What makes a DAC better than another

Quote:

Originally Posted by XTTX 

Aside from quality discrepancies, what characteristics make one DAC IC better than another? Is it the resolution, sampling rate, slew rate, accuracy, noise/distortion? For the first two, I wouldn't think that there is much distinction due to the limitations of human hearing, but what characteristics really distinguish one DAC from another?

Better is a loaded term, but here is my take on it.

The job that a DAC has to do is to accurately reconstruct a continuous waveform (changing voltage) from discrete samples. If a DAC can do that with negligible added noise/distortion and with no added phase difference between channels (perceived as wider soundstage)  it is a good DAC whether it costs $50 or $25,000 - if it cannot do that it is flawed. It it adds it's own character to the sound it is flawed, if it rolls off before 20K it is flawed, basically if it is not razor flat from 20 - 20K it is flawed, IMHO of course. If two DACs are very different, to the point of audibility then at least one of them is faulty, IMO.

Of course rigorous testing of audible differences is rarely performed here, mostly we have (IMO) unhelpful, poorly controlled sighted tests which are open to all forms of cognitive bias. 

The biggest single difference you will ever find betwen two "good" (i.e flat) DACs is volume pure and simple, the line outs on different DACs can vary by several db , my own Entech is a good 0.7db hotter than my Marantz, this is a difference easily detectable and often mistaken as greater dynamics and so on.

Once you adjust for level many times the differences between good DACs become very very hard to detect.

As for audibility of other differences between good DACs, well normally unless you are talking about NOS jobbies then distortion/noise are so low as to be undetectable in normal listening, jitter has never been sufficiently proven by any rigorous testing to be an audible issue, so you can forget that.

What is left ?

One word.

"HYPE"

Well they do sound different from one another, some better some worse, its all about synergy.

But generally speaking the hype surrounding a product can contribute to its perceived performance.

Quote:

Originally Posted by XTTX 

In that case, why the need for external DACs, sound cards (aside from multi-channel capabilities), etc.

External DACs have better power supplies, are in a quieter environment (noisy rfi/emf) and only do one thing. But I am interested to hear what the big guns have to say to your question.

The shorter the signal path and the less active components in the audio path, the less colored it'll sound. The best DAC's are not complicated...they only use few good components off a good PSU.

Ideally it should be totally transparent. You shouldn't know it's there.

Test this by plugging the output back into an input. (i.e send the signal round the 'analogue hole'. You can use S/PDIF outs and ins as well). Can you tell which is which? How many iterations does it take before it becomes obvious?

This is easy to do with a pro-am grade multi channel audio interface. If you have a single stereo channel audiophile grade device then you will need to record the original from the output on something like Audacity then play it. Again how many times can you do this before you notice?

Other things to look for on a superior product are ongoing support, particularly as regards drivers, reliability and feature set. That's basically what you pay for in the modern world. RME are particularly strong in this regard. As far as I know they are so far the only company with the resources and dedication to have done a deal with M$ so that their software analyses  your PC configuration and then optimises it's installation automatically. If anyone knows otherwise i's be pleased to hear from them.

I bought my first outboard audio interface (M-Audio Fast Track Pro) on the basis of HappyCamper's rationalisation above. It seemed to make sense isolating the PC and Interface from one another.

On a personal note having this proved to be a revelation. I hadn't fully appreciated the power and utility you gain when you get access to Windows audio multi channel features. So much so that the FTP was very swiftly replaced by the 10 in/14 out all @ up to 24/192 MOTU Ultralite. You also get 7 band parametric EQ, 2 compressors, Reverb  and up to 8 independent mixers. 2 mic/instrument inputs with preamps. It operates in stand alone mode (not connected to a PC)  and the full feature set is available using the front panel. Portable too. All this for substantially less money than equivalent hi-fi grade gear.

Quote:

Originally Posted by nick_charles 

 

 

Better is a loaded term, but here is my take on it.

. It it adds it's own character to the sound it is flawed, if it rolls off before 20K it is flawed, basically if it is not razor flat from 20 - 20K it is flawed, IMHO of course.

 

Every heard of phase distortion?   If you see a DAC with a razor flat frequency response odds are good that it has a lot of phase distortion in the treble.    Same goes for speakers,  typically speaker designers don't shoot for a flat frequency response (ie +- 0.1dB) because doing so would add so much phase distortion as to prove unlistenable.  A DAC has a crossover in it just like speakers.  So your opinion is correct such a DAC is flawed as is every DAC ever designed,  its all a delicate balance of compromises to give good sound.   This is why smart designers don't bother publishing meaniless specs,  'cause untrained buying public don't understand how to interpret the results.
 

How about the DC offset at the outputs? And whether there are capacitors in the output path? Should these be things to consider when deciding which is "better"?

Quote:

Originally Posted by aspenx 

How about the DC offset at the outputs? And whether there are capacitors in the output path? Should these be things to consider when deciding which is "better"?

There shouldn't be any reputable DAC's on the market with DC offset.  I wouldn't rule out a DAC just cause it has coupling caps.  Some of the best have them.   Every DAC is a compromise,  you just have to pick your poison. 

Originally Posted by aspenx 

How about the DC offset at the outputs? And whether there are capacitors in the output path? Should these be things to consider when deciding which is "better"?

DC decoupling caps are fine. Some say that they kill bass...My Firestone Spitfire DAC has them and its DC offset is null. That's its FR off its Supplier DPS, I can't see any bass roll off: http://www.firestone-audio.com/rmaa/SPITFIRE.htm

All roads lead to Rome, smart designers can overcome any problem...some just seem to design them to go for the specs and don't even listen to them, like that Benchmark DAC1. It's exclusively using LM4562 and NE5532 opamps, this is not acceptable for such an expensive unit.

Quote:

Originally Posted by regal 

There shouldn't be any reputable DAC's on the market with DC offset.  I wouldn't rule out a DAC just cause it has coupling caps.  Some of the best have them.   Every DAC is a compromise,  you just have to pick your poison. 

Ouch. That either makes mine a not very reputable DAC or my DMM a lousy piece of hardware. I can't be sure of either now...

 

Quote:

Originally Posted by leeperry 

 

DC decoupling caps are fine. Some say that they kill bass...My Firestone Spitfire DAC has them and its DC offset is null. That's its FR off its Supplier DPS, I can't see any bass roll off: http://www.firestone-audio.com/rmaa/SPITFIRE.htm

 

All roads lead to Rome, smart designers can overcome any problem...some just seem to design them to go for the specs and don't even listen to them, like that Benchmark DAC1. It's exclusively using LM4562 and NE5532 opamps, this is not acceptable for such an expensive unit.

This leads me to something else that is baffling to me as a newbie... Is having the DC decoupling caps at the DAC's output the same as having them in the input signal path of the amplifier? (Since all roads lead to Rome?)

Quote:

Originally Posted by aspenx 

Ouch. That either makes mine a not very reputable DAC or my DMM a lousy piece of hardware. I can't be sure of either now...

 

 

This leads me to something else that is baffling to me as a newbie... Is having the DC decoupling caps at the DAC's output the same as having them in the input signal path of the amplifier? (Since all roads lead to Rome?)

Are you using a fluke?  If you are using a $20 special don't put to much weight on its accuracy, a few mV's of course is not an issue unless your amp has a huge gain.

Yes coupling caps on eith place means- not DC coupled,  same difference.

 

Quote:

Originally Posted by regal 

Are you using a fluke?  If you are using a $20 special don't put to much weight on its accuracy, a few mV's of course is not an issue unless your amp has a huge gain.

 

Yes coupling caps on eith place means- not DC coupled,  same difference.

 

I used Fluke back in school, but the current DMM I got doesn't even have a brand and it's about 12 local dollars, approximately US$9. Oops.

The DMM does show that the final output of my current setup to the headphones is well below tangent's threshold of +-20mV. I set my opamp for unit gain most of the time but sometimes push it to 3. Thanks for the reminder, my headphones have a sensitivity of 100dB/mW and the impedance is only 42ohms so I should look out for it. I do have a problem with predicting the DC offset though. I used the calculator on tangent's site and predicted a very low DC offset but ended up with insane (~100mV) values with 2x LT1028ACN8s... I DIY-ed the adapter so there might be a problem there, but I don't know.

Dang, that was all off-topic. So, from what I understand from you guys' helpful posts, the DC offset at the output of the DAC doesn't matter at all, because it can be filtered and solved before the amplification stage anyways? And whether it happens in the DAC or amp doesn't matter either? That really puts a lot of DACs on the same plane doesn't it?