Jump to content

NAD M51 & C390DD Owners & Discussion Thread


Recommended Posts

What happens with 32/96 music - is there not much room for attenuation in the digital domain for a 35bit DSP? Not sure if this is 24bit in a 32 bit wrapper or native 32bit information but question still stands.

 

That's right.    So, with 32bit audio into a 35bit DSP .... You can only attenuate 18dB, before you start to affect the SNR of the signal.

 

 

If a 24 bit file is padded with zeros to 32bit then does information at the 'end' get lost when attenuating the signal in the digital domain?

 

Yes.   At the end with the zeros added.    The extra zeroes added to the 24bit file (to make it 32bit) represent quieter sounds than the 24bit system can store.     So adding the zeros, is saying, there are now all these quieter levels (where we can store sounds quieter than the 24bit system) .... and they're all fillled with zeros  (ie.  there ARE no quieter sounds).

 

The original 24bit data  (now padded with zeros to make 32bit) are exactly the same as they were.   Unchanged, and still representing the same levels.

 

 

It was an example sorry hypothetical. I meant to ask from which end does information get lost: from the more important information 'front' end or the zeros 'back' end? Or am I looking at this too simply?

 

No.   You're not looking at it too simply.     The information gets lost from the "quiet" end.    So, if you had the 24bit signal ... and then you converted it to 16bits .... you would lose the 8bits representing the quietest information.     24bit stores down to -144dB ... and 16bit stores down to -96dB .....  So what would happen is any sounds quieter than -96dB would be lost....   You would lose 48dB of range  (144 minus 96).

 

 

 

I  think its safe to assume that he was concerned about people judging Accuphase  CD players  without using a preamp. (There was one review where this happened). Personally I think he was either trying to dumb things down for the masses or the answer was paraphrased by the interviewer.  Now, I know I am not the only one wondering this but could you kindly explain what exactly happens to the digital stream when it is attenuated, and how does this reduce the voltage output  and hence the audio volume?

 

Sure.... but saying that the reason why is "digital bitstripping" .... is very bad.....   when the real reason is more likely to be that the "preamp" (ie.  the analog circuitry after the DAC)  inside the CD player .... is inferior to the standalone preamp.

 

 

Fair enough - but why does  the voltage drop after the DAC?  In my world when I looked at this in the early 90's, I was given an explanation  which  was along the lines that  the output voltage was dependant upon the amount of information being fed into the DAC.  If you reduce the information (Bits etc), then there is less info being fed to the DAC and the Voltage drops. It seems there is more to it based upon what you and Dave are saying.

 

Yes... and that type of explanation can lead to the common misunderstandings.

 

In the digital file.... Let's say a 16bit digital file....  sounds are represented from 0dB (loudest) down to -96dB (quietest).

The DAC chip will always output 0dB at a certain volts (or current) ..... and -1dB, as a certian specific volts .... and -12dB as a certain specific volts.

 

 

So, let's say we manipulate the volume with a 16bit DSP  by -12 dB...     we shuffle all the data downwards in level .....  So what was 0dB before... is now -12dB  (what was -32dB is now -44dB, etc).      Now the DAC will output what used to be stored at 0dB at the voltage which corresponds to -12dB    (and this is how the "voltage drops")

 

 

What we find is that for each 6dB of level .... it takes 1bit to represent.    So by shuffling the data downwards to -12dB .... we will have pushed the bottom 2 bits worth of the quietest sounds down too far.   They are lost....  and thus we now have "14bit of resolution"

 

 

BUT.....     If we manipulate the volume of the 16bit file... using a 24bit DSP .....   the first thing that happens is the 16bit file has it's bottom 8 bits filled with zeros .... so it becomes a 24bit.     Then, same as above.... The data is shuffled downwards 12dB... which is 2bits .... and we now have the exact same data as before .... but in a 24bit system .... and now only the 4 lower bits have zeroes in them.

 

 

 

I hope that helps.   These things are difficult to explain clearly without being too quick... or too long.

  • Like 2
Link to comment
Share on other sites



Actually that is a good question considering we are talking about the M51 which is likely to be hooked up to a HTPC of some sort. If for example someone decided to use their HDMI output from their GPU and decided not to bitstream stream untouched there is an option in JRiver to output all 16 and 24bit data in to 32bit data so this question interests me too.

 

The extra bits allow you store quieter sounds.

 

Adding 8 bits to a 16bit system  (ie.  going to 24) .... is saying....   You have amplitudes between 0 and -96dB.     Now you have exactly the same data, but you also have room underneath to store down to -144dB.

 

The zeros go at the end of the file where the "quieter sounds" go.    ie. in the place you don't actually have any information  (because your original signal didn't contain those quieter sounds)

 

 

Hope that makes sense.

  • Like 1
Link to comment
Share on other sites

Actually that is a good question considering we are talking about the M51 which is likely to be hooked up to a HTPC of some sort. If for example someone decided to use their HDMI output from their GPU and decided not to bitstream stream untouched there is an option in JRiver to output all 16 and 24bit data in to 32bit data so this question interests me too.

 

It won't matter, as the M51 treats everything as 35bit audio.

Link to comment
Share on other sites

 

 

BUT.....     If we manipulate the volume of the 16bit file... using a 24bit DSP .....   the first thing that happens is the 16bit file has it's bottom 8 bits filled with zeros .... so it becomes a 24bit.     Then, same as above.... The data is shuffled downwards 12dB... which is 2bits .... and we now have the exact same data as before .... but in a 24bit system .... and now only the 4 lower bits have zeroes in them.

 

 

 

I hope that helps.   These things are difficult to explain clearly without being too quick... or too long.

 

Thanks  Dave, that  does make sense.  So I take it that the process used in the 35 bit  M51 and the Accuphase 48 bit digital preamp is different to digital up-sampling  performed  by CD players/ PC software?

Link to comment
Share on other sites



Thanks  Dave, that  does make sense.  So I take it that the process used in the 35 bit  M51 and the Accuphase 48 bit digital preamp is different to digital up-sampling  performed  by CD players/ PC software?

 

By "upsampling" ... I assume you mean changing the sample rate   (as opposed to changing the bit rate, like we've been talking about)

 

Yes, it's different.   It's more complex, but in some basic ways the concepts are similar  (it's all digital manipulation).     Due to sampling theory (the mathematics behind it)  re-sampling requires filters, which make the practical implementation of them more difficult compared to changing the bitrate     (perfect filters don't exist) .....   This is why upsampling is normally better performed "offline"  (ie.  not in a time sensitive setting) .... because then the filters can be super-powerful, and it doesn't matter how long they take.... and you normally have access to more horsepower (eg.  in a "computer").

  • Like 1
Link to comment
Share on other sites

By "upsampling" ... I assume you mean changing the sample rate   (as opposed to changing the bit rate, like we've been talking about)

 

Yes, it's different.   It's more complex, but in some basic ways the concepts are similar  (it's all digital manipulation).     Due to sampling theory (the mathematics behind it)  re-sampling requires filters, which make the practical implementation of them more difficult compared to changing the bitrate     (perfect filters don't exist) .....   This is why upsampling is normally better performed "offline"  (ie.  not in a time sensitive setting) .... because then the filters can be super-powerful, and it doesn't matter how long they take.... and you normally have access to more horsepower (eg.  in a "computer").

 

Although I'm a fan of NOS. Time sensitive setting...CPU speeds are starting to skyrocket now. Even quantity of cores. I mean there are dual quad cores in phones now...why not a dual core dedicated to upsampling in real time. No need to wait. Could this be possible? Moore's law for transistors. Why not DACs and Hifi. Sorry this is getting O/T.

Link to comment
Share on other sites

By "upsampling" ... I assume you mean changing the sample rate   (as opposed to changing the bit rate, like we've been talking about)

 

Yes, it's different.   It's more complex, but in some basic ways the concepts are similar  (it's all digital manipulation).     Due to sampling theory (the mathematics behind it)  re-sampling requires filters, which make the practical implementation of them more difficult compared to changing the bitrate     (perfect filters don't exist) .....   This is why upsampling is normally better performed "offline"  (ie.  not in a time sensitive setting) .... because then the filters can be super-powerful, and it doesn't matter how long they take.... and you normally have access to more horsepower (eg.  in a "computer").

 

Thanks again!   My Accuphase DP-700 upsamples DSD  data by 2X for a sampling frequency of 5.6448 MHZ/1-bit. 24 bit material is not upsampled or converted into a higher bitrate. Therefore the Accuphase CEO comment would be accurate in that there would be bit losses with digital attenuation because there are no layers of zeroes added for padding.  His comment ( in 2006) was in relation to CD players only.

Link to comment
Share on other sites

Although I'm a fan of NOS. Time sensitive setting...CPU speeds are starting to skyrocket now. Even quantity of cores. I mean there are dual quad cores in phones now...why not a dual core dedicated to upsampling in real time. No need to wait. Could this be possible? Moore's law for transistors. Why not DACs and Hifi. Sorry this is getting O/T.

People will always write more complex filters to take advantage of the extra power

 

xxhighend are using 12 core i7 for their digital filters.      :party

 

 

 

EDIT:   I'm having a haphazzard night.   I edited my post, and got rid of everything, whoops.

 

 

What I said was:

 

Not really

 

People write more powerful filters to take advantage of the power

 

Upsampling in a computer is normally "not realtime" even if done at playback .... because it happens in a process which is independent of playback (or even before playback begins)

 

Using "more horsepower" in the DAC.... is what companies are doing now with the FPGA / XMOS type processors .... but it is not always clear if these are independent enough from the audio clocks  (ie.  they're "realtime") .... where as in a computer (before or at playback) it's completely separate.

 

Putting a more powerful processor  (ie.  a computer) inside the DAC .... is difficult.      Computers are noisy.

Edited by davewantsmoore
Link to comment
Share on other sites

Thanks again!   My Accuphase DP-700 upsamples DSD  data by 2X for a sampling frequency of 5.6448 MHZ/1-bit. 24 bit material is not upsampled or converted into a higher bitrate. Therefore the Accuphase CEO comment would be accurate in that there would be bit losses with digital attenuation because there are no layers of zeroes added for padding.  His comment ( in 2006) was in relation to CD players only.

 

You have up to 6dB of attenuation then  (EDIT:   before your increased SNR causes you to effectively "lose a bit").    There are other factors which mean you might get more.     Your comment assumes the DSP operates at 24bit  (which I'll assume is correct).

 

 

Also, depending on how much slack we want to cut him... his comment was general  (in relation to all digital audio, not just accuphase).    Like you say he was obviously dumbing down a marketing message to make the point that their SOTA solution is CD player + preamp   ..... It's just a shame it fuels misunderstanding... but I probably would have done the same if I were him  *wink*

Edited by davewantsmoore
Link to comment
Share on other sites



Bought one of these the other day to replace my dac magic. Have an Arcam A28 integrated amp and PMC TB2S+ speakers, so no doubt a little overkill but for the price it was an unbeatable choice.

 

Was expecting a bit of a difference as I know the speakers are quite capable, but the improvement in imaging and instrument separation revealed just what an excellent buy this is at that price point. Especially given the dacmagic while not fantastic, isn't bad. Biggest difference is on more complex tracks with multiple instruments with the separation and on vocals. A lot smoother and I suspect more accurate as well. Can see this dac lasting for quite a while and several system iterations.

  • Like 1
Link to comment
Share on other sites

Yep, the M51 is good (by all accounts, I haven't heard one) ..... but the Dacmagic is terrible IMO.    It was only a C+ in it's day... and is pretty old now too.

 

I wouldn't class it as terrible. I had an arcam CD73T cd player which was superseded by the dac and flac.The dacmagic just wasn't quite as good as the cd73t. The M51 is spectacular though for the price range.

Link to comment
Share on other sites



I had M51 direct onto a few power amps and it was fine

I was just reading a few posts saying it sounded a hell of a lot better plugged in to a pre-amp. Is it just a tiny improvement?

 

Would need to find suitable power amp for my Osborns with the NAD M51

Edited by roh008
Link to comment
Share on other sites

In my case the improvement was quite significant. It seems to boil down to this:

If the analog output stage of the preamp is better than that in the M51 there is likely to be an improvement.

To my mind, it is very unlikely that the M51 at $1300 will have a better analog output superior to a high quality dedicated preamp.

  • Like 1
Link to comment
Share on other sites

In my case the improvement was quite significant. It seems to boil down to this: If the analog output stage of the preamp is better than that in the M51 there is likely to be an improvement. To my mind, it is very unlikely that the M51 at $1300 will have a better analog output superior to a high quality dedicated preamp.

 

Even if true, if the M51 is in the chain, then the limiting factor will be the M51 anyway. So putting a preamp in the path will simply have an additive effect - so can't improve upon the M51 analogue output. So given they are connected in series, the M51 is likely to be superior if just connected by itself as it takes one component out of the chain. Also remembering that even on a 24 bit source, up to 66dB of attenuation can be applied with absolutely no loss of quality.

Edited by adprom
Link to comment
Share on other sites



Also remembering that even on a 24 bit source, up to 66dB of attenuation can be applied with absolutely no loss of quality.

 

There's a caveat to that....   (Just so people don't think I'm a one-eyed-pusher of digital)

 

No loss in the digital stage ... yes.

 

But   ... when the M51 is outputting a -66 dBFS signal  ... that signal is 0.0005 volts ... and if the analog circuitry (including it's own SNR) does not deal with this signal well, then it may be inferior to outputting fullsized (2V) signal from the M51, and then attenuating it somewhere else   (or ideally just only amplifying it by the correct amount .... attenuation is bad)

Link to comment
Share on other sites

 

I was running the M51 directly into a 200wpc power amp and not happy with the sound at all. I felt it wasn't doing justice to the ML2 LEs at all. There was little body and weight to the sound and the sound stage was small and recessed. Then a friend loaned me an AR REF5 valve preamp and we couldn't believe the difference it made. Sound stage was larger, the music had more presence, vocals were more textured, bass improved and surprisingly I lost nothing in clarity, detail and resolution.

 

I couldn't afford this preamp so settled for a Ming Da MC300-PRE and I'm getting a similar  result.

 

This was not a volume issue. With the direct connect of the M51 to the power amp the volume control on the M51 was never above -25db. With the preamp, I set the vvolume control on the M51 to fixed, -1 db and the voulme on the MC300 is never above the 9 o'clock position.

 

I could never go back to not using a preamp.

 

I totally agree. I tried the same to Modwright KWA150 and was not happy. When running through the Modwright LS100, It was so good. Warm sound, bass more solid and open sound stage.

Link to comment
Share on other sites

I wonder if this is my problem as well. I had the M51 driving an Emo XPA-5 and now a pair of Consonance Cyber 800 monos. *** sounds a bit too lean and without much warmth and body to music. But transparency is first class though. Probably time to hunt for a good valve preamp :)

Edited by eyesofra
Link to comment
Share on other sites

Interesting regarding the NAD M51 not sound great when directly plugged in to a power amp.. I was leaning towards the NAD M51 and a power amp set up!

 

Its system dependent.  I have heard it both ways on a few systems.  On mine and systems similar to mine it was better to much better direct.  On a vintage type system it was tried on a simple pot was better than direct connecting.

 

Thanks

Bill

Edited by bhobba
Link to comment
Share on other sites



  • Recently Browsing   0 members

    • No registered users viewing this page.




×
×
  • Create New...
To Top