[b0bb]

  Dear B0bb,
Thank you again for your advise. I have read this forum again, and take notes for the upgrade and their parts number, it is all there in details. I eventually like to have a XO socket installed for comparison before and after the baking in. Have you try switch back to XO 575 and compare with 950X after finalize the up-grades ? . If you did, please tell us your findings. So may goes to the OP to draw a conclusion.  I think, it is troublesome but will be fair for have a final, all small change add up may have effects including our hearing. 
All for the best,
 
fjc36

I sometimes put the CCHD575 back in, it never stays in for long, makes the LKS sound like a common delta sigma DAC.
Here is a picture of a few other XOs I tried.
 
If you like the warm tubey sound the Epson XB42 will make the LKS sound like a vinyl record on a turntable.

 

[b0bb]

That's avery useful calculator @b0bb, many thx.

Other than phase noise & stability, are there other clock-properties that may influence the sound?

And also what would be your choice of a highend ocxo for an ess dac? I did check some but the ocxo's either need to much power (>1w) or have too long warming times (>1min) or are too expensive. Usually all three are true.

 
The type of XO is also important, for example there are XOs that do not use quartz like the Epson XB series, these are surface acoustic wave oscillators, very warm sound.
These is also the MEMS, micromachined silicon pendulums, they are very robust but no one seems to use it for audio, might get around to trying it.
 
I do not have a specific recommendation beyond the Abracon AOCJY1, it turned out better than expected, as this was experimental I did not want to spend too much money on it.
 
The next step is to use an OCXO with a SC cut crystal, these have much better jitter characteristics and they start from 250 USD.
The power is not a real issue unless it is a drop-in replacement, fast warm up ( < 30s) OCXO just cost more money, 350 USD and up.
 
The Pulsar Clock stuff is nice but they have gone quiet, their design appears similar to the one from Magic Crystal, a Russian company.

 

[b0bb]

The CCHD575 has a 50ppm stability, this means the carrier frequency changes by 50ppm for each 1 degC in temperature (more accurately it is per degree Kelvin).
 
This translates to 500fs change in the timing period.
 
The AOCJY1 OCXO is rated at 50ppb, this translates to 0.5fs.
 
 
3 degC is about the difference in temp change when a window is opened or the AC  turned on, on a warm day.
 
CCHD-575@50ppm will see up to a 1500fs change in the timing
CCHD-950@25ppm will see  change of 750fs.
AOCJY1@50ppb will change by 1.5fs
 
If we look back at the r.m.s jitter calculated at a controlled temperature
CCHD-575 is 151fs
CCHD-950 is 227fs
AOCJY1 is 1596fs
 
The number in a real application is as follows
CCHD-575   1651fs
CCHD-950    977fs
AOCJY1       1598fs
 
I think this is partly why the OCXO holds its own despite an apparent worse rms jitter number.

 

[abartels]

  The CCHD575 has a 50ppm stability, this means the carrier frequency changes by 50ppm for each 1 degC in temperature (more accurately it is per degree Kelvin).
 
This translates to 500fs change in the timing period.
 
The AOCJY1 OCXO is rated at 50ppb, this translates to 0.5fs.
 
 
3 degC is about the difference in temp change when a window is opened or the AC  turned on, on a warm day.
 
CCHD-575@50ppm will see up to a 1500fs change in the timing
CCHD-950@25ppm will see  change of 750fs.
AOCJY1@50ppb will change by 1.5fs
 
If we look back at the r.m.s jitter calculated at a controlled temperature
CCHD-575 is 151fs
CCHD-950 is 227fs
AOCJY1 is 1596fs
 
The number in a real application is as follows
CCHD-575   1651fs
CCHD-950    977fs
AOCJY1       1598fs
 
I think this is partly why the OCXO holds its own despite an apparent worse rms jitter number.

 
Hi B0bb,
 
Just curious, most OCXO's need 5V to operate, Crysteks need 3.3V. Does the Abacron need 3.3V too or did you feed it with 5V?
 
 
Regards,
 
Alex
 
BTW, I tried replacing Xpresso's on Hifiberry DAC+Pro with NZ2520SD which delivers a LESS detailed soundstage, not is big and not as wide too....
But, NZ2520SD have better bass control.......
 
Regards,
 
Alex

 

[b0bb]

   
Hi B0bb,
 
Just curious, most OCXO's need 5V to operate, Crysteks need 3.3V. Does the Abacron need 3.3V too or did you feed it with 5V?
 
 
Regards,
 
Alex
 
BTW, I tried replacing Xpresso's on Hifiberry DAC+Pro with NZ2520SD which delivers a LESS detailed soundstage, not is big and not as wide too....
But, NZ2520SD have better bass control.......
 
Regards,
 
Alex

The Abracon is a 3.3V part.
 
The NDK may not be able to drive the HiFiBerry sufficiently.
 
The Xpresso has a 3.3ns rise time, for 3.3V logic into a 15pF load, this gives about 15mA.
The NDK datasheet quotes a 6ns risetime, this puts the NDK current capability at 7.5mA

 

[fjc36]

 Dear Bobb,
If not going to the expensive OC type, you do recomend the Abracon Ablno fix freq 100mhz, better than the Crystek -950X and much better than the 575 right?

 
I think you do mean the regular XO, not the variable VCXO, the cost  of  standard XO is very reasonable around US $ 10-15 only. Abracon VCXO  will cost around $ 30-40.
Types of XO to chose

OCXO Oscillators (517)	TCVCXO Oscillators (980)	VCSO Oscillators (73)
Programmable Oscillators (25,821)	TCXO Oscillators (2,681)	VCXO Oscillators (1,703)
Standard Clock Oscillators (37,555)	VCO Oscillators (594)

 

[b0bb]

   Dear Bobb,
If not going to the expensive OC type, you do recomend the Abracon Ablno fix freq 100mhz, better than the Crystek -950X and much better than the 575 right?
 
 
I think you do mean the regular XO, not the variable VCXO, the cost  of  standard XO is very reasonable around US $ 10-15 only. Abracon VCXO  will cost around $ 30-40.
Types of XO to chose

 
I recommend the 950X as the first choice. The ABLNO is a mixed bag, it is quite an aggressive, bright and forward sounding presentation.
Get both if you want to experiment but use the 950X if have to chose one.
 
The ABLNO is the fixed frequency, ABLNO-T and ABLNO-T2 are VCXOs.

 

[fjc36]

   
I recommend the 950X as the first choice. The ABLNO is a mixed bag, it is quite an aggressive, bright and forward sounding presentation.
Get both if you want to experiment but use the 950X if have to chose one.
 
The ABLNO is the fixed frequency, ABLNO-T and ABLNO-T2 are VCXOs.

Thanks, In our case, we should select the fixed type XO right?  (since there is no intension of utilize voltage control at pin 1 in this DAC circuit design), though VCXO provides a way through feed back to stablize the OSC freq shift or swing.

 I will try purchase a 950X, and wait till your latest discovery
. 
fjc36

 

[b0bb]

  Thanks, In our case, we should select the fixed type XO right?  (since there is no intension of utilize voltage control at pin 1 in this DAC circuit design), though VCXO provides a way through feed back to stablize the OSC freq shift or swing.
 
 I will try purchase a 950X, and wait till your latest discovery
. 
fjc36

 
950X is a fixed type XO, avoid VCXO unless you want to experiment.

 

[prot]

http://www.ndk.com/en/ad/2013/001/pdf/c_NH47M47LA_e.pdf
Pretty amazing phase noise numbers but the 5ppm stability is nothing special (or I did misread something).
Anyone who tried that new wonderclock already?

 

[fjc36]

Dear B0bb,
What is your idea about the development of MQA? A new codec system developed by Meridian Audio claim to be solution for the high-res music? What will happen to the conventional DAC such as the LKS MH-DA003? Redesign the DAC IC chip to include the MQA codec?
thanks,
 
fjc36

 

[b0bb]

  Dear B0bb,
What is your idea about the development of MQA? A new codec system developed by Meridian Audio claim to be solution for the high-res music? What will happen to the conventional DAC such as the LKS MH-DA003? Redesign the DAC IC chip to include the MQA codec?
thanks,
 
fjc36

To get the best out of MQA a redesign is required. Playback on non MQA specific hardware appears possible provided you are willing to accept 15.8bit playback resolution on a 24bit track.
 
MQA has a lot to prove, especially in areas like being DRM free, MQA insists it does not contain DRM, MLP its older sibling, partially failed because of this.
The other big one is watermarking. This is an intentional corruption of the audio signal to embed tracking data, MQA says no watermarking, I will wait and see, this killed DVD-A as a viable commercial product.
 
Lastly, I do not know if the decoder will be free to use in things like foobar.
 
Paul McGowan from PS Audio posted a short impression of MQA
http://www.psaudio.com/pauls-posts/mqa-thoughts/
 
More info from MQA's inventor, Bob Stuart
http://www.computeraudiophile.com/content/694-comprehensive-q-mqa-s-bob-stuart/
 
 
More info about watermarking on audio.
http://www.mattmontag.com/music/universals-audible-watermark

 

[fjc36]

To B0bb,
If a redign of the DAC happens, how long do you think it will happen? Is that mean the current purchased LKS has to be retired on the high res digital stream object? Shall we delay the buying and observe?
 
fjc36

 

[fjc36]

I am trying to following the threads of MQA, it seems logical to transform digital data by log scale( like we do on the volume control)  to avoid waste. A linear scale on sampling and on precision may not be necessary. But in a long run we may find another type of error happens in MQA just like we found jitter, digital filter fault, and fidelity on CD format etc.
 
fjc36

 

[b0bb]

  To B0bb,
If a redign of the DAC happens, how long do you think it will happen? Is that mean the current purchased LKS has to be retired on the high res digital stream object? Shall we delay the buying and observe?
 
fjc36

 
That is an open question, if you are looking to buy a DAC now, get a conventional DAC like the LKS and check back in a few years to see if there is sufficient market interest in MQA  ( things like DACS supporting MQA and MQA music )