Ayre QB-9 USB DAC problem

[jarek]

Hi,
I have Ayre QB-9 USB DAC and Vista 32-bit PC. No matter what I do, I it does not play or plays with cracks with y6 and y7 XXHighEnd versions.
1. On engine 4 and specyfying DAC as 24 bit, XXHighEnd plays (track mark is moving) but none sound is heard. No matter what other settings are.
2. On engine 4 and specyfying DAC as 32 bit, XXHighEnd displays error: "Device allocated but can't play. Check DAC settings, applied sample rate, etc."
3. On engine 3 and specyfying DAC as 32 bit, XXHighEnd displays error: "Endpoint create failed -? The DAC does not accept this byte sequence. Try switching DAC needs from 32 to 24 bit"
4. One engine 3 and specyfying DAC as 24 bit, XXHighEnd playes with a lot of cracks, no matter what other settings (Q1,2,3) are.

What might be the reason ? Is it solvable ?
Best regards
Jarek

[PeterSt]

Difficult ...
First off, I have heard the problems with these DACs before; I didn't look it up, but I recall to have ask questions (don't ask me which right now) and never got answers. The real problem is though, that this is not about async DACs, because they play.
Btw, the PWD seems to be similar, though played somewhere, I asked questions again, but no answers (this was 2 or 3 days back only).

An always useful test (for me) : what does Foobar make of it ?

Besides this, currently your best option would be Engine#4, DAC Needs 24 bits, set the Device Buffer Size in XXHighEnd to 4096, choose KS Adaptive Mode, set Q1 to 0 and tell me what happens. Additionally post the X3PB log file of that.
If this setting doesn't work, try Q1 = 30, and maybe 20 and 10.
If this works, follow the procedure as explained in the Tooltip of the KS Modes "Adaptive Mode", in order to find the buffer size.

I hope we can work it out ...
Peter

[jarek]

Foobar works fine. J.River Media Center works excellent. I already tried manipluating Q1 and set to 4096, still is the same. This is powerful computer, Dual Core, 4GB RAM. Processor utilization is in either case below 5%, no matter the Ayre plays or not.

I will test what you suggested and send a log tomorrow.
Cheers

[jarek]

Dear Peter,

I am not DAC profesional, but I hope these two articles below might be helpful to you. They are about asynchronous USB comminisation DACs. They appear to bahave differently than other USB dacs. Both Ayre and dCS (and Wavelongth) use these technology.

AYRE white paper:
http://www.ayre.com/PDF/Ayre_USB_DAC_White_Paper.pdf

Everything seems to come from Wevelength idea about those dacs:
http://www.usbdacs.com/Concept/Concept.html

I hope this may help somewhat to solve this issue...

Jarek

[PeterSt]

Hi Jarek,

I will look into it and let you know !

Thank you,
Peter

[jarek]

Dear Peter,

I did what you suggested, and I have found that Ayre plays well on Engine 3 and Engine 4, but on Q1 set near 0 on adaptive mode. Greater values cause either distorted or none sound. I managed to get down with buffer to 384, but on 512/768/1024 is goes just more and more fluently without any dropouts. I marked Double, no upsapmpling.

The best sound is for Q1 set to minimum, which is 15 in Engine 4 special mode, buffer 1024, special mode. Greater Q1 seams to make sound worse. Is it normal ?

Best regards
Jarek

[PeterSt]

Hey Jarek, great ! What I don't understand though is that you say that Engine#3 also works ... ? (I mean, I thought you had tried that ?!).

Your other observations from Adaptive Mode and the Q1 setting sound very logic to me, although I didn't know it.
The Q1 on Special Mode says nothing much because it is completely dependend on system behaviour (once you have found the buffer size it will cope with).

One question, before I may misunderstand :
Is it correct that you can set the buffer size for the QB-9 (outside XXHighEnd I mean). And the dCS ?


PS: When I saw your Wavelength link (and the Ayre paper) I already knew it had to work, just because the "Gordon" stuff works. So, not knowing what to do else, I tried to get a QB-9 from the importer here, but I guess I'll cancel that ... (a bit stupid, but alas).

[jarek]

Hi Peter,

yes Engine 3 works too. I just tested it on Q1 set to zero, and it played without dropouts. I did not put too much attention to this, since Engine 4 plays better and more stable, at least on Ayre QB9 and this system.

Can you explain why those asynchronous mode USB dacs have problems to play on higher Q1, and generaly speaking they play better on minimum Q1 ? This seams opposite to other dacs. I have also another system with RME HDSPe card, and I had no problem to set whatever Q1/Q2/Q3 even on minimum buffer.

Jarek

[PeterSt]

Hi Jarek, thank you. I was waiting for that. But allow me to ask ...

4. One engine 3 and specyfying DAC as 24 bit, XXHighEnd playes with a lot of cracks, no matter what other settings (Q1,2,3) are.

Can you tell me where it went wrong on this one at first. I mean, this is important for me to know, so I may be able to guide a next person in the proper direction. Notice this is not at all about you, and even if you (e.g.) not tested it before, I want to know.
So, what can people do wrong here (at first glance or so) ? Thanks !

Can you explain why those asynchronous mode USB dacs have problems to play on higher Q1, and generaly speaking they play better on minimum Q1 ? This seams opposite to other dacs.

I really don't know, but that it works differently on the transfer of data is obvious (to me). But there's also the internal working of the OS, which (for example) for USB is different for XP compared to W7, and from what I have read, W7 is again different compared to Vista. This is all about the low latency MS tries to achieve, and in Vista they made a mistake or something (again, I have only read about this ... the difference between XP and W7 is my own experience).
Anyway, the transfer of data with async will be more under control of the DAC, although in either case it is the driver which tells when buffers run empty (and those buffers are on the PC side). I think that the amount to transfer is more fixed than with other connection means (like isochronous), and you can't hussle (that much) with the buffer size on the PC side, meaning it should match the transfer amount which is - or may be a "redundant" operation, hence out of (my, the programmers) control. If one (me) would behave normal, nothing is the matter I guess, but here we're not behaving much normal ...

Repeated question (related to the above I think) :

Is it correct that you can set the buffer size for the QB-9 (outside XXHighEnd I mean). And the dCS ?

Thank you,
Peter


PS: I really love your website and I'd like to copy it (or at least the concept of it ) ...

[jarek]

Dear Peter,

so you managed to identify my business ? You are talented man really

As regards to buffer size setup for DACs, one does not install any software on OS for Ayre nor dCS. DAC is automaticly recognized by any OS when you plug it into USB port. So, nowhere to set buffer size outside of XXHighEnd. I maniplated only values in the XXHighEnd program.

I will check exactly Engine 3 on Ayre on Monday and let you know.

By the way, I have to say this. Today I spent some time calibrating switches for my home system with RME card and external dCS Scarlatti DAC connected by dual AES. I received the sound to die for !

Quad, no upsampling, Q1=6, Q2,3,4,5=0. Engine 4, Mode normal, buffer=32 (set on RME and XXHighEnd). PERFECT !

For web site, you can contact me outside forum, if you already identified me

Cheers
Jarek

[jarek]

Peter,

I have checked Engine 3 on Ayre QB9 and it plays stable and nice at least on low Q1 up to 4 and lower.

For Engine 4, on buffer 512 plays stable only to Q1 < 6 , for higher buffers : 768 to 4096 plays well for Q1 < 29.

How can we identify the right buffer in this case ?

BEst regards
Jarek

[PeterSt]

Hey, thank you Jarek.

Ok, now you're overasking me. There may be a best buffer setting, but I am not that far that I can tell. Not for my own soundcard/DAC, and thus sure not for everyone's. Ask me again after a year, and maybe then I'll have an answer.
It *is* true that I can quite confidentally head towards something which sounds better lately (like I didn't even listen to Adaptive Mode really before bringing it out), but there is still a lot more to learn. It may go faster than a year, and maybe that happens because you all out there get the hang of what sounds better. And again, how could I do *that* on my own. It would need 1000 times of my time.

In either case, I am happy to learn it is all running as should now. And thanks for the extensive feedback.
Up to the next round !
Peter