[Viper2001]

Quote:

Received my Aqvox power supply. After a couple of days I honestly can not hear any difference. So either the Anedio D2 USB is really well designed or the SBT USB power is pretty clean.
I am using DH Labs newly designed USB cable.
Note, I have no cans but by my two channel system is pretty revealing.

 
Now, try a PC USB cable and then use your Aqvox power supply.  Do you hear any difference?

 

[cky8]

Viper, why don't you build your own DAC? I am sure you will come out with something good 

 

[Orangecrush]

Quote:

 
Now, try a PC USB cable and then use your Aqvox power supply.  Do you hear any difference?

 
Tried it tonight for a listening session. I discerned no difference with a mid level Belkin USB cable. 

 

[Theogenes]

I've just finished reading this thread in it's entirety (I'm interested in the Anedio and have a bit of an OCD streak ), and I have an utterly noobish question if any of the many more knowledgeable guys here has a moment: I know for a while, NwAvGuy's DAC really seemed to be moving the "better measurements can correlate with better sound" concept, and it seems the guy's design measures pretty well. (Again, please forgive me if I'm mistaken on any of this, I am very much an enthusiast and not at all an expert). This is going to be a naive question I know, but where did he go wrong? It seems nearly everybody that listens to a D2 walks away impressed, but many people that hear the ODAC just don't like it. I understand that two different design goals can result in very different outcomes, but it seems like the two camps (Anedio and NwAvGuy) had at least somewhat similar goals and both measure well, but one gets a vastly differing set of responses. Anybody want to take a stab at explaining this to a hoi polloi fella like myself?

PS: I know the guy mentioned above caused some kind of ruckus in the past, so if my post causes any trouble or breaks any rules I'll gladly remove it. This just seems like an interesting question to me!

Thanks to all!!

 

[veritas aurum]

As Von Recklinghausen (of H.H. Scott and one of my heroes) once said "If it measures good and sounds bad, your are measuring the wrong things". I just think James at Anedio is measuring the RIGHT things. Regards

 

[LeoDuran]

...Anybody want to take a stab at explaining this to a hoi polloi fella like myself?

Another take away is perhaps: What really matters is how it sounds.
(so it's great when you have an opportunity for an in-home trial)

 

[DaveBSC]

Quote:

I've just finished reading this thread in it's entirety (I'm interested in the Anedio and have a bit of an OCD streak

), and I have an utterly noobish question if any of the many more knowledgeable guys here has a moment: I know for a while, NwAvGuy's DAC really seemed to be moving the "better measurements can correlate with better sound" concept, and it seems the guy's design measures pretty well. (Again, please forgive me if I'm mistaken on any of this, I am very much an enthusiast and not at all an expert). This is going to be a naive question I know, but where did he go wrong? It seems nearly everybody that listens to a D2 walks away impressed, but many people that hear the ODAC just don't like it. I understand that two different design goals can result in very different outcomes, but it seems like the two camps (Anedio and NwAvGuy) had at least somewhat similar goals and both measure well, but one gets a vastly differing set of responses. Anybody want to take a stab at explaining this to a hoi polloi fella like myself?

PS: I know the guy mentioned above caused some kind of ruckus in the past, so if my post causes any trouble or breaks any rules I'll gladly remove it. This just seems like an interesting question to me!
Thanks to all!!

 
A DAC that measures well in traditional metrics is actually not that hard to make. The Benchmark measures very well. The analog output stage has a lot to do with the sound you hear from a DAC as opposed to what the instruments pick up, and its what tends to separate the good from the bad.

 

[Viper2001]

Quote:

 
A DAC that measures well in traditional metrics is actually not that hard to make. The Benchmark measures very well. The analog output stage has a lot to do with the sound you hear from a DAC as opposed to what the instruments pick up, and its what tends to separate the good from the bad.

 
I've learnt over the years that certain things are not quite measureable in the traditional sense.  For example, I told Pat @ AR-T that there is a hint of glare on vocals for the Legato.  He acknowledged what I heard, and believed the cause was from the noise in AC.  I replaced the soft recovery diodes in the Legato with Cree SiCs, the same ones used in D2.  The glare is gone.  I know a lot of the AC noise we thought we had from the power line is actually generated from inside the piece of equipment, right at the silicon diodes.  Replace them with something with the least switching noise will improve sound.
 
James did a measurement of the Cree SiC and found that it did have lower switching noise.  This is why he adopted my recommendation on the Cree.  If you replaced the Cree SiCs with some cheap and noisy 1N4001 diodes, you'll hear pretty bad sound out of D2.

 

[Sapientiam]

Quote:

This is going to be a naive question I know, but where did he go wrong?

 
Claiming it was audibly transparent based on traditional measurements and comparison with the Benchmark only. No evidence has so far been offered that the Benchmark is audibly transparent and no correlation between transparency and measurements has been demonstrated other than appeal to Ethan Winer's authority.
 
Incidentally Thorsten Loesch has made (not so traditional) measurements of the ES9023 which don't show it in a very good light : http://www.diyhifi.org/forums/viewtopic.php?f=2&t=2229&start=30#p46715
 
....between 0dBFS and -60dBFS there is around 20dB of noise floor modulation.

 

[Sapientiam]

Quote:

 
A DAC that measures well in traditional metrics is actually not that hard to make. The Benchmark measures very well. The analog output stage has a lot to do with the sound you hear from a DAC as opposed to what the instruments pick up, and its what tends to separate the good from the bad.

 
I reckon - but have no measurements of my own to offer as evidence - that S-D DACs add stuff in that no degree of output stage magic can take out. Take a look at Thorsten's post I just linked to. A decent output stage can't polish a t*rd. ESS's S-Ds are better than most and they have measurements which show this, but its a bad bunch overall.

 

[Theogenes]

Thanks for all the responses. It seems like different designers use some pretty different measurements when trying to numerically evaluate performance. Is there any standard on what measurements matter most for DACs, or is it more a matter of the opinions of the designer? I know each engineer will obviously have their own leanings and thought processes, but is there any agreed-upon standard which pretty much all DAC designers would say is crucially important? Jitter is obviously a large part of the focus of this thread, and in many others I've seen, but I'm just curious if there is any general consensus on the matter. 
 
And thanks again for humoring noobie questions!!!

 

[veritas aurum]

Component selection certainly plays a large part in SQ. There are many new technologies available-Cree SIT diodes, Niobium caps, smd bulk foil resistors, etc. Then there is layout. particularly in digital devices. Others would cite power supply, output stage, low jitter, or other attributes. What Anedio has done (and very well) is the spend the money and put the expertise where it actually makes a difference in SQ. The D2 is not perfect, but it is a good balance of cost and performance, perhaps the best out there. and that is a function of design expertise and compromises where they do not affect SQ. Regards

 

[Sapientiam]

Quote:

Thanks for all the responses. It seems like different designers use some pretty different measurements when trying to numerically evaluate performance. Is there any standard on what measurements matter most for DACs, or is it more a matter of the opinions of the designer? I know each engineer will obviously have their own leanings and thought processes, but is there any agreed-upon standard which pretty much all DAC designers would say is crucially important? Jitter is obviously a large part of the focus of this thread, and in many others I've seen, but I'm just curious if there is any general consensus on the matter. 
 
And thanks again for humoring noobie questions!!!

 
Noobie questions are the most fun because most people are too scared to ask them 

 
Some people's opinions are that full-scale THD+N is the most important measurement for a DAC. The manufacturers headline this figure, but I reckon its the figure with the least relevance to SQ out there. A little closer to reality would be low-level distortion performance - say at -60dB, however TDA1543 posts an amazingly poor figure here while still sounding very acceptable.
 
My own view is that noise modulation is the closest measurement to perceived SQ, trouble is no-one can agree how to measure it. All noise figures are subject to some averaging, yet we need to know the instantaneous noise levels in the presence of music (or a music like signal like a high crest factor multitone signal)/. FFTs as a measurement tool don't cut it so I think its only when the whole industry has ended its love affair with FFT we'll get progress on this. Wavelets IMO are the way forward.
 
I had a mini-epiphany recently on jitter - I now reckon its important to S-D type DACs only. Multibit are much more immune as there's very little potential for noise modulation to occur when there's no out-of-band noise emanating from the DAC.

 

[DaveBSC]

Quote:

I reckon - but have no measurements of my own to offer as evidence - that S-D DACs add stuff in that no degree of output stage magic can take out. Take a look at Thorsten's post I just linked to. A decent output stage can't polish a t*rd. ESS's S-Ds are better than most and they have measurements which show this, but its a bad bunch overall.

I'm not sure that's true. MSB would argue something very similar, but I don't necessarily think that S-D can't be salvaged. The PCM1792 is possibly beyond saving, but I think there's the potential to make extremely good DACs out of the AD1955, AD1853, and AKM4399. The Accuphase DC-801 stacks eight AD1955s per channel, and I think you'll find that it'll put up a tough fight against anything using four or eight PCM1704U-Ks.

 

[Sapientiam]

Quote:

I'm not sure that's true. MSB would argue something very similar, but I don't necessarily think that S-D can't be salvaged. The PCM1792 is possibly beyond saving, but I think there's the potential to make extremely good DACs out of the AD1955, AD1853, and AKM4399. The Accuphase DC-801 stacks eight AD1955s per channel, and I think you'll find that it'll put up a tough fight against anything using four or eight PCM1704U-Ks.

 
MSB would argue something that will sell their femtoclock though won't they?

 I've played with the AD1955, yes its a good DAC but for dynamics its a non-starter compared to even the humble TDA1545. Hard to get the soundstaging/ambience retrieval of the AD1955 from multibit but I'm getting there. Having pored over the datasheet AD1955 at length I can't for the life of me see the virtue of stacking them - what rationale did Accuphase give? I'm not a fan of PCM1704 - not because its SQ sucks but because its very expensive and second it always seems to be on the verge of going out of production. Oh its also R2R rather than segmented current source, that counts against it given the phenomenal price.
 
<edit> AKM4399 looks interesting, I need to explore this one.
 
<pps> ESS's Martin Mallinson has already demonstrated some of the things wrong with ADI's S-D parts. Check out the video from RMAF last year. Of course he doesn't mention ADI by name.