I'm just not finding that big a difference in A1, A2, A3.
Yes, A3 is about 0.5 dB below the other 2, but the spectrum is remarkable the same, as far as I can tell.
Below is a plot of all three spectrums with 12th octave smoothing. They look pretty much the same with no smoothing, just harder to see. No odd spikes or anything. I've boosted A3 by 0.5dB to line it up.
Is this wrong? The files look very similar to me. The only noticeable differences are in A2 and some small subsonic level differences on all 3. That's below 7Hz
Yes, A3 is about 0.5 dB below the other 2, but the spectrum is remarkable the same, as far as I can tell.
Below is a plot of all three spectrums with 12th octave smoothing. They look pretty much the same with no smoothing, just harder to see. No odd spikes or anything. I've boosted A3 by 0.5dB to line it up.
Is this wrong? The files look very similar to me. The only noticeable differences are in A2 and some small subsonic level differences on all 3. That's below 7Hz
Attachments
Do you mean you listened directly to the analog output (without ADC and compression)? That is the difference (as I can hear so clearly), what can I say? As usual: "pick your poison".
But nothing so far explained the mechanism involved unfortunately. And my listening was affected by my computer DAC or Wolffson1xxx hand phone DAC which may have Sigma-Delta.
Next time you want to listen and decide whether you want the mod or not, try drum solo (or at least Dire Strait).
Catch up time, beem distracted by personal/medical/old lady stuff.
Back alive. For the life of me, I could not get the LM4562 to play nice..
Feel free to ask any questions. Just don't ask me to use the LM4562 model in LTSpice.
Yes, that is strange as LM4562 does work in real life, indeed very similar is used in the Oppo 105/D. So I understand you going over to LT1115 and that tallies up OK.
Indeed - just don't assume it will be OK. But with those values, in the real world, so-far-so-good.
Last edited:
Joe,
My issues with the LM4562 stemmed from the differences in how PSPICE and LTSpice manage the model. I just couldn't get good convergence, not that the opamp in its physical self is broken.
Also, the compensation scheme as given in the ES9018 datasheet schematic starts getting a good amount of gain peaking with any opamp that has a dominant pole much below the 50-60 MHz of either the LT1115 or the LM4562. It's definitely something to keep in mind.
My issues with the LM4562 stemmed from the differences in how PSPICE and LTSpice manage the model. I just couldn't get good convergence, not that the opamp in its physical self is broken.
Also, the compensation scheme as given in the ES9018 datasheet schematic starts getting a good amount of gain peaking with any opamp that has a dominant pole much below the 50-60 MHz of either the LT1115 or the LM4562. It's definitely something to keep in mind.
Thanks for that - your point about A3/B3 is well taken - I will be honest, it's the one I would like you to pick, but are you using a delta-sigma DAC.
Clearly the comparison should be between A1/B1 and A3/B3.
I would ignore A2/B2 - that is my fault.
The filter's impedance is too low.
Mooly made me aware of that - and indeed I have gone back and altered the 'Scenario 2' in Post #1 and posting similar values to his.
But, the question is that Mooly's files were created via ADC after the post-DAC filter has been applied, but now adding another delta-sigma DAC can be seen as undoing the potential benefit. Like one giving and the other giveth away. In this instance it is not a level playing field.
I am not question A1/B1 and A3/B3 as files - there is no reason that the ADC is not doing a capable job. So checking A1/B1 versus A3/B3 should be done with a DAC that does not need the 'fix' for delta-sigma DACs (as this is what we are testing for, whether there is or is not).
I don't have a multi-bit DAC here and those files can only be tested that way. I think my opinion is the correct one because it is the logical one.
But at least we are on the right track, thanks to Mooly, we are listening - anybody with a really good non d-s DAC out there to do the listening test?
Is this averaged out measurement.
I just want to repeat, ignore A2/B2 - my bad. But if a difference @ 1KHz and flat response, then 0.5dB is unacceptable - we know which file will be favoured. But an average out up to 20KHz?
Chris, you are still the best man for the job, burn those files A1/A2 and A3/B3 to a CD-R?
That should work.
OK, need to get these files under some control, according to Mooly.
Genesis:
> File A1 Genesis is un-modified player.
Good.
> File A2 Genesis is modified player.
Yes, but filter too low in impedance.
> File A3 Genesis is player modified to 'final spec'.
OK. This is the one we should have been listening (or chosen) for.
Ignore A2.
Now for Natalie:
> File B1 Natalie C is modified player.
Modified how? Low impedance - no good. Again, my apologies.
> File B2 Natalie C is un-modified player.
OK, unambiguous.
> File B3 Natalie C is player modified to 'final spec'.
OK. This is the one we should have been listening (or chosen) for.
It should have been a comparison between A1 and A3, where A1 is standard and A3 is the correct filter implementation
The other should have been between B2 and B3, where B2 is the standard and B3 is is the correct filter implementation.
Genesis:
> File A1 Genesis is un-modified player.
Good.
> File A2 Genesis is modified player.
Yes, but filter too low in impedance.
> File A3 Genesis is player modified to 'final spec'.
OK. This is the one we should have been listening (or chosen) for.
Ignore A2.
Now for Natalie:
> File B1 Natalie C is modified player.
Modified how? Low impedance - no good. Again, my apologies.
> File B2 Natalie C is un-modified player.
OK, unambiguous.
> File B3 Natalie C is player modified to 'final spec'.
OK. This is the one we should have been listening (or chosen) for.
It should have been a comparison between A1 and A3, where A1 is standard and A3 is the correct filter implementation
The other should have been between B2 and B3, where B2 is the standard and B3 is is the correct filter implementation.
While that is promising and I suppose I should be happy about it, truth is that I am not convinced this is the best way - but thankful for your efforts anyway.
Now you say "deeper" is interesting, because that has definitely been commented on before, seeper and more dimensional, so why not play it directly on your system before it gets to another ADC and see if that aspect gets heightened even more. I can't hurt to listen directly to what the filter done, since if we are going to come to a final conclusion that it actually works, then that is how you would end up listening to it always - none of us ever listens to something that has gone through two DACs, do we? Testing, maybe, but after that?
Makes sense?
Last edited:
Seems to me that two things would be desirable to do:
1) Chris 'anatech' burn files to CD-R and listen, that way his system will not be disturbed, indeed that is his reference point.
2) Mooly actually connecting up his 'final spec' DAC directly to his system and form an opinion on that.
But I would like to thank all for their efforts - at least we are listening for something.
OK, up-to-date. Now I have to wait to you guys wake up.
1) Chris 'anatech' burn files to CD-R and listen, that way his system will not be disturbed, indeed that is his reference point.
2) Mooly actually connecting up his 'final spec' DAC directly to his system and form an opinion on that.
But I would like to thank all for their efforts - at least we are listening for something.
OK, up-to-date. Now I have to wait to you guys wake up.
Previous trials (nothing to do with this) have used Bach, Veljo Tormis, Fairground Attraction, the list goes on... and all drew some kind of criticism.
What would be your choice ?
Yes, I appreciate you can't please everybody.
Not wanting to gild the lily, but what if you were asked to be part of a DBT and hated the music?
Maybe on dangerous ground there, so better not say too much.
- Status
- Not open for further replies.