• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Is a master clock needed?

Keith_W

Major Contributor
Joined
Jun 26, 2016
Messages
2,937
Likes
6,808
Location
Melbourne, Australia
Yes, I know that a master clock makes no difference if you are using a single digital device, like a DAC. However, my question is about multiple digital devices.

I read this article (in a pro audio blog) about whether your studio needs a master clock or not. My take-away from that article was contained in this quote:

SoundonSound said:
In most compact project studios, there's little need for a master clock. The required system clocking can usually be achieved by interconnecting the equipment directly and, as explained above, where there's only one A‑D in the system, it's generally best to use that as the clock master anyway.

Typically, with a stand-alone A‑D configured as the master, its digital output would be passed on to the audio interface as either an S/PDIF or AES3 signal (for a stereo A‑D), or an ADAT signal (for a multi-channel A‑D). All of those protocols include embedded clock information which the interface can be configured to accept via the appropriate audio input as its slave clock reference. Alternatively, a word clock output could be taken from the A‑D and connected to a word clock input on the DAW interface (remembering to ensure the correct 75Ω termination is in place — see the 'Interface-induced Jitter' box), with the interface set to use the external word clock as the slave reference.

However, in more elaborate and expansive systems, where there are several A‑Ds and lots of other digital outboard, it's often more convenient and practical to have a centralised master clock source, and to distribute clocks from that to all of the other devices, all of which are configured as slaves. All master clock units provide numerous word clock outputs, and often several AES11 clocks too (AES11 is basically a silent AES3 signal, intended specifically for clocking purposes). In this kind of system, though, it would be worth ensuring that the A‑D converters all work well when operating on external clocks, to maximise their audio quality.

The only situation where a dedicated master clock unit is truly essential is in systems that have to work with, or alongside, video, such as in music-for-picture and audio‑for‑video post‑production applications. It's necessary here because there must be a specific integer number of samples in every video picture‑frame period, and to achieve that, the audio sample rate has to be synchronised to the picture frame rate. The only practical way to achieve that is to use a master clock generator that is itself sync'ed to an external video reference, or which generates a video reference signal to which video equipment can be sync'ed.

The application is for a friend of mine who has a system configured like this:

- Turntable --> Phono stage --> ADC
- ADC --> MiniDSP
- MiniDSP --> 3 Topping DAC's --> rest of the system

My understanding of using multiple digital devices is that each device might latch on to the signal at different times, and variations in clock accuracy between each device might cause clock drift over time. After some time, the difference in timing between DAC's might become audible, particularly if one DAC is driving the tweeter, and another DAC driving the midrange, etc. When I was configuring my own system, a friend of mine who is an audio engineer told me NOT to use multiple DAC's for multichannel digital output because of clock drift. Or if I wanted to, I had to slave them all to a master clock. Because of his advice, I purchased an 8 channel DAC.

However, that article I linked to mentions that a master clock is not needed in "simple" studio setups, and is only essential for complex setups involving multiple ADC's or if video is involved. Because that is a pro audio blog, I am guessing that they do not sit down for an hour listening to a single album played from start to finish, so clock drift may be less of an issue for a "simple" studio where they stop and start tracks which will give all digital equipment in the chain a chance to resynchronize.

As far as I am aware, my friend's ADC (I don't know what brand) and his Topping DAC's do not have clock outputs or inputs, so it would be impossible to slave the DAC's to the ADC, or even slave all the digital devices to an external master clock. I suppose this may not a problem if he was using a digital source, because the signal would stop and start at the beginning of each track, meaning that the DAC's would have an opportunity to resynchronize. But he is using vinyl, which means noise might be transmitted to the DAC's even between tracks, so the DAC's might not have an opportunity to resynchronize.

I did some "back of the napkin" math, and this is what I came up with. Assume we have a DAC with a deviation of 50ppm, and a "worst case" scenario where the difference between the first and second DAC is 50ppm.

- 44.1/16 * 2 channels = 44100 * 16 & 2 = 1,411,200 bits per second
- DAC clock runs at double speed = 1,411,200 * 2 = 2,822,400 cycles per second (or 2.8MHz)
- 50ppm variability at 2.8mHz = 50/1,000,000 * 2.8224 = 0.00014112 seconds (or roughly 0.1ms) every second.
- Clock drift in 60 seconds = 6ms
- Clock drift in 1 minute = 360ms
- Clock drift in 30 minutes = 21,600ms (or 21.6 seconds)

I did not study maths beyond high school so there is a very high probability that I made a mistake in my math. I would appreciate correction, because clock drift of 21.6 seconds at 30 minutes seems astoundingly high to me. I am no match for you engineering types! So please be nice to me if I got my math wrong! Also, according to my calculations, the difference at just 1 minute of playback is 360ms so it should easily be audible. I subjectively did not hear anything amiss after listening for several minutes.

Advising him to change all his equipment to allow slaving to a master clock would be a major expense for him, so I want to check with ASR whether there is any truth to the assertion that clock drift between DAC's can cause group delay, what the magnitude of the problem is, and what it can potentially add up to over time. I do not wish to give bad advice, so your input is welcome.
 


JSmith
 
The application is for a friend of mine who has a system configured like this:

- Turntable --> Phono stage --> ADC
- ADC --> MiniDSP
- MiniDSP --> 3 Topping DAC's --> rest of the system

I don’t think there are Topping DACs that can take external clock input?

My understanding of using multiple digital devices is that each device might latch on to the signal at different times, and variations in clock accuracy between each device might cause clock drift over time. After some time, the difference in timing between DAC's might become audible, particularly if one DAC is driving the tweeter, and another DAC driving the midrange, etc. When I was configuring my own system, a friend of mine who is an audio engineer told me NOT to use multiple DAC's for multichannel digital output because of clock drift. Or if I wanted to, I had to slave them all to a master clock. Because of his advice, I purchased an 8 channel DAC.

I didn’t want to buy an 8-channel DAC for my DLBC setup (4 channels), but my two DACs are connected to a Mac which can do drift compensation for devices that aren’t hardware synced.

- Clock drift in 1 minute = 360ms

- Clock drift in 30 minutes = 21,600ms (or 21.6 seconds)

That’s a lot! I wonder how that would sound. I suppose you’d get drop-outs because of buffer over/underruns?

I could try running my DACs without drift compensation, but how do I know they actually drift apart, what should I listen for?
 
Mistake in your math. Assuming 50 ppm timing mismatch you have:
30 min x 60 second/minute or 1800 seconds.
1800 seconds x 50/1,000,000 is .09 seconds.
You might hear that as a mismatch for sure.

Firstly why are you saying clock runs at double speed? The 2nd mistake is you have bits and samples mixed up. You still only have 44,100 sample periods per second. You could go about it this way in samples per second, and you still end up with .09 seconds from 3969 samples which 3969/44,100 samples per second is the same answer of .09 seconds for 30 minutes.

44,100 samples/second x 60 seconds/minutes x 30 minutes =79,380,000 samples.
79,380,000 samples x 50/1,000,000=3969 samples.
3969 samples/44,100 samples/second=.09 seconds.

Also noise between tracks has nothing to do with this.

The ADC clock via Toslink or coax SPDIF feeds the miniDSP. The miniDSP unit has to synch up to the incoming SPDIF signal, and if it is feeding three Topping DACs via SPDIF, they'll all synch via the SPDIF and all of them are synched to the ADC clock all the way through. You don't have to have separate clock lines in such a case. You won't have any drift.
 
Last edited:
Mistake in your math. Assuming 50 ppm timing mismatch you have:
30 min x 60 second/minute or 1800 seconds.
1800 seconds x 50/1,000,000 is .09 seconds.
You might hear that as a mismatch for sure.

Firstly why are you saying clock runs at double speed? The 2nd mistake is you have bits and samples mixed up. You still only have 44,100 sample periods per second. You could go about it this way in samples per second, and you still end up with .09 seconds from 3969 samples which 3969/44,100 samples per second is the same answer of .09 seconds for 30 minutes.

44,100 samples/second x 60 seconds/minutes x 30 minutes =79,380,000 samples.
79,380,000 samples x 50/1,000,000=3969 samples.
3969 samples/44,100 samples/second=.09 seconds.

Also noise between tracks has nothing to do with this.

The ADC clock via Toslink or coax SPDIF feeds the miniDSP. The miniDSP unit has to synch up to the incoming SPDIF signal, and if it is feeding three Topping DACs via SPDIF, they'll all synch via the SPDIF and all of them are synched to the ADC clock all the way through. You don't have to have separate clock lines in such a case. You won't have any drift.

Thank you! I suppose even in the worst case scenario, and two DAC's are off by 50ppm in the exact opposite direction, we would end up with 0.18 seconds after 30 minutes. Or 0.36 seconds after an hour. And probably nothing if the MiniDSP, as you say, syncs to the ADC and passes on the clock signal to the DAC's. I guess it is theoretically possible, but not anything to worry about. Just a question, how do you know that the MiniDSP syncs to the ADC? Are all digital equipment designed to sync to the incoming signal? Also, does this hold true if he does not use SPDIF? For example, USB?
 
I don't know for certain. How does the Adc connect to the minidsp and which minidsp unit is it? Does the minidsp connect to the Topping dacs via spdif? If it's spdif or toslink all the way through it is in the format that the receiving device synchs to the incoming signal as it would not work otherwise. PLLs are used to stay in time with the incoming signal. So the 1st device on the chain becomes the master clock.
 
Ok, I got to ask: why bother with turntable if going adc?
Also why not use a minidsp with multichannel outs, why use 3 Topping dacs?
To me this seems like another unnecessarily complicated pathway? How much cabling with all this?

Sometimes we create the problem for the solution
 
Ok, I got to ask: why bother with turntable if going adc?
Also why not use a minidsp with multichannel outs, why use 3 Topping dacs?
To me this seems like another unnecessarily complicated pathway? How much cabling with all this?

Sometimes we create the problem for the solution
For years I used an adc to handle various analog media. It feed a nice dac with volume control on the output. Works just fine and eliminated an analog pre amp.
 
Are all digital equipment designed to sync to the incoming signal? Also, does this hold true if he does not use SPDIF? For example, USB?

I think that spdif has a clock signal embedded in the data - it is self clocking, which can drive downstream consumers.

Not so with usb audio, as that is “pull” based: sinks pull from sources.
 
I don't know for certain. How does the Adc connect to the minidsp and which minidsp unit is it? Does the minidsp connect to the Topping dacs via spdif? If it's spdif or toslink all the way through it is in the format that the receiving device synchs to the incoming signal as it would not work otherwise. PLLs are used to stay in time with the incoming signal. So the 1st device on the chain becomes the master clock.

Thank you. I have pointed him towards this thread. If he doesn't register on ASR and answer the question himself, I will ask him and get back to you.

Ok, I got to ask: why bother with turntable if going adc?

Because he has about 1,500 - 2,000 records which he has collected all his life. If he digitizes each record, that would be 60 minutes each. Or 1,500 - 2,000 hours of ripping. Would you like to volunteer? :) He needs an ADC because he is using a digital crossover for his DIY speakers.

Also why not use a minidsp with multichannel outs, why use 3 Topping dacs?

I don't know the answer to that question. I was not involved in that decision. I suggested that the cheapest way to avoid the clock drift issue would be to use the ADC and DAC's built in to his MiniDSP. He might have reasons to do so, but I am not going to defend them on ASR. So let's just say it is his preference, or belief that his route delivers superior quality, and leave it at that. I do not feel like getting into a DAC debate with him or ASR.
 
Since the DSP has its own clock domain, any incoming data is resampled to that. The DAC are then fed from the DSP clock domain via SPDIF. Since SPDIF contains the clock, no additional sync is needed between the DAC’s.
 
Since the DSP has its own clock domain, any incoming data is resampled to that. The DAC are then fed from the DSP clock domain via SPDIF. Since SPDIF contains the clock, no additional sync is needed between the DAC’s.

I don’t think clock syncing is an issue with serialised digital devices, there is usually buffering between them anyway. For devices acting in parallel on the other hand…
 
Because he has about 1,500 - 2,000 records which he has collected all his life. If he digitizes each record, that would be 60 minutes each. Or 1,500 - 2,000 hours of ripping. Would you like to volunteer? :) He needs an ADC because he is using a digital crossover for his DIY speakers.
Ah, that makes sense. Wow, thats some collection. Dont mention we have more on tidal/qobuz and get lots of other data with roon...
Someone figured out how to get vinyl playback through roon dsp as well.

If I remember correctly Merging has a simple but expensive solution for multichannel active setups with analogue inputs
 
I have signed on - 'so Keith doesn't have to answer for me. :)

Sorry, I can't see how to quote individual posts - so I'll just have to copy various people's text into my post - and then reply:

@AudioJester asked: "why not use a minidsp with multichannel outs, why use 3 Topping dacs? To me this seems like another unnecessarily complicated pathway?"

Using the miniDSP 'nanoDIGI' - a digital only device - allows me to use external A2D & D2A converters which are of higher quality than those which miniDSP use in their products. (My nanoDIGI was an upgrade from the 10x10HD which I used initially.)

My mate suggested I buy 3x Topping E30 DACs, after their excellent write-up on ASR.

@Blumlein 88 wrote: "How does the ADC connect to the minidsp and which minidsp unit is it? Does the minidsp connect to the Topping dacs via spdif? If it's spdif or toslink all the way through it is in the format that the receiving device synchs to the incoming signal as it would not work otherwise. PLLs are used to stay in time with the incoming signal. So the 1st device on the chain becomes the master clock.".

No, unfortunately - since starting on a 'ripping' journey ... it is no longer SPDIF/TOSLINK all the way. :(

Previously, I had:
  • phono stage --> A2D converter --> 4-way digital source selector --> nanoDIGI --> 3x Topping E30s ... all connected with coax SPDIF cables.
However, now that a computer is involved - for ripping LPs and playing them back from SSD, via Roon, I now have:
  • phono stage --> A2D converter --> 4-way digital source selector ... connected with coax SPDIF cables.
  • 4-way digital source selector --> SPDIF-to-USB converter (a miniDSP 'USBstreamer') --> PC (as the PC only works with USB input, not SPDIF).
  • PC --> USB-to-SPDIF converter (DDHiFi TC100-COA) --> nanoDIGI --> 3x Topping E30s. So the nanoDIGI and the E30s are connected with SPDIF.
Hence Keith's concern about clocking.
 
Someone figured out how to get vinyl playback through roon dsp as well.

Yes, I think he was aware of that. Something about Icecast. I couldn't really understand what he was saying, but then I wasn't really paying attention.

If I remember correctly Merging has a simple but expensive solution for multichannel active setups with analogue inputs

They do? What solution is that?

BTW if I wanted to, I also have a solution for multichannel active setups with analogue inputs. I have an RME Fireface UC, which is what I use for my 8 channel active system with digital crossover. It has an ADC, 8 DAC outputs, mic preamp, and several digital inputs. It was quite pricey, about USD$1400. Though probably not as pricey as what Merging would charge (I know, I also have Merging in my system).
 
Last edited:
I don’t think clock syncing is an issue with serialised digital devices, there is usually buffering between them anyway. For devices acting in parallel on the other hand…
Yes, there is. The ADC clock is not the same as the DSP clock. You'll need to convert those clock domains.

Parallel is easier in this case. The signals already have the clock embedded, and the DACs just extract and stabilize that. The result will be three DACs that remain in sync.
 
Aah, @AudioJester - your comment "Someone figured out how to get vinyl playback through roon dsp as well" ... is another important question! :rolleyes:

I am using Roon to play my vinyl because I understood it's possible to use its DSP capabilities to put a "FIR filter overlay" over the top of the IIR filters that the nanoDIGI provides. This was implemented a few days ago and it is indeed an upgrade to my SQ!

I was told that Roon would have to pull the ripped vinyl files from hard disc - hence the requirement to rip my vinyl.

However, I have been made aware of a method whereby Roon is able to play my digitised vinyl "on the fly" - so ripping them to disc is not required! :D

This is to use Icecast to set up a radio station on my PC - which Roon can accept as a "Local Radio" station! Brilliant! But do you know how to set this up?
 
Yes, there is. The ADC clock is not the same as the DSP clock. You'll need to convert those clock domains.
I read this often here but still struggle to understand it. Anyone care to explain or link to where it is explained? Thanks a lot!
 
Yes, there is. The ADC clock is not the same as the DSP clock. You'll need to convert those clock domains.

Parallel is easier in this case. The signals already have the clock embedded, and the DACs just extract and stabilize that. The result will be three DACs that remain in sync.
To add to that slighlty. The dacs will remain in sync, but if you want them also to have the same delay, the dacs should be identical.

Different dacs would likely have different PLL buffer depths which would result in a different latency through them. Sound through one would be time shifted with respect to the other.
 
Back
Top Bottom