or Connect
Head-Fi.org › Forums › Equipment Forums › Dedicated Source Components › December 2013 Mid-Level DAC Comparison
New Posts  All Forums:Forum Nav:

December 2013 Mid-Level DAC Comparison - Page 83  

post #1231 of 1331
Quote:
Originally Posted by yfei View Post


Benchmark Media's official statement regarding DAC2 was that when they designed DAC1, the analog stage is far superior than it's digital part, the dac chip was the bottleneck. DAC2's main change is the digital section, now it is on par with the analog section.

 

Bear in mind that being Benchmark, they'd be talking measurements as performance, not SQ. I haven't looked at the design of the DAC2's analog stage, but from a subjectivist design pov their DAC1 DAC's output stage sucked for sure.

post #1232 of 1331
Quote:
Originally Posted by Sapientiam View Post

Bear in mind that being Benchmark, they'd be talking measurements as performance, not SQ. I haven't looked at the design of the DAC2's analog stage, but from a subjectivist design pov their DAC1 DAC's output stage sucked for sure.

According to Benchmark the headphone amp is exactly the same in DAC2 as it was in DAC1.
I'm not sure if there were any changes regarding analog inputs and the other outputs.
post #1233 of 1331

I haven't looked at the headphone amp, I was only referring to the opamp stages immediately post- the DAC chip.

post #1234 of 1331
Quote:
Originally Posted by Barry S View Post
 


Thanks for the thoughtful response. To paraphrase Winston Churchill--USB is the worst type of audio data transfer except all the others that have been tried. :beyersmile: 


I think it's just a case of most drivers sucking badly. I remember one time I got a blue screen mentioning some abcdefg.sys driver - it belonged to a E-MU USB soundcard or similar - and I didn't even have it connected to the computer! So it makes you think why the heck that driver was "somehow active" even if I was using a whole another device. Lots of crap running under the hood. Uninstall everything you don't use.


Edited by hekeli - 1/8/14 at 11:52pm
post #1235 of 1331
Quote:
Originally Posted by Sapientiam View Post

I haven't looked at the headphone amp, I was only referring to the opamp stages immediately post- the DAC chip.

Interesting, care to share some more details on this?
post #1236 of 1331

Glad to - pic attached of the relevant parts.

 

Points of note - groundfill used as power ground but inadequate supply decoupling for the opamps (two small

MLCCs fairly distant from each dual opamp).

 

Using slow opamps (10MHz GBW) with almost zero RF filtering between them and the DAC chip.

 

I don't have access to the other loyers of the PCB but it wouldn't surprise me if signal gnd hadn't been kept separate from power gnd.

post #1237 of 1331
Quote:
Originally Posted by Sapientiam View Post

Glad to - pic attached of the relevant parts.



Points of note - groundfill used as power ground but inadequate supply decoupling for the opamps (two small
MLCCs fairly distant from each dual opamp).

Using slow opamps (10MHz GBW) with almost zero RF filtering between them and the DAC chip.

I don't have access to the other loyers of the PCB but it wouldn't surprise me if signal gnd hadn't been kept separate from power gnd.

I guess this is getting a bit OT and perhaps is more suitable for the DAC2 thread or sound science forum, apologies in advance, but still wink.gif

Interesting. Surely this can be measured if you have access to the relevant equipment and a DAC2, even without knowing the other layers of the PCB.
I must admit I'm a bit rusty on my electronics, but why would 10MHz be insufficient speed for the opamps in this scenario?
And what kind of RF filtering would we be looking at for this use case anyway? Assuming RF is indeed radio freq filtering, aimed at >MHz range, not kHz that analog audio usually operates at.
Given that one of their marketing points is 'Distributed Power Regulation' implying a complex power supply I would assume they took measures to ensure proper grounding where required.

I am probably a bit biased here since I bought one after comparing it to multiple others and to my untrained ears it sounds very good biggrin.gif

Anyway, if it trails way off topic, or it's posted elsewhere, send me a message? smily_headphones1.gif
post #1238 of 1331

The comments and pic refer to the DAC1. I have no idea what improvements were put into the DAC2.

 

10MHz is the 'gain bandwidth product' - so at a frequency of 10MHz the gain is 1. More to the point though the opamp gets overloaded with RF (very much higher than audio) signals in these kinds of applications but it doesn't show up in THD+N measurements.

post #1239 of 1331
Quote:
Originally Posted by Sapientiam View Post

The comments and pic refer to the DAC1. I have no idea what improvements were put into the DAC2.

10MHz is the 'gain bandwidth product' - so at a frequency of 10MHz the gain is 1. More to the point though the opamp gets overloaded with RF (very much higher than audio) signals in these kinds of applications but it doesn't show up in THD+N measurements.

Yes I am familiar with the GBP. What I was wondering about was why you feel 10MHz wasn't enough for this usecase?
post #1240 of 1331

Well a typical rule of thumb is that you'd like a factor of at least 100 between the top end of the audio band (the signals you're processing) and the gain bandwidth of the opamp. But what comes out of a DAC has a bandwidth easily exceeding 10MHz because it consists of very fast rise-time steps. Lynn Olson did some measurements with a spectrum analyser on what came out of a PCM63 chip - much older (and therefore slower) part than today's S-D chips. He found something like a 20MHz -3dB bandwidth. So to use a normal audio opamp like 5532, those steps should be greatly attenuated to approximate a normal audio band signal. Or alternatively use a much faster opamp which can handle the steps without filtering. The latter option I've seen used on at least one design.

post #1241 of 1331
Well yes, I would think pretty much any opamp requires some form for a LP filter? And what if the target gain was not 100?
With DSD capable DAC's I would think we're looking at outputs with way way faster rise times than what we're used to look at anyway?
So even a 20MHz GBP would fall short?
Surely they made sure the gain/LP filter spec accounted for this?
post #1242 of 1331

There is an LP filter of sorts - just not very sharp, being single pole - but its action is dependent on the opamp having some gain available. The stage in the pic is a transimpedance stage - current to voltage. Its using the loop gain to provide a low impedance 'virtual ground'. DSD dacs have risetimes similar to S-D DACs, no real difference there, very fast.

 

Yes 20MHz GBW falls very far short. I really have no idea what their thinking was - possibly that 'it measures fine, so its fine'. The DAC chip makers publish recommended circuits, this looks very close to what ADI suggest be used.


Edited by Sapientiam - 1/9/14 at 3:02am
post #1243 of 1331
Quote:
Originally Posted by Sapientiam View Post

There is an LP filter of sorts - just not very sharp, being single pole - but its action is dependent on the opamp having some gain available. The stage in the pic is a transimpedance stage - current to voltage. Its using the loop gain to provide a low impedance 'virtual ground'. DSD dacs have risetimes similar to S-D DACs, no real difference there, very fast.

Yes 20MHz GBW falls very far short. I really have no idea what their thinking was - possibly that 'it measures fine, so its fine'. The DAC chip makers publish recommended circuits, this looks very close to what ADI suggest be used.

I thought the 5532 was just 10MHz, and the "fix" was to use a faster one with 20MHz?
I'll see if I can find the relevant circuitry in the DAC2 when I get home.
post #1244 of 1331

No - my 'fix' (which wouldn't actually fix the sound, hence the scare quotes) would be to install a passive (LC) filter before the opamps. Then a 10MHz part would do fine. However ADI explicitly warn against using such an approach (passive I/V) in their DS for the AD1955.

post #1245 of 1331

Guys, please stay on topic.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Dedicated Source Components
This thread is locked  
Head-Fi.org › Forums › Equipment Forums › Dedicated Source Components › December 2013 Mid-Level DAC Comparison