NDS without UnitiServe?
Posted by: J Saville on 03 February 2014
Just wondering why people bother with an NDS without buying a US? I have seen that a lot of people on the forum try the NDS and state that it 'isn't as good as their CDS3', yet nobody seems to mention what their source is or aren't trying a Serve with it.
The way I see it the Serve is effectively the transport, thus is a crucial part of the system.
All thoughts welcome.
Cheers,
J Saville
One thing I dud notice though is that the rips are different... More tomorrow ..
Yes, I understand.
However I am out 
. you can never say never. I need to hear the US someday, but I am in no hurry.
next purchase if I can ever afford it will be NDS or Linn Klimax/1
In the meantime, the effect the DAC V1 has had on me is that I have started listening and enjoying tracks I haven't heard in years and even started buying some new music. I feel strange saying this but I don't feel sad anymore about the CD's that will not be available on Vinyl
And it sounds very musical and enjoyable to my ears. Maybe its the fantastic audiophilleo software in the DAC V1 with the lower ultra jitter fixed clocks or I don't know
all I know is that .. USB asynch just works
Talking about fixed clocks : the Linn Klimax/1 has an individual clock for each frequency, each with it's own power supply? talk about attention to detail.
The rips - it makes nonsense for them to be different. However, dbpoweramp and naim rips don't line up exactly in audacity. I believe they don't like up because of offset issues, and I haven't been able to do the invert 1 file + merge tracks trick to unequivocally prove they're the same.
Ok this I can understand... yes maybe that's why the US sounds better.
If its the TCP parameters driving this within the TCP/IP stack, then some of us are perhaps hearing the interaction of the TCP protocol machine sending and receiving its TCP windowing and flow control signalling (which is of course is driven into the bidirectional frames and then physically encoded on the wire) which is established between peers - in this case the Network player and UPnP server (irrespective of what is in between on a home LAN)
Is the difference worth the US - I don't think I can answer that. Is the difference enough to close my other servers down for Naim streaming duties , and use a dedicated RaspberryPI for my Naim UPnP - you bet!
Now I expect the same thing to happen with async audio over USB. Here the parameters will be driven by the USB peering chipset and drivers - I have yet to look in to this.
The point about USB is one I am thinking through, I.e. will a Mac mini with a linear PSU via asynchronous USB sound as good as a US with a linear PSU via SPDIF into a V1?
Wonder if anyone has done this comparison?
Jude
MM, if you are using correctly calibrated rippers, i.e. it understands the offsets imposed by the CDROM the US, dBpoweramp, EAC rippers create identical WAV file data chunks - down to the very last bit with no zero padding inserted - I posted some results of tests I undertook on this with other forum members on this site a couple of years ago - in the ripping wars era......
However iTunes seems to not have any ability to calibrate itself for the CDROM its using, and therefore introduces a few milliseconds of offset at the start of each track rip. (measured on Windows with a Hitachi CDROM)
Simon
I haven't done it, but I have compared the Squeezebox touch through optical and then through USB into DAC v1
I can't do it side by side and have to rely on memory.
it's a close call. The issue is, both are buffered ! Hence both sound very, very good.
Anyway with USB, the DAC V1 controls the flow of data, whereas with SPDIF, the V1 has to wait for data from the touch and then buffer it.
I just use USB with the touch, and am quite happy with the result.
The SPDIF input is no slouch, and even with my very jittery source like airport express, sounds still amazing.
Nice to have both types of inputs.
End result and bottom line :through USB the V1 is a source that I can live with for a long, long time.
Will I buy a US for the DAC V1. No need for it, and the US doesn't have USB out anyway.
would I buy a US for NDS. due to cost reasons, no. the NDS is expensive enough as it is.
Jude, I have not a side by side - but have noticed that the quality of asynchronous audio transfer over USB varies on the source with other non Naim DACs.
Async transfer is more in common with TCP transfer in terms of flow control than that of SPDIF. Modern SPDIF implementations are quite benign IMO outside any possible issues with RFI and transport clock modulation.
Async USB and TCP have more signalling flowing between the peers to control the window sizes/data buffering, and both use quite detailed protocols to convert the chunks of data. Further with USB there is quite a marked change in the profile of the signalling between USB3 and USB2 (currently Naim only supports USB2 - USB2 is half duplex and USB3 is full duplex). I would expect therefore USB2 and USB3 to 'sound' different on most implementations that supported both.
And to your question - I would be genuinely surprised if async USB did not 'sound' different to SPDIF on a DACv1 within a revealing system.
Simon
To my ears, FLAC files have a slighter softer presentation than WAV. The leading edge of transients are ever so slightly blunted.
That's my impression also. I could get used to it and live with it, but happily I don't have to.
Jude, I have not a side by side - but have noticed that the quality of asynchronous audio transfer over USB varies on the source with other non Naim DACs.
Async transfer is more in common with TCP transfer in terms of flow control than that of SPDIF. Modern SPDIF implementations are quite benign IMO outside any possible issues with RFI and transport clock modulation.
Async USB and TCP have more signalling flowing between the peers to control the window sizes/data buffering, and both use quite detailed protocols to convert the chunks of data. Further with USB there is quite a marked change in the profile of the signalling between USB3 and USB2 (currently Naim only supports USB2 - USB2 is half duplex and USB3 is full duplex). I would expect therefore USB2 and USB3 to 'sound' different on most implementations that supported both.
And to your question - I would be genuinely surprised if async USB did not 'sound' different to SPDIF on a DACv1 within a revealing system.
Simon
Understood, Simon.
It occurs to me that the more and more I explore 'enhancements' the differences in use between asynch USB and stream via NAS seem to merge.
For example, in the case of a Mac Mini, I would need (in a pure sense) to switch off the wifi and bluetooth radios and connect to my home network via an ethernet cable . In doing so, I would then need to consider whether the router and/or switch would benefit from a 'quiet' PSU, as many have done with streamers.
@analoguemusic. My view is essentially the same as yours regarding the V1, however, I am just exploring future possibilities....
Also, I am conscious going too 'off pist' from the OP's point.
Cheers
Jude
I'm one of the early UnitiServe adopters, and I'm coming around to Tog's view.
My latest review projet is a RIP-NAS-PLAY server using VortexBox (VB) that rips to FLAC on an internal 2TB WD Red HDD (3 or 4TB optional). Logitech Media Server (LMS) is the playback software. Output is via USB. The unit runs on an external 12V switching supply.
The unit also rips DVD audio and video, and BluRay (with purchase of a license). Essentially all types of file are played up to 384 kHz, and DSD (.dff) files are supported. Optional software includes Bit Torrent, Bliss, Plex, and Subsonic.
Control is through free or purchased apps running on Android or Apple tablets or phones / pods.
Using the HiFace USB to S/PDIF converter to feed the Naim DAC, then the SuperNait 2 into the Amphion Argon 3 speakers, after two weeks, I've narrowed the sonic differences down (they're slight) to the differences between FLAC (VB) and WAV (U/S) files, rather than differences between the units.
To my ears, FLAC files have a slighter softer presentation than WAV. The leading edge of transients are ever so slightly blunted. It remains an engaging presentation, but misses out on some rawness and edge that WAV files capture better. Because of the transient blunting, subtle timing information can be less clear on FLACs.
That said, I could certainly live with the FLAC presentation, and if I'd never lived with WAV playback, probably wouldn't know the difference. FLACs are just a little easier on the ears.
The linear PS that I use on the U/S also lifts the performance of the VortexBox unit.
The server experience lives or dies on the user interface. In this case, I was using iPeng on an iPad, and while not as slick as nServe, it does the job, I got to like it and could live with it. Going back to nServe, though, I realize just how blazingly fast it is, and how good the U/S is in aggregating all sources (the U/S picked up the VB server immediately) into one coherent library.
VB ripping is slower than the U/S (10 minutes on average) and indexing is daily, although a scan can be forced.
The VB unit is a third of the price of the U/S and does require quite a bit more effort in setup, but overall is a convincing package.
The next iteration of VB 2.3, based on Fedora 20 is highly anticipated and available on beta at present. SOtM sell a number of internal power filters for drives and a fantastic USB card to improve things still further.
Tog
I'm one of the early UnitiServe adopters, and I'm coming around to Tog's view.
My latest review projet is a RIP-NAS-PLAY server using VortexBox (VB) that rips to FLAC on an internal 2TB WD Red HDD (3 or 4TB optional). Logitech Media Server (LMS) is the playback software. Output is via USB. The unit runs on an external 12V switching supply.
The unit also rips DVD audio and video, and BluRay (with purchase of a license). Essentially all types of file are played up to 384 kHz, and DSD (.dff) files are supported. Optional software includes Bit Torrent, Bliss, Plex, and Subsonic.
Control is through free or purchased apps running on Android or Apple tablets or phones / pods.
Using the HiFace USB to S/PDIF converter to feed the Naim DAC, then the SuperNait 2 into the Amphion Argon 3 speakers, after two weeks, I've narrowed the sonic differences down (they're slight) to the differences between FLAC (VB) and WAV (U/S) files, rather than differences between the units.
To my ears, FLAC files have a slighter softer presentation than WAV. The leading edge of transients are ever so slightly blunted. It remains an engaging presentation, but misses out on some rawness and edge that WAV files capture better. Because of the transient blunting, subtle timing information can be less clear on FLACs.
That said, I could certainly live with the FLAC presentation, and if I'd never lived with WAV playback, probably wouldn't know the difference. FLACs are just a little easier on the ears.
The linear PS that I use on the U/S also lifts the performance of the VortexBox unit.
The server experience lives or dies on the user interface. In this case, I was using iPeng on an iPad, and while not as slick as nServe, it does the job, I got to like it and could live with it. Going back to nServe, though, I realize just how blazingly fast it is, and how good the U/S is in aggregating all sources (the U/S picked up the VB server immediately) into one coherent library.
VB ripping is slower than the U/S (10 minutes on average) and indexing is daily, although a scan can be forced.
The VB unit is a third of the price of the U/S and does require quite a bit more effort in setup, but overall is a convincing package.
Hi Jan,
The next iteration of VB 2.3, based on Fedora 20 is highly anticipated and available on beta at present. SOtM sell a number of internal power filters for drives and a fantastic USB card to improve things still further.
Tog
@Simon: I tried to tweak the tcp/ip parameters on my PC. It just caused the vpn connection to my work PCs to become flaky. And I'm not sure it did anything to the sq.
MM - hmm the TCP/IP parameters could mean anything. Its specifically the transport control protocol (TCP) you want to look at - and specifically the window send and receive sizes.
It might also be worth seeing what TCP options are exchanged when the TCP flow is setup - here WireShark is useful.
Although for Linux, this might be of use.
http://kaivanov.blogspot.co.uk...inux-tcp-tuning.html
However if you can borrow a Unitiserve and watch it connect with your network player and look at the values and steady state sizes that are being selected using wireshark and use that as a template to aim for.
On windows these values are selectable within the registry.
On the SQ front it is subtle - kind of like differences between FLAC and WAV. Once you have got comfortable with a certain sound then moving to a new UPnP server with different parameters grates.
BTW Netstat -tpn -c on a linux machine can be quite useful give you a realtime look at the UPnP TCP streaming buffers.
Simon
As related to OP, I have not tried many sources into NDS other than via U/S or HDX. Streaming file to NDS was superior to same file coming off of a 64 Gb PNY Brand USB drive into the front of NDS.
Hi Connon,
Yes S/PDIF into the Naim DAC. What I've retained from earlier discussions here is that the difference between the U/S and HDX S/PDIF outputs are largely down to the quality of the power supply. I''ve not compared the U/S with linear PS to the HDX ; the SQ is just so good with the augmented U/S into the Naim DAC, that I've not felt the need to look elsewhere.
As for comparisons of the U/S S/PDIF output vs streaming, yes, streaming has the edge. One day, the NDS...
Jan
That's at odds with HiFi Critic's (Martin Colloms' ?) findings. In their case, USB came out ahead, but they had taken care to shut down unused NDS functions prior to the test.
I have vague recollections of that article. Can't remember exactly what 'extra functions' they turned off or which brand if USB drive they used, etc. It is just my humble experience.
Naim were using the NDS streaming as the source into their Statement amps in Las Vegas at the CES and not a USB into the front of it. I would assume they were trying to maximize the performance of the system by using their best possible source.
I have vague recollections of that article. Can't remember exactly what 'extra functions' they turned off or which brand if USB drive they used, etc. It is just my humble experience.
Naim were using the NDS streaming as the source into their Statement amps in Las Vegas at the CES and not a USB into the front of it. I would assume they were trying to maximize the performance of the system by using their best possible source.
There is no reason why in a particular setup that USB memory playback might sound preferable, indeed IMO there is less extraneous processing with USB memory playback than with TCP/IP or (or even async USB for that matter), so you can see why it might or should potentially 'sound' better. However there are indeed many variables with USB and USB memory stick playback such as chipset and design - all of which might have an affect. (which is why I like to keep my DAC separate from my digi source processing - but thats another story).
As far as the trade show commentary - I doubt Naim would have been promoting its top of the range network player by not using it on a network - and that probably had more bearing than anything else.
Simon
That's at odds with HiFi Critic's (Martin Colloms' ?) findings. In their case, USB came out ahead, but they had taken care to shut down unused NDS functions prior to the test.
Please, read also The Absolute Sound Feb 2014 number #240 - good review & really different USB vs. Ethernet quality outcome.