advertisement


NOS DACs vs the rest

Er, no. The micro mega stage 3 uses a TDA1305 which is a bitstream DAC and outperforms the TDA1543 on every level, it’s in a different league.

Slightly off topic but the Naim CD3.5 uses the 1305 as well and has a seven pole bessel filter anologue stage on the end. I bypassed the anologue stage leaving just the data sheet first order filter and much preferred the sound. IMO the anologue stage makes it sound artificial and processed. Added plus was I could get rid of two psu's as well :)
 
Do you have a rough amount in mind? And is there an itch scratching component to the decision?

I just ask because if all you want to do is get a similar result to your Orchid, but on a cheaper rung of the ladder, perhaps a MHDT Canary would be the ticket, without even getting into different technologies?
Two good questions there. I posted this thread as I was thinking about Metrum Onyx that was in the classifieds until recently. More than I can pay really, but, as you say, there is a certain amount of itch scratching. I have changed the layout on my office. I’m now retired and don’t need all the workspace so have re designed the office to accommodate a space suited to listening. I’ve been so surprised at how much I’ve enjoyed listening to music in the office that I’ve started to think about optimising the office for digital and keep the lounge for vinyl.
 
An someone explain the differences, in language an idiot would understand, the pros and cons of a non oversampling DAC over the other non non oversampling DACS?

My understanding so far Is that a NOS DAC will sound more natural/analogue, whereas a ‘normal’ DAC will have a bit more sharpness about it, a more ringing sound? But I’m sure I’m missing something.

If only it was so simple to make generalisations. If I were to try, i'd say almost the exact opposite, that normal DACs sound analog and smooth, whilst NOS DACs sound crisp and digital by comparison.

The normal DAC is correct, the NOS DAC is wrong (as in all the differences are distortion not present in the recording), but the vast majority of the distortion is beyond most people's hearing. Now which you prefer is personal, and will I think very much depend on what qualities you listen to, and frankly I think the differences are easier to measure than to hear, if you see what I mean, so don't get hung up on the tech, just concentrate on whether you like what it sounds like.
 
If only it was so simple to make generalisations. If I were to try, i'd say almost the exact opposite, that normal DACs sound analog and smooth, whilst NOS DACs sound crisp and digital by comparison.

The normal DAC is correct, the NOS DAC is wrong (as in all the differences are distortion not present in the recording), but the vast majority of the distortion is beyond most people's hearing. Now which you prefer is personal, and will I think very much depend on what qualities you listen to, and frankly I think the differences are easier to measure than to hear, if you see what I mean, so don't get hung up on the tech, just concentrate on whether you like what it sounds like.
Yes, a large part of my anterior motive is to learn a bit about the technology. Graphs and numbers do not facilitate learning for me, I need a picture. From what you’ve said the picture I have is that distortion is an inherent problem of converting a digital signal into one that’s analogue and OS cleans up that distortion to one degree or another and NOS says, fook the distortion, let’s stay as close to the original signal as poss???
 
It is probably more accurate to say that converting analogue to digital is where the distortion occurs. NOS says fook it, let's ignore that and let the analogue reconstruction filter have a go at cleaning up the mess.

Early A-to-D was a little crude, modern A-to-D is really designed in line with the assumption oversampling will occur during reconstruction.

Oversampling exists primarily to enable that distortion to be moved far above the 20khz nominal limit on our hearing, then minimise it with filtering.

Both can sound good, because our ears are quite forgiving.
 
It is probably more accurate to say that converting analogue to digital is where the distortion occurs. NOS says fook it, let's ignore that and let the analogue reconstruction filter have a go at cleaning up the mess.

Early A-to-D was a little crude, modern A-to-D is really designed in line with the assumption oversampling will occur during reconstruction.

Oversampling exists primarily to enable that distortion to be moved far above the 20khz nominal limit on our hearing, then minimise it with filtering.

Both can sound good, because our ears are quite forgiving.

Excellent, I get that.

Does that mean a modern DAC will beat an old DAC in a fight?
 
That probably depends on numerous factors - the trend is for better technical performance, but many people prefer a sound that does not correlate with that. There are also many products that are extremely popular which seem to pay no heed to technical performance (measured performance) - Total Dac, (arguably Schiit too) for example.

Also, input bitrate plays a part. NOS at a samplerate of 44.khz has potentially a very different noise profile than 96khz samplerate data. It has nothing to do with the original (analogue) signal, but the artifacts of the process of converting that to digital data. The further away you can get those from the passband (audible output) of the reconstruction filter the better. Metrum's last generation of dacs (Pavane/Adagio), also Holo Spring or Sonnet Morpheus are examples of NOS dacs that accept high bitrates.

Which sounds better (to an individual) is ultimately about either preference (does this dac fit my preferred sound profile) or doctrine (does it measure well?, Does it have the buzzwords that conform to my idea of what technology makes the best sound or conformance to the theory as I understand it). Subjective vs Objective is an argument that never ends well in forums though!
 
  • Like
Reactions: mjw
Also, input bitrate plays a part. NOS at a samplerate of 44.khz has potentially a very different noise profile than 96khz samplerate data. It has nothing to do with the original (analogue) signal, but the artifacts of the process of converting that to digital data. The further away you can get those from the passband (audible output) of the reconstruction filter the better. Metrum's last generation of dacs (Pavane/Adagio), also Holo Spring or Sonnet Morpheus are examples of NOS dacs that accept high bitrates.

Yes, none of these issues really apply when we are at 96kHz and above... you can use an old TDA1541A at 768kHz if in Simultaneous Mode.
 
I've struggled to get digital sounding how I would like tbh.
The closest I've got is the NOS Lite DAC-ah that I have now, which uses 8 x TDA1543 in parallel. I think using multiple DACS is some way of 'averaging out' any errors?.. I don't really know, all I do know is it sounds more analogue than a lot of others I've had (Beresford, some Naim CDP, JVC 1010 internal DAC).
And it's proper cheap (£150).
The analogue output on my AIWA Excelia cdp is quite nice too - not sure what DAC chip that uses. It misses out on a lot of detail but it's very not 'digital' sounding at all.

I do find a lot of older DACs sound more analogue. I've always fancied a DPA, something like a little bit 3 or one of the better Audio Alchemy setups.
 
Excellent, I get that.

Does that mean a modern DAC will beat an old DAC in a fight?

Even if it does, metaphorically speaking, you may still prefer to stroke Bagpuss than a CatBot5000.

Since it sounds like there's a bit of an oversampling itch to scratch, and you've already got a good un-modern NOS DAC for comparison, how about trying a really up to date design like a recent-ish Chord or maybe Mytek? Various price levels/trials/low risk resale options available.
 
Yes, My main system has the HX1.2 with L300, LP12. Also use a Rel S5. It’s a sound I like very much.
Nice. Do you have the hen’s teeth IKEA stands? The Elves sound fab on them. Infact most small speakers I’ve tried on them (some good ones) sound better on the IKEA jobbies than conventional stands.

Anyway, back to DACs, my RME is in the bedroom, the Chord 2Qute does the TV. Audial S4 showed them the door and then bolted it shut but does cost twice as much.

BTW I work from home (sort of) and the Audial is playing music almost all day every day; totally non fatiguing and so well implemented I want for nothing.
 
Nice. Do you have the hen’s teeth IKEA stands? The Elves sound fab on them. Infact most small speakers I’ve tried on them (some good ones) sound better on the IKEA jobbies than conventional stands.

Anyway, back to DACs, my RME is in the bedroom, the Chord 2Qute does the TV. Audial S4 showed them the door and then bolted it shut but does cost twice as much.

BTW I work from home (sort of) and the Audial is playing music almost all day every day; totally non fatiguing and so well implemented I want for nothing.

You got me curious with that mention of the Audial (I'm currently using a little SPS DAC, and considering eventually looking at something like Audio Note/SW1X/??? to replace it. One day...).

So I had a search, and in case it's useful for KS, found your thread here.
 
Yes, a large part of my anterior motive is to learn a bit about the technology. Graphs and numbers do not facilitate learning for me, I need a picture. From what you’ve said the picture I have is that distortion is an inherent problem of converting a digital signal into one that’s analogue and OS cleans up that distortion to one degree or another and NOS says, fook the distortion, let’s stay as close to the original signal as poss???
When you sample at a particular sample rate, you can perfectly (ish) capture information up to half the sample rate (22kz in the case of red book). But the samples contain spurious information repeating the real information around multiples of the sample rate. If you reproduce the sample values using "sample and hold" you retain those false high frequencies, and you also roll off the high frequencies in the audioband a bit. That's how old dacs used to work. You can deal with this by oversampling which is the only way to filter out the hf digitally (and you can address the roll off) or you can use an analogue filter. Confusingly NOS dacs usualy omit the analogue filter too. Quite why that is supposed to be a good idea eludes me. It isn't really closer to anything except old technology, only done wrong. But some people think it sounds better.
Equally confusingly modern dacs don't just use oversamping to apply the filter, but then go on to convert to anlogue by reducing the 16 bits or 24 to a handful of bits, or even one, at a really really high frequency. This is for a different reasons connected with the problem of making 56k different outputs at exactly the right level. It's easier to make outputs at a few different levels but do it faster. I suppose at this point the NOS dac might be said to be closer to the data on the cd. But is it closer to the "original signal", well it depends what that means.
This is one area where it is difficult to avoid graphs and numbers. Digital sampling really is about maths. But "more accurate" doesn't necessarily mean "sounds better".
 
Months on I stand by everything I said in that review although I might try to be less pretentious if writing it again. It is difficult to explain the magic of a superbly implemented, no expense spared, TDA1541 chip but the best I can do is: a piano sounds like a piano, a violin sounds like a violin and a guitar sounds like a guitar. Well made Nos dacs tend to do this.
 
Many thanks to everyone. This thread has been very informative for me. I even understood some of it.

Now listening to a Benchmark DAC that stayed unloved in a drawer after a couple of issues. The issues were fixed sometime ago, but I still had negative feeling towards it. It’s now in my system and sounding pretty damn good.

Subjectivity eh?
 
It is probably more accurate to say that converting analogue to digital is where the distortion occurs. NOS says fook it, let's ignore that and let the analogue reconstruction filter have a go at cleaning up the mess.

And what analogue reconstruction filter would that be?
 


advertisement


Back
Top