XXHighEnd-09-z7.4 Stop issue

[Diede]

Hi Peter,

I just downloaded and installed XXHighEnd-09-z7.4.
Everything works perfect and first impression is that it sounds even better then XXHighEnd-09-z7.3.

I do have one issue though, when I hit the 'Stop' button the last sequence continous in loop. The only way to stop it is to kill engine 3 with the 'Off' button.

I tried with Fade-out steps 0-3, nothing changes.

Best regards,
Diede

 

[PeterSt]

Hi Diede,

With what further settings is this ? (SFS, Q1, Q1F, DevBufSize, input sample rate, output sample rate)

[Diede]

Hi Peter,

See signature data:

Engine#4 Adaptive Mode / Q1/2/3/4/5 = 6/0/0/0/0 / Q1Factor = 1 (Dev.Buffer = 4096) / ClockRes = 15ms / Straight Contiguous / SFS = *2*/*2* / not Invert / *Ph minus (strength 0-1)* / Playerprio = Low / ThreadPrio = Real Time / Scheme = 3 @ / All Services Off except RDC and PAN / No Running Time / Minimize OS = On / Double Octo Arc Prediction Upsampling / -> 24/768 Phasure NOS1 async USB DAC, Driver v1.0.2 (16ms)

I'm running in attended mode for the time being.

BR,
Diede

[PeterSt]

Diede - you are completely right. With an SFS under 8 it "loops" at the end. Not good ...

Something to solve.

Thanks !
Peter

[JohanZ]

- you are completely right. With an SFS under 8 it "loops" at the end.

Same loping effect here with a SFS =12/12

[PeterSt]

I will solve it anyway, but Johan ... just curious, what other settings go along with this ? (SFS, Q1, Q1F, DevBufSize)
So, before I focus on that "under 8" while other situations incur for the same, I rather know the other situations ...

Thanks !
Peter

[JohanZ]

It's all in my profile : ......Q1=14 /  Q1Factor = 1 (Dev.Buffer = 4096) / ClockRes = 15ms / Straight Contiguous / SFS = 12 /.......

[PeterSt]

LOL. So Johan, are you saying that SFS=8 works and SFS=12 again not ?
(you never said that of course, but it should be my conclusion)

[JohanZ]

SFS 12 is the lowest workable value for me!

[PeterSt]

Aha, interesting ...

What is the latency (ms) in the Driver Control Panel ?
I tested this with 4ms, so it is here where the difference must be ...

[JohanZ]

Can not verify it at this moment. I'll inform you this afternoon.

[praphan]

So, before I focus on that "under 8" while other situations incur for the same, I rather know the other situations ...

Thanks !
Peter

Hi Peter,

Since I am also as curious as you, I hooked up my spare bookshelf speaker to my tube amp and check this "loop" phenomenon out. I have developed a habit of checking the DC and LF AC pulse at the hit of stop on all new XX release before feeding the signal to my more expensive speakers.

This LOOP from the stop is related to :

(1) SFS : in my case below 6-8 is already a problem

(2) Rate of AP upsampling : At SFS = 2 and 1xAP (or no upsampling at all), you will have a very short loop. At SFS = 2 and 4xAP, the loop will be a little longer. Then at double octo AP , the loop seems to be forever.

(3) The level of digital attenuation : the louder the digital volume, the longer the loop. At -40 dBFS , 8xAP and SFS of 10,  very short loop.

All other fixed settings are : net buffer size = 4096x14x1, Phasure driver latency of 4 ms, PA - engaged.

I found this loop syndrome in z7-4 (implying low SFS, high volume and high rate of up sampling) to be dangerous to my system. Killing XXEngine 3 to end the loop will follow by strong punchy crack. I tried rebooting my audio PC to see if things will change. And yes after one of the reboots, I encountered the loudest crack ever (with a AC pulse of +/- 1.5v reading) immediately after the first start of the track. This pulse triggered the protection circuitry of my amp and hence cutting the power immediately.    Luckily, I tested this with my bookshelf speaker. 

Switching SFS to 60 seems to bring crack at the start and plop at the stop to be on par with z7-3. One improvement that I found on z7-4 is that *at SFS = 60* the click and/or crack from the change of digital volume is disappeared (no AC pulse on the meter)

Without engaging PA, XXHE behaves nicely. Haven't got a chance to check the SQ though. I have to be comfortable with bookshelf speaker test before connecting my B&W to the amps.

Hope the above helps.
Kind regards,
Praphan

[PeterSt]

Thank you so much Praphan.

Kind regards,
Peter

[JohanZ]

What is the latency (ms) in the Driver Control Panel ?
I tested this with 4ms, so it is here where the difference must be ...

The latency shows 4ms.
I have the impression when i abort playback i got the loop. When i play until the end in unattended playback the wallpaper remains at the end and xxhe won't go to its innitial state. Phasure USB audio control panel isn't active anymore. XXHE won't start anymore. I need to remove the audio engine from the memory to start XXHE again.
Johan

[PeterSt]

Below there's a version of the sound engine that won't go into the loop. BUT, it also doesn't go into any second loop of buffer playing. This means :

- Whatever helped a kind of falsely in previous 0.9z-7 versions now isn't there anymore;

- That it is totally clear that Q1 x Q1F x DevBufSize determines the length of the fade-out granted (when too short : TICK);

- That the Fade-Out parameter in XXHighEnd's settings will only make it worse when set longer than the default of 3 (the fade-out will always be cut by the buffer length). Notice that shorther won't help either once the standard "3" does not work.

- That the SFS (size) can't have an influence really anymore (good by itself) UNLESS it implies a shorther playback time than the buffer length depicts.

- That because of the eliminated anomaly in the first place, now sooner a small tick may happen than a larger which latter actually even repeats (there is no repeat anymore).

- That people using KS Special Mode may run into special strangeness which I myself can't test. But the main rule remains : increasing the total product of the buffer size must help.

And keep in mind : the less Digital Attenuation (so, the louder the volume), the more difficult it will be.

Let me know what happens ...
Peter


PS: Be careful that you (later) can differentiate between this version of XXEngine3.exe and the one originally belonging to 0.9z-7-4; you won't be able to see it on anything except by the date/time of the .exe (this one being from the 27th of September).

[PeterSt]

Thank you very much Johan.
You were just ahead of me ...

[boleary]

Leaving the fade-out step at 3 and with the settings in my signature, the new Engine3 works well. Small ticks at volume changes and sometimes a small tick at the beginning of a playlist but that's about it.

[praphan]

Leaving the fade-out step at 3 and with the settings in my signature, the new Engine3 works well. Small ticks at volume changes and sometimes a small tick at the beginning of a playlist but that's about it.

Hi Brian,

As we are in the same league on low SFS and net buffer size, I would appreciate your sharing of experience at the stop of track.

With your current set up, I wonder if you had any inaudible "plob" at the "stop" or "next" of the track at all. You can use DC volt meter to measure this strong LF AC pulse (approx +/- 1 volt depending on digital attenuation level) at the binding posts. Alternatively, watch the excursion of the woofer to see if it traveled so much.

Thanks,
Praphan

[praphan]

Below there's a version of the sound engine that won't go into the loop. BUT, it also doesn't go into any second loop of buffer playing.

Let me know what happens ...
Peter

Hi Peter,

The loop problem is completely solved.

In addition, this version of XXEngine3 does give me a better solution to minimize the PLOP at the stop of the track. It did turn scary PLOP to small plob at the stop or changing track.  However, this can only be achieved with very narrow setting range ie SFS at 2 and only 2 and net buffer of 14(or 15)x1x 4096. Tick by changing digital volume can be indirectly solved by using normalized volume. To minimize tick when changing volume, SFS has to be 60 and this is not good since it will bring back strong PLOP at STOP.

So far so good for me. Thanks so much.

regards,
Praphan

[boleary]

Hi Praphan, its too early here to test right now, but as I recall, I wasn't getting any ticks or plops at stop or at changing tracks (Alt N) in the middle of one playing. Also I didn't test any other SFS. Will play around with different settings this evening.

[praphan]

Hi Praphan, its too early here to test right now, but as I recall, I wasn't getting any ticks or plops at stop or at changing tracks (Alt N) in the middle of one playing. Also I didn't test any other SFS. Will play around with different settings this evening.

Thanks a lot Brian. No hurry. I still have acceptable plop at stop and changing track. If you don't then I will try to leverage from your experience.

Regards,
Praphan

[boleary]

My bad. Yes I have sometimes two small ticks, and sometimes the last sample or samples briefly repeat, when I stop or change tracks. They are not loud or dangerous soundin. Setting the SFS to 60 and I get the same results as you as well.

[praphan]

Brian thanks,

As for me the plop at SFS 60 is larger. So I settle at SFS of 2 which results in a more favorable SQ in my system.

Regards,
Praphan

[boleary]

In  my system voices are too flat with a SFS of 60. They are more harmonic or dimensional with a SFS of 2: previous harshness with a really low SFS has been replaced by this more complex, "realistic"  sound. This has been the case for me since the initial release of 9Z7. Then again maybe I've become addicted to a kind of distortion, but I sure do like it.