XXHighEnd

Hi,

I am using https://sites.google.com/site/hifacemods/home-1/jkdac32 (http://JKDAC32) which built on the famous HiFace unit. Lately, it was discovered that it sounds better if used with Young's drivers (Young is 32/384 unit). Also, lately it was announced that John's remakes based on HiFace work well with Young's drivers and can be run at 32/384.
Problem: when I try to do so, the sound is somewhat 2x times slower. I run at KS Special 32/352 which is 8x upsampling from XX. Tried out many variants with buffers, Q1 and other. Nothing helps. Any ideas? Thanks.

Ivo

Wrong URL. This is right:

https://sites.google.com/site/hifacemods/home-1/jkdac32

Hi Ivo,

Did you also try Adaptive mode (IIRC the Device Buffer must be 2048 (and then Q1 at 1) ?

And what about WASAPI (Engine#3) ?

Regards,
Peter

Yes, I tried Adaptive with 2048 and Q1 = 1. The same. When I tried Engine3, the whole thing crashed.

Ivo

And the same problem with the normal HiFace drivers ?

Well, I have not tried 'normal' drivers because those are 24/192. As John mentioned, only with Young's we can try up to 32/384.

Everyday I play in 4x (32/176) and it is ok with both normal and Young's driver.

Have not tried Engine3 thou.

Ivo

The normal HiFace also allows for 24 bits (DAC Needs). Does that make a difference ? (but be careful at trying)

I was using Hiface initially at DAC = 24/192.  All was fine. Now, for a while run at DAC = 32/192. All fine also. Sound the same.

Ivo

OK, got it. Yes, I guess tried also that, but not yesterday. Let me do some more tests tonight. @ work now.

EOF

Peter,

Tried just. DAC needs 24 and 32 bits. Both sound the same - slow. I checked, I have latest Young driver.

Could it be a problem with XX?

Hey, others! Has anyone tried feeding 32/352 or more into John Kenny's modified Hiface with Young driver from XX?

Ivo

Peter,

With WASAPI (Engine3) is exactly the same. While 32/176 sound OK, 32/352 is slow.

Do you need log files?

Ivo

Peter,

I guess I found it...

Got 24/352 native DXD file from 2L. Played it back 'identically' on XX and on JPLay. Sound is OK, not slow. This proves that driver works OK.
I guess there is something in the way XX up-samples to 352 and 384 with ArcPred.

Ivo

Odd.
Can anyone with an NOS1-USB confirm this by means of flipping Sw#5 up, deactive "Is NOS1" in XX Settings and next upsample 8x ?

I can't test this myself because in this mode my own NOS1 doesn't do the intended job for other reasons.

Thanks in advance !
Peter


PS: Ivo, you are using 0.9z-6-c1, right ?

Peter Hi,

Just test using NOS1 USB with the setting you asked for.

No problem at all. Normal stereo sound at 352.8K
Turning on sw5 will kill right channel as usual.
Best regards,
Praphan

Thank you very much Praphan, and I didn't want to say it but that is sure what I recall (and should have tested anyway of course).

But now what ? I really can't imagine what could work out on any such driver differently while sure knowing that some kind of trick must be applied to get 384 (352.8 ) to let a DAC work at that rate while normally it can do 192 only. Or it just must be a commercial trick to sell another DAC (the Young) while it's really the driver coming with that which allows it (the 384).

Anyway, now what. Well, maybe that you don't use the latest XXHE version.

But Ivo, something must be wrong somewhere. Look here : Play settings for M2Tech Young DAC (http://www.phasure.com/index.php?topic=1935.0). That is you in there too you know. I guess you must elaborate somewhat ...

Peter

Praphan, just to be sure (because it is so ununderstandable for me) : This is 16/44.1 A.P. upsampled to 352.8, right ?

Thanks,
Peter

Affirmative !
Red book 16/44.1 AP x 8 to 352.8 K. "Is NOS1" unchecked; SW5 off. Confirm no problem.
Thanks.
Praphan

Thank you again Praphan.

Yes, I am using latest XX. with 1c at the end.

Strange? Will read the thread you pointed and see if anything is not right.

Ivo

In short:
It seems I have fooled myself a bit and you as well. But maybe not at least to some extent.

Details:
That file I got from 2L (24/352) and which I played via JPlay and XXHighend actually sounded equally wrong. The tricky thing is that it was symphonic music track unknown to me and it sounded ok, however, yesterday I noticed that while its original length is 3:03, actual playback lasted for approx. 5:30. Slow sound again. I cannot believe I did not notice this from the first time.
So, it seems that my system (PC) is capable of playing anything up to 24/192. If I go further, I get slow sound. Now, I think, I tried various settings last night and slow sound anyway. Even in MinOS mode unattended. What surprises me is that I have Core 2 Duo CPU (E6300) which actually is utilized just for few % while playing back 24/192. So I wonder why do I get slow sound if I double that.
At the end my theory is that my PC is probably too weak for material > than 192 kHz. Maybe you have other thoughts on this?

Ivo

Did not mention what is "slow sound":

It is like when playback is slower. Imagine drums get hit slower and voice is stretched and rude. The tonality goes wrong, high frequencies are very attenuated. The easiest way to hear the slow thing is if there is voice. That is why symphonic stuff sounded kind of OK to me. Violins sounded as cellos and bass was catastrophic deep and muddy.

Peter,

When running 24/192 material without any upsampling, is it still a benefit to use "Arc Pred"?

Ivo

Well, nobody blames you for falling into that 2L "pitfall", or at least certainly not me. I experienced that myself with an earlier version of XXHE while I didn't know. In that case only the "not-upsampling" played too slow and I never noticed it with that 2L stuff. When I finally knew I presented it to two other persons for fun, and they didn't notice it either. The track lasted somewhat longer though (exactly twice  :)).

Now your case;
I can't imagine that this happens because of a too slow system, especially if two players exhibit exactly the same while both ustilize not the same CPU amount. A driver issue ? then you shouldn't be alone. A hardware problem then ? not likely, but the best option if not a general driver problem;
When the "firmware" is not fast enough and/or bandwidth problems are implied, you get exactly what you have (an uneven speed (meaning : not exactly half or so)). Ask John, because he will know what his "trick" is - and I have some ideas about it (which I won't exhibit).
Btw, the firmware will be related to the driver, so it could be either which influences the other in a wrong way, so to speak.


No, it won't do a thing. Arc Prediction is only engaded when real upsampling is in order (so 2x at least for 192).

Good luck, and let me know when you have new information.
Peter

Peter,

While waiting on John's reply:

Yesterday I recalled that some time ago I have been changing USB polling setting in registry.Thought that might affect the thing. I re-installed USB drivers and so did with Young driver and still sound is slow.
Could there be problem with USB cable? I think less likely.

When you mean "firmware" are you talking about some John's addition to the whole HiFace env? I think there is no a thing like that. I guess he uses M2Tech's HW and does not add anything? OK, just thinking loud.

Thanks,

Ivo

No, John won't be doing anything with firmware I guess. But he could be doing something which increases the bandwidth while the hardware (or firmware !) used is too critical hence can't cope with that.

Notice Ivo :
I already sort of said it, but in my view nothing that is related to the software side of things at the PC side (nor the hardware there) can incur for slower passing of the samples, unless it is an even multiplier (and then it will be a sheer bug)). So, only at the "drawing end" (the interface, the DAC - where the clocks are) can do this in a "stepless" fashion. It could even be a(n unintended) "filter" after the clock output, which makes the clock pulses run slower in effect.

Peter

I bought JKDAC32 from him in January 2012. Recently in his webpage he announced that "now" JKDAC32 runs at 32/384 using Young driver.

I am starting to think: maybe he has started to sell new revisions of the same DAC which have this little "bug" solved. Maybe my early revision is just not running > 192 by design.

He is not answering yet, but if that is so, I will be very upset.

Thanks for support,

Ivo

What confuses me on his web pages (two days back at least) is that when you dive deeper, two JKDAC32's are there, one being able to do 192 and the other 384. This seemed redundant and wrong data to me, because somehow I associate the "32" to 384 at the same time. But this is not true of course. But anyway this is why I referred to that topic from January/February from this year and told you "you were in there too", making the same mistake : thinking you got the "384" in the first place. But apparently this is not true, as I see it now.

Anyway both versions carry the same price, and I would not be happy as a customer. But you can almost bet that John applied "some" hardware small change, related to the bandwidth perecived problem I talked about.

Regards,
Peter

Peter,

Thanks for looking @.
I think we can close the topic for now. Will inform if I hear anything back from John.

Ivo