UnitiServe Vs Memory-stick into nDac
Posted by: MartinCA on 05 August 2011
Mostly I listen to music streamed to my nDac using iTunes via an airport express. SQ is good - in the sense that it is several orders of magnitude better then using the airports on-board DAC and streaming straight into the pre-amp. But it falls well short of the SQ when I play a WAV file off a memory stick into the DAC.
So I have been watching the various streaming options develop with the hope that (a cost-effective) one will come along which gives the sort of SQ possible with the memory stick. The NDX looks good - except that its DAC component, which must account for half the price, is completely redundant for me.
Anyway, today I dropped into my local dealer to listen to a Unitiserve into an nDac. I took along a couple of CDs and a good quality memory stick with the same CDs ripped into WAV using EAC. I had seen comments on here that the SPDIF interface from the UnitiServe was 'not audiofile quality', so I was hoping that this was relative to 555 kit only, and that the Unitiserve would be close to the memory stick in SQ.
In theory there ought to be little difference - the same CD ripped to the same format as close to bit perfect as dammit, with just the SPDIF connection to contend with. But in fact I thought the unitiserve sounded noticeably better - more bass, and a fuller less bright sound. The person with me thought there was more detail in the unitiserve, but I didn't really hear that.
I was wondering whether anyone out there with a Unitserve can comment on whether their experience is similar?
Also, as an aside - with the forthcoming introduction of internet radio on the UServe and HDX - does anyone know whether that can be retrospectively enabled on this kit bought before it was made available?
Posted on: 06 August 2011 by DavidDever
Yes, all Naim servers are capable of utilizing the Internet Radio features once released.
As an aside - each of the five models (NS01 / 2 / 3, HDX, UnitiServe) has its own software build, but a unit from three years ago can be brought up to date with the appropriate sequence of software updates.
Posted on: 06 August 2011 by Stoik
Originally Posted by MartinCA:
I had seen comments on here that the SPDIF interface from the UnitiServe was 'not audiofile quality'...
Well, let them stay as they are; comments. Here are the facts now:
The two pictures following are the innards of the UnitiServe (first pic) and the HDX-SSD (second pic). Circled in red are their SPDIF interface card... and they look pretty similar to me.
And why your WAV sounds best on the UnitiServe is plain simple: More system ressources and processing power makes more advanced softwares to run more smoothly. See a UnityServe like a small computer, and the nDAC like a mix of a microcontroler and a signal processor.
Bye.
Posted on: 06 August 2011 by AMA
Originally Posted by Stoik:
Originally Posted by MartinCA:
I had seen comments on here that the SPDIF interface from the UnitiServe was 'not audiofile quality'...
Well, let them stay as they are; comments. Here are the facts now:
The two pictures following are the innards of the UnitiServe (first pic) and the HDX-SSD (second pic). Circled in red are their SPDIF interface card... and they look pretty similar to me.
And why your WAV sounds best on the UnitiServe is plain simple: More system ressources and processing power makes more advanced softwares to run more smoothly. See a UnityServe like a small computer, and the nDAC like a mix of a microcontroler and a signal processor.
Bye.
Stoik, the photo session is not complete from engineering point of view.
Let the S/PDIF boards be the same - have you compared the PS feeding the two boards? 
I guess this is the point which may (or may not) lead to the difference in SQ.
But you are absolutely right that nDAC is struggling in processing the data from memory stick on the fly, especially if this data is hi-res and compressed. The only solution is to run processing with a dedicated power rail from XPS (providing that existing CPU and SW is fast enough to process the data flawlessly).
I'm not sure if it's implemented -- AFAIK my XPS is only used for analogue output of nDAC 
Posted on: 07 August 2011 by Simon-in-Suffolk
I am not sure the NDX DAC components would count for half the price. Readiing the White papers the DAC and analogue filtering is relatively modest compared to the NDAC, albeit very effective. The SHARC DSP algorithms look largely shared between NDX and NDAC, however the control circuitry and interface circuitry on the NDX is a lot more advanced, and I suspect much design time went into that aspect so as reduce EMI coupling affecting clocks and digital signal stability thereby reducing harmful RFI. This is where I suspect a significant proportion of the 'value' of the NDX is.
Therefore the NDAC in my mind is a natural progression for the NDX, as the relatively simple NDX DAC is replaced, and allows the NDX SHARC DSP and it's clocks to create a rock solid and clean SPDIF stream from the TCP payload fed to it by the NDX control circuitry. To do this simply is relatively straightforward, to do it well with minimal compromises is a different story hence why as an engineer I bought into Naim.
As far as USB and Unitiserve, I think the issues that will affect SQ here is RFI. USB isby design a loose standard to keep costs down. However different USB memory sticks/leads produce varying
levels of RFI, and reflections in the serial lines. This has the ability to pollute the nDAC. You hear this with different USB sticks 'sounding' different
The Unitiserve produces the SPDIF via a 75ohm source, however RF currents from it's power supplies and DSP will conduct through the SPDIF into the NDAC. It is not clear whether Naim have gone to the same extent as the NDX here, but being a small PC based computer it is going to be electrically noisy compared to using optimised micro controllers as used by the NDX and we know the Unitiserve power supply is more modest than the HDX, so it is possible there may be SQ compromises here. But as said many times we all listen to things differently and these differences may not to be relevant to you.
Simon
Posted on: 07 August 2011 by KRM
The "comment" that the digital out on Unitiserve is "...not of audiophile quality" came from Naim, via my dealer. Admittedly, no photographic evidence was provided. We did listen to it though, verses the NDX and the Linn Akurate.
Keith
Posted on: 07 August 2011 by MartinCA
The comment about Unitiserve not being audiophile quality was made by KRM in the NDX Vs Linn DS thread. I think KRM was just relaying what he had been told, and without evidence I think neither he nor I were reading too much into it. My dealer has certainly told me a whole lot of incorrect things (such as the "The Unitiserve needs another piece of Naim kit like the UnitQute to provide a user interface to it").
But if true, I didn't interpret it to mean that the UnitiServe SPDIF is inferior to the HDX - on the contrary I would guess they are the same. However, I assumed that it meant that more attention had gone minimising any possible interference in the NDX.
The reassuring thing for me, thinking of buying a UnitServe, is that the SQ of the UnitiServe into the nDac is better than the best I can currently achieve (that being via memory stick).
Posted on: 07 August 2011 by MartinCA
Originally Posted by Stoik:
And why your WAV sounds best on the UnitiServe is plain simple: More system ressources and processing power makes more advanced softwares to run more smoothly. See a UnityServe like a small computer, and the nDAC like a mix of a microcontroler and a signal processor.
Originally Posted by AMA:
But you are absolutely right that nDAC is struggling in processing the data from memory stick on the fly, especially if this data is hi-res and compressed. The only solution is to run processing with a dedicated power rail from XPS (providing that existing CPU and SW is fast enough to process the data flawlessly).
Well ..... except that the tests I did were with16 bit WAV, which is neither hi-res nor compressed, so the DAC shouldn't have been struggling, so I think Simon's explanation of RF interference is more likely, although the Corsair memory stick I used is supposedly a high quality one.
Posted on: 07 August 2011 by MartinCA
Originally Posted by DavidDever:
Yes, all Naim servers are capable of utilizing the Internet Radio features once released. As an aside - each of the five models (NS01 / 2 / 3, HDX, UnitiServe) has its own software build, but a unit from three years ago can be brought up to date with the appropriate sequence of software updates.
Thanks David - appreciated.
Originally Posted by MartinCA:
I was wondering whether anyone out there with a Unitserve can comment on whether their experience is similar?
I want to understand what is going on technically, but there comes a point when I give up and think who cares provided it sounds right. I appreciate the replies trying to explain what is going on, because I am interested in that. But I'd like to hear from anyone with a UnitServe/nDAC combination who has compared this with straight memory stick input.
Posted on: 07 August 2011 by KRM
MartinCA is exactly right. At the time, I had intended to connect the Unitiseve to my nDac. A friend of mine has an HDX and he had found, like many others, that it really starts to sing when pair it with the Dac. On the face of it, the Unitiserve is an HDX without the bits I don't want (pretty screen, analogue out etc.), so I thought I could get a source that would be similar to the HDX + nDac and better than my CDX2 for £2.5k less than the cost of the former. In my experience, this not the case. We still bought the Unitiserve to use as a UPNP server to the NDX. So the comment applies when comparing the Unitiserve to more expensive Naim sources. If (when!) Naim release a 5 series streamer it may or may not be better than the Userve via SPDIF. I think it's a great product. You may be happy with it's SPDIF out and if you get it first you can upgrade latter to an NDX (or whatever) and have a very neat solution streaming solution.
Keith
Posted on: 07 August 2011 by KRM
I suspect that the technical differences between a memory and the Userve into the NDac are many and, well, technical. My hi res music is on a 0.5tb usb hard drive plugged into the back of the Userve. I haven't compared it to plugging directly into the dac (or the NDX) because I'm lazy, I don't like and mistrust A/B comparisons and it's so much easier when you can't control the music properly directly to the nDac.
Keith
Posted on: 07 August 2011 by AMA
Originally Posted by MartinCA:
Originally Posted by AMA:
But you are absolutely right that nDAC is struggling in processing the data from memory stick on the fly, especially if this data is hi-res and compressed. The only solution is to run processing with a dedicated power rail from XPS (providing that existing CPU and SW is fast enough to process the data flawlessly).
Well ..... except that the tests I did were with16 bit WAV, which is neither hi-res nor compressed, so the DAC shouldn't have been struggling, so I think Simon's explanation of RF interference is more likely, although the Corsair memory stick I used is supposedly a high quality one.
Martin, the data processing load is not a switch which is OFF at 16/44 and ON at 24/96.
The data reading is constantly stumbling at certain degree at 16/44 and at higher degree at 24/96, less so at WAV and more so at FLAC.
Reading data from a memory stick can be flawless providing the power supply is OK, processor is fast enough, software is not lagging.
It's quite possible that nDAC built-in PS is not good enough to keep up with CPU for flawless reading.
But I do admit the original hardware design was not up to the needs of the flawless reading.
Other than this -- I don't see a problem to read the data from memory stick with no bit flips
Regarding RF -- it equally contaminates USB and S/PDIF receivers.
My personal tests showed no difference between USB and UnityServe when nDAC was powered with XPS.
I also powered US through a dedicated spur away from Naim amps and nDAC/XPS because US might contaminate the power bus with switching noise.
Posted on: 07 August 2011 by aysil
The comment that the digital out on Unitiserve is "...not of audiophile quality" was a careless statement; it is a pity that it is being quoted so often. Audiophile quality is very relative! I think UServe into nDAC sounds pretty amazing, and I would call it audiophile quality. However, HDX into nDAC sounds even more amazing (with the same music file), which is for me an indicator how the same electronic card can sound different in another mechanical and electromagnetic environment and with a different PS. The difference was obvious for me. I did not compare them to USB, though; so I can not answer Martin's original question.
Posted on: 07 August 2011 by aysil
Originally Posted by KRM:
... At the time, I had intended to connect the Unitiseve to my nDac. A friend of mine has an HDX and he had found, like many others, that it really starts to sing when pair it with the Dac. On the face of it, the Unitiserve is an HDX without the bits I don't want (pretty screen, analogue out etc.), so I thought I could get a source that would be similar to the HDX + nDac and better than my CDX2 for £2.5k less than the cost of the former. In my experience, this not the case. We still bought the Unitiserve to use as a UPNP server to the NDX. So the comment applies when comparing the Unitiserve to more expensive Naim sources. ... I think it's a great product. You may be happy with it's SPDIF out and if you get it first you can upgrade latter to an NDX (or whatever) and have a very neat solution streaming solution.
Keith
Keith,
The UServe-NDX combination which you finally opted for may make sense if you already own one of them and want an upgrade - because of resale costs. However, normally HDX is a better option than this combi because it costs less, and has all the features you would get from this combi - and according to my experience, some even in better quality.
If you like the dac section of NDX so much that you will use no external dac, then it's a different story; however for an external dac, HDX is a better option. I wanted to note this for possible new buyers.
Posted on: 07 August 2011 by MartinCA
Originally Posted by AMA:
Martin, the data processing load is not a switch which is OFF at 16/44 and ON at 24/96.
The data reading is constantly stumbling at certain degree at 16/44 and at higher degree at 24/96, less so at WAV and more so at FLAC.
Reading data from a memory stick can be flawless providing the power supply is OK, processor is fast enough, software is not lagging.
It's quite possible that nDAC built-in PS is not good enough to keep up with CPU for flawless reading.
But I do admit the original hardware design was not up to the needs of the flawless reading.
Other than this -- I don't see a problem to read the data from memory stick with no bit flips
Regarding RF -- it equally contaminates USB and S/PDIF receivers.
My personal tests showed no difference between USB and UnityServe when nDAC was powered with XPS.
I also powered US through a dedicated spur away from Naim amps and nDAC/XPS because US might contaminate the power bus with switching noise.
Hi AMA
I didn't say, or even imply, that the data processing load was like an On/Off switch. What I do think that reading a 16/44 uncompressed file from a memory stick should be well within the capabilities of the nDAC. I don't mean this aggressively - but is there any evidence that the nDAC is constantly stumbling with 16 bit WAV (or even 24bit FLAC)?
But it is interesting that you hear no difference between USB and Unitiserve when the DAC has an XPS.
The demo nDAC I heard with the Userve didn't have an external PS, but I do at home. So if/when I get a UServe I'll compare them again. As far as I can tell from the DAC white paper the DSP which processes the USB input remains powered by the onboard DAC PS rather than the XPS, though I guess it benefits from the load taken away from it.
On RF - surely the degree of contamination would be different between that generated by a memory stick and that coming from the UServe into the SPDIF feed?
Heck - when did enjoying music suddenly require a PhD in audio-electronic engineering? Oh hang on - it doesn't!
Posted on: 07 August 2011 by Guido Fawkes
Naim DAC plays USB sticks superbly in my experience, but they are a bit of pain to use.
Posted on: 07 August 2011 by Simon-in-Suffolk
Hmm, well i can usually hear a difference with USB in the nDAC/555ps and SPDIF from an NDX with the latter sounding usually better - although different memories sound different for reasons I said earlier. Also the the rear USB seems to sound better than the front one. Short of seeing a detailed internal design of the nDAC i can only speculate why that is the case. However with USB, the nDAC is driving the clock, which is an advantage, as opposed to SPDIF where it attempts to lock with it.
BTW I don't buy this 'processing load running out of steam' theory - i'd be surprised if NOPs dont feature highly in the operand execution cycle.. What is more likely - and what occurs on other designs that I am very aware of that are not Naim, ie my own, is that digital noise induced on the power lines from the logic gate circuity in the micro controllers needs careful decoupling. The signature of WAV decoding is regular and less intense compared to FLAC or mp3 which is more variable and more non NOP operations per unit time are required.
A computer CPU running other processes as well as the required audio algorithm will typically induce a lot more powersupply noise and EMI than a modest microcontroller, not to mention noise caused by CPU interrupt load handling.
Simon
Posted on: 07 August 2011 by Noogle
Originally Posted by MartinCA:
Originally Posted by AMA:
The data reading is constantly stumbling at certain degree at 16/44 and at higher degree at 24/96, less so at WAV and more so at FLAC.
Reading data from a memory stick can be flawless providing the power supply is OK, processor is fast enough, software is not lagging.
It's quite possible that nDAC built-in PS is not good enough to keep up with CPU for flawless reading.
But I do admit the original hardware design was not up to the needs of the flawless reading.
Other than this -- I don't see a problem to read the data from memory stick with no bit flips
Regarding RF -- it equally contaminates USB and S/PDIF receivers.
I'm sorry, but I think it's craziness to suggest Naim can't design a DAC that decodes high bitrate sources or can't read from a USB stick. I do buy that there may be slightly different digital noise levels from different bitrates and different decompression algorithms, but Naim go to some lengths to isolate the analogue and digital parts of the DAC.
Also, if you use Toslink for S/PDIF, doesn't that provide RF isolation?
Posted on: 07 August 2011 by Noogle
Originally Posted by Simon-in-Suffolk:
However with USB, the nDAC is driving the clock, which is an advantage, as opposed to SPDIF where it attempts to lock with it.
With (asynchronous) S/PDIF, doesn't the DAC just clock in data at a fixed rate from the FIFO buffer? Not sure I understand how this is a whole lot different to clocking in data from USB?
Posted on: 07 August 2011 by Simon-in-Suffolk
Hi Noogle. With SPDIF the clock is asynchronous, and so the SPDIF receiver will lock onto it if it can - ie it is an accurate recognised frequency. If not it will try and adapt to the sender clock. Either way the clocks are not directly coupled or synced. If the send and receiver clock are extremely accurate then the issue of synchronisation largely goes away.
With USB, it has no clock of its own, it is driven by the clock of the host, ie the nDAC drives the clock synchronously. Here accuracy of the clock (within reason) doesn't affect data integrity from the USB, and there are no synchronisation issues. Partly because of this synchronous approach far higher clock speeds can be used with USB than SPDIF.
Remember the SPDIF is at the stream bit level which need decoding. This is different to the bits or bytes fed into the DSP/DAC.
Simon
Posted on: 07 August 2011 by Noogle
All I'm saying is that once the DAC is locked on the S/PDIF source (which presumably is the case most of the time) then the data input to the DAC is driven by the DAC's clock. It will only lose sync if there is buffer overflow or underflow. This is analogous to the USB data input being driven by the DAC's clock. I don't see how the S/PDIF processing comes into it as all the logic will be synchronised with the DAC master clock?
Posted on: 07 August 2011 by Stoik
After reading your posts, I've just passed the last hour browsing for infos regarding the NDX streamer, as I had a few questions loose in my mind about it. Reading the white papers and looking at a few innards pics, I now understand better why it sounds already nice as is, and why its SPDIF output is very good.
First, the SPDIF output. You probably know that the NDX is equiped with the same SHARC DSP and buffer that you'll find into the nDAC too. Well, the SPDIF output is generated by it, and sent to an output buffer, according to the white papers. So one of the finest DSP available is taking charge of this delicate task.
Secondo, the mystery behind the way it sounds. First statement; The NDX is clearly NOT a stripped down HDX, the internal layout and components are completely different. Second statement: The NDX shares a LOT with the nDAC regarding the internal layout strategies, mainly into the electrical decoupling of the control, DSP and analog sections.
Into the NDX, the D/A conversion, analog filtering and output stages are difffering from the nDAC by using a stereo DAC instead of dual monos, and an impressive 7 lines OP amp filtering stage (it looks like a beefed up version of the one found into a CDX2) instead of the full discrete components one found into the nDAC. Some special attentions, like curving the resistors rods like a loop in key areas, tells me that it's very well built.
Now back to your regular programming.
Bye.
Posted on: 07 August 2011 by AMA
Originally Posted by Noogle:
I'm sorry, but I think it's craziness to suggest Naim can't design a DAC that decodes high bitrate sources or can't read from a USB stick.
Noogle, nobody saying that. I own nDAC and it DOES decode hi-res and it DOES read USB.
With a certain degree of accuracy, of course. Others DACs do the same.
It's really surprising to see that many people believe that all DACs are capable to pick up the S/PDIF bitstream on the fly without a single bit error. How exactly this myth is created?
Posted on: 07 August 2011 by Simon-in-Suffolk
Noodle, conceptually at the high level you may be right. But the engineering, design considerations and side affects are quite different between the two.
Simon
Posted on: 08 August 2011 by Simon-in-Suffolk
Further, although Naim don't go into detail on the SPDIF receiver, it might be based on the elastic buffer with low bandwidth loop correction method. [AES 2006, Paul Lesso]. If the receive buffer is large enough and the rate of reclocking frequency correction low enough, typically less than 1 Hz, then artefacts from deterministic and non deterministic jitter of the SPDIF signal are significantly reduced in the real world when converted into the analogue domain by pushing the artefacts outside the audio band. USB synchronous clock does not need to worry with any of this.
Posted on: 08 August 2011 by Jan-Erik Nordoen
Hi Martin,
I have the nServe and the nDAC and periodically compare files ripped by the Serve with the same files played back through USB keys (Kingston and LaCie) from the DAC. To my ears, the USB playback always seems a little more detailed at first compared to the Serve, but not on repeated comparisons. The Serve provides a fuller, more fleshed-out presentation, a sense of forward momentum and the feeling that all the musicians are in the same room. USB playback doesn't do these things as well.
Jan
