Master Clock Talk

[bodiebill2]

Interesting measurements by Golden One of the Gustard X26 with an external 10 MHz clock:
https://goldensound.audio/2022/04/14/gustard-x26-pro-measurements-f20-firmware-10mhz-clock/
Basically he is saying that external 10 MHz clocks are only needed for pro use and actually increase jitter. I do not believe he mentions the actual external clock used.
Whatever the theory, in my experience an external clock considerably improves things, and that was the case for all I have tried (Afterdark, Mutec and LHY).

 

[Dandoudou]

Interesting measurements by Golden One of the Gustard X26 with an external 10 MHz clock:
https://goldensound.audio/2022/04/14/gustard-x26-pro-measurements-f20-firmware-10mhz-clock/
Basically he is saying that external 10 MHz clocks are only needed for pro use and actually increase jitter. I do not believe he mentions the actual external clock used.
Whatever the theory, in my experience an external clock considerably improves things, and that was the case for all I have tried (Afterdark, Mutec and LHY).

I like to read the reviews of Golden One, and I read this review sometime ago. I was also surprised that Golden One says that.

 

[DACLadder]

No one (that I have seen) ever stated that with 10M clocking the DACs actually measure better. I was resistant to 10M for the very reason until I gave it a listen with a good quality 10M clock. 10M is probably adding it own "color" that is pleasing to the ear. And one manufacturer's method for 10M clocks is greatly different from the next. Golden One used the AP555 10M clock which he says does not make a difference in jitter measurements. But we know it can make a difference in sound.

The X26 pro is quite impressive on the measurements. Greater than -160dB noise floor. Even with 10M the jitter artifacts are better than -145dB. How does it sound internal versus external? Wish I had the Audio Precision 555 analyzer!

 

[JaMo]

No shade on Golden Sound. The guy reports what he can measure. I haven't been able to get info on what external master clock he used. It had been interesting to know. Anyway, measurements are not the full truth as @DACLadder wrote above regarding sound performance. Some parameters are relevant as IM (intermodulation) and of course, analog distorsions.

I think we have a very exciting and interesting period ahead when machine learning or AI can help us to find the combinations of parameters connected for the sonic results. I personally think that todays methods of measurements still are too primitive to really find the answers we all want to understand. I think AI can and will help us to find out what is in between todays measurements and hopefully find new areas to investigate.
/Jan

 

[sajunky]

Basically he is saying that external 10 MHz clocks are only needed for pro use and actually increase jitter. I do not believe he mentions the actual external clock used.

He is doing measurements in ideal conditions, removing ground loops and other noise as much as possible. This a right approach, as results are repeatable, can be compared directly with other devices. On the other side, it doesn't give information how these devices are performing in the real-world noisy environment. Adding such tests would be complicated and there is no standards to follow.

A comment on the external clock may be taken based on his particular (clean) environment, I don't know. His knowledge on clock performance and synchronisation issues is limited, he is technician, not a designer, you must remember that. Performance of a clock should be evaluated in a full spectrum of frequencies and external clocks deliver better performance in the range 1Hz-10Hz where phase noise parameter is the worst. Internal clocks are 20-30dB worse in this range and are not specified below 10Hz. It is true that clock synthesiser degrade performance, question is how much. Only listening tests can tell.

A second aspect on pro-use vs. home-use: I would say, when you synchronise two devices with the external clock, it is pro-use.

[EDIT] I looked at the site, and a surprise... He didn't do phase noise tests vs. frequency. So how he can make such conclusion, attaching his own theories? it is a typical Goldenone.

 

[DACLadder]

@JaMo How does your X26 Pro sound internal vs. external clocking? JaMo has some good clocks, Mutec Ref 10 SE120 and OCK-2. thnx

 

[JaMo]

@JaMo How does your X26 Pro sound internal vs. external clocking? JaMo has some good clocks, Mutec Ref 10 SE120 and OCK-2. thnx

The X26 Pro was decided and bought based on the very good master clock response of the U18 with the K2 clock syntheseizer. The answer is the implementation of the K2 clock syntheseizer and how good it let the master clock conduct. It is brilliant sounding with the Mutec SE120. Firmware is the vF.20 (Beta). I did a quick test with the LHY OCK-2 aswell and it answer really good also on it but not really up to the level of the Mutec. The X26 Pro doesn't sound like the other DS dac's I have and have listened to.. It is a very nice tuned dac where the ESS sound is absent. The dac is detailed and very dynamic with authority and just brilliant with U18 and with both on a good master clock.
I use Gustard's own I2s (HDMI cable)
Edit. I forgot. I have Gustard fuses in both.
/J

 

[DACLadder]

A comment on the external clock may be taken based on his particular (clean) environment, I don't know. His knowledge on clock performance and synchronisation issues is limited, he is technician, not a designer, you must remember that. Performance of a clock should be evaluated in a full spectrum of frequencies and external clocks deliver better performance in the range 1Hz-10Hz where phase noise parameter is the worst. Internal clocks are 20-30dB worse in this range and are not specified below 10Hz. It is true that clock synthesiser degrade performance, question is how much. Only listening tests can tell.

Wise words indeed.

It has been my assumption that 1Hz and 10Hz phase noise of the `10M generated audio clocks may be slightly better than say the Accusilicon's TCXO 1Hz/10Hz performance. With the uber best 10M clocks of course. 5dB better 10M clock may yield only 1 dB better performance of the synthesized clocks. But it could beat the TCXOs. If you look at the Accusilicon datasheet 1Hz and 10Hz phase noise (with Kingwa's 90/98M types) spec -59/-58dB/Hz at 1hz offset. A synthesized clock with -110dB/hz at 1Hz 10M may provide better 1Hz and 10Hz performance With a -120dB clock a few dB better synthesized clocks. Above 100Hz phase noise of both methods very low. (-130 to -145dB) In my experience 1 and 10Hz phase noise affect sound the most yet are hardest for any audio clock to achieve. An audiophile's lament....
http://www.accusilicon.com/docs/AS318BM.pdf

 

[gimmeheadroom]

Don't you guys mean 1 KHz and 10 KHz?

 

[JaMo]

Don't you guys mean 1 KHz and 10 KHz?

No, they meant what they wrote. 1Hz and 10Hz. Some OCXO makers announce phase noise even below 1Hz. Ex. 0.1Hz
/Jan

 

[DACLadder]

1Hz and 10Hz. 1kHz and 10Khz phase noise is usually very low compared to 1 and 10Hz.

Next question why 1Hz and 10Hz important for audio.? Some other technology like high speed data bus may prefer optimized phase noise at 100kHz for lowest data error rate. My theory is these low frequency phase noise are the worst specs from any oscillator. 100Hz and above phase noise are generally 60 to 80 dB lower (better) than 1Hz. And 1Hz/10Hz tend to affect bass, pace/ rhythm, and staging. That is why we fuss over 1Hz.

IMHO. Phase noise tends to modulate directly the D/A process. If your clock has noise at 10hz then the sound will be imprinted with the 10Hz noise. A 100Hz bass note modulated by 10Hz varies the frequency by a full step in either direction. Sound now more vague and warmer. 1Khz note and 10Hz modulation not as bad.

What are your thoughts?

 

[gimmeheadroom]

Thanks @JaMo @DACLadder , I had not considered this

 

[Jake2]

I've been reading in the word 'offset' (from clock carrier frequency, typically 10mhz in our discussed ext clock use case) into all the references above to 1hz and 10hz. Correct?

In which case in simple terms the greater the offset (the X axis deviation/error) the lower the desired/acceptable amplitude (error/deviation in Y axis). So in that general sense the 10hz offset spec is perhaps more important than 1hz offset.

Edit... which I guess is all implicit/ assumed knowledge here, with @DACLadder 's interesting speculations as to how the audible effect of phase noise manifests in music. I concur... that's how I've envisaged it too.

 

[Dandoudou]

I use Gustard's own I2s (HDMI cable)

I also use it. It's a very good cable, it's a pity that it was discontinued.

 

[DACLadder]

IMHO 1Hz phase noise affects pace and rhythm and perhaps bass notes down to 20Hz. . 1Hz, 10Hz, etc are just measurement offsets and easy for comparison. The oscillator phase noise is really bouncing between all frequencies in between. A spectra of noise so to speak.

Some say 0.1Hz phase noise makes a differences. Haven't evolved that far.