PDA

View Full Version : NOS verses Oversampling ? Choosing a Dac



don
18-05-2009, 13:28
Gidday all after introducing my self I now wish to post my first question to you guys! I need to make a choice about purchacing my first dac to play PC based music. My budget is about $400 Australian. After much research i have come down to 2 choices in NOS

Both NOS Dacs, which are known for there analogue type easy listening tones, use the Philips TDA1543 Dac chips( 8 chips in parrallel). They do not utilize opamps.

Valab NOS DAC (2009 Version) (http://cgi.ebay.com/Valab-NOS-USB-DAC-TDA1543-Low-Jitter-1ppm-TCXO-Inside_W0QQitemZ300307438536QQcmdZViewItemQQptZLH_ DefaultDomain_0?hash=item300307438536&_trksid=p3286.c0.m14&_trkparms=72%3A1234%7C66%3A2%7C65%3A12%7C39%3A2%7C 240%3A1318)

Gigalab NOS USB DAC TDA1543 1ppm TCXO +Re-clock Circuit (http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=280332293768&ssPageName=MERCOSI_VI_ROSI_PR4_PCN_BIX&refitem=280331665148&itemcount=4&refwidgetloc=closed_view_item&refwidgettype=osi_widget&_trksid=p284.m263&_trkparms=algo%3DSIC%26its%3DI%252BC%252BP%252BS%2 52BIA%26itu%3DIA%252BUCI%252BUA%26otn%3D4%26ps%3D1 0)

Beresford 7520 built by Stan is my choice for a Oversampling DAC after reading excelent reviews on this forum and others.

I have no experience with DACS so any advice on these Dacs or opinions on NOS verses Oversampling would be much appreciated.
Don

DSJR
18-05-2009, 17:39
Greetings!

I don't know the two DAC's you mention and can only speak from positive reviews here and elsewhere on the Beresford. So why bother posting? Well, all I can say is that the more complex you make something, the more room for errors IMO.

The other point I'd like to make is that DAC chips these days aren't very expensive and, if implemented well, don't perform too differently IMO, many Top End designers adding a flavour to suit their tastes (and very expensive casework too).

I suppose my recommendation would be for the Beresford. Try this for a while and then try the others to see what you think.

I'd be interesting to read what others have to say...

StanleyB
18-05-2009, 19:25
My budget is about $400 Australian.
At AUD265 odd for the TC-7520 AoS Export, that leaves you quite a bit of beer change.

NRG
18-05-2009, 20:31
I've not heard the 1543 but my experience of the 1541A in NOS mode is of a very lucid and detailed midrange, great bass and an ever so slightly rolled off treble. Given a good implementation, which is very key to getting the best out of these chips as Leo will tell you, they are capable of a very captivating and engage performance much more so than the Burr Brown and Wolfsen DAC's which sound a bit too incisive, bright and lean in comparison.

Trouble is you are asking for a recommendation but only you really know what is best suited to you and your choice in music, I would say the Ebay DAC's are a bit more of a risk / lottery sound quality and reliability wise unless you are prepared and capable of driving a soldering iron to fix things if you find out you've bought a bad one.

Modern DAC's can and do sound similar mainly IMHO due to datasheet engineering that results in similar layouts and component use. The 154x DAC's can sound very different amongst similar DAC's (layout, PSU etc) and certainly sound different to the Burr Brown etc. DAC's. I prefer the 1541A sound but that's not to say it would suit you.....tricky choice...

ReachtheSky
18-05-2009, 23:47
Hi Don. I can’t help you with your NOS choices, but I can reaffirm all the good things that have been said about Stan and his 7520. It’s a great product at a great price (with suitable opamp upgrades - only A$10 to A$20 each). It is a top DAC, headamp and digital preamp for power amps all in one box. IMO I don’t think there is a product to compete with it in this price range. And Stan is very accommodating to deal with and provides prompt delivery to Australia. (However your listening tastes may prefer the easy laid back TDA154x analogue sound!)

StanleyB
19-05-2009, 05:44
dac to play PC based music.
Both the TC-7510 and TC-7520 were developed to get the best from musc files and streamed audio. It's a gap that I detected in the DAC market some years ago. In the main, most DACs out there appear to be aimed at the CD and DVD market, where the player is used as a transport and the D to A decoding is done by the external DAC.
From my experience, the NOS DAC chips are squarely aimed at CD audio. But that is not surprising since they were designed to decode CD audio in the first place. High resolution PC audio files did not exist in the days when NOS chips sets were developed. To get an idea of what I mean, play back a low resolution mp3 file through a TDA1541 and through an oversampling DAC chip and compared the two in a listening test. The main thing that sticks out is that the OS chip sounds far more like the original uncompressed audio sound. The NOS chip just doesn't sound anywhere as detailed and shows up the imperfections in the mp3 file far more. That disrupts the listening pleasure.

STan

StanleyB
19-05-2009, 05:51
Modern DAC's can and do sound similar mainly IMHO due to datasheet engineering that results in similar layouts and component use.
Really:scratch:? The facts do not support your assertions. There are current and voltage chips out there. There are also different types of DAC chips format as well. You have single bit, 8x, 64x, etc oversampling. Some chips have inbuilt filters of different kinds. Output levels varies. Some chips need 1, whilst others need two supplies. Some have an inbuilt clock, whilst others need an external clock.
All of the above affects the sound of each chip and design of the DAC circuitry. It is far removed from your suggestion that they follow similar layouts and component use.

Stan

Marco
19-05-2009, 07:22
I've not heard the 1543 but my experience of the 1541A in NOS mode is of a very lucid and detailed midrange, great bass and an ever so slightly rolled off treble. Given a good implementation, which is very key to getting the best out of these chips as Leo will tell you, they are capable of a very captivating and engage performance much more so than the Burr Brown and Wolfsen DAC's which sound a bit too incisive, bright and lean in comparison.


I don't normally participate much in this area of the forum, but I think this requires some comment.

Neal, I fully concur with most of the above, as that's exactly how my DAS-R1 sounds, apart though from the "ever so slightly rolled off treble". I have found no evidence of this myself when the 1541A has been optimally implemented and used in the right system. There of course could be a multitude of variables responsible for why this is what you've experienced, and none of them necessarily related to the chip itself.

I've compared Burr Brown and Wolfson-based DACs head-to-head in my system with the DAS-R1, and the former sound exactly as you describe. However, as far as treble goes, I would describe it as the former having a sonic signature which unnaturally over-emphasises the upper frequencies, or perhaps rather that their leaner nature in the bass draws more attention to the treble, therefore making one feel that it is over-emphasised.

The DAS-R1 (using 1541s) has no such treble over-emphasis: it is simply more 'natural' sounding in that area, rather like vinyl in that respect, and/or perhaps as above, it also has excellent bass which gives music proper 'foundation', thus not drawing one's attention to the upper frequencies and laying them bare for 'forensic inspection', in the way of BB or Wolfson-based DACs (although these can of course be successfully implemented so that the effect is minimised and the overall result is very musical sounding).

Choose whatever explanation you think best fits the truth, but there is no evidence in my (extensive) listening experience which suggests that the 1541A has an "ever so slightly rolled off treble", *providing* that it has been implemented optimally, and by "optimally" I mean something at least of the calibre of my Audiocom-modified DAS-R1, which no doubt your own designs, and those of Leo's, for example, ably emulate.

If you're hearing an "ever so slightly rolled off treble" with a 1541A, then IMHO something else in your system is responsible, restricting or influencing the signal in some way, which could be set-up related (cables, stands, mains - your room - goodness knows what!), 'cos it ain't happening chez-moi :)

Marco.

StanleyB
19-05-2009, 08:00
The 'rolled off treble' has absolutely nothing to do with the TDA1541. The roll off comes from the anti aliasing filter network that bolts on to the audio output of the TDA1541.

Marco
19-05-2009, 08:18
................ *if* indeed you hear this "rolled off treble" in the first place! ;)

Marco.

NRG
19-05-2009, 09:31
You can measure it as ineed Leo will tell you. Perhaps 'closed in' would be a better description.



Both the TC-7510 and TC-7520 were developed to get the best from musc files and streamed audio. It's a gap that I detected in the DAC market some years ago. In the main, most DACs out there appear to be aimed at the CD and DVD market, where the player is used as a transport and the D to A decoding is done by the external DAC.
From my experience, the NOS DAC chips are squarely aimed at CD audio. But that is not surprising since they were designed to decode CD audio in the first place. High resolution PC audio files did not exist in the days when NOS chips sets were developed. To get an idea of what I mean, play back a low resolution mp3 file through a TDA1541 and through an oversampling DAC chip and compared the two in a listening test. The main thing that sticks out is that the OS chip sounds far more like the original uncompressed audio sound. The NOS chip just doesn't sound anywhere as detailed and shows up the imperfections in the mp3 file far more. That disrupts the listening pleasure.

STan

:scratch: What's different in the SPDIF stream between a decoded MP3 file vs a WAV file? I certainly have not heard this, I prefer the 1541A NOS presentation of MP3 even at 128Kb vs an OS DAC, the 7510 included. I think anybody would really struggle to detect a difference no matter what DAC you are using between 320Kbs and WAV, may be a golden eared few with very revealing systems.

Your sentance above Stan doesn't make sense.

lovejoy
19-05-2009, 09:45
Hi Don,
I've had some of the best listening sessions I've ever experienced in digital audio with a DAC using 1543 chips in parallel in Non-OS mode. The music just flows, it's lush, focus is beautiful, there's real weight and authority to instruments, vocals have that up close and personal feel, bass is warm, deep and well controlled. It's like a very good turntable.

I've moved on from there to the 7520 for an element of future proofing - I can now play 24/96 audio which anything 1543 based is not going to do. It's also incredibly good as a pre-amp, especially with new op-amps and to my ears, sound quality, whilst quite different to the 1543 - i.e. it's not as 'analogue' sounding - makes up for it with bags of dynamics, more accurate bass and fine detail.

To me, it's the digital source that makes all the difference. Either DAC is made or broken on what computer you're using, the operating system, whether your files are compressed or not, how your computer is connected to the DAC and how good the interconnect is between the two. If you don't pay attention in all of these departments, you'll have a hard time telling the difference between DACs anyway.

Marco
19-05-2009, 09:59
You can measure it as ineed Leo will tell you. Perhaps 'closed in' would be a better description.


LOL. There's nothing "closed in" about the sound of the DAS-R1 (using the 1541s you describe), either! :)

With respect, Neal, I'm not really interested in what can be 'measured', simply because measurement apparatus is often not compatible with what's considered as natural (or 'accurate') when processed through the auditory complexities of the human ear (and the brain's interpretation of it).

Basically, things aren't that simple because we (as 'complicated' humans) don't always hear music in the way an oscilloscope measures sound, which makes one question how relevant some scientific measurements actually are in hi-fi applications.... ?

With hi-fi systems, too, the sum of its constituent parts is always greater and of more relevance than the individual effect of each component. However, that aside, I stand by my comments earlier in reference to your "slightly rolled off/closed in" remarks regarding the 1541A:


I've compared Burr Brown and Wolfson-based DACs head-to-head in my system with the DAS-R1, and the former sound exactly as you describe. However, as far as treble goes, I would describe it as the former having a sonic signature which unnaturally over-emphasises the upper frequencies, or perhaps rather that their leaner nature in the bass draws more attention to the treble, therefore making one feel that it is over-emphasised.

The DAS-R1 (using 1541s) has no such treble over-emphasis: it is simply more 'natural' sounding in that area, rather like vinyl in that respect, and/or perhaps as above, it also has excellent bass which gives music proper 'foundation', thus not drawing one's attention to the upper frequencies and laying them bare for 'forensic inspection', in the way of BB or Wolfson-based DACs.


That's what my ears and experience tell me, not an oscilloscope ;)

Marco.

StanleyB
19-05-2009, 10:31
Let's correct a massive misconception here once and for all.
The NOS chip sounds different in the bass and treble region compared to a OS chip for the simple reason that in a NOS chip the sampling rate of the lower bass frequency is the same as in the treble region. So for each time period you are going to get the same amount of samples in the top and bottom. What that means is that say a 100Hz signal is going to be sampled a lot more times for the same sampling period, compared to a 10KHz signal. The sampling rate is a fixed 44.1KHz for audio in a NOS chip.

In the case of a OS chip, the sampling rate is dependent on the sampling rate of the chip. In many cases it is 8X. That means a OS chip is sampled 8X more than in a NOS chip. The end result is that there are more sampled bits in the treble from a OS chip compared to a NOS chip. More sampled bits equates to better signal resolution, which equates to a more accurate reproduction of the original analogue audio signal before AD conversion.

It is unfortunate that some folks attribute the AD/DA technology to the chips used, instead of to the technology used.

Stan

Marco
19-05-2009, 10:34
Hi Stan,

Thanks for that. Can you translate the above into 'laymens terms', though, so a non-techie numptie like me, who just listens to music, can understand the point you're making?

Ta! :)

Marco.

StanleyB
19-05-2009, 10:39
:scratch: What's different in the SPDIF stream between a decoded MP3 file vs a WAV file?
Are you being serious???

In order to convert a WAV file to mp3, the WAV file is digitally filtered and some parts of the frequency response is even removed. It's a lossy compression method. Decoding a mp3 file back to WAV does not recover the bits that were discarded during mp3 compression. However, using a OS DAC, a lot of the information discarded during mp3 conversion can be reasonably recreated from the 'before and after' audio bits. A NOS chip in unable to do that.


Stan

StanleyB
19-05-2009, 10:44
Hi Stan,

Thanks for that. Can you translate the above into 'laymens terms', though, so a non-techie numptie like me, who just listens to music, can understand the point you're making?

Ta! :)

Marco.
In broad terms, and without resorting to even more tech talk, a 200HZ signal sampled at the 44.1KHz sampling rate is going to have more bits than a 20KHz signal sampled at 44.1KHz.

don
19-05-2009, 10:47
At AUD265 odd for the TC-7520 AoS Export, that leaves you quite a bit of beer change.

Thanks for the advice from everyone seems I have started a lively debate.
I have decided to purchace my first DAC based on positive feedback and pricepoint. The TC 7520 is in my budget with money left over for quality interconnects.(and a few beers)
I have not yet listened to enough dacs to draw my own conclusions. Although as far as NOS goes I do have an old Marantz CD player which utilizes the TDA1541A and it does sound sweet and natural.

As far as purchasing a TC 7520 it appears I can PM stan to arrange this ?

Marco
19-05-2009, 10:54
In broad terms, and without resorting to even more tech talk, a 200HZ signal sampled at the 44.1KHz sampling rate is going to have more bits than a 20KHz signal sampled at 44.1KHz.

LOL. Although I understand more what you mean now, I was looking for a reference in terms of how you think music is handled by both (as this is what I can relate to). I'm afraid that most of this 'techie stuff' is pretty meaningless to me, but not to worry :)

Given your undoubted talents with modern chipsets, I'd like to see what you could do, commercially, with a 1541A-based design. If Neal and Leo can build excellent sounding examples relatively inexpensively, so could you, surely - what do you think? Expand your product range to show the best of what old and new technology can do! :smoking:

Marco.

NRG
19-05-2009, 13:24
Are you being serious???

Yes.



In order to convert a WAV file to mp3, the WAV file is digitally filtered and some parts of the frequency response is even removed. It's a lossy compression method. Decoding a mp3 file back to WAV does not recover the bits that were discarded during mp3 compression.

Never said it did.



However, using a OS DAC, a lot of the information discarded during mp3 conversion can be reasonably recreated from the 'before and after' audio bits. A NOS chip in unable to do that.


Stan

No it can't, the data is lost, no matter how much you oversample it's not going to come back. Oversampling eases the requirements of the Anti-Aliasing filter and makes better use of the 16bit information available, IE it makes the DAC more linear...but you know this already.

NRG
19-05-2009, 13:33
LOL. There's nothing "closed in" about the sound of the DAS-R1 (using the 1541s you describe), either! :)

With respect, Neal, I'm not really interested in what can be 'measured', simply because measurement apparatus is often not compatible with what's considered as natural (or 'accurate') when processed through the auditory complexities of the human ear (and the brain's interpretation of it).

Basically, things aren't that simple because we (as 'complicated' humans) don't always hear music in the way an oscilloscope measures sound, which makes one question how relevant some scientific measurements actually are in hi-fi applications.

With hi-fi systems, too, the sum of its constituent parts is always greater and of more relevance than the individual effect of each component. However, that aside, I stand by my comments earlier in reference to your "slightly rolled off/closed in" remarks regarding the 1541A:



That's what my ears and experience tell me, not an oscilloscope ;)

Marco.

Good for you Marco, I'm glad the DAS-R1 is as good as you say, no doubt due to Sony knowing how to implement the 1541 correctly. However, measurement is important and while you may not place any merit in it I bet the Sony engineers who worked on the DAS-RA bloody well did.

You need measurement to give a base line you can't design without....well you could try but....:doh:

Marco
19-05-2009, 13:37
Neal,

I didn't say that measuring stuff wasn't important (of course it is); merely that some measurements may not be as conclusive as you might think...

The reason why measurements aren't of interest to me personally is because I don't design stuff, D.I.Y or otherwise. I just buy equipment and listen to music on it, hence why I will always trust my ears and (quite extensive listening experience in this area) before the readout on anyone's oscilloscope, no offence :)

Based on that, I maintain that the 1541A (when correctly implemented) is not in any way "rolled off" or "closed in" ;)

Marco.

Tom472
19-05-2009, 15:55
Without wanting to flare up any arguments, I was under the impression that when you oversample a signal, you don't get any more or fewer samples - rather you get more "gaps" between the samples. The reasoning being that when you unsample the higher frequency signal, the unwanted sidebands are shifted further away from the audio range (and so easier to filter out).

What strikes me is how little difference it actually makes (there is a difference) but I've A/Bd and old CD player switching in and out the oversampling and the difference is quite subtle. Makes you wonder why the Philips engineers went to all that trouble when designing early CD players.

Anyhoo, back to the original question - I guess the answer is you'll have to try different DACs to really see which you prefer (cop out). The Beresford does come so highly recommended however that you probably can't go wrong as a starting point.

Cheers

Tom

leo
19-05-2009, 16:33
It gets confusing guys;) I tried to stay away from this thread for as long as I could because the only thing I can add takes it way off topic from the original post

I have not listened to a 7520 or the Valab mentioned in the first post.

From experience I don't like using older dacs in parallel like the Valab uses (I have my reasons) I've tried it before to satisfy curiousity in diy designs but didn't stick with it.
I've also heard that the Valab had a few quality issues with the capacitors it used, this may now have been sorted out in the later batches?

A few of the older dac chips can be used as NOS not just the TDA's because unlike newer stuff the digital filter is external rather than internal in the actual dac chip

Having the digital filter, I/V etc etc inside the actual dac chip makes designing cdp's and external dacs much cheaper, easier and far more compact.
Like most things cramming as much technology into a single device can have compromises.
This is not to say all of the all in one chips are shite because they are NOT, like everything else you have to find the best performing part to suit your needs

TDA1541 is still one of my favourite chips but it also has a lot of downsides

Circuitry needs to be much more complex, larger and expensive, it needs more powersupplies with different voltages including two which are -v.
The chip itself is no longer produced so does not comply to ROHS (it contains lead) so a lot of countries with these ROHS regulations can't use it anyway in their new products.

Upside is get it right and it sounds fantastic IMHO! but at a price:eyebrows:
The downside mentioned above is an advantage to the tinkerer as we have more freedom to fine tune having all this circuitry external to the chip

The mistake some people make is that they buy say a cheap old cdp using the TDA1541 , take out the SAA7220 digital filter and expect perfect NOS, not surprising it does not sound that great because the output stage was originally designed for oversampling.
Theres quite a lot involved;) One of the diy dacs I use was designed to be used with NOS, nothing I've heard yet can produce piano sounds as realistic as this one!
Swapping op-amps and using fancy arsed caps is not going to help either a simple converted dac or cdp made to be used with OS, it requires re-design

So without trying to drone on anymore if you want something cheap, convenient and not needing any diy I don't think I could recommend a NOS design.
The 7520 has had nothing but praise on here from what I've seen so maybe its the safest bet.

I've still yet to hear a cheap commercial dac to outperform my diy dac, soon as I do I'll be sure to mention it.
Some of us are not into diy just to keep things cheap, its because we want the best sound possible TO US and this is not always possible with commercially made units

StanleyB
19-05-2009, 17:15
Without wanting to flare up any arguments, I was under the impression that when you oversample a signal, you don't get any more or fewer samples - rather you get more "gaps" between the samples. The reasoning being that when you unsample the higher frequency signal, the unwanted sidebands are shifted further away from the audio range (and so easier to filter out).

There is a lot of misunderstanding about oversampling. Some folks even confuse upsampling and oversampling.



What strikes me is how little difference it actually makes (there is a difference) but I've A/Bd and old CD player switching in and out the oversampling and the difference is quite subtle. Makes you wonder why the Philips engineers went to all that trouble when designing early CD players.
The best test record I have found to check out the effect is Marvin Gaye - What's Going On. Listen to the vinyl with a MC cartridge, and listen to the CD on a 14 bit NOS, 16 bit NOS, 8X OS, and 64X OS. The appearance and disappearance of certain instruments and sounds is the one thing that is so striking. Why it is, I have no idea.

nb2
19-05-2009, 18:38
I guess a lot of you already know that, but there is very useful information about sampling and oversampling on
http://www.lavryengineering.com
Have a look at the "White Papers" in the "Support" section.

ReachtheSky
20-05-2009, 01:02
Hi Don,

To me, it's the digital source that makes all the difference. Either DAC is made or broken on what computer you're using, the operating system, whether your files are compressed or not, how your computer is connected to the DAC and how good the interconnect is between the two. If you don't pay attention in all of these departments, you'll have a hard time telling the difference between DACs anyway.

Hi Rich, I’m very interested in you comments about the effect of digital sources. I recall you did comment on this matter in another thread, but I’d appreciate a sort of summation of your views on digital sources.

I am in the early stages of moving from CDP/DVD source to computer/HD based digital source and would be most grateful for your views based on your experiences.

Thanks Granville

lovejoy
20-05-2009, 07:18
Hi Granville,
I'd be more than happy to share my experiences. It's all still very much in the work in progress stage, which I shall elaborate on, but I think it will require a new thread...

ReachtheSky
20-05-2009, 12:54
Yes a new thread would be appropriate.

becseattle
04-06-2009, 19:46
Well I actually use the 2009 Valabs DAC, and I have compared it to many others, as documented on other fora. The whole Ebay shippping and Paypal was smooth, and the producer quite responsive to questions. It's an excellent NOS DAC, and I prefer it to the usual list of NOS candidates.

I have also been through umpteen other DACs, and while the limitations of NOS DACs are pretty clear, I like it better than the many upsamplers/oversamplers I have put into my system. On paper I see the appeal of a Benchmarks etc, but in practice I usually prefer the NOS sound.

I use DACS with a Macbook and a SB2 wirelessly, and the rest of the system is YBA Passion Integre and Audio Physic Virgos. I found that the addition of a linear power supply made a bigger difference with the SB than any particular DAC of a class. Head to head I still prefer my Naim CD5/Hicap, but I seem to use the SB most of the time.

Bruce