[doktorsteve]

I agree the effects are probably minor. I like to think of it as being able to see the bottom line of the eye chart rather than the line above when I use the LM4562. At my age I should be happy to see the line above so the extra acuity is quite seductive! Unfortunately the price is a touch of glare and I switched back to the JRC2068 this morning to calm things down.
 To my ears there is not much difference between the JRC2068 and the NE5532.  I am pretty happy to use the JRC5536 with a gain of 2 as the preamp for my speaker system so I might try that too. Its low bandwidth could be an advantage.
At the other extreme I am curious to hear how the O2 sounds without the gain amplifier. Hopefully the ODAC can drive the paralleled load resistor and volume pot and there is enough volume from my 32R phones.

 

[doktorsteve]

On the basis of listening to the first side of Ziggy Stardust  3 times I declare the NE5532 is my favourite (for today). It is easy to overlook this device as it is not so immediately impressive but grows on you over a period of time.

 

[frenchbat]

Currawong, I can bring the JRC2068 I'm not using to the heaphone festival in 2 weeks. I'm not using it right now. Just drop me a PM.

I've tried whatever OPAMPs I've had that would fit (some are singles and a Browndog DIP adaptor is too big). The feel the differences between them are very minor. My attitude is that I'll just leave whatever I have spare in there, though I'll probably drop the NE5532 back in it if I decide to review any headphones for consistency (I can't get the JRC2068 from Mouser here and haven't gotten around to sourcing it locally). At some point I will grab an ODAC out of curiosity, but also to have a basic reference system that a large number of people are famliar with when I write impressions of headphones.

 

[doktorsteve]

So I am now making progress. I changed the resistors in the post filter of the ODAC from 220R to 1k (not straightforward when the originals are 1206 and I only have 0805 - am I on the wrong forum?). I think this reduces the -3dB point from 100kHz to 50kHz (I'm guessing as I don't know the resistance built into the DAC chip) and gives a 1 dB droop at 20kHz. As my hearing takes a dive at 12kHz this is not going to bother me 1 iota. Well the difference is huge. The sound is smoother and more detailed with the NE5532. I have yet to try the  other opamps but I am guessing that the differences between them will not be as great because they are not going to get so much out of band junk. I'm currently enjoying Jessie J for the first time since I was given the CD  for Christmas.

 

[Battou62]

May I ask the point of changing the ODAC maker's meticulous designs?

 

[adydula]

Battou62....
 
They do it because they can!! 
 
Boys being boys.....its interesting to watch and see what they come up with.
 
I myself just use is the way it is until I can really hear some "imporvements' that matter to my ears!
 
all the best
Alex

 

[doktorsteve]

You ask why I modded the ODAC? Because I had paid for it and I did not like it. Just like the guy on page 1 of this thread. Luckily I have the means to change some parameters.
 
Maybe you should be asking RS why he left the bandwidth so wide. I don't think that reducing the bandwidth increases distortion. In fact the lighter load on the output amp of the DAC may even reduce it. Ok so the gain at 20kHz falls by 0.5dB. Big deal. And the phase change due to the filter? Apparently there is a big phase change through the DAC anyway. 
 
Clearly there is a compromise because the 1k output resistance increases noise (how much?) and makes the ODAC more sensitive to cable loading ( really 1m cable is not much capacitance compared to 2200pF) and the output is not blue book level. RS already added the series resistor and changed the cap from the reference design so I just took it a bit further. 
 
Returning to the O2 I changed the 5532 to the 4562 this morning. The clarity is back but not the offensive edge. I expected the sound to stay the same as for the 5532 as I had made the changes to the ODAC but the increase in clarity remains. I have been listening for 3 hours, tapping my feet and enjoying every minute - it just sounds right now. If that is just expectation bias then I am dead impressed.
 
I don't think I am deriding RS's objectivist philosophy - I would have thought that minimising the bandwidth of the ODAC output signal would make perfect sense in order to avoid high frequency signals outside the bandwidth of the  4556s. Without the test gear this is just conjecture.

 

[doktorsteve]

I watched this video about the Sabre DAC. Watch it and you might want to reconsider your faith in objective testing.
 
http://www.youtube.com/watch?feature=player_embedded&v=1CkyrDIGzOE

 

[zilch0md]

Excellent lecture!  Thanks for that link!

 

[willmax]

 

[stv014]

Quote:

I watched this video about the Sabre DAC. Watch it and you might want to reconsider your faith in objective testing.

 
To be honest, it sounded mostly like marketing material for the ESS DACs, rather than really explaining and proving that delta-sigma DACs are flawed. Even if the NPSS noise is true, intermittent noise at a -100 dB or lower peak levels is not audible (I tried a test where complex music is upsampled to 96 kHz sample rate from 44.1 kHz, and the audio information is shifted to the frequency range of 25950 to 48000 Hz, leaving only silence in the audible range. I played this back at a volume that would be too loud for normal listening, but heard nothing on my non-ESS DAC; if there really was a noise issue with the DAC, I think I should have heard something). Also, on the "non-linear excess phase" picture, it is normal that the time between the edges increases near the maximum and minimum levels (this happens with any dithering of a full range signal to 1 bit; multi-bit delta-sigma DACs - which are the most common currently - allow for better dithering that makes the quantization noise uncorrelated to the input signal, and avoid the problem); the HyperStream graphs basically "cheat" by running at a higher clock frequency.

 

[stv014]

Quote:

Originally Posted by doktorsteve /img/forum/go_quote.gif
 
Maybe you should be asking RS why he left the bandwidth so wide.

 
I assume he wanted a flat (+/- 0.1 dB) frequency response up to 19 kHz ? The maximum variation of 0.1 dB in the frequency response is mentioned in his criteria for transparent audio reproduction.
 
Quote:

Originally Posted by doktorsteve /img/forum/go_quote.gif
 
And the phase change due to the filter? Apparently there is a big phase change through the DAC anyway.

 
If the DAC uses a linear phase filter (I think it does), then it does not necessarily change the phase much. Nevertheless, the phase shift in the filter is not an issue if it is the same on both channels, the roll-off will become audible much sooner.

 

[doktorsteve]

Did I miss a train wreck somewhere?
When did we go off topic?

 

[stv014]

Quote:

Originally Posted by adydula /img/forum/go_quote.gif
 
They do it because they can!! 
 
Boys being boys.....its interesting to watch and see what they come up with.
 
I myself just use is the way it is until I can really hear some "imporvements' that matter to my ears!

 
The problem with most DIY "modding" or "tweaking" of audio equipment is that it usually relies on sighted comparisons against the previous/unmodified state of the device, which are also affected by the limitations of auditory memory (only in rare cases do the modders buy two identical devices and perform a level matched double-blind ABX test between the original and the modified version). Objective tests with audio analyzers and oscilloscopes, null tests, etc. are also rarely performed, often even RMAA is not used. It is not surprising that major improvements can be heard after "op amp rolling" or other changes, even if the sound really remains the same, or even becomes slightly worse.

 

[K_19]

For anyone with ODAC with both RCA and standard line out:
 
Is it okay to have both of those connections connected at the same time or not? I ask as I only have the ODAC with the standard 3.5 line out and I could really use a multiple output for it right now...