo ground which causes your "shield" to become part of the problem? Usually addressing the source, beads, or some other form of filtering are easier than creating a proper shield when faced with a cable radiating common mode energy.
The details of the nature of the shielding seem especially important to me in the case of situations where noise problems are occurring during periods of little to no sata bus activity as it does sound like a common mode problem.
Edit: deleted text from here.
Edits: 07/12/14 07/12/14
Why and how would you assume that a shielded sata cable may have an improperly applied shield? It's in the cable specification and shielding noisy signal pairs is standard in electronics.!
Seems to me that you have the time to argue for the sake of it and it would be much easier for you to try a shielded v unshielded cable in your system to see if there are any changes. Then you can postulate about common mode effects to your heart's content.
There is no reason to think that motherboard ground planes, component specification, and trace location satisfy the criteria you postulate about.
As I explained in that post, if the shield tie point is not actually ground as can be the case with typical common mode problems where the entire chassis, signal ground etc. is changing potential with respect to earth at the resonating frequency then tying your shield to it does no good to solve these common mode problems and maybe even makes things worse by tie-ing in even more conductor and effectively enlarging the antenna.
"There is no reason to think that motherboard ground planes, component specification, and trace location satisfy the criteria you postulate about."
No reason unless you happen to be living on planet earth and are subject to the laws of nature that is. This will be happening in every PC at least to some degree. As I've said before, if this is the most significant noise problem amongst the noise problems attempting to be solved then shielding from the sata bus signalling activity might not even noticeably help. It's just one of many examples I could have used. It just happens to be one which is likely to be a more significant problem to more readers here than sata bus signaling radiation and is most often the result of bad power supply design but can be fairly effectively dealt with in most cases by proper rf grounding of the audio gear,including the PC, as much as possible.
That is why I brought it up by attempting to get more details about your system. Again, I have no way of knowing your system. Even when asked you seem reluctant to share the details.
For an illustration about the pc chassis quietness try using your scope to measure your pc chassis to a another unloaded breaker circuits ground. You will not need the most sensitive ranges of your scope to see signal even in an otherwise quiet rf ambient environment if your pc is typical.
you seem to have a lot of time postulating and imagining. CMRR issues are common not only due to grounding but poor CMRR in active components and trace routing. It is not the only issue in computer audio and the issue I posted was that of an unshielded sata cable either radiating or receiving interference.
No need to postulate or imagine.
Fact is you neglected to give proper context to your statements and so I tried digging deeper.
I hope I get the chance to repeat this thing I've stated over and over some more because you seems so great at comprehending it.
It would be too much to expect reports that correlate listening observations with measurements, especially since these reports don't contain detailed particulars of what was listened to. I would not expect measurements from those audiophiles who don't claim to be technical, but those who have already posted pictures of their test equipment and boasted of escapades involving scopes and soldering irons are certainly capabible of such.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
I should never post before drinking my coffee. My second reading of your post made me realize this statement I made above."In summary, based on what I've heard so far I'd be more inclined to try some cable beads than a shielded cable in your situation. Of course it's damn near impossible to make a good armchair assessment on this kind of thing so I'd definitely take my words for what they are worth...not very much. "
does not make much sense to say to a someone who is saying adding a cabled setup reduces system noise. Not that beads wouldn't help but I'd be less inclined to suggest they might in light of my new found comprehension of your post.
Sorry for confusing the topic. I'll edit the previous post accordingly to remove this unneeded comment.
Edit: by the way I definitely concur with rickm's statement about crosstalk above. There is only so much in layout which can be accomplished to prevent the near field coupling due to high di/dt current loops, and high dv/dt hot nodes. Presumably the biggest culprit would be the current loops in low voltage computers like this but those edge rates can be very extreme. In cases where layout is optimized there is not much left to do to it than shielding which would be tough/expensive to pull off well in those close confines and so proximity to the fields is your enemy.
Edits: 07/12/14 07/12/14 07/12/14
If there is no data going over the SATA cables (as in my case) and the signal drivers for these cables are not enabled (as per the SATA specification) then changing the SATA cables is akin to changing any other wire in the box that is not directly carrying a digital signal. It will probably make less difference than changing a power wire, including a SATA power wire to a device that has active digital processing (e.g. a SSD or spinner, both of which have a microprocessor inside).
The problem is that this is all speculation. Unless some method can be devised to give rapid, repeated and reliable measurements that can be correlated to long term listening results, making progress by trial and error listening tests is going to be slow, akin to searching for one's car keys in a dark room without a flashlight.
IMO effort would be better spent on simple devices to provide analog digital separation using a narrow interface between the DAC and the computer. The narrowest possible interface today is the IIS interface with clock at the slave end or the DSD equivalent. Here one is concerned with the operation of a few hundred transistors and the necessary ground and power wiring, not billions of transistors doing unknown functions according to impenetrable software. If a USB DAC is used it must have this isolation built in, since the USB attachment in the DAC is itself a computer due to the complexity of the USB protocol.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Exactly, even if it's 20 years ago, the data still went thru the cable
There were probably SATA cables the data went through when the album was mastered. If a CD it probably went through a SATA cable when the album was ripped and stored on hard drive. And if it was a download, it probably went through a SATA cable in the remote server that provided the download. Are you suggesting that these (historical) cables affect sound quality?
I can be reasonably confident that no audio data went over a SATA cable 20 years ago, as SATA came into use in 2003. :-)
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
you are seeking to change the subject again.
Any data you send to RAM has to come thru the Sata cable.
is DIFFERENT depending on the hard drive cable and that the data sitting in RAM overnight is different data depend on the cable?
As that data sitting in RAM is just zeros and ones you should be able to compare that data.
Done that yet?
"As that data sitting in RAM is just zeros and ones you should be able to compare that data."
Fred hasn't, it was in my RAM disk. :)
And yes, I am certain that the data in my RAM disk was not corrupted by my SATA cable. I am not going to bother to explain how I know this, because some inmates probably wouldn't understand and there would be even more useless discussion.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
for your refrigerator and be done with it.
Whether you unplug the refrigerator or not, I'm sure the new power cable will make a significant improvement in the sound of your system.
The power cord to the refrigerator likely makes no difference in the sound quality of the your system as long as it's unplugged!
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: