Digital Drive

Upsamplers, DACs, jitter, shakes and analogue withdrawals, this is it.

Return to Digital Drive


Message Sort: Post Order or Asylum Reverse Threaded

Possible to split a 10Mhz clock output?

42.2.210.87

Posted on December 15, 2016 at 19:42:19
eduardoo
Audiophile

Posts: 1080
Joined: August 14, 2002
I'm currently using a custom made external 10Mhz Rubidium master clock with my Esoteric K-03 SACD player. The clock has one single output only. If I'm to get another piece of equipment that also accepts such clock input, is it possible to split that clock's output into two? If so, how do I go about doing it?

Thanks.

 

Hide full thread outline!
    ...
RE: Possible to split a 10Mhz clock output?, posted on December 15, 2016 at 23:06:34
knewton
Audiophile

Posts: 421
Location: Mid-Atlantic/Northeast
Joined: May 18, 2010
Yes. You need to obtain a special 3-way coaxial connector called a T-splitter/adaptor. A Google search will quickly produce some pictures which will immediately explain their name. Be certain that the one you obtain is specified to have an characteristic impedance (either 50 ohms or 75 ohms) matching the clock coax cable, which itself should match the impedance of the master clock generator's output. T-splitters are common devices.
_
Ken Newton

 

RE: Possible to split a 10Mhz clock output?, posted on December 16, 2016 at 02:10:50
eduardoo
Audiophile

Posts: 1080
Joined: August 14, 2002
Thanks for your input. Didn't realise it would be so simple.

Are such BNC splitters "good" enough for hi-fi, particularly since the idea about such clocks is all about precision?

Thanks.

 

RE: Possible to split a 10Mhz clock output?, posted on December 16, 2016 at 03:07:31
PAR
Audiophile

Posts: 1784
Location: South London
Joined: April 15, 2013
Contributor
  Since:
January 12, 2014
The subject of using external or system clocks is more complex than finding a stable generator for the signal. The whole subject is way beyond my competence but I understand that issues such as phase behaviour and the PLL for distribution are important.

All dCS system clocks have multiple outputs so what you want to do is achievable. However they use " low noise, low skew clock distribution systems that get the clock to the point of use exactly as it was generated." - whatever this entails.

What is subjectively observable is that dithering the signal seems to improve the distribution (apparently has something to do with " dead spots" occuring in regard to the PLL). It is switchable on/off with their clocks so comparison is simple.

I have also experienced the significant changes in sound that comparing different generations of their clocks brings even though basic clock stability specifications are similar.

Unfortunately the choice of 75 ohm cable seems also to affect the resulting sound even given cables claimimg quite tight observance of the spec and with due attention paid to screening and using "audiophile approved" dialectrics such as PTFE.

I have also seen arguments from pro recording engineers that the use system clocks is worse then just letting each component run with its own inbuilt clock. I don't necessarily agree but the argument is out there.

So, as the subject itself is not very straightforward it is hard to guess whether BNC splitters are "good" enough for hi-fi.

Suck it and see. If you are happy with the result then that is what counts.

 

RE: Possible to split a 10Mhz clock output?, posted on December 16, 2016 at 12:16:12
knewton
Audiophile

Posts: 421
Location: Mid-Atlantic/Northeast
Joined: May 18, 2010
Yes, they are good enough so long as they are accurately and stably impedance matched to the rest of the clock distribution network. The impedance matching will minimize jitter introduction by the splitter. Don't obtain some mysterious no-name part here, but there's also no benefit in locating some boutique part produced by the audio industry either. Simply source a quality constructed part of the correct impedance from one of the established RF connector vendors. The cost still will be quite minimal. Also, be sure it has the type of connector ends you require for mating to your coax, which are probably BNCs. Connector gender (male/female) will also matter.
_
Ken Newton

 

RE: Possible to split a 10Mhz clock output?, posted on December 16, 2016 at 18:28:07
eduardoo
Audiophile

Posts: 1080
Joined: August 14, 2002
Thanks for your detailed reply. I'm aware of the controversy of using these clock implementations, and that's one of the reasons why I'm not going crazy with esoteric 's expensive matching clocks. The cheap custom made rubidium clock seems to work reasonably well with the player, giving a more expansive and solid sound.

Thanks again.

 

RE: Possible to split a 10Mhz clock output?, posted on December 16, 2016 at 18:29:34
eduardoo
Audiophile

Posts: 1080
Joined: August 14, 2002
Thanks. Will keep that in mind when I need to get them.

 

RE: Possible to split a 10Mhz clock output?, posted on December 17, 2016 at 10:30:08
AbeCollins
Audiophile

Posts: 27441
Location: USA
Joined: June 22, 2001
Contributor
  Since:
February 2, 2002
As knewton already mentioned, you might get by with a simple BNC T adapter depending on the loads and how they are terminated at the receiving ends, and the cable lengths. Another option is to use something called a distribution amplifier.

Personally, I'm not convinced that the 'accuracy' of a Rubidium clock is of any benefit in a home audio setup. The output of a Rubidium clock actually comes off an ovenized voltage controlled crystal oscillator that is locked to the hyperfine transition state of the Rubidium atom (~6.834 GHz). The phase noise of the crystal oscillator is probably more important than the clock's absolute long term accuracy.

I worked for a company called Efratom in the 1980s on crystal oscillators, Rubidium clocks, and hydrogen masers in a past life. And I stayed at a Holiday Inn Express a few weeks ago. ;-)

Didn't we talk about this almost 4 years ago? What 'custom made' Rubidium clock do you have?



 

RE: Possible to split a 10Mhz clock output?, posted on December 17, 2016 at 19:40:33
eduardoo
Audiophile

Posts: 1080
Joined: August 14, 2002



Yes, that was me.
Got one made with an fe5680a with a nice linear PS in a decent box for a few hundred dollars. As said, despite the likely limitations, it does seem to sound better with it in the system, so I just left it there.

 

RE: Possible to split a 10Mhz clock output?, posted on December 18, 2016 at 22:11:35
AbeCollins
Audiophile

Posts: 27441
Location: USA
Joined: June 22, 2001
Contributor
  Since:
February 2, 2002
Very cool! But how do you know that it's output is precisely 10.00000000 MHz as shown on the HP frequency counter?

Is the reference clock in your HP frequency counter more accurate than your Rubidium oscillator? ;-)

We used a reference Rubidium oscillator that was calibrated and traceable to the National Bureau of Standards (now called NIST). The signal from the reference Rb oscillator was input into a Tracor frequency difference meter and compared against new Rb oscillators that we produced. The reference Rb oscillator would be sent back to a calibration lab periodically to ensure that it's accuracy was certified and traceable to NIST.

It's interesting to note that the freq diff meter needle would swing in one direction and then the other as the crystal in the Rb oscillator would be swept in each direction (with an error voltage applied to a varactor diode which changes it's junction capacitance which would of course alter the frequency of a tuned circuit) until the physics package warmed up and disciplined the crystal output and locked it to the hyperfine transition state of the Rb atom.

Here's an old Tracor Frequency Difference meter similar to the one I used to use. Fun stuff back in the day. Enjoy your Rb oscillator.











 

RE: Possible to split a 10Mhz clock output?, posted on December 18, 2016 at 22:34:13
eduardoo
Audiophile

Posts: 1080
Joined: August 14, 2002
That frequency counter belongs to the guy who made the thing for me. I actually asked him the same question. He said that his counter had been calibrated with some reference facility (can't remember the name now, but it sounded credible to me when I heard it), so I guess I'll just have faith in it. :P

 

RE: Possible to split a 10Mhz clock output?, posted on December 18, 2016 at 22:34:42
gusser
Audiophile

Posts: 2322
Location: So. California
Joined: September 6, 2006
10mhz does not equal any standard audio sampling rate.

So does your player accept 10mhz? If so it must have an internal PLL with a divider chain to lock to 10mhz. And that PLL chain will probably have more jitter than a simple CXO at the proper frequency.

Also a BNC tee is a poor way to split an RF signal. Just as paralleling two 8 ohm speakers makes a 4 ohm load, two 50ohm loads in parallel is 25ohms, so much for maintaining proper cable impedance. Also the signal will be at half the launch amplitude.

At a minimum you need a 1x2 pad to maintain the impedance on both loads. But ths will still cause a 3db drop in signal level. If you can't tolerate that, than a distribution amplifier is needed.

If you want to distribute video or RF and don't want losses via pads or resort to an active distribution amplifier - here's the trick.
You need to do a "loop through" topology. This was very common with analog video systems. Also 50ohm Ethernet "Thinnet" worked the same way.

If your devices have two BNC connectors wired in series for the external clock input, then you are set to loop through. If not, some modification will be required.

Remove the 75ohm termination resistor on both device's clock inputs. Then put a BNC Tee right at the 1st device connector. If it's an RCA jack, it's ok to use a BNC to RCA adapter, the impedance mismatch at this short distance is insignificant at 10mhz. Then run a RG58 cable from the clock generator to the first BNC Tee. Then another cable to the second device with another BNC Tee attached. Then pout a 50ohm terminator on the end of the last BNC. This will maintain the proper cable impedance as well as the proper signal level. You can repeat this loop for as many as 10 devices and over 100 feet of cable at 10mhz.

Note this assumes your clock source is 50ohms as is most likely the case with an atomic standard. If however it's 75ohms, just use 75ohm cable, RG59 or RG6, and terminate with a 75ohm terminator resistor rather than 50 ohms.

 

RE: Possible to split a 10Mhz clock output?, posted on December 23, 2016 at 19:57:03
Posts: 16
Joined: December 28, 2007
It should still work, but the use without a clock distribution circuit for more than 1 output will cause a change in the load that the rubidium oscillator sees and could possibly cause some issues with the module itself.

You do really need to see the output, which typically is 0.5v. Voltage will drop with a t-piece in place and some equipment may not lock properly.

 

Page processed in 0.031 seconds.