Word clocks, any experience?

[Reid Malenfant]

Having done a little bit of reading i think i'm beginning to understand why these word clocks can make a good bit of difference to reproduction of digital signals (in my case CD).

It appears that anything daisy chained in a digital system is going to receive & extract the clock from the data stream & this already has a delay due to extraction & cable length from the source & this is of course the dreaded jitter :eek:

An accurate word clock can syncronize everything as long as equal length cables are used from the word clock itself to all digital components.

Has anyone here had experience of using one of these? I can only ask ;)

[Welder]

I’ve not tried one Mark but isn’t it a bit of an overkill for a single Dac and source replay system?

My understanding is they may be useful for controlling jitter in multiple processing system such a recording studio by acting as a master clock for multiple slave clocks but most Dacs these days have accurate enough clocks for jitter to be negligible.
If the incoming data is buffered and stacked and clocked out to analogue conversion by the Dac clock rather than the source clock then isn't jitter in the incoming data stream is rectified at the buffer :scratch:
If jitter in a two clock set up is a problem then wouldn’t it make mores sense to just get a better Dac with a more accurate clock?
Idk, maybe the above is only relevant to USB data transfer.

Intersting though.

[Reid Malenfant]

Hi John, hope you are well mate ;)

You should know i'll be going to a three point system, Krell DT10 transport, dCS Purcell upsampler & then to a dCS Delius DAC. That looks a bit like a daisy chain to me :eyebrows:

I'm simply asking, i have no knowledge of these things myself, but from what i have been reading a word clock makes some sense. dCS actually sell one, the Verona :)

Now just because i have stepped onto the dCS bandwagon doesn't mean i'm a mug. What i'm wondering is if i can get my Krell DT10 modified with a superior clock & then buffer that & output it to the other kit?

There is method in my madness, maybe :lolsign:

[Welder]

Hey Mark, I’m good thanks :)

Right, I see ;) Maybe if you’re going to have multiple clocks then there may be some benefit. I have no idea tbh. I’m just going on what I’ve read.

I must admit I get a bit confused by all digital ailments being blamed on jitter.
While it may be, from what I understand, a more serious issue using SPDIF it seems like well implemented USB data transfer reduces jitter to really negligible amounts and the problems in fact lie with the analogue stage which tends to get neglected a bit on a great many Dacs :scratch:
Looking at various jitter specs for Dacs (if you can find them) and then reading what the experts write concerning what amount of jitter will adversely affect the analogue output. It seems that many USB Dacs achieve what can be considered as negligible jitter :scratch:
The problem is, very few people have the equipment to actually measure jitter and any problems just get attributed to it :(

Idk mate. I remember when CD players came out and there wasn’t all this concern about jitter then. CD players are just a transport and Dac after all.
As soon as separate Dacs and computers joined in jitter was what the Hi Fi press used to slag computer audio off with. Strange really.

[Audiocom AV]

Hi John, hope you are well mate ;)

You should know i'll be going to a three point system, Krell DT10 transport, dCS Purcell upsampler & then to a dCS Delius DAC. That looks a bit like a daisy chain to me :eyebrows:

I'm simply asking, i have no knowledge of these things myself, but from what i have been reading a word clock makes some sense. dCS actually sell one, the Verona :)

Now just because i have stepped onto the dCS bandwagon doesn't mean i'm a mug. What i'm wondering is if i can get my Krell DT10 modified with a superior clock & then buffer that & output it to the other kit?

There is method in my madness, maybe :lolsign:

Hi Mark

Does the Krell DT10 have a word clock input?

If so you can use a clock sync generator to the DT10 and DCS DAC.

Cheers,
Mark

[Vincent Kars]

Sound on Sound has an article about master clocks
http://www.soundonsound.com/sos/jun10/articles/masterclocks.htm

[Audiocom AV]

Sound on Sound has an article about master clocks
http://www.soundonsound.com/sos/jun10/articles/masterclocks.htm

Thanks for the information.

This article does not cover the real issue of locking the source of the SPDIF to the receiver with a common clock.

[Lodgesound]

Hi there;

Work clock generation is fine except that you will need a very stable source for the clock itself.

In my studio I use a broadcast SPG (sync. pulse generator) to feed the word clock generator's reference input. The SPG provides a stable reference to 6 decimal places of colour subcarrier (4.433619 Mhz) and thus the clock is extremely stable.

In professional situations such stability is essential for maintaining absolute sync. across a huge variety of production platforms where a standard speed must always be adhered to - to be honest in the home it could be a little OTT however as you say having all devices locked to a common reference would reduce jitter to a minimum.

The cost of the above equipment on the second hand market is around £1000-£1500 depending on the type of SPG in use.

[Reid Malenfant]

Hi Mark

Does the Krell DT10 have a word clock input?
No, but i reckon it shouldn't be too difficult a modification to give it one ;) Effectively it'd be a case of selecting internal or external clock, that may prove to be a tad simplistic though when it come down to it :eyebrows:


Cheers for all the input so far guys :) I'm keeping an open mind obout it but i definately need to implement an external word clock if i use the Purcell to convert to DSD to feed the Delius. There is a difference between the clock in the Krell & the one in the Delius & this results in a pause in the proceedings every so often (10 minutes or so for a second) where it looks like the Krell plays catch up :scratch:

I reckon going over to AES3 will eliminate that as the kit will be receiving the clock from the Krell & locking to it rather than having to run the Delius in master mode & having two slightly different clocks :rolleyes:

[Lodgesound]

Can you not set the Delius to slave lock to the Krell?

[Reid Malenfant]

Ummm, the Krell does happen to have a clock output, the only problem is is that it's an optical AT&T connector & all the word clock connectors on the dCS kit are 75 Ohm BNC connectors :lolsign:

Not so easy as it at first sounds :doh:

[Lodgesound]

Understand that but is it possible to set the Delius to just lock to it's input rather than to the word clock?

If it is trying to lock to a reference that is'nt there this would explain your occasional muting problems - setting it to lock ONLY to it's input will solve this problem.

[Reid Malenfant]

I understand what you are getting at, but to run DSD via firewire (IEEE1394) the Purcell upsampler needs to be locked to the Delius, so it's kind of arse about face. So the problem is the difference between the Delius clock & the Krell clock but at the Purcell upsampler & not the DAC.

I have had an idea though that might just work :scratch: If i take a look at the wordclock waveform it shouldn't be too difficult to reproduce that in the Krell from it's own clock & give the Krell a couple of clock outputs via BNC connectors & lock both Purcell & Delius to it.. With any luck it might be damn near exactly the same as that feeding the optical (AT&T) converter which would make life very easy indeed :)

[Reid Malenfant]

Jeez, i figured something was a bit of a cock up (not clock up :eyebrows:) so i wondered if i could slave the Delius to the Purcell, after all the Purcell will lock itself onto the datastream from the CD player.

When i originally connected it up i followed the wiring diagram in the Delius manual, but i thought i'd now take a look in the Purcell manual...

What did i find :scratch: :rolleyes:

I found the opposite of what i saw in the Delius manual :steam:

Looks like the problem is about to be solved, i just need to connect the word clock out of the Purcell to the word clock in of the Delius :cool: Apparently the IEEE1394 connection produces a lot of jitter & the word clock connection syncronizing them sorts it out. Something tells me the sound quality is going to improve a tad & these occasional drop outs will be a thing of the past :)

Fancy messing up the diagram in one of the manuals :mental:

[Reid Malenfant]

Right, that's sorted it :) In more ways than one though as all of a sudden i need to get to know my whole CD collection again :lolsign:

Stuff buying a word clock, this is more than good enough & frankly far better than i was ever expecting 3551

[Steve H]

When i originally connected it up i followed the wiring diagram in the Delius manual, but i thought i'd now take a look in the Purcell manual...

What did i find :scratch: :rolleyes:

I found the opposite of what i saw in the Delius manual :

Mark, the Delius manual shows the word clock out to a dCS transport. In that config, the Delius clock bypasses the Purcell. If a non dCS transport cannot take a word clock signal, then dCS requires the Purcell to deliver the clock to the DAC, which is what you now have.

To my ears, a dedicated clock delivers a subtle change but one which is hard to live without once you've heard it. Equally, in your system, I'd upgrade to an Elgar Plus before buying the clock.

Steve Dickinson reviewed the Puccini in HiFi Plus and, whilst praising it, questioned the quality of the analogue output stage. Later, he reviewed the USB Word Clock with the Puccini and felt his questions about the analogue output stage were answered. The review is on the dCS web site.

Steve

[Reid Malenfant]

Hi Steve, cheers for the reply & the information ;)

The combo has been working perfectly since about 4PM, i haven't stopped listening to stuff with vocals since then as i want to make sure i get a bit of a grasp of the things i'm hearing :eyebrows:

My last DAC had a pure class A output stage & these things are a reasonable leap up the evolutionary chain of DACs etc, if this is a poor output stage then i'd like to hear it with a good one :lol:

Warranty runs out in less than 2.5 months so i might take a look inside & see what i can see ;)

[amdismal]

A master clock makes no sense when you are changing sample rate from the 44.1k of the Krell to 192k from the Purcell. The Purcell should lock onto the Krell at 44.1k, and the Delius onto the Purcell at 192k.

If you want to link clocks, you don't need a master in this setup - you need to clock out from the Delius to Purcell at 192k, and 44.1k from Purcell to Krell. Unfortunately the dcs kit does not have this functionality.

[Reid Malenfant]

Actually it would make some sense if the Krell had a word clock input, but it hasn't ;) One clock which has the same length cables feeding to each bit of kit would result in everything clocking at precisely the same time with zero delays between each bit of equipment.

You are correct that the Purcell has locked onto the Krell, but the Delius has locked onto the word clock output of the Purcell at 44.1KHz. The output from the Purcell isn't 192KHz, it's actually 2.822MHz (using the firewire connection & DSD output) which if you do a bit of math is precisely 64 times 44.1KHz :)

[amdismal]

Oh my Purcell was non-firewire, so I used 192 into my (also non-firewire) Elgar.

I must say that I was never convinced by the theory of the master clock. It clearly makes sense in a pro environment, where you have all sorts of components and they do need a single clock. But when you only have one source and limited or no processing, then the theoretical best place to have the clock is right by the DAC, and to have the transport act as slave to this. This is a perfect solution, whereas a master clock is less so, as both DAC and transport have to deal with clock recovery.

But that's theory, and practice is always different: presumably they engineered out the issues with clock recovery. I never got to try a clock with my setup, but I would be keen to try it.

The other problem with using master clocks is much more important now than it was: the different sample rates of source music. I know there was a thread on the Slim Devices website where there were all sorts of problems playing 88.2 and 96k source material at half speed, as the clock was a 44.1k one, and the Transporter was therefore transferring at that speed! You'd need some metadata transferred to the master clock (or remote control, I guess) to control the speed. I imagine that USB async in modern DACs sorts this.

[Reid Malenfant]

Oh my Purcell was non-firewire, so I used 192 into my (also non-firewire) Elgar.
I'm going to try out the dual AES3 connection so i can go & upsample to 176.4 KHz & 192.. I can leave those connected & effectively decide which i want to connect to via the Delius. I'd have loved the Elgar, but it just didn't make sense that they made the thing a different shape :scratch: Not so important in all honesty :eyebrows: But a consideration due to the Elgar being more expensive as well, & none are cheap :eek:

Must say i'm absolutely loving what the combo is doing, i'm sure you know yourself :eyebrows:

Not a huge audience for these due to the cost, but worth every single penny imo, i wish i'd seen the light earlier ;)

[Steve H]

[QUOTE=amdismal;
You'd need some metadata transferred to the master clock (or remote control, I guess) to control the speed. I imagine that USB async in modern DACs sorts this.[/QUOTE]

The dCS USB clock is manually switchable between 44.1 and 48Ks/s

Mark, I wondered about the Elgar's different shape but it's a really elegant thing and it has a liquidity of sound which I thought the best thing I'd ever heard. Also, the granite top matches perfectly with Vertex AQ Kinibalu supports.....;)

[Reid Malenfant]

Hi steve, cheers for the reply :)

Trust me i'm in audio heaven right now :eyebrows: I'm using an Oyaide IEEE1394 firewire cable of 60cm length between the relevant ports. Standard 2M or whatever it is gold plated BNC cable for the word clock between the Purcell & Delius....

I'm going to get a shorter BNC to BNC cable (cheap as chips on fleabay) & see if it makes any difference.

This setup makes the older Monarchy Audio dual 20 bit DAC sound rather dated, there is an incomparable difference. Somewhat like from going from 2D to 3D & i'm not comparing things to these cack 3D TVs out there ;)

Apart from my upgrade to Krell DT10 transport this has been by far the biggest improvement in sound quality & imaging i have ever encountered :)

Loving it :lolsign: All the best :cool:

[nat8808]

Can I ask a quick (haha!) question?

When you use a word clock generator does it have to provide different clocks for different sample rates? Or does the slave work out what it needs from just one incomming clock rate?

I have a studio word clock that outputs 256fs (don't really know what that means) so if I wanted to get a future DAC to slave to it but at different sample rates (e.g. 44.1 and 48 or say 96 and 88.2) would it work or would I need a more flexible word clock and have to keep switching clock rates to match the format?

For example, I use sources at both 44.1 and 48 and in the future will want to record/playback at 88.2 and 96 AND I have a format convertor with word clock in. Can I use my clock with 256fs outputs to be master of all sources and DAC?

I do have a 44.1/48 DAC with word clock input so perhaps I should experiment but does this work with 256fs?

[Reid Malenfant]

Hi Nat, it more than likely depends on the equipment it's feeding. For instance this dCS kit requires a 44.1KHz word clock & it converts from that for DSD which is 64 times the frequency.

I have seen studio type DACs that need different word clock frequencies when using different sampling frequencies. When i was looking into the word clocks i couldn't help but notice :eyebrows: If you take a look at something like the Apogee Big Ben i think you'll see what i mean ;) There were others i found & every one could output multiple frequencies (though obviously not at the same time) :)

I have a feeling that if your clock only happens to output one frequency you might have to upgrade, though don't take this as gospel, it'll depend on the DAC as well :cool:

[Audiocom AV]

Also try the Aardvark Aardsync, it was £1695 but can be purchase S/H for around £350.00.

[nat8808]

Also try the Aardvark Aardsync, it was £1695 but can be purchase S/H for around £350.00.

That's what I've got see! Aardsync II - £120 Ebay bargain.. That's kind of why I bought because of the price but err... don't know how to use it or if I can very well. Ebay prices are actually more regularly ~ £250 rather than £350 - whilst the Apogee Big Ben that was released only a year later for near the same price but with lots more features which implies lower quality in exchange for quantity, still fetches £600+ no matter its age.

I reckon slaving my PC would be easy as semi-pro/pro cards are nice and flexible. The dCS I have might be ok too - will have to check. I guess it's the future dac I'm yet to buy will be the difficult one, especially as I wanted to go DIY - a word clock receiver may not be something readily available to add to a DIY project, let alone one for my TASCAM CD700 transport.

I have too seen clocks with multiple outs like the Big Ben so perhaps something with multiple outs is what I need.

The Antalope gear looks very tasty at £5000 plus... with an atomic clock add-on!

[Reid Malenfant]

The Antalope gear looks very tasty at £5000 plus... with an atomic clock add-on!
I saw something earlier today with a Rubidium clock, just wish i could remember what the hell it was as i might be able to find it again :doh:

[nat8808]

Was it silver and shiney with bright numerical display in the centre? Yep, that's the Antelope Rhobidium Atomic clock add-on to their sync clock generator.. £4300 list price add-on to their £2400 universal clock.

[nat8808]

Just found the (brief!) Aardsync II manual online..

Seems I can simultaniously output say 44.1 and 88.2 and 256xFS (which looks like it is 11.28 MHz) or 48 and 96 and 256xFS (does this 256xFS still remain at 11.28MHz?) but not output different base frequencies at the same time like 44.1 AND 48 - for that the manual shows two Aardsyncs synced themselves to an external clock generator...

So, looks like I have to manually switch between 44.1 and 48 (and so their 88.2/96 multiples) from the front panel.

With my current set-up, upsampling with the dCS may well be limited to 88.2 from CD unless it can derive its own clock multiples from 256xfs.

[firedog55]

Sound on Sound has an article about master clocks
http://www.soundonsound.com/sos/jun10/articles/masterclocks.htm

The Sound on Sound article is about synching devices using a master word clock. It is correct as far as it goes, but if that is not what you are talking about, the conclusions of the article don't necessarily apply.

I asked Steve Nugent of Empirical Audio about the article, as he makes a "master clock" device for home audio. This was his response:
Studio master clocks are not the same as how my products work. They are generally used for synchronizing events for editing. They are generally not used for playback. His article confuses the word-clock with the frequency I call master clock, which is generally 128 or 256 times the word-clock. Some of the things he says about master clocks are true, like with the A/D converter. However, all of the "master-clocks" that he tests are actually word-clock generators.

The reason that I use the term "master-clock" is that this is the term that is used at the D/A chip. Unfortunately, the word also describes a system word-clock in a studio environment. Totally different.

The tests that he performs are interesting primarily for studios. IMO, providing even a superb low-jitter word-clock is not very effective for reducing system jitter. The clocks need to be low-jitter true master clocks at 256X the word-clock frequency. This is how my devices work. When you provide only word-clocks, then the device must synchronize its internal clock to this and then the jitter is a function of the internal high-frequency clock, not so much the external word-clock.

[DSJR]

For general playback, wouldn't a buffer do it all, retaining so many frames and then "clocking" them out to the DAC, or are DCS so above this sort of thing that they have to make it complicated? :lolsign:


I dunno, a QED Digit with two supplies worls so well it staggers me sometimes - an excellent upgrade and still "musical" sonics...

[sq225917]

Yeh, if your dac, the last part in your chain buffers the signal, and removes jitter then why bother with a clock?

[Reid Malenfant]

:lolsign: I have no idea other than the fact that the clock is much more accurate timing wise. Jitter reduction will only ever be as good as the timing in the DAC.

Frankly i really don't know what's going on here now in my system. The dCS kit should be slaved to the transport but it isn't as the transport isn't dCS kit. Reading the dCS instructions it states that there may be pops & clicks due to dropouts if this isn't done :scratch:

No such animal here i'm afraid. The odder thing is the Krell CD transport is now hooked to a Monarchy Audio DIP 48/96 upsampler that is outputting 24 bit 96KHz & this then feeds the Purcell which converts that to DSD (44.1KHz 64x fs)...

The Purcell is feeding a word clock signal to the Delius to sync the two though, it's the bit near the beginning that has me scratching my head..

Still no pops & clicks & better sound by a noticeable margin :confused:


I also discovered that the Verona master clock can sync to a GPS satellite & thus use the onboard atomic clock as long as a GPS clock is purchased & an aerial.. Didn't know that ;)