[b0bb]

@b0bb

Would I be right in thinking that you currently feed the LKS with a Singxer SU-6?

Was the move away from the onboard usb or Singxer SU-1 purely motivated by a desire to improve sound quality or to eliminate operational issues?

I have modified the Amanero USB board and I agree that it sounds easily as good as my modified SU-1 however I still get loud pops at times when dealing with upsampling in hqplayer, even the SU-1 isn't completely immune to this. It resulted in a burned out a woofer earlier this week, so even though I love the sound of dsd256 with the EC modulators I feel like I'm treading on eggshells each time I launch a track!

Out of curiosity, have you attempted dsd1024 (cpu permitting)?

The SU-6 does not experience the loud bang when going from 44k to 48k based DSD, there is an incremental improvement in sound staging but the improvement does not justify the 80% price premium over the SU-1

48k DSD is not part of the official DSD spec, so this problem is not entirely unexpected, the 004 does not reliably lock to the 48k DSD rates causing a burst of noise during sample rate basis change (44<-->48k), this can be partially mitigated by setting the DPLL bandwidth to 9 or higher.

The SU-6 data stream is more stable so there are less issues with the 004 aquiring 48k DSD PLL lock (provided you use a short enough HDMI cable)

The hqplayer problem can be avoided by specifying conversion to a fixed sampling frequency eg 22.58MHz for DSD512, you will lose the use of some filters that cannot do non-integer conversion eg sinc-M, turn off auto-rate family to enable this.
This will make hqplayer use a constant rate basis, and there will be no more loud bangs on the Amanero or SU-1.

I am waiting for a decent DSD1024 DDC unit to show up in the market, both the Singxer SU2 and Denafrips Iris and Gaia use non Xmos proprietary solutions requiring custom windows drivers.
(The marketing sheet says driverless on linux/mac but somehow neglects to mention the top sampling rates require the custom windows drivers)

The sound quality from the NAA running the Linux 5.8 low-latency kernel is so for ahead of Win10, it was almost unbelievable, moving back to Win10 made me wonder if I was listening to the same audio chain.
This was achieved by stripping down the Ubuntu linux distribution to its bare essentials, this is not possible on Win10 and all that extra bloat from Microsoft takes a heavy toll on the sound quality, this makes Win10 a non option.

 

[piaseczek]

Is this crystal better than the stock ? And how many should be ordered ? 1, 2 ?

The jitter figures are the best for the stock but it's the 50PPM clock.
According to b0bb and many others 950x sounds better despite worse parameters.

One piece can be ordered.

 

[piaseczek]

Interesting comparison of clocks:
https://www.audiosciencereview.com/forum/index.php?threads/close-in-jitter.1621/page-4#post-40559

 

[rettib2001]

The SU-6 does not experience the loud bang when going from 44k to 48k based DSD, there is an incremental improvement in sound staging but the improvement does not justify the 80% price premium over the SU-1

48k DSD is not part of the official DSD spec, so this problem is not entirely unexpected, the 004 does not reliably lock to the 48k DSD rates causing a burst of noise during sample rate basis change (44<-->48k), this can be partially mitigated by setting the DPLL bandwidth to 9 or higher.

The SU-6 data stream is more stable so there are less issues with the 004 aquiring 48k DSD PLL lock (provided you use a short enough HDMI cable)

The hqplayer problem can be avoided by specifying conversion to a fixed sampling frequency eg 22.58MHz for DSD512, you will lose the use of some filters that cannot do non-integer conversion eg sinc-M, turn off auto-rate family to enable this.
This will make hqplayer use a constant rate basis, and there will be no more loud bangs on the Amanero or SU-1.

I am waiting for a decent DSD1024 DDC unit to show up in the market, both the Singxer SU2 and Denafrips Iria and Gaia use non Xmos proprietary solutions requiring custom windows drivers.
(The marketing sheet says driverless on linux/mac but somehow neglects to mention the top sampling rates require the custom windows drivers)

The sound quality from the NAA running the Linux 5.8 low-latency kernel is so for ahead of Win10, it was almost unbelievable, moving back to Win10 made me wonder if I was listening to the same audio chain.
This was achieved by stripping down the Ubuntu linux distribution to its bare essentials, this is not possible on Win10 and all that extra bloat from Microsoft takes a heavy toll on the sound quality, this makes Win10 a non option.

Thanks for providing such an insightful reply b0bb.

As fas as HQplayer is concerned, I'll impletent the settings suggested.

Yes, bloated OS's do seem to have a big impact on sound quality.

I've actually been jumping between distros of linux, namely audiolinux and gentooplayer over the past few years.
Strangely enough I've also had great results with a corperate version of windows (LTSC) installed on a small optane drive with a few latency mods and using Audirvana + to upsample to DSD512 and send the audio via upnp to a linux endpoint.

I'll also be keeping an eye out for a DDC that can deal with DSD1024 natively, hopefully one will come along before too long!

 

[piaseczek]

Inside of the 575 Crystek.

 

[piaseczek]

It looks like the specification for the ABLNO was updated:
https://pl.mouser.com/ProductDetail/ABRACON/ABLNO-100000MHz?qs=%2BrH4t%2BeVZ2PJm9z5yYDaAg==

Typical Phase Noise (100MHz Carrier) Vdd=3.3V
@ 10 Hz offset -88 dBc/Hz
@ 100 Hz offset -118 dBc/Hz
@ 1,000 Hz offset -141 dBc/Hz
@ 10,000 Hz offset -160 dBc/Hz
@ 100,000 Hz offset -161 dBc/Hz

All parts are tested:
"Maximum Phase Noise is verified on 100% of the parts at 25ºC ± 3ºC."

"The above specified Phase Noise & Jitter is with the oscillator device configured as a VCXO. In XO
configuration, the Phase Noise will be slightly better at each offset between 10Hz and 10 kHz, by
approximately -3dB to -5dB."

 

[b0bb]

Inside of the 575 Crystek.

That is the CCHD950 not the CCHD575, image turned around 180degrees

More pictures here : http://bbs.hifidiy.net/thread-1052376-1-1.html

The CCHD575 internals:

More pictures here: https://hallmanlabs.com/2018/01/23/revived-the-amanero-close-up-of-crystek-cchd-575/

The 950 crystal is approx 4x the size of the 575, this gives better mechanical stability and has higher thermal inertia making it less temp sensitive.
This is the key to why the 950 performs so well and why it is a common premium upgrade option in high end audio DACs.

The 2 types of internal construction is shown below, the 575 quartz crystal is very much smaller
http://www.txccrystal.com/term.html

 

[b0bb]

I've actually been jumping between distros of linux, namely audiolinux and gentooplayer over the past few years.
Strangely enough I've also had great results with a corperate version of windows (LTSC) installed on a small optane drive with a few latency mods and using Audirvana + to upsample to DSD512 and send the audio via upnp to a linux endpoint.

The critical thing is to have the NAA endpoint on headless (no-GUI) linux and not windows to get the absolute best performance.

I originally used Signalyst's HQPLayerOS running a near-realtime Linux 5.4 kernel.
Ubuntu Server 20.x running the low latency 5.8 kernel actually produces a better sound quality, I am currently running 5.8.11.
Ubuntu Server is the most stripped down version they offer, I use this as a starting point to continue stripping down the distribution.
5.8 is still quite young, for example 5.8.12 is a step backward compared to 5.8.11 in the sound quality dept.

HQPlayer server is quite tolerant of the OS as long as there are enough CPU cycles.
If you shutdown and disable some of the more superflous junk like Windows Telemetry, Windows Search, Win10 starts to behave properly

 

[oldearwax]

...
If you shutdown and disable some of the more superflous junk like Windows Telemetry, Windows Search, Win10 starts to behave properly

Use
https://github.com/Sycnex/Windows10Debloater
script to debloat windows 10.

 

[piaseczek]

That is the CCHD950 not the CCHD575, image turned around 180degrees



More pictures here : http://bbs.hifidiy.net/thread-1052376-1-1.html

The CCHD575 internals:

More pictures here: https://hallmanlabs.com/2018/01/23/revived-the-amanero-close-up-of-crystek-cchd-575/

The 950 crystal is approx 4x the size of the 575, this gives better mechanical stability and has higher thermal inertia making it less temp sensitive.
This is the key to why the 950 performs so well and why it is a common premium upgrade option in high end audio DACs.

The 2 types of internal construction is shown below, the 575 quartz crystal is very much smaller
http://www.txccrystal.com/term.html

Great info, thanks b0bb.

Do you know what is the difference between the crystek 950x and the 950?
Jitter and the PPM values are the same, only the temperature range is wider for the 950x.
Do you think that 950x is a selected version of the 950 and jitter could be lower?

950x:
https://pl.mouser.com/ProductDetail...HD-950X-25-100000?qs=L4klnTtkofObrfUJUtC8ig==
950:
https://pl.mouser.com/ProductDetail...CHD-950-25-100000?qs=SZbbnLfOXtbuC9IKo91wAw==

 

[b0bb]

Great info, thanks b0bb.

Do you know what is the difference between the crystek 950x and the 950?
Jitter and the PPM values are the same, only the temperature range is wider for the 950x.
Do you think that 950x is a selected version of the 950 and jitter could be lower?

950x:
https://pl.mouser.com/ProductDetail/Crystek-Corporation/CCHD-950X-25-100000?qs=L4klnTtkofObrfUJUtC8ig==
950:
https://pl.mouser.com/ProductDetail/Crystek-Corporation/CCHD-950-25-100000?qs=SZbbnLfOXtbuC9IKo91wAw==

The wider temp range crystals show less frequency deviation with temp.
Regular crystals have the AT curve, the wider temp crystals have the flatter curve.

Assuming these were the actual curves for the Crysteks, the the X variant is represented by the flatter curve
CCHD950 has a max deviation 25 ppm@70ºC
CCHD950X has a max deviation 25 ppm@85ºC

Drawing a horizontal line at -25ppm, this is 70ºC for the lowest curve, the curve above it at 70ºC shows a deviation of -22ppm. (These are not the actual Crystek curves but it illustrates the point).

 

[Gilberto 62]

Good afternoon gentlemen. Please see if anyone can shed some light on my digital path. In our dear LKS there are three aspects of configuration with which I am somewhat confused. Are:
-Swap DSD Channel ---------- Yes or No ...?
-DE-EMPH ----------------------- Yes or No ...?
-DoP --------------------------------Yes or No ...?
Please, if anyone is kind enough to give me feedback, please note that I play ALL my music on Foobar by upsampling to DSD512.
Thanks for your valuable help. A greeting from Barcelona....

 

[b0bb]

Good afternoon gentlemen. Please see if anyone can shed some light on my digital path. In our dear LKS there are three aspects of configuration with which I am somewhat confused. Are:
-Swap DSD Channel ---------- Yes or No ...?
-DE-EMPH ----------------------- Yes or No ...?
-DoP --------------------------------Yes or No ...?
Please, if anyone is kind enough to give me feedback, please note that I play ALL my music on Foobar by upsampling to DSD512.
Thanks for your valuable help. A greeting from Barcelona....

No to all 3 questions
De-emph was used in very early CD releases (early 1980s), this is a form of equalization best applied once at the time the CD is added to the library.
If you are using the Amanero windows driver there is no need for DoP

 

[Gilberto 62]

As always, thank you very much b0bb and thanks to all of you who help others. By the way, I see no one talking in the forum about the Musetek da005, which appears to be the successor to our LKS 004 and from what I read it has a number of "creme de la creme" finishes and components. Do you really think that he is a big step forward from our current dac ...? It is a serious option as a replacement for 004 for which we do not have the capacity for the upgrades that are seen here in the forum ...? In other words, that Musetek 005 can be like a LKS 004 at the maximum upgrade level ...? Thank you for your valuable opinion and point of view ....

 

[piaseczek]

How to desolder the Crystek 575 without a hot air gun, great video: