My deep dive into media storage interfaces: Musical differences heard between chipsets, Firewire 400/800, USB, SATA, flash drives, SD cards, and network shares (Warning: may cause seizures in the DBT crowd and flat earth naysayers)

[Superdad]

Greetings everyone:

 

First off, I noticed a couple of similar threads started this past week, but given my approach and the information I have been working towards presenting, I'm taking the liberty of starting a new one. This is going to end up being a VERY long post, but it sure won't be as long as the process it took to test this all out. I have spent at least 15 hours on this particular adventure so far.

 

There may be a lot of boring details in the middle of this report, but be sure to stick around to the end where the big surprises are. Well, some of the surprising conclusions should be no surprise at all…

 

 

It all started because I need more drive storage space for my music library (all Redbook and higher-res in AIFF, some DSD tracks as well).

I won't bother with much system/room background--many of you already know my set up or can look at profile album pics--but I will refresh on the computer setup: Headless 2010 Mac mini 2.4GHz Core 2 Duo w/8GB RAM, Mavericks stripped, Audirvana+ 1.5.10 w/custom iZotope SRC settings feeding 176.4/192 (soon to be 352.8/384!) to WaveIO XMOS-based async-USB>I2S interface (mounted inside DAC and not powered at all by USB bus); Rest of DAC is Hovland/Swenson NOS PCM1704 and discrete FET/JFET I/V output stage; multiple Swenson choke-filtered power supplies inside DAC chassis.

 

I have always liked the OWC (Other World Computing) Mercury Elite series of hard drive enclosures. They are fanless and quiet, they use proven Oxford Semi chips for their drive bridging, and they have standard barrel connectors taking external 12VDC. Back in the IDE/ATA days I used their plastic/steel shell enclosures (even quieter), but since the takeover by SATA drives it's all aluminum.

 

I had a 1TB drive on hand to replace my 500GB main music drive, but I discovered it was starting to fail on reads (kind of scary since that 1TB had been my family back-up drive on our LAN). Since Jud here at CA, and a few others before him started reporting on some sonic differences between drive connections, and since this was one area I had not gotten around to testing, I figured I'd give it a go. Mavericks and A+ 1.5.10 my system has been so revealing--and incredibly sensitive--if something was to show up, now would be the time.

 

The multiple generations of OWC drive enclosures I own each have a different compliment of ports. Initially I figured I'd just try the one drive with Firewire 800, FW400, and USB ports. Of course my Mac mini has only a 9-pin Firewire 800 port, so a little 6pin>9pin 400>800 adapter dongle would be used during the process.

 

MORE PREAMBLE AND DISCLAIMERS:

------------------------------------------

I went into this exercise not expecting to hear much, and figuring that FW800 would be what I would use with the larger drive, simply for speed of copying tracks to it and backing up my library (the 500GB drive my library has been on was in an enclosure with just FW400 and USB2.0; and it was always daisy-chained into the FW400/800 1TB).

If I had any expectation, it would be that USB would not be the preferred attachment since conventional wisdom says so (for various reasonable reasons) when using a USB DAC. Also, Jud had reported not liking the sound when he was forced to use USB3.0 temporarily this week.

 

Regardless, my general methodology of using my suite of most difficult and nuanced test tracks, changing only one variable, and switching A/B/A/B/A--and allowing for "no difference" as an option--gives me confidence in what I hear. (I request that all the DBT people and sighted-listening naysayers stay away from this thread.)

 

I do want to be clear that I still believe that checksummed, bit-identical files--no matter how many times they get copied back and forth and around the world--still sound exactly the same. (Sorry SandyK)

 

But I am am now fully convinced, and I hope my experiments show and inspire others, that, just as playback is effected by the realtime interface between computer and DAC (ground plane noise, jitter, etc.), whatever tiny noise currents that go on in the INTERFACE between the actual storage media (hard drive, SSD, flash) and the computer do play a small role in the overall presentation.

 

Before I get into all the tech talk, I want to assure you all that my listening tests occurred BEFORE I made any deeper examination of chipsets, etc. I really did not have a bias or agenda towards any one interface, and if I did, I did like I always do and check that once the music starts. One really can not discern subtle musical differences at this level if ANYTHING other than the music is going through your mind while listening. And the NO-DIFFERENCE option is always an option for me! (There were a number of instances during the process where I heard no difference--the USB 2.0 connections all sounded about the same.)

 

To be honest, and I don't mean this to sound elitist, I suspect that some of the differences I'm reporting will be tough to pick out on a majority of even high-end systems. Yet I am sure that most of you would hear and understand them if you spent a half hour in my room. Actually, those of you who have had no trouble both discerning characteristics of the progression of Audirvana Plus versions this year, and who are also able to tune your iZotope SRC settings in small steps--you're hearing is already dialed in to the sort of subtle, but realism-gestalt significant differences between the various interfaces.

And I hope and expect that most everybody will easily be able to hear the difference between the best and the worst data interface. You can start there and then see if you can hear the steps between the middle-ranked ones.

 

PROCEDURE AND DETAILS OF INTERFACES TESTED:

-----------------------------------------------------------

I took a freshly formatted hard disk mechanism (7200rpm Hitachi Deskstar with 32Mb cache) and copied 8 selected AIFF music tracks to it (mix of 16/44, 24/88, and 24/192). In the end I settled on listening with 3-4 of the tracks once the particular effect became readily apparent. I moved this one mechanism into each of the OWC enclosures (SATA interface is so easy with no cables!), and each mechanism got powered by the same stock 12V/3A external switching power supply (hooked up a big linear supply at the end of the trials, but that mostly just helped the bass).

 

So each time, only one drive and the same one power supply was plugged in. For each drive that had multiple interfaces, I listened to each one (except eSATA interface since Macs don't offer that port--too bad, it is in a way the most native, not requiring bridge processing between the SATA drive mechanism and the motherboard.

 

Playback was always via A+ playlist, with Integer, Direct, and Exclusive modes all on, Redbook upsampled to 24/176.4. I always loaded the track fully into memory before playback (select track, press play and then pause immediately, wait for grey bar to finish track preload). It is pretty crazy that even with a memory player I could hear all these differences. John Swenson and I have theories, but no measured explanation or proof--other than the fact that this is all repeatable (which I did over multiple days) and I would bet a good sum of money that I can pick the best two interfaces out of the crowd blind.

The fixed-resistor stepped attenuator level control of my Hovland HP200Pi was never touched between devices (there are a couple of tracks that I have to raise or lower the level by one step, but I am always consistent).

 

Here are some pictures so you can see some of what I was doing:

 

 

The enclosures were as follows (with notes about how they feel in the rankings):

 

OWC Mercury Elite Pro quad-port (FW800, FW400, USB2.0, eSATA) with Oxford 944SE bridge, LSI FW843-07 Firewire PHY. This was the worst sounding of the quad-port enclosures, pretty much regardless of port used.

 

OWC Mercury Elite Pro tri-port (FW800, USB3.0, eSATA) with Oxford 944 bridge and Oxford 3100 USB3.0 chipsets. I did not open this drive enclosure as it is new 2TB unit with a warranty seal sticker, so I just copied my file set to it. Oddly, this one ranked better--via FW800--than the above quad-port, even with the presence of USB3.0.

 

OWC Mercury Elite Pro quad-port (FW800, FW400, USB2.0, eSATA) with Oxford 924DSB bridge and T.I. TSB81BA3E Firewire PHY. This was the best sounding of the quad-ports.

 

OWC Mercury Elite Pro dual-port (FW400 and USB2.0) with Oxford 934SSA and LSI FW802C Firewire PHY. This was far and away the best sounding of the OWC enclosures. None of the above via FW800--or 400 where available--even came close. OWC still sells this as an empty enclosure for $40 (the web page still lists the Oxford 934 for it, I hope it still has the LSI PHY); I plan to buy one or two of these while still available. Some of you might want to also.

 

OWC Mercury-- This is the original plastic clamshell over thin steel case. It is for IDE/ATA drives only, has just FW400 ports and uses the classic Oxford 911 chipset and a T.I. PHY. It was okay, better than the quad and tri-port units above, but not in league with the still-available dual-port SATA Elite Pro.

 

MacAlly model PHR-100AF-- This is a discontinued (but still available on the web) Firewire 400 ONLY enclosure with a IDE/ATA interface (we call that PATA--parallel ATA--versus the now dominant SATA--serial ATA). This one is a bit odd because the only chips in it are an LSI PHY (the exact same one as in the good sounding dual-port OWC above--hmm…) and a big SST brand chip that is just firmware memory (those appear in the other drives too). So while I am not sure how this ATA interface works, I can tell you that it POSITIVELY SMOKED every other drive I tested! I really was not prepared for this. The only reason I tested it was because after all my other port and enclosure tests in which I ended up with the clear winner of the OWC dual-port, I looked around the house for one more drive with just a couple of ports to be sure I was preferring FW400 and not just a fluke chipset. I took this one off my wife's desk. Were it not for the fact that it is really hard to find IDE/ATA raw drives over 500GB these days, I'd take this enclosure for my music system. Also, its external PS supplies both 12V and 5V. I suppose it could be contributing sonically, but it is a hassle for when I go for full linear supplies on my computer/drive system.

 

 

INTERFACE RANKING AND CONCLUSIONS:

------------------------------------------------

Here is the ranking of the interfaces (again, all played off the same drive mechanism with a couple of noted exceptions, and all using the same cables for a given interface). From worst to best:

 

7) USB 3.0 -- (Obviously just via the one drive I had with 3.0, and of course using a proper 3.0 cable--but feed into the USB2.0 port of my mini as only the newest machines have 3.0 ports. So maybe I was just listening to 2.0 from a drive with a 3.0 chipset).

Just dreadful. All the "sameness" and "getting in the way of the music" confusion of USB2.0, combined with a nasty hashiness. Piano (on the Bromberg and elsewhere) was awful, as were cymbals.

 

6) USB 2.0 -- It is funny. In some was USB 2.0 was not terrible, and if you have a Firewire DAC then it will likely be a good sounding choice. But horns did not have the right "blattiness" and a lot of timing cues seemed off. Its effect across multiple tracks was of making then sound more the same than different. And the whole "air of the room" sensation I got with the better interfaces could be noticed missing from the first 2 seconds of the track. In my system I could name the USB2.0 sound blind--any day of the eek.

 

5) Firewire 800 -- Played fine without seeming to bother the computer/player s/w much, but fell far short of the "oh shit" grin I get with Firewire 400. Lowest level details missing, but were I really hear it is in the realm I talk about elsewhere: The degree to which the harmonics of the instrument seem separate or together with the playing of the note itself. Once I keyed into that it was easy to tell the difference. At first the FW 800 seemed to have this big and full bass with lots of overtones. But they were too much and not real. Also, the brushes and hits on cymbals revealed differences--and not in the favor of FW800.

As reported in the enclosure specific section, there was a difference between the FW800 ports, and I now attribute this to the different chip sets. I think it is more likely the PHY chip difference rather than the Oxford drive bridge chip differences.

By the way, for the FW800 test, I used a 24 inch cable that had the two wires for power cut. The nice folks at USB | FireWire - Angled - Short - Custom - Cables - 877-522-3779 will do this service for $10 on any cable they sell (apparently a lot of the super high-end video production people need this; the guy asked me if I was using a RED camera).

 

4) Firewire 400 -- (And of course this is with the passive little 6pin>9pin adapter; Power wires were still present--my 24" 6pin>9pin cable with power wires cut comes on Tuesday.) There is no doubt that for an external hard drive interface for Macs, FW400 is the winner. While this was especially the case for the enclosures with fewer ports, even the quad-port enclosures sounded best from their FW400 port.

 

I think this could have as much to do with the speed and processing required at the computer end. It all seems crazy because we are talking memory play after the track is loaded. But the computer still talks to the drive briefly during playback--just watch the drive light and you will see that some sort of connection "keep alive" signaling is still going on.

 

3) SATA -- The internal drive and motherboard interface of the 2010 Mac mini is 3Gb/sec. SATA. So at the end of the drive comparison it was the obvious next up for testing. No extra interface bridging between the drive and the controller, and right there on the motherboard.

In most ways it was indeed better than the external drives, but there was something about it that bugged me. There was nowhere near the distance between it and the dual-port FW400 drive that there was between the FW400 and the FW800. For the difference I could live with the FW400. And there was something bothersome about it--different than any of the other effects I had heard this week. I did not compare the internal SATA to the very seductive MacAlly FW400-only drive I had liked so much (it was already back on my wife's desk and I was tired at that point.)

 

 

THE POT OF GOLD:

---------------------

I did revisit some of the above the next day, and spent quite a lot of time enjoying the great sound from the chosen FW400 drive with numerous other tracks. And I hooked up this huge 12V/10amp Astron linear power supply to the hard drive. Modest improvement in the bass--with no doubt--take it out and it changes--(but why?!).

 

Then, after tuning in again with my test tracks (this is how one ends up hating the tunes they love), I took the next obvious step: The Mac mini has an SDXC card slot! And unlike USB thumb drives which have clock crystals in them and still have to go through the entire USB protocol stack when data is transferred, SD cards have simple one-bit transfer modes and require very little processor attention. USB ports have to accommodate all sorts of devices; SD cards are as close to just plugging in nonvolatile RAM as you can get.

 

Of course I need to compare a USB thumb drive here in the mix. And I know that some people really like flash drives, even going so far as to provide them with an external power supply. I am going to try a flash drive tonight, but based on what I heard this week, I'll risk giving USB thumb drives a provisional spot at #2, but that may in fact be a mistake since I was not at all partial to USB from the drives and thought they were somehow grunging up the USB to the DAC. (I'd test it right now but my music server is in the middle of cloning my 500GB library drive to the new single platter 1TB WD I just received and which will be swapped into the best of the OWC cases.)

 

UPDATE: I did just test a Kingston 4GB thumb drive compared to both my FW400 external and the SD card. I very definitely found it to be a mixed bag. On the one hand it seemed to have the both the "quiet" and the "harmonic singularity" of the internal SATA or the FW400, but layered on top of that was some of the same things I heard with the rest of the USB trial. It just not not get that last bit of coherence and misses some details. I'm guessing it's competing on the USB bus. I bet USB sticks would be a great choice for someone with a Firewire DAC.

 

#1) The big winner for me, and by no small margin, is the SD CARD! Holy cow. I almost could not believe it. So much is so right. All the qualities that kept improving as I moved up the chain through the quieter processing, lower speed interfaces were exactly the nature of the top of the mountain that the SD now sits on. The sound is lightening fast! No hash, infinite detail. I am so smitten! It really is a long way ahead of all that came before it.

 

For fun I went back to a couple of the lesser ranked interfaces, FW800 and USB from a drive. Yep, that's the comparison I'd play for a visitor--and even for a hardcore skeptic.

It is funny: While the differences are not at all of the kind you get when changing analog cables, I find it--on my resolving system--to be just as obvious, maybe even more so. However, it's been a while since I changed cables, with the exception of a modest USB cable that I added a few months ago. And I find this SD card thing versus a FW800 drive connection to be a much larger difference that the USB cable was (again, it was not a spectacular or expensive USB cable).

 

Since my Firewire 400 cable and adapter dongle still had power lines form the computer connected, for fun I unplugged it from the mini and from the wall. It helped only a tiny amount--nothing at all like the other very obvious differences I was sorting out.

I was using just a 2GB SD card I pulled out of my old camera. Yesterday I ordered a 32GB card for $23. Presently they go up to 256GB but get pricey. I'll just rotate some playlist tracks to it.

 

Aside from the many variables of room/system/aural sensitivity, etc., there are some computer-related reasons why some of you may have a different experience if you test some of the things in this report. Most specifically, in the case of SD card slots: My 2010 Mac mini has a single Broadcom chip controlling both the Ethernet and the SD card slot. So the SDXC card interface on my machine does not appear on the USB bus. However, as I found out from Jud the other night, on his 2009 MacBook Pro, the SD slot (not SDXC so limited to 32GB) appears on the USB bus, and thus SD might for him not be far ahead of other interfaces as it is for me.

 

 

SONIC DESCRIPTIONS:

--------------------------

I'll try to describe the most obvious sonic attributes that I heard vary between the interface methods:

On the worst-ranked interface the harmonics seemed separated from the fundamental attack of the note. On the big fast played acoustic bass of Brian Bromberg (the virtuoso All Blues track is what I am thinking of), the notes almost tripped over one another as the very prominent overtones formed a rich and overly ripe bloom. And on the slow-played, delicate solo piano of Bill Evans "Peace Piece," a similar effect where the contact of the hammer, the sound of the wood key, the forming of the note, and its harmonics were markedly less real than with the better interfaces. Evans' piano is not a especially good on this recording, but his nuanced and emotion-filled playing is. So when the system allows the plink and wood to sound real along with tone and overtone of the strings, you know it right away.

 

What was funny with the Bromberg, and with bass attack tone in general (such as drums, etc.) was that at first with the better interfaces I thought that bass weight and fullness were missing. After going back and forth a few times, it became quite obvious that the bass was just faster and more of a singular propagation. For example, on the Bromberg, the overtones no longer bloomed, rather the bass notes all came out as a simultaneous whole, allowing me to hear more of the low-level, in-between elements, practically to the point of seeing his fingers fly over the strings (certainly hearing more of his skin scraping the strings).

 

Two other words keep popping to mind as I listen tonight: Mellifluousness and speed. The best of the interfaces startle me both with the natural smoothness and true character of an alto sax or clarinet, and also with astounding start/stop speed. All overhang is gone (especially with the SD card) and the music pops into the room from out of nowhere. And yet when the notes played do sustain, they trail off in the air and I hear the faintest tail as the piano note falls away. I have been chasing sonic nirvana all my life and with the SD card this is the closest I have ever come.

 

The other major effect I noticed, especially with the very best interface, was the very quiet background. I know "black backgrounds" are kind of an audio cliche, but in this case it was true in an obvious way, and one that I think accounts for a lot of the extra realism I am hearing. Some of the recordings I use are quite old, and although I don't notice much tape his, I do notice the "air" of the studio in which they were made. This manifests itself in two ways: My room is sealed, VERY quiet (out here in the country) and my speakers go into the low 20s. So the low frequency pressure of the venue, and any trucks that rumbled by the studio during the session, are apparent. Secondly, there are things going on with people in the background of these recordings. I'm not referring to obvious clinking or mumbling, I'm talking about stuff that is buried WAY down in the recording. And on some tunes which I have listened to for decades (the aforementioned Bil Evans as one), I have started hearing the faintest wisps of new noises. I am familiar with the effect where once you have heard something, you can switch your system back the other thing and will still hear it. Well, not this time! I can hear things with the SD card that really do go away with the other interfaces.

 

Frankly I was and am incredibly surprised (and even a little dismayed) that the media data storage transport interface has revealed itself as a real factor. Of course many users and manufacturers have been experimenting with this for some time with C.A.P.S. servers, dedicated music server systems, SATA filter cards, etc.

So maybe I am late to the party. But there are a lot of Mac users here at ComputerAudiophile.com and some of our interface options are limited. So I hope that this report can serve as a pointer for your own experimentation. Since I'm not about to give up Audirvana+, I am staying with Mac for now. Fell asleep with my iPad in bed at 3:00 a.m. the other night researching motherboards for Hackintoshs that work with Mavericks and are low power like Chis's C.A.P.S. choices...

 

So you Mac owners with machines that have built-in SD card slots--give it a go! (I would not bother with external SD card readers as they have to attach via USB. And do first check in System Profiler to see if you machines SD is handled by the USB controller. If it is I'm not sure how that will sound.)

 

Oh, crap on a cracker! (as my wife is fond of saying): I just spent another 40 minutes shifting between the SD card, my FW400 drive, and the same tracks shared from my desktop i7 mini's hard drive Public Folder. Audirvana was just as happy to retrieve them from over the network (and through my new Cisco Gigabit ethernet switch). With my 2010 Mac mini, the data is served to the processor via the very same Broadcom chip which handles the SDXC card slot of this machine. And guess what? It sounds damn great! I was pretty shocked, and went back and forth a bunch--mostly in disbelief. But I can tell you that via the shared drive the music sounded WAY more like the SD card than it did like the FW400 drive. Did it beat the SD card? Not quite. I hesitate to describe its shortfall, because it was not at all like those of the USB or FW800 connections. All the detail and speed is there, but there almost a faint hash to cymbals and a lack of the last bit of space and ease.

 

But hey, this was just a first quick attempt for the drive on the network. And yes, I know that everyone has been using NAS setups for a long time. Suppose I have to teach myself how those can work with the Mac and iTunes/Audirvana. But sticking an extra machine on my network and sharing it is no big deal. Uh oh, I thought I would never have to tread into those really crazy waters of custom ethernet cables! I'll kick and scream before I go there. I am hoping the nonsense really ends at the back of the music server (i.e. the one attached to my DAC) and that this is really all about things being quietly processed inside the machine and my preference for the Broadcom chip handling the ethernet and SD card.

 

Madness to be continued. Thanks for reading.

I welcome questions and suggestions (other than that I be committed to an asylum).

 

Once again, please help me keep this thread from devolving into a debate over the audibility of this unmeasurable stuff. I'd rather go over to their threads than to sidetrack this one. I only wrote the truth of everything I heard. If some don't believe me that's fine. If you want proof, then you will have to come to my house for a visit to enjoy some tunes (I'm near Yosemite National Park, and my offer is serious unless you are a creep.)

 

Cheers,

Alex Crespi

[sandyk]

The Mac mini has an SDXC card slot! And unlike USB thumb drives which have clock crystals in them and still have to go through the entire USB protocol stack when data is transferred, SD cards have simple one-bit transfer modes and require very little processor attention. USB ports have to accommodate all sorts of devices; SD cards are as close to just plugging in nonvolatile RAM as you can get.

 

I would have loved to try that route too, but my Windows PC doesn't have a card slot.

[mayhem13]

It would be nice to get a comparison of SD and the Sata interface with an SSD since the similarity is no moving parts. I've heard similiar accounts from some using mSata equipped laptops top.

[sandyk]

The Mac mini has an SDXC card slot! And unlike USB thumb drives which have clock crystals in them and still have to go through the entire USB protocol stack when data is transferred, SD cards have simple one-bit transfer modes and require very little processor attention. USB ports have to accommodate all sorts of devices; SD cards are as close to just plugging in nonvolatile RAM as you can get.

 

 

Alex

I just did an ebay search to see if it was possible to retrofit a card reader to my W8/64 PC, They all seem to connect to onboard 9 pin USB headers. Does yours use a USB connection too ?

Regards

Alex

[Snowmonkey]

deleted

[Superdad]

Alex

I just did an ebay search to see if it was possible to retrofit a card reader to my W8/64 PC, They all seem to connect to onboard 9 pin USB headers. Does yours use a USB connection too ?

 

No, the 2010 Mac mini's SDXC card is controlled by a Broadcom Ethernet/SD Card controller chip: NetXtreme® Gigabit Ethernet Controller with Integrated SDXC Card Reader, ASF and EEE Compliant - BCM57765 | Broadcom

 

Here is a pic of the Mac mini board. Oddly, the Broadcom chip (highlighted in red square), is at the opposite end of the PCB from the SD card slot.

http://d3nevzfk7ii3be.cloudfront.net/igi/eBGMrGURJuJjcmlh.huge

 

I need to research other Mac models' SD card slots further to report which are on the USB bus and which use a separate controller like mine. It is easy for users with their Macs to just look up in About This Mac>System Report>Card Reader, and to also look to see if the card read shows under a USB branch.

 

Mayhem: I would have tested an SSD if I had one. I have been considering a small one for a while--just for the OS and player s/w. I read that the early ones had a lot of capacitor switching noise, but that newer ones don't. I do think the bulk of what I was hearing is caused by the bridging interfaces themselves, not so much by the drive.

 

So while I would not expect an external SSD connected via FW or USB to be any better than an outboard HD, an internal SATA-connected SSD might get rid of some of what I did not like about playback from the internal spinning SATA drive. That one already ranked pretty high at #3 (maybe #2 if I do another direct against the USB stick which also had flaws just because it was another process on the USB bus). It would not surprise me at all if an internal SSD sounded darn great. And while the internal SD card is the simplest, least protocol intensive interface, the network connection's strong showing makes me believe this is not over yet.

[Superdad]

Did you try Ethernet or WiFi? What was #2?

 

I did try ethernet--see 4th paragraph from the end--but not wifi.

 

#2 was provisionally USB stick, but read to see the limitations I heard. Right now #2 really is a drive shared from my i7 mini via ethernet. It is not perfect, but it is the closest to the sound of the SD card. I was really surprised. It bodes very well for an exciting project that John Swenson and I have been sketching out and buying development boards for for over a year.

[Snowmonkey]

Oops, I deleted my post, but you had already quoted me. Sorry, I didn't read your post thoroughly. Still, it might be interesting to try WiFi if only for completeness. I have excellent WiFi reception and I get the impression it sounds slightly better than Ethernet from my router. (I use a NAS). Cue debate on the imperfectness of cables for signal transmission.

[Jud]

I read somewhere that Apple switched from USB to PCIe for the card reader a couple of years ago, at least for the MacBook Pro. The Broadcom chip sounds like yet another controller/bus variation for Minis.

 

Because my MBP uses the USB bus, I wouldn't expect to get all the sonic benefits you found from the SD card. Still, I'm curious.

[Jud]

Something I'm not sure about: Discussion on Apple support article for the SD card slot implies to me that I might be able to use SDXC cards in my mid-09 MBP. Anyone know for certain?

[BobP63]

Really not sure why this should be a factor when A+ plays from preloaded memory. Excluding running the DAC and external storage on the same interface, how would a storage drives interface produce "after the fact" audible differences to data already loaded into memory? Not looking to start another objectivist vs subjectivist debate here, but this seems a stretch for substantial audible gains to be found here.

[Melvin]

Thanks for posting this in such detail Alex. I've been curious about using this storage medium with my Mac mini but the cost factor has been prohibitive. Inspired by your post I played around a bit with an old 2gb SD card I had laying around with very good results. I loaded McCoy Tyner "Nights of Ballads and Blues" DSD (amongst others) on the card, used A+ without iTunes integration, and the QuteHD via SOtM converter (powered with a King Rex PSU). Perhaps the difference I hear is simply the removal of my spinning Oyen Digital FW drive but nonetheless, a difference there is. I think it's definitely worth more exploration so I just placed an order for a SanDisk 128gb (Class 10, Extreme SDXC UHS-I Memory Card, 45MBps Read/Write Speeds). Apparently said costs have come down a bit since I last looked!

[souptin]

Really not sure why this should be a factor when A+ plays from preloaded memory. Excluding running the DAC and external storage on the same interface, how would a storage drives interface produce "after the fact" audible differences to data already loaded into memory? Not looking to start another objectivist vs subjectivist debate here, but this seems a stretch for substantial audible gains to be found here.

 

I'll have a go at this, even though I'm an evil naysayer, part of the DBT in-crowd (as is obvious from my personal charisma, style and panache), and am afraid of seizures. Good job the voodoo dolls don't work on tins.

 

Firstly I reckon any data alteration before during or after a file is transferred from storage to ram is way, way down the list of likely sq culprits. Probably the prime suspect is leccy noise.

 

Secondly, the hard disk does not spin down after memory is preloaded, it remains powered up and spinning like a wee top, man! Probably similar thing with other storage (albeit without the moving parts of hd). Hence preload is not complete pancetta, pannini, or even panacea. Hopefully I spelt one of those right. Logically if my theory has any merit, I'd expect some benefits from files stored elsewhere on the network (as reported by SuperTed, I think?). Though from personal experience of a narcoleptic Time Capsule, network storage can create a few problems all by itself.

 

Other wee but relevant experiments (am I allowed to use bad words like 'experiment' here? ) might be:

 

1. Is it detrimental to have storage devices attached and powered up even if they are not accessed? Fire hazard warning though.

 

2. Or, to go to the opposite extreme, if you have loads of ram, load the test file onto a ram disk and detach everything else except the os disk.

 

 

Finally (one for the nitpickers - and I know you're out there) - who else lolled at a topic title with the words "Warning: may cause seizures in the DBT crowd" and adds I request that all the DBT people and sighted-listening naysayers stay away from this thread" when it also includes in the very same OP the words "In my system I could name the USB2.0 sound blind" and "I would bet a good sum of money that I can pick the best two interfaces out of the crowd blind."

 

SuperTed, I hereby accuse you of being a sneaky DBTester! Come over to the dark side where you truly belong. Our cookies are baked in hot ovens with tasty brown sugar, not cryogenically frozen igloos.

[BobP63]

Hi Souptin, I can understand issues related to electrical noise as well as the intrinsic noise of the drive itself. But this discussion relates to interfaces which I suspect is barking up the wrong tree so to speak. Despite the efforts taken I still think there are too many variables and externalities to isolate the interface protocol as the source of improvement.

[souptin]

Hi Souptin, I can understand issues related to electrical noise as well as the intrinsic noise of the drive itself. But this discussion relates to interfaces which I suspect is barking up the wrong tree so to speak. Despite the efforts taken I still think there are too many variables and externalities to isolate the interface protocol as the source of improvement.

 

Actually I agree. Obviously I wasn't there and have only limited mind-reading skills, but yeah SuperTed lost me at that point on the same issues as you - other possible variables.

 

What say we hold off on his execution for the moment though? We are objective, reasonable people aren't we - could be there was something thrown up by the test that hasn't been mentioned.

 

Barking up the wrong tree though? Too many damn dogs on this forum as it is.

[thesurfingalien]

I do want to be clear that I still believe that checksummed, bit-identical files--no matter how many times they get copied back and forth and around the world--still sound exactly the same. (Sorry SandyK)

Alex Crespi

 

Hi Alex,

 

Alex Kethel (sandyk), besides himself, has quite a few reports (including ones under DBT-conditions) of people that actually confirm his findings of bit-identical files sounding different. I wonder how it is possible for you to not believe their reports, but are willing to dedicate the longest posts in existence on C.A. (I did not check that, BTW) based only on what you can hear...

 

Kind regards,

Peter

[BobP63]

Hi Alex,

Snip -

I wonder how it is possible for you to not believe their reports, but are willing to dedicate the longest posts in existence on C.A. (I did not check that, BTW) based only on what you can hear...

 

Kind regards,

Peter

 

Safe bet on longest post...

[Superdad]

I'll have a go at this, even though I'm an evil naysayer, part of the DBT in-crowd (as is obvious from my personal charisma, style and panache), and am afraid of seizures. Good job the voodoo dolls don't work on tins.

 

Firstly I reckon any data alteration before during or after a file is transferred from storage to ram is way, way down the list of likely sq culprits. Probably the prime suspect is leccy noise.

 

Secondly, the hard disk does not spin down after memory is preloaded, it remains powered up and spinning like a wee top, man! Probably similar thing with other storage (albeit without the moving parts of hd). Hence preload is not complete pancetta, pannini, or even panacea. Hopefully I spelt one of those right. Logically if my theory has any merit, I'd expect some benefits from files stored elsewhere on the network (as reported by SuperTed, I think?). Though from personal experience of a narcoleptic Time Capsule, network storage can create a few problems all by itself.

 

Other wee but relevant experiments (am I allowed to use bad words like 'experiment' here? ) might be:

 

1. Is it detrimental to have storage devices attached and powered up even if they are not accessed? Fire hazard warning though.

 

2. Or, to go to the opposite extreme, if you have loads of ram, load the test file onto a ram disk and detach everything else except the os disk.

 

 

Finally (one for the nitpickers - and I know you're out there) - who else lolled at a topic title with the words "Warning: may cause seizures in the DBT crowd" and adds I request that all the DBT people and sighted-listening naysayers stay away from this thread" when it also includes in the very same OP the words "In my system I could name the USB2.0 sound blind" and "I would bet a good sum of money that I can pick the best two interfaces out of the crowd blind."

 

SuperTed, I hereby accuse you of being a sneaky DBTester! Come over to the dark side where you truly belong. Our cookies are baked in hot ovens with tasty brown sugar, not cryogenically frozen igloos.

 

 

Souptin, you put a smile on my face this afternoon! Everyone's comments and questions are appreciated. One question, about what you speculate as being the prime suspect: what is "leccy noise?" (BTW, my handle is Superdad, not SuperTed; strange not that I think about, in my whole life I don't think I have ever personally met a person named Ted.)

 

I realized that my ridiculously long report probably caused some confusion. And I see that I did not go very far in speculating after the fact about the mechanisms responsible for what I hear. I think there are a couple of reasonable takeaways from my experience with the storage interfaces (I partially leave out the HD itself because I moved the same raw drive from enclosure to enclosure):

 

a) The enclosures that had the least going on electronically and which were clocking out data at the slowest rate sounded best. Thus, FW400 sounded best even from the multiport interfaces, and the fewer other bridging interface ports the units had the better they sounded (and that is very repeatable);

 

b) While "a)" above is about whatever reduced minuscule "noise" gets onto the cable to the computer (it is still hard for me to fathom it--every scientific fiber screams its crazy--but I hear it), I think a big factor concerns what the computer itself has to do. And there too, the slower, simpler, more direct interfaces sounded better. Key proof of this, and possibly not spelled out in my report, is that there was a bigger sonic difference between using FW400, FW800, or USB (an outlier for other reasons) on a given quad-port enclosure, than there was between say FW400 on one enclosure and FW400 on another (there was one exception, but I chalk that one up to a bad design--one of the quad-ports sounded awful on all ports).

 

Further support of "b)" is that:

1) the internal SATA drive does not require the computer to get the data through some other bridging controller other than internal SATA one it is already using;

2) the SD card is a very simple, almost passive interface (unlike thumb drives that have a PHY with PLLs and a crystal), and again the ethernet stack is already running so data through it was pretty good--and not chance of much external noise getting through that one (I think my Mac mini has a built-in LAN filter chip; maybe they all do).

 

Oh, Souptin, I did try and report on the sound playing from the SD card while the FW400 drive was still hooked up and power on verus unplugging it from the computer (and its SMPS from the wall). It was sort of a yeah, a tiny improvement, but nothing at all like the interface comparison differences. Perhaps if I compared with a FW800 connected/powered-on versus off the result would be more significant.

 

And I will try your RAMdisk suggestion (though I have just 8GB installed).

 

 

BobP63:

Yes, I feel your bafflement at the whole thing since we are playing from preloaded memory. It makes no stinkin' sense to me as a layman! My friend/design partner John Swenson has--since I told him about all this last week--expressed an interest in finding a way to probe the computer while these interfaces are running to see if some noise, grunge, RF, whatever can be found. The possibility also exists that the computer is using some extra cycles in continuing to talk to the external interfaces.

 

While you are correct that there are a lot of " variables and externalities to isolate the interface protocol as the source of improvement," my report was a dedicated attempt to do just that in as controlled a way as possible. I was quite careful to change one variable at a time and to A/B/A/B/A t be certain I was neither mis-judging or fooling myself.

 

Of course what I found is not likely to be universal, owing to different computer internals, and of course to different systems/rooms/ears. In addition, some DACs may be more or less sensitive to all of these computer related factors. However, susceptibility or lack thereof in a given DAC does not confer superiority either way. In other words, a DAC/system that is not sensitive to these variables (USB cable, computer s/w, power supplies. etc.) is not necessarily the better, more musically accurate product. On the other hand, there are and will be DACs that are both less immune and superior sounding. John keeps insisting to me that it is/should be possible to reduce DAC susceptibility to everything that come before it.

 

In the meantime, network drive access does seem to have promise, and for the Mac systems whose SD card slot is not on the USB bus, that is a fair, if limited capacity, choice.

[sandyk]

Bob

Perhaps you should check out Peter St's forum XXHighEnd - Index where many C.A. members are also members ?

There are numerous reports there of files saved at different locations sounding different despite being played from System memory.

There are also plenty of reports here about it being inadvisable to use the OS HDD for music storage.

Alex

[Superdad]

I read somewhere that Apple switched from USB to PCIe for the card reader a couple of years ago, at least for the MacBook Pro. The Broadcom chip sounds like yet another controller/bus variation for Minis.

 

Because my MBP uses the USB bus, I wouldn't expect to get all the sonic benefits you found from the SD card. Still, I'm curious.

 

Hi Jud:

The Broadcom chip is a PCIe host interface, ethernet transceiver, and 5-in-1 memory card reader all in one. http://www.broadcom.com/collateral/pb/57765-PB00-R.pdf

 

There certainly have been some questions about which machines put the SD card reader on the USB bus versus with a controller on a faster bus. See here: https://discussions.apple.com/message/23131660#23131660

 

Even Apple's help page seems to have some wrong information about the SD/SDXC card slots since it implies that my 2010 mini's slot is not SDXC when I know that it is. About the SD and SDXC card slot Still, the page is worth checking out.

 

From my nosing around late last night between Mac Specs, Prices, Answers & Comparison @ EveryMac.com, Est. 1996 and motherboard photos and chip i.d. notations on iFixIt.com, it seems apparent (and kind of obvious) that the Macs with card reader slots on the USB bus are considered "SD readers," and that the machines called out as having "SDXC" readers are on the faster bus. But the specs of SD, SDHC, and SDXC have more to do with compatibility with ever larger card capacities. The speed rating of the cards--and the speed capability of the reader/bus--are separate.

 

In our case, for audio, the speed is not important. What seems important is if the reader is on the USB bus or not. Then again, I don't have a Mac here with a SD reader on the USB bus to compare to. You can let us know how your SD slot fairs versus other interfaces, but that won't tell us how it compares to an SDXC slot using a non-USB-bus chip.

[souptin]

First, and most important - this is SuperTed:

 

 

Are you sure you don't want to change your name now? You can get pyjamas and outfits and everything. Though I'd think twice about the full outfit if I were you - remember that episode of CSI?

 

Other less important things:

 

leccy noise = electrical noise. Think you've covered this anyway. Anecdotally I've had more musically satisfying computer systems when I don't have everything but the kitchen sink hanging off of every available port. Am pretty sure that any computer will poll all attached devices at regular intervals regardless of whether they are actively in use or not.

 

RAM disks - I messed around with that a bit when ssds started to come on the market. I figured with a bit of applescripting I could get iTunes to play selected tracks from RAM. But long before I got my finger out the proper programmers had come up with the real deal. The main flaw I see in the experiment (of ram vs another storage interface) is that having more ram makes just about everything run better, so reserving some for ramdisk use would be a backward step unless you have masses of it installed.

[Superdad]

Hi Alex,

 

Alex Kethel (sandyk), besides himself, has quite a few reports (including ones under DBT-conditions) of people that actually confirm his findings of bit-identical files sounding different. I wonder how it is possible for you to not believe their reports, but are willing to dedicate the longest posts in existence on C.A. (I did not check that, BTW) based only on what you can hear...

 

Kind regards,

Peter

 

Well, I still have an open mind, and Alex K. did send me another pair of files the other night. Certainly my system has advanced since the last round I tried with his files. (Last time it was also very frustrating for us both because most of the tracks he sent were either very electronic or oddly recorded; I may need to send him a CD of something I use and know intimately.)

 

The core aspect of what SandyK is presenting I can almost wrap my head around: that the rips from a very optimized and clean-powered optical drive can sound better than from a lesser unit. What I can't quite fathom is the sensitivity of the data afterwards. Alex puts them in zip files (sends them over the web) and says one must be very careful to unzip them in place on the media that you will play back from, and that they should not be copied back and forth or the differences will be diminished or destroyed.

I took a track of mine that I know very well and copied it (uncompressed) up to the cloud and back and then from drive to drive on my LAN. Then I compared it by ear to the original and could hear no difference at all.

 

So again: better rips I can believe. Checksum-identical files sounding different when played back from the same interface? I'm not there yet. But I'll try again with a VERY open ear/mind.

 

I think my report is less controversial in that I am claiming that the electronic activity--which goes on in and between both the computer and the storage medium--has an audible effect and the methods sound different. That is very different than what Alex K. is proposing. Funny thing is, I very much thought of him--and of his USB sticks and retrying his experiments--while I was doing my interface research. I would think that what I found would be of great interest to his cause.

 

Cheers,

Alex C.

[sandyk]

Alex C.

Unfortunately, I have found that this kind of dated and annoying material from the Sony BluSpec samplers highlights these differences more than most due to the overblown HF content, (tinny ) just as the Sony engineers apparently did. The best way for me to go forward is obviously to provide high quality recent stuff.

(including perhaps one of B.D's 24/192 tracks ) on a USB memory stick, along with the special +5V low noise linear USB PSU and modified USB cable.

The problem there is the bulk, and need for a special transformer with selectable 120V and 240V mains input.

The weight then makes it a costly exercise to post overseas ( and get it back again after being passed around!) It is much easier to demo these things locally .

In coming weeks I should be able to demo these things directly to Sydney E.E Ionwyn B.,who has worked for both DEQX and Cochlear, as well as hopefully David L. (Audiophile Neuroscience) from C.A. in early December when we hope to have another listening session via B&W 802's and other expensive gear, as well as some DIY efforts at a friend's place.

I am glad to see that you are now reporting hearing many things that the closed minded types insist is not possible.

Your views should matter more than most, as like Barrows you have been heavily involved in the industry.

 

Kind Regards

Alex

[elcorso]

Hi Alex,

 

I have similar experiences when I play some music from my internal SSD OS hard disk.

 

I don't have too much room to store more, but the music I store there is for testing some things, like A+ versions, OS X versions, etc.

 

BTW, do you know the brand name & model of the lesser noise new SSD?

 

Very nice report, thanks,

 

Roch

 

PS/Alex (sandyk): I read the reports about this in the XXHighEnd forum, but must of their users are under Windows, with some different HD structure than Mac?

[wdw]

Hello Superdad,

 

GREAT POST...many thanks for this effort...as you freely admit, you are walking well into the seriously nutbar side of human behaviour but this work is of great benefit to all here.

Thanks.

Will be buying an SD card when back at work next week as I am very curious to hear if my experience follows yours.

 

Did you consider a "bus" powered SDD on firewire? Wondering what sonic signature this would have in comparison.

 

Is the method to use A+ in non-integrated mode and drag a playlist to the card?

 

Regards,

Warren