Normal Phase & Inverse Phase And PA +/-

[PeterSt]

Hi Peter, I am not sure but I think not. Anyway, how I can check that?

Look at the date/time of both; they should be fairly close to eachother, and neither of them should imply the date that 0.9z-7-3 (or earlier) was posted. If you by now also downoaded the newer XXEngine3.exe from the 0.9z-7-4 version then of course that one is newer.
But anyway, if you recognize that you put a new version nicely in its own folder, this can't have gone wrong.

I am using no usb or SPDIF because I am stealing the signal directly from balanced soundcard DAC chip legs (through  coupling caps) to a tube analog circuit that works as buffer and converts the balanced signal in SE.

Anything can go "strange" here (especially at converting from balanced to SE) but I tend to focus on that Crack Detect thing ...

I could try with  a Cambridge DacMagic via usb or SPDIF but I think it works only 16 bits 48 Khz via USB and 24 bits 96 Khz via SPDIF

Please do. But careful, because when that is a "DAC Needs" = 24 device, for sure PA won't work (it is just not supported at this time).
16/48 will never work with PA.

Peter

[phantomax]

Hello Peter,
sorry for the delay but I've been quite busy and I couldn't try the Cambridge DAC yet. But I'd like to tell you a strange thing: except with Iphase off an PA -, that play without problems, when I stop the music with the other options the rigth channel replays the buffer and ends with the familiar plop, the left channel not.
I don't know if it is the rigth procedure but I have deleted all the previous engine3 archives and I have placed the latest version in the Z7-4 folder.
It seems that IPhase off and PA + plays properly except for the forementioned rigth channel problem . No way with IPhase on : PA + ends with the crack detect message and PA - with the distorted sound and crack.

Maxi

[PeterSt]

Maxi,

Can it be so that you by accident have "Is NOS1" activated ?

Peter

[phantomax]

Peter,
No, it isn't. I am wondering about the possibility that there is something wrong with the soundcard or the MoBo. This soundcard replaced one that was fried by accident (a poor electrical contact on the PC inlet).
This is being an "annus horribilis" with hardware. I recently lost one 300B tube after a bias adjustment and so on...

Maxi

[PeterSt]

Maxi - I can't tell. All I can say is that -somehow- the "bytes" seem to get mixed up. Not knowing more than I do, the first thing I would say is that it has to be a bug on my side. But I think I checked all too thoroughly plus you are the only one ?

Okay, when this would be the result of a fried card - is 0.9z-7-3 still working (sorry if you already talked about that and I forgot).

Anyway, two indications for bytes being mixed up :
1. Crack Detect tripping;
2. Your replay of the buffer from one channel.

Do you ever have normal sound still without PA ? Is it stereo ?

Really, I think of a bug on my side. Can't you try that DacMagic ?? it would be convenient you know ...

Regards,
Peter

[phantomax]

Peter,
with 0.9z-7-3 all settings were running rigth (apparently). With earlier 0.9z-7 versions there was only the plop issue now solved.
The problem is PA related: no problem with Iphase on or off at all when PA is not active. In fact the sound is fantastic ...  and stereo ... I think. 

Regards

Maxi

[PeterSt]

with 0.9z-7-3 all settings were running rigth

Were, yes. But is that still so ? I mean, it is you who refer to a fried soundcard ...

Anyway please remember, by now I start to blame myself. No clue how yet. Maybe you can post a screenshot of all your settings from the Settings Area ? and mention Q1 etc what applies here ? then I can exactly mimic your situation here.

Thanks !
Peter

[phantomax]

Peter,
the soundcard died before any 0.9z-7 version  (R.I.P.) 

I'll post later the screenshots because I have to disconnect now.

Thank you very much!

Maxi

[phantomax]

Hello Peter,

Here are the snapshots:

[PeterSt]

Great Maxi, thanks.

At this moment I can't test your settings myself, but I see something which makes me wonder ...
You have "Do not start Playback at all" active. This shoudln't matter, but I recall a report from someone (form fairly recently) who had a for me unexpected issue with that activated. And, I don't recall that I solved it and actually I completely forgot what it was about. Anyway, this is not a setting I ever use so it may be worth to try it with that deactivated.
Otherwise I will try it myself (activated) of course, later today.

Regards and thanks for your efforts,
Peter

[phantomax]

Hey Peter,
no hurry with this because I can play flawlessly with Iphase off and PA- and, of course, with PA deactivated. Otherwise I'll try deactivate "Do not start Playback at all" and we'll see.

Thanks

Maxi

[phantomax]

Peter,
Deactivate "Do not start Playback at all" didn't change nothing except this time there was no crack detect message before that hideous noise.

Best regards

Maxi

PD. I'll try the DacMagic this weekend.

[phantomax]

Hello Peter,

I'm still mulling over the matter and I wonder why everything works (almost) fine with non invert phase and doesn’t work at all with phase inverted.

Premises:
   1 – It’s PA related
   2 -  It’s phase invert related
   3 -  There is some kind of decompensation between channels inside the digital part of the DAC or before the DAC (due to the Phase Alignement way of working). This is shown in the panel meters of the soundcard when the music stops.
   4 – This “decompensation” provokes after a while a total failure in the analog circuit of the DAC except with  non invert phase and PA – that seems to counteract that “decompensation” and sounds wonderful. Besides, the failure is progresive with Phase inverted and PA -. Not so with phase inverted and PA+.

I am not 100% sure but my DAC  has something to do in all this. Above all the analog part.
My knowledge is quite limited since I only apply ready recipes so I am going to resume the circuit behaviour if anyone could explain where is the problem. Although I think is a combination of both, the Phase Alignement and the analog circuit. It seems they are not compatible except in the forementioned cases.

The differential balanced signals from DAC are "summed" up inside the tube stage. Each phase of the differential output of the DAC chip sends out two components: music and conversion errors. Other phase sends “minus music” and the same conversion error. The signals then are summed out of phase.
The circuit offers common-mode signal rejection; this means that passes differential input signal ( music), but ignores what is common to both input signals. Common-mode signals are extraneous to the actual input signal and usually consist of hum, power-supply noise, and RFI. The point is scrub away the added electrical contamination

Conclusions:
   - I am totally lost.

Maxi

[PeterSt]

Okay Maxi. What I am going to write I already wanted to write, but it seemed less complex to wait for your DACMagic result. However, since you said you are going to do that in the weekend, I thought to write it anyway. But it turns out that you were just ahead of me ...

First off, remember, pre-0.9z-7-4 only Normal Phase and Ph- were considered to be okay. From off 0.9z-7-4 all is okay. So, this should mean that while now e.g. Inverted Phase and Ph+ work, you encounter the problem; it also means that when you before used that without problems, it just did not engage, or at least not properly. Now :

Phase Alignment implies DC Offset. Thinking of a speaker driver it is easiest to envision that the diaphragm is going to play to the back when Ph- is activated. This counts for Normal (absolute) Phase. With Inverted Phase and Ph+ activated, playing is to the front. Notice that actually the Normal/Inverted phase is unrelated to this whole scene, but, what you would normall want always, is playing to the back. And, when something in your chain is inverting Absolute Phase (I think that is what is happening in your chain, right ?) the Normal Phase and Ph- will play to the front. So you Invert Phase and activate Ph+ and now all plays to the back again.
Remember, this is only the envisioning of things and it doesn't happen to your loudspeaker at all. Well, it shouldn't as per the mandatory descriptions and advised tests.
But in electronics similarly happens.

As I said, Inverted Phase it totally unrelated. Thus also : no matter how the common mode rejection mechanism works, it works the same for Absolute and Inverted phase. Only the plus and minus signals from the the differential signal as a whole swap. That's all. But ...

Positive and negative Voltage rails need not to act the same. Actually, "severe" things need to happen to create negative Voltage. It uses more current, which proves it. And *here* the culprit must be sought somewhere ...

This is what brought me to ever thinking about this area, but actually I wondered from the start what would happen (and it left my memory because nobody complains) :
Contrary to a PCM D/A converter chip, a sigma delta (I think always) works with positive Voltage only. This already looks dangerous when we apply Ph- which incurs for a "too negative" Voltage. So, I thought about this, and thought to let it go, until people started to complain. Well, nobody did so far. And thus, while you do as the only one, I think that can't be it. It shouldn't because music data is about plus and minus data (Voltage) anyway. And a bit more plus or minus easily happend in music itself. Take 1Hz as an extreme example, and you'd have DC Offset for half of a second.

As said, the differential setup in general can't be causing troubles either; it just doesn't matter.

That you have less of a problem with Inverted Phase and Ph+ seems logic (I did not know that yet, but ity helps now) because that would be a natural setting (just ment for Inverted Phase in general, as explained above). But, it still gives you problems, though less.

With Normal Phase (and properly recorded recordings) transients will go to the plus side. So, loudspeaker driver faster goes forward than backward is the analogy. This is a but of a moot thing because of swing-up needed, but say on average this is no. Same in electronics (including the swing up). With Normal Phase and Ph- something fights that. Well, not really, but the "tension" will go to the back (speaker speaking) and the transient will again go forward. Also notice that apart from the average being forward, the largest "excursion" of the transient will also be forward.
Now make it Ph+ and strange things happen. Tension is forward and excursion is also forward. But, transient voltage is also plus and thus forward. Keep this in mind.

Your most severe case is the other way around. Inverted Phase and Ph-. Remember, this is not the natural situation to begin with. So, we must assume that this enters your DAC like I just desribed, because the inversion in your chain happens further down the line. Thus same story : tension is forward and excursion is also forward. One major difference though, transient Voltage is negative now. Remember, that more difficult Voltage to create.

What about this more current using Voltage - which will also mean less under control, is overshooting (digital volume doesn't matter I think). Just a wild guess, but I think it can happen. But, only because you dialed in Inverted Phase and Ph- which actually is wrong. With Inverted Phase and Ph+ ?

Well, tension will be forward (at DAC input) and transients will go to the negative side. *Now* it can overshoot because of that tension (you can just envision that as in a loudspeaker as I tried to explain). To, tension is already there, and the transient is more forceful than when started from neutral.

I must stop here. Why ? because this would exhibit analogue distortion opposed to digital distortion. And I don't know what you hear. And while I would hear the difference, you may not know how it sounds differently.


Anyway, something like this I thought of this morning.
We could also settle for : something in your soundcard can't bear DC Offset. It would be the most normal thing actually. Not that I really know why, but we all know it doesn't belong in audio systems ...
One thing before you come up with it : when overshooting would really be in order (and it may even clip because of a very low slew rate (back to the other side)) it will NOT clip because of too much offset; this is in the sofware ...

Let's wait for you DACMagic trial ...
Peter


PS: I just (audibly) again tried 16/44.1 to 32/88.2, 176.4 and 705.6 in all combinations. No anomalies here.

[phantomax]

Peter, thank you very much for such a extended response for a particular (and unique?) case like mine. I am going to read it carefully to fully understand... You know, my english...

Maxi