Listening to digital music storage devices

Posted by: Jan-Erik Nordoen on 21 September 2011

The latest issue of HiFi Critic contains an enticing piece by Andrew Harrison and Stephen Harris (of Naim) on sound quality differences between various digital music storage technologies. In a nutshell, they noted marked SQ differences according to the manner in which music was stored ; differences that were considered in some cases « akin to changing loudspeakers ».

 

The reference system comprised a dCS Purcell Upsampler and Delius DAC feeding a Music First system controller and Chord SPM 1200C power amp, into B&W 802D speakers. Cabling was mostly Nordost ; power conditioning by Isotek. The NAS units were linked to the audio system by a UnitiServe and an NDX, the Serve used as a ripping tool and network server. The NDX’s DAC was bypassed to use the dCS two-box DAC.

 

Storage devices were sited in another room connected to a Cisco Linksys E4200 wireless gigabit router in the listening room via Belkin Cat 6 cable. Another gigabit switch in a remote room (NetGear Prosafe GS108) enabled several NAS units to be online at the same time, each connected to the switch with a high-quality Cat 5e patch cable.

 

First, two 4-bay QNAP boxes were compared: the QNAP TS-439 Pro equipped with four 2-TB Seagate LP drives (QNAP1), vs the QNAP TS-419P+ equipped with four 2-TB Hitachi Deskstar 7K3000 disks (QNAP2). QNAP 2 came out ahead with better tunefulness, « lines of melody and rhythm cooperating better », better instrument distinction and tonally less messy than the QNAP1. The difference was considered as substantial as upgrading from a ₤500 DAC to ₤2000 one.

 

The second comparison was on one NAS, a Synology DS411 Slim, comparing four different drives, two hard disk drives and two solid-state drives: (HDDs : 500 G Hitachi Travelstar 7K500 and a 500 G Seagate Momentus 7200.4 ; SDDs : 128G Kingston SSDNow and a 120 GB Corsair F120.)

 

The overall ranking placed the SQ of the Kingston SSD ahead of the QNAP2 – though the latter had the best bass performance -- followed by either of the HDDs in the Synology NAS.

 

The kicker in the article though was a sidebar comparing different RAID configurations on another Synology unit, the DS 211 equipped with two 2TB Western Digital RE-GP HDDs set up in RAID 0, i.e., with the data striped across them to augment performance. The verdict ? Possibly the best result of all the configurations listened to, with sustained pace and drive, body and richness to music -- that through the Kingston SSD seemed lightweight  -- and an overall relaxed quality that enticed further listening.

 

The tests were preliminary and in no way intended as a buyers’ guide, yet they certainly provide food for thought and ideas for endless hours of experimentation.

 

Jan

Posted on: 23 September 2011 by Jan-Erik Nordoen

Thanks Simon for doing the test. Now we have a sample of one. I would love to hear from others who are able to do the comparison, so here’s an idea ; please bear with me ! The hypothesis to be tested would be the HiFi Critic finding that RAID 0 configuration (striped data) present music more coherently than other RAID configurations. As the other configurations were not specified, we might as well start with RAID 1.

As David has mentioned, one NAS is required to store the test library, and a second NAS to configure as RAID 0 (striped *) or RAID 1. This would be the only variable to change. I would suggest long-term listening in each mode and not quick A/B comparisons, which are misleading. Since we’re all experienced in hearing differences between various pieces of audio equipment, I would suggest applying the same outcome measure here : which of the two modes, if any, sound better to you and if so, what differences are you hearing, or in what way is one better than the other ?

Given that prior beliefs can influence the subjective outcome, it will be useful to note, upfront, whether you believe that different RAID configurations can sound different or not. If I get enough responses, I’ll stratify the results by this variable. Should be interesting.

Thanks in advance !

Jan

* David : any suggestions for optimal block size, based on *typical* average file size ?

Posted on: 23 September 2011 by Hook
Originally Posted by DavidDever:

       

         class="quotedText">
        ...and you have to tune this for best performance based on the average size of the files you are writing to disk.



But this simply determines how hard the NAS has to work.   I understand that if a NAS is not tuned for optimal performance, it will have to work harder, and could generate more noise.

But if the NAS is in a different room, airborne noise won't be an issue.  If ferrite chokes are used, or even better, if the network player is connected using a wireless antenna, then cable-borne noise will not be an issue.

So what's left?   Assuming there are no setup issues, and the two NAS boxes are delivering the same bits to the network player, and we've eliminated the effects of RFI, then I fail to see how the two NAS boxes can have any effect whatsoever on sound quality.

Always willing to listen to new theories though.

Hook

PS - I wouldn't be shocked if turning on a nearby 60" plasma TV had some effect...
Posted on: 24 September 2011 by DavidDever
I'm guessing here, but it may have to do with any of the following factors: power supply, streaming buffer size, file type-who knows? And I'd also have tried it with a Linn DS for sake of completeness. Keep in mind that the role of journalists is to stimulate discussion, whereas good scientists try, as best as possible, to define the discussion around repeatable and empirically verifiable results....