[Exocer]

In staying on topic (I almost only posted about the netgear switch, not a dig at anyone else) - I decided to purchase a clock with measured specs of -137dbc/hz at 10hz from Ebay along with a pcb from AliExpress to mount it. It provides a sinewave output, so I will bring the minicircuits adapter back in, on the Etherregen.

I ended up liking the Netgear GS103v3 a lot. I replaced the 220uf cap immediately following the dc input in both of them,
with a non polarized AudioNote Kaisei of 330uf value and it was a nice uptick. They aren’t in my chain presently but I plan to revisit them in the not so distant future. A no brainer purchase. I was powering them with a Sean Jacobs DC3 dual rail lps. I built it myself using Mundorf SG awg wire, and a pair of mundorf sg 15.5 based dc cables with jssg360 shielding.

Pics or it didnt happen:

 

[MartinWT]

With the OCK-1 I concluded I don't like the effect of connecting one of these tiny 45 gram ground tubes to it, it's not an improvement vs the sound of the native OCK-1 after it has been left on for 24 hours with stabilised oscillators. Sound is leaner and edgier with these ground tubes attached using a crocodile clip to the spare square wave BNC (the only attachment point).

I found, too, that my AfterDark clocks don't like grounding boxes. Every stage of the digital chain, from router to DAC, loves them.

 

[MartinWT]

More generally my working theory is the more modest one's streamer (like say the one built into my R26), the greater the potential upside of upstream network path improvements.

My streamer is the Sonore Signature Rendu SE optical, which is not modest. Yet improvements I have made further upstream still have had a great effect on sound quality.

In digital, every part of the chain counts.

 

[Jake2]

My streamer is the Sonore Signature Rendu SE optical, which is not modest. Yet improvements I have made further upstream still have had a great effect on sound quality.

In digital, every part of the chain counts.

Yes it was a rather simple theory with a mere two points on the curve, my R26 & the Grimm.

I guess I was thinking a basic streamer’s noise tolerance, and ability to rebuffer and error correct without getting stressed in its electrical operating characteristics thus losing its timing precision to generate perfectly clean synchronous square wave digital audio stream to pass to the DAC etc, is likely particularly low.

 

[MartinWT]

What surprised me most is that the Signature Rendu SE takes optical input. Despite the optical link, noise and jitter still get through (although greatly reduced). Things that have improved the upstream feed include power supplies, grounding boxes, clock (to EtherREGEN) and cables.

 

[Jake2]

Curious indeed. I expect this ties to the way the sound of SFPs differs, which presumably also involves the introduction of jitter somewhere in the process of conversion to optical, though I’ve yet to read a compelling explanation of the specific mechanism of action for this. (To be fair this more likely reflects my limited reading) As given my rudimentary knowledge of TCP/IP, data packets contain timing and amplitude info along with a checksum will be checked at each handoff in the chain and rejected and resent if imperfect. And yet clearly it’s neither perfect nor that simple as hard coded jitter (corrupted packets?} and/or noise-inducing jitter on arrival vault the optical moat as you say.

 

[Dandoudou]

Curious indeed. I expect this ties to the way the sound of SFPs differs, which presumably also involves the introduction of jitter somewhere in the process of conversion to optical, though I’ve yet to read a compelling explanation of the specific mechanism of action for this. (To be fair this more likely reflects my limited reading) As given my rudimentary knowledge of TCP/IP, data packets contain timing and amplitude info along with a checksum will be checked at each handoff in the chain and rejected and resent if imperfect. And yet clearly it’s neither perfect nor that simple as hard coded jitter (corrupted packets?} and/or noise-inducing jitter on arrival vault the optical moat as you say.

It's the same thing as for the switches.

 

[gimmeheadroom]

TCP/IP was not designed for low latency, it just passes a mostly correct bitstream.

 

[skylinekursk]

is Cybershaft Precision premium better or worse than LHY OCK-2?

 

[Jake2]

is Cybershaft Precision premium better or worse than LHY OCK-2?

​

Apparently a top model discontinued in 2016 per discussion on a similar clock that was for sale here, with some speculation on likely equivalent current Cybershaft models.

https://www.stereonet.com/forums/to...ock-and-shunyata-alpha-50ohm-bnc-clock-cable/

If you could get more info on claimed phase noise for this unit/model @1hz offset, or better yet a phase noise chart or certificate for this particular unit, members here might be able to better assist.

There’s also a factor that I understand crystal oscillators lose accuracy over time - but I’ll defer to others as to whether this could be material over a 7+ year timeframe.

Claimed phase noise for LHY OCK-2 for comparison is -115dBc/1hz.

 

[piguetpolo]

What surprised me most is that the Signature Rendu SE takes optical input. Despite the optical link, noise and jitter still get through (although greatly reduced). Things that have improved the upstream feed include power supplies, grounding boxes, clock (to EtherREGEN) and cables.

From what I've read, the weak link in optical is the quality of the fiber, and the transmitting, receiving and translation circuits—the specifications of which are mostly unknown. That's where the jitter and phase noise is introduced. The advantage is that one gains galvanic and RF noise isolation.

 

[Exocer]

I have my Ock-1 back and connected with the LMR-400. The sound quality improvement is undeniable (compared to using the internal clock). I am guessing the OCK-2 will sound even better. More tests arriving in the next few weeks.

 

[MartinWT]

From what I've read, the weak link in optical is the quality of the fiber, and the transmitting, receiving and translation circuits—the specifications of which are mostly unknown. That's where the jitter and phase noise is introduced. The advantage is that one gains galvanic and RF noise isolation.

Yes, I went through quite a few SFPs and fibres before I settled on the combo I'm using now. Single-mode fibre sounds best with well chosen 10Gbps SFPs both ends. The isolation is excellent in my 5m run and sound quality considerably higher than when I was using ethernet cable. I have a small collection of discarded SFPs and fibre!

 

[bodiebill2]

To ground or not to ground?

Some have reported that attaching a ground box to a reference clock harmed rather than benefited the sound. This is not what I experienced with the Mutec Ref10 SE120 though: grounding it via an RCA-BNC adapter gave a clear improvement.

However, I just tried with the LHY OCK-2 and there indeed I prefer not to ground.

Not sure, but could this be related to the fact that on the Mutec one can deactivate individual clock outputs? My grounding was done on a deactivated output. On the OCK-2 this is not possible. Maybe grounding the OCK-2 other than via a BNC output would improve the sound?

 

[MartinWT]

To ground or not to ground?

I tried grounding boxes on my AfterDark clocks (single output) and the sound quality was adversely affected.

A grounding box on the component being fed the clock, though, works wonders.