Measuring XXHighEnd ...

[PeterSt]

Mani,

Right after Q1 was introduced a few of us were seriously testing it and we were shouting to eachother to notch down after starting at the default of 14 (IIRC this was merely off line). At 4 things started to happen, and at 1 we were shouting things got wild and crazy. This was all in a time span of a week or so (slowly getting the grasp), and while at first being under 0 we were all impressed, we later all came back on that. Now, some 2 years further down the line, try violins. They don't work at these low levels. Things seem more detailed, but violins sound digital. Now :

I have always known (and told about it when it was necessary) that the values below 0 are special because they imply some randomness. And although I was talking into the blue to some extend, I knew about that randomness, that by itself creating a resonance. As you can see in the pictures where Q1=-4 is involved, you can clearly see it does something with a pattern of 3/100 of a second. How that sounds ? I guess more detail is perceived but in the end it sounds like more digital. Watch the violins.

Peter

[SeVeReD]

Mani,

Right after Q1 was introduced a few of us were seriously testing it and we were shouting to eachother to notch down after starting at the default of 14 (IIRC this was merely off line). At 4 things started to happen, and at 1 we were shouting things got wild and crazy. This was all in a time span of a week or so (slowly getting the grasp), and while at first being under 0 we were all impressed, we later all came back on that. Now, some 2 years further down the line, try violins. They don't work at these low levels. Things seem more detailed, but violins sound digital. Now :

Peter

Just wanted to quickly say I agree with you above Peter.  I got Q1 down to -2 for a long while and liked that... then, moved it up to 0 and felt it gave it a not so see through crystal clear view as -2, but did add some weight/meat to the 'notes'.  Then Pedal posted about 4 and I moved it up there and liked the balance I heard there and have stayed there.  I've listened to 14 for awhile and to me, it is 'woolly'/fuzzy  compared to  4,(don't take those terms to too much extreme, just in comparison).  14 not as see through clear as 4.  I HAVE NOT listened  to settings inbetween 4 and 14, but we should no doubt.

Also, as my sig says, I haven't been able to stay at Q2-0 Q3-0 and have moved them both up to 26 - 30.  Keeping them at 0 sounds like low Q1 settings compared to 26-30.  Moving up to Q2Q3-26 starts putting meat on the notes.  26 compared to 30 =     26 makes it sound more like lower q1 settings and 30 makes it sound more like heading toward 14 ... i thinks
Also, moving from Q2Q3-0 to Q2Q3-26-30 changes phase.  At Q2q3-0 I like phase ~ (normal?) and at Q2Q3-26-30 I like phase I (invert).  Even so with 26-30 there are slight phase changes going on (also phase preference is also recording dependent, but most recordings sound correct at those above basic phase settings given those certain Q settings).

[manisandher]

I'm sure you're both right.

In all honesty, I've never really explored any of the Q settings. I started with Q1=0 (I think) and then tried -4, liked it and stuck with it. Q=-4 seems to bring my FF800 (used as a DAC, not an spdif pass-through) more to life. But I will explore all the Q sliders further.

My comment about Q=-4 being the gold standard was a bit 'tongue-in-cheek' - it seems to provide the biggest difference with Foobar WASAPI... ergo must be the best setting. But of course, this is flawed logic

I'll give some violins a play and try to hear the difference for myself.

Really interesting stuff Peter...

Mani.

[Telstar]

My comment about Q=-4 being the gold standard was a bit 'tongue-in-cheek' - it seems to provide the biggest difference with Foobar WASAPI... ergo must be the best setting. But of course, this is flawed logic

Wait, wait, wait.

I remarked that Q1=4 is indeed the gold standard because that's the setting that i like the most, not the one most different from foobar. That would be just silly.

I experimented with several settings. Anything under 0 was plain cr*p (=digital)
Tried 5, 6, 8, 10, 14 and the max. Above 14 things went too fussy. Between 4 and 6,8,10 I ended to prefer 4. Between 4 and 14, it is more a matter of taste. with some recording 14 (more smoothing) can be more pleasant, but I believe to be less true to the original.

Now, my comparisons with Foobar, which started about one year ago are interesting too.
I told Peter that I felt xxhe to be a BIG improvement over it only from version v (or w now i dont remember).
So they come after, choosing the Q1 and the Q1 has not been changed after that. I used 4 since the first time Peter said that it was his favourite. I tested again with x1 and my preference was still Q1=4. In that occasion i briefly tested Q2 and Q3, but didnt like them particularly.

I have to say that I didnt do much tests with those, and after the comments on those, I will do more evaluations.

[Telstar]

(1) 1st trace 4-0-0-0-0 Unattended compared to Foobar WASAPI;
(2) 2nd trace -4-0-0-0-0 Unattanded compared to Foobar WASAPI;
(3) 3rd trace 4-0-0-0-0 Unattended compared to -4-0-0-0-0 Unattended.

???
I dont understand. You must have done a typo with some attended/unattended in the list above.

[PeterSt]

Telstar, no ... but you may be missing the -4 vs. 4 ?

[PeterSt]

Moving up to Q2Q3-26 starts putting meat on the notes.

Dave, FYI : As you may recall I deliberately set Q2/Q3 back to 0 to get the real merits from the stopped services. I kept that up for quite a while. However, the other day at testing with these analysis, Q2/Q3 by accident got stuck on 30/30, and it was only the next day that I started to recognize just what you said in the quote. Yesterday I payed attention to it explicitly, and it just seems to be so.
Even "meat on piano notes" would be a good one to state. Like I earlier said that a piano would get less dry from it, I now hear the same for bass. It's more spatious, and to me this comes as a good thing.

[AUDIODIDAKT]

and Peter sets his specs back at Q1/2/3/4/5 = 4/30/30/0/0 (must be that left ear) haha

Whatever I tried I liked Q1/2/3/4/5 = 4/30/30/0/0 the most !
0/0/0/0 is simply to much dynamics for me, like it for 10 minutes then always go back to 30/30/0/0.
But the dynamics are amazing!

roy

[Telstar]

Telstar, no ... but you may be missing the -4 vs. 4 ?

Yes, I did.
I have one more comparison request then

+4 attended vs +4 unattended

[PeterSt]

http://www.phasure.com/index.php?topic=692.msg6125#msg6125 

[Telstar]

http://www.phasure.com/index.php?topic=692.msg6125#msg6125 

ok, i'm blind :D

[leifchristensen]

i think som weird "echo/decay"  appears when I move above 20 on q2&3
sound gets a little unstable
not bothered by too much dynamics at any setting
best
Leif

[SeVeReD]

i think som weird "echo/decay"  appears when I move above 20 on q2&3
sound gets a little unstable
not bothered by too much dynamics at any setting
best
Leif

Could be,,, this is kinda what I might describe what I hear going from 26 to 30 with Q2Q3

Ok, let me throw this out so Jeffc doesn't feel all alone, (btw, I haven't listened to bluray ripped vs non yet, I will try sometime...)
Could you measure this for me?
When an album doesn't have a back cover, XXHE mirrors the front cover and pulls the song info and tracks/highlights which song is playing in a generated list displayed on the mirrored cover.
I would like to see measurements of this situation above vs the old way of just showing the front cover and info in bottom and top margins ()without a generated mirrored cover song list.
did that make sense?

[PeterSt]

Well, I now have accomplished the most difficult task of it all : compare the DACs output with the original digital data. Took me two weeks only, and although not directly to the benefit of XXHighEnd, I hope we audiofreaks will benefit in general. Ah, no, I'm sure you will, because I'm also 100% sure we're al hoaxed by official measurements. 0.00002% THD blahblah ... prrrrt.

Ok, my DAC is Non Oversampling, so it will produce 40% THD (or whatever measurements try to fool me with), but one thing : NOS will be the most representative for what's in the WAV file, and what that brings you can see below. And you could say that already brings nothing much.

The red line is where the digital data is, and the black lines show where the DAC plays (or where it's off). Top is the left channel, bottom the right.

Btw, if you hear a kind of p*ssed voice here, yeah, that's a kind of how I feel. This all s*cks so much all over that it seems a better idea to make my reel to reel noise free, somehow. On the other hand, it shows a. the truth of digital, and b. the enormous amounts of improvement to achieve.


Ok, what do we actually see ?

The white small graph around the red line is the representative of the transients. Transients are the more or less steep changes in volume, like a smash on the rim of a snare drum will create a high transient. The second picture shows the transients better, because they are higher there. It may take looking at your monitor from more below or above to clearly see them (because of the white on grey).

These pictures weren't taken to show explicitly what I saw : there is a relation between the direction the transient goes and the direction the DAC is off. When the transient shows below the red line, the "distortion" (the black line) will be above the red line. Look at the first lower dip of the black line (both channels) in the first picture, and see that the transients are at the opposite of the red line.
What functionally happens is that the transient goes to the minus (volt) direction, but the output wave by the DAC stays behind and is more posititve than the digital file. Remember, the digital file is at the red line, so when the black line shows above it, its voltage is too high.
The other way around, when the black line is below the red line, its voltage is too low.

In general you can say that there is no - no - no - way the DAC can follow anything of what it is dictated.
And remember, this is unlike the earlier graphs which showed differences between settings and players ... this is just the absolute reference compared with the digital file.


Now let's look in more detail to the first picture. At the second time the black line (both channels) come above the red line, you can see that the transients are not in one direction only (like more and more and more positive amplitude), but in both directions (high plus, high minus, high plus, high minus) which might happen with a fast frequency like from a snare drum. Now look at the transients from left and right, which are just about equal. However, the "distortion" is not equal ! The left channel gets inconcistent, and the one few samples it is more plus than an imagineable average line, and the next few samples it is more minus. And oh, that imagineable line shows at the right channel (bottom).
The bottom picture shows this even more clear, and I can tell you this is throughout; The left channel of my DAC is even more incorrect than the right channel. Something is just wrong, and it shows clearly. Note though this is at the "unmeasurable" (hahaha) uV micro level.


Allright, this is just an example of what can be done now, and as you can imagine it is now possible to find the anomalies of players or settings in an absolute sense. However, thinking twice you can also see that it is not said that it is more easy this way, and this is just because the DAC itself is so much off in the first place.

On a side note, think about this : when the DAC would better follow those transients, various other things would go wrong, like bursting speaker driver diaphragms when the amp can follow in the first place. So, it is clear to me this can be much much better, but to a certain extend only. When things become real good, other parts of the chain will collapse, I'm sure.
This is also the reason why hires material really doesn't make it better. The inividual digital steps will be smaller, but the general transients will stay, and the DAC won't be able to follow anyway. Not by a mile !

Oh, obviously this is from my "super duper" NOS1 Phasure DAC, and I can tell you that a new version of it is coming up. This time it will have a kind of infinitly more "fast" analogue stage. Yep, that's the attention point now : very fast analogue parts. And we will see what that brings ...
(it has been a kind of quiet around the NOS1, right ? ... not for long anymore).

Peter


PS: Before one starts to compare with the earlier pictures ... the horizontal (time) scale is equal to some of the earlier pictures, but the amplitudes you see in the black lines have been devided by 10 here !! (or otherwise the graph would be outside of the picture all over ... so bad is it).

PPS: I must put forward a very small disclaimer : The ADC (Fireface800 in my case) is involved too of course. But I think this can follow everything, just like the digital data and the high transients in the WAV file at some stage came from an ADC. This is just measuring voltage from one sample to the other, and making that a digital number. I don't think much can go wrong with that, as long as the "measuring device" in there can measure fast enough for the time it has for the one sample; I don't think a greater (high transient) or lower (low transient) voltage difference between samples will influence the value the measuring device presents. But maybe it does, and if so we're looking at the ADC for a more or less part, and all doesn't say much.

[manisandher]

... if so we're looking at the ADC for a more or less part, and all doesn't say much.

My own experience and everything I've learned from reading about that of other people suggests to me that AD conversion is much more difficult to get right than DA.

I really hope the RME ADC isn't affecting the results. If so, what you're doing is amazing...

Mani