External Word Clock

[Y-my-R]

Yeah... if it works in slave mode, then I'd just clock the Behringer from the D8B or HDR (whichever gives the best results), and use it the way you suggested.

While that article I linked to explained why it's usually preferable to use the clock in the A/D you're planning to use, it also made a point of the quality difference mostly happening at a very low threshold that shouldn't be "obvious" - and pointed out that something like a less-than-ideal microphone placement would have a MUCH bigger impact on the resulting quality.

So, not really THAT big a deal, and I'd just use the Behringer in slave mode, in that case.

(Or try some contact cleaner in that switch, after all. Maybe it's just dirty).

[Old School]

Hi,
One final question. I don't see a bit depth selector on any of the master clocks I've looked at, are the master clocks all 24 bit?

Have a blessed day,
Mike

[Y-my-R]

From my understanding, the clock signal itself carries no "data" in that sense, which should mean that there is no "data package size" such as 24-bit or 16-bit.

It's just a representation of a 0 and 1, for "Pulse on" and "Pulse off" that is supposed to happen at a fixed and stable interval of 44100 pulses per second for the 44.1 kHz sample rate and 48000 pulses for the 48 kHz sample rate, etc.

So, in terms of clocking, the bit rate of the audio signal being transmitted over "separate" digital audio cables like ADAT optical, S/PDIF or AES/EBU shouldn't matter.

I'm just assuming this from my general understanding of the topic, and never really read anything about it, specifically, though. So, if I'm wrong, somebody correct me, please.

On the Lucid GenX192 I have, there also is no bit depth setting, though. I'm happy with that one, btw, but have been using it only as a clock-distribution device, lately, with my Apollo 8 providing the clock signal.

(I got that one in a trade for a "less-than-$200" value (...traded a 100' 16-channel XLR stage-box with strain relief for hanging from the ceiling for it). I saw those pop up for under $200 in other places, too (Reverb, Craigslist, etc.)).

[doktor1360]

Hi,
One final question. I don't see a bit depth selector on any of the master clocks I've looked at, are the master clocks all 24 bit?

Have a blessed day,
Mike

Hey Mike...

I've been (successfully) using a Rosendahl NanosyncS for about two years now (late Nov 2021). I don't know what your budget is, but I picked one up for $125 - the front of the unit needed some straightening on one of the rackmount ears... nothing I couldn't straighten out with 3 taps of a heavy hammer and a hard surface - it works perfectly... but I digress...

I use it as a Master Clock. it provides clock syncing for a Cranborne 500R8 and a Mackie 800R preamp, and has five additional bnc coax outputs for any future expansion (if I need clocking for some other gear). It's completely configurable, and solid as a rock. The one I linked below is less costly than everything I pulled up doing a search this morning...

https://www.ebay.com/itm/266384578575?hash=item3e05c2180f:g:ffUAAOSwQ0Bk5lsg&amdata=enc%3AAQAIAAAAwOaq0rylRZoz23Ygbkt4%2FAHcxrv55kFJtGE53Y1mN%2B2siQEl86L6%2FFg%2B8YxU0FQm%2BhBLKMgXFBi1zj5COZ8xz%2BNHoQyU%2F6mUYegvSASDeRN%2BFGTxnokFYGkuRjIgCd6EpGE2wXoAQfvwFrnZNmjNbrInzVsQ9%2BUPWFnPF8R%2F%2FrPN0aHZhL%2BuvAUrB7gFG8dEB2pRW%2Bxnss%2BCRYwoxj%2F3W%2BKL59JtcNm5RaL%2BmasA0Eq3fKNAFbFDTQZE%2B4tSPdMgBA%3D%3D%7Ctkp%3ABk9SR8Sq2dGsYw

It pays for itself pretty quickly, all things considered. I personally think this would work flawlessly for you, regardless of clock speed or bit rate depth requirements deemed necessary.

I trust you'll get everything worked out correctly...

Of course, here it comes...
[Standard Mgmt Disclaimer] - "Your actual mileage may vary..."

\w/

[Old School]

Hi Dok,
My budget is strained right now to say the least. The op-amps on most of my Furman HMS-16 studio monitor mixers are shot and I found out that they are soldered in and not snap in, (and there are 24 of them in each unit, I'm not even going to try to solder ONE of those tiny buggers to a PCB) so to replace this system with a modern equivalent will be be almost $5,000. Also I just spent $1600 on a D8B I bought that had been in storage for 14 years, so low mileage. The Rosendahl is a great clock, but I couldn't find one at that price. I bought an Aardvark Aardsync II for $60. It supposedly has a very low jitter of 100 picoseconds. We will see how it does. Oh, and I finally found out from a Behringer representative (just try contacting Behringer, they hide behind a wall of no phone numbers and no email) that the ADA8200 does NOT ouput BNC word clock. Did not all of us who looked at the rear panel of this device not get the impression that it did? Behringer says in their advertising that "this unit can be used as a master clock for your studio". I don't know of many studios who snyc things through adat connections only. I think that at best, this was false advertising and at worst, fraud. But I digress. I will let you know if the Aardvark lives up to it's hype.

Have a blessed day.
Mike

[magicbuss]

Hi, I've used those Behringer units. If I remember correctly they don't generate word clock via the BNC connector. It is just for receiving clock in slave mode. Took a bit of sorting to find that out. Works fine as a slave.

[magicbuss]

The Aardvark works well, I think I still have one here as well as an Aardvark Clock DA which gives you up to 6 clock outputs.
They are available if needed.
Good luck with your set up.

[csp]

Slightly off topic but not being an expert on clocking --- my system uses Optical sync and I have never had any problem doing this. I have no other item connected that requires sync.

If I have (as in my case) a d8b and Apogee clock card and an Alesis HD24 that has clock in only, irrespective of whether the two units are internally terminated or not and I want to use the d8b as the master clock, can I simply run a BNC cable from the Apogee output to the HD24, or do I have to use a 75ohm terminated T-Piece at either one or both ends?

Naturally setting the d8b as the clock "master" and the HD24 to "word clock".

David

[Y-my-R]

The general rule is:

- Dedicated Word Clock BNC connections need to be terminated via a 75 Ohm Terminator on both ends. If daisy-chaining devices, only devices at the very ends of the chain should be terminated with a 75 Ohm Terminator. Devices that aren't at the end of a chain (i.e. a device in the "middle" if daisy chaining 3 devices) should NOT be terminated.

- Some devices have termination built-in, on their BNC connectors. Some don't. Check the manual, and if it doesn't say (which they VERY often do not), contact the manufacturer if they're internally terminated or not.

- Some devices allow you to switch the built-in termination on or off as needed. For example, the Mackie HDR/MDR has a button for this on the rear panel.

- Some other devices do NOT allow to turn termination on or off and always have it on. The Apogee Clock card, for example, is such a device. You would have to physically modify the card, if you would want to use the Apogee Clock card at any other position in a daisy-chain than the very ends of it, because if it were positioned somewhere "in the middle" the termination should not be active, and will likely cause problems if it is.

- MANY other devices are NOT internally terminated but also do NOT allow you to turn the termination on. This makes it easy to use such devices "in the middle" of a word clock daisy chain, since you don't have to do anything to use them that way. However, if you want to use a non-terminated device with a BNC word clock connector at the very end of a word clock daisy chain, you need to add a T-Piece and a terminator.

- Digital audio connections that also contain the audio data (and not ONLY word clock) do NOT give any termination options. So, none of the above applies to ADAT, S/PDIF, TDIF or AES/EBU connections.

- As mentioned in an earlier post above, most digital audio devices that connect to a computer-host (e.g. most computer audio devices), allow you synchronize to an external, incoming embedded word clock signal that is part of the ADAT and S/PDIF and TDIF and AES/EBU data stream. Many other non-Computer devices also allow this - such as your HD24.
(That the Apogee clock card can't do this, is more of an exception than the rule... or much more, it's SO FRICKEN OLD that it didn't do all the stuff, yet, that was pretty commonplace for most newer devices. At least that's my take on it).

- Syncing off of the embedded word clock signal that is embedded in the ADAT or S/PDIF etc. data stream is perfectly fine, for devices that allow this. However, I do think (this is an assumption, not actual knowledge) that the word clock "pulse" is simply derived from the frequency of when the "audio data packages" arrive at the receiving device. These "data packages" can have different sizes. Typically 16-bit or 24-bit of "size" for each such "pulse" that arrives at a frequency of 44,100 such pulses per second, or 48,000 pulses per second (for devices that only support 44.1 or 48 kHz sample rates, like the old devices we usually talk about on this forum).
So, it's the incoming "data packages" that create the sort of "square wave" that a dedicated word clock connection via BNC usually sends for "pulse on" and "pulse off." For devices carrying such "data packages" it would look more like "actively receiving data = pulse on" vs "not receiving data/transmission-silence = pulse off"... and that would alternate at the frequency of the sample rate (or double, if you want to count the no-data/data-silent parts separately).

This concludes the "dedicated word clock" vs "embedded word clock in audio data streams" bit I wanted to share.

Since we're already "almost" there, let's talk about what the frequency, or "sample frequency" actually does, to give a more complete picture of what's happening in a digital audio transfer:

For digital audio signals (with embedded word clock), such as ADAT, S/PDIF, etc., each such "pulse" as described above, consists of a single "data package" or "sample" of a certain size (16 or 24 bit, usually), that are sent at a certain speed (i.e. 44,100 such "samples" or "data packages" sent per second... or 48,000 for the 48 kHz sample rate).

The data package size, determines the dynamic range, between the loudest, and the quietest signal that can be reproduced, because of the amount of data that is available to represent this (16 bit or 24 bit).

Now let's think about what we're actually trying to "describe" with this type of data. For simplicity, let's thing about a sine wave in the "real world" that comes out of your speakers, while examining what the speaker cone actually does, when playing back that (analog at that point) sine-wave. At the "zero" point, the speaker cone is in a resting position - so, it isn't pushed out, nor "sucked in" by the voice-coil/magnet that is attached to it.
Now while playing the sine wave an looking at it in VERY SLOW motion, the speaker pushes OUTWARD while the slope of the sine-wave rises... then travels back INWARDS after the sine wave reaches its top and then returns to the "zero/resting" point, but continues to essentially get "sucked in" further than the resting position, as the sine wave starts reaching BELOW the zero/resting point of the speaker.

So, a sine wave makes the speaker cone travel outward from the zero point, and inward from the zero point, along with the position of the signal at the sine-wave (above/below the zero aka "resting point" line).

Now, we want to "digitize" such a sine wave to make it usable in computers or digital devices. Because of how "digital (aka 1s and 0s) work to describe data, we can't reproduce a fully "smooth" sine wave as it would be created by an analog oscillator. What is being done instead, is to describe many small sections of the sine-wave in a row, to more or less give the "position" above or below the "resting point/zero-line" in our speaker example, above.

If you think of a sine wave like a nice and even hill in the landscape, you'd essentially have to built "steps" into the hill that are all the same height, to go up the hill on one side, and down the hill on the other side (...and technically, also down into a deep valley that is as deep as the hill is high... but for simplicity, let's just look at the "above the zero/resting-point" for this "hill" example).
The "height" of the "stair-steps" represents our "sample rate." If you carve 44,100 steps into our hill, and look at it from a certain distance, the "steps" have a certain visible size - the hill, at least where our "steps" are, is not longer "smooth" but there are 44,100 steps that "describe" how quickly the hill rises (or falls on the other side).

Now, if you'd carve twice as many but "half as high" steps into the same hill of the same physical height, you'd have to carve 88,200 steps into the hill. The result is a much "smoother" looking representation of the hill via those steps. That's the difference between different sample rates (and the reason why 44,100 steps vs 48,000 steps don't make a big difference in terms of quality).

I don't want to move too far away from this example, but the more steps, the higher the (analog) audio frequencies you can capture or represent. So, even though 44.1 kHz "should" in theory cover the roughly 20 kHz max the human ear can hear on the high end, a SIGNIFICANTLY higher sample rate could theoretically reproduce the "hill" much smoother in the digital realm and capture higher frequencies, clearer harmonic overtones and create less disharmonic distortion.
(Personally, I think that recording at 44.1/48 kHz is perfectly fine, though... most people (...or ALL in my example) who claim to hear a difference, still fail a 44.1 vs 88.2 / 48 vs 96 blind test every single time... we tried that at a company I used to work for... was pretty funny... the arrogance about that dropped by a lot, afterwards, hahaha).

So, the more steps we got to describe "our hill" or our sine wave, the smoother our digital "waveform" representation of it will be, and the higher the audio frequency and clearer the high-frequency overtones we can capture.

So at the maximum travel outward or inward, that our speaker can go at the furthest before it's overextended and no longer sounds as it should (i.e. distortion/break-up, etc.), we have reached our maximum "dynamic range" - or the "loudest point" at the peak of our sine-wave/top-of-the-hill/maximum-speaker-extension.

Now, you might ask... are all the hills the same size? They certainly aren't in the real world. Good point!

Well, no matter how high the hill is, you still only have either 44,100 or 48,000 steps to represent and describe it "digitally."

What if you want to describe a GIANT mountain instead of a hill... well, you ALWAYS want to describe the peak of the mountain... otherwise, you'd have to "cut" our representation off at the top, when running out of "steps" available to describe it. So, what you have to do, is to start describing the mountain from a bit up (aka turn down the gain in the real world)... rather than from the "flat land" on the bottom, so you can still reach the top.

...and that's where the "package size" or "bit depth" of each transmitted sample comes in. Each sample can essentially say "I'm so-and-so far away from the bottom of the hill - or much more from the starting point of our description of the hill."

A larger bit-depth/package size can describe a greater distance without running out of available data-space to describe it, while a lower bit-depth can't cover as much "height."

That is our dynamic range. The more data that is contained in each of the samples/data-packages (16 or 24 bit) that get transmitted 44,100 or 48,000 times a second, the more "dynamic range" can be reproduced.

Now, a common misconception is, that a higher bit-depth allows you to capture LOUDER signals. But that's not the case. In digital audio, there's an absolute "maximum loud" ceiling that you can't go above. And if you hit that ceiling, the top of the hill would appear "flat" while it isn't. And in digital audio, that means that the speaker would essentially "stand still" at it's maximum extension for a period of time, instead of moving out and in... and that results in at least ugly "digital distortion" of if you keep doing it for extended periods of time, speaker damage (the voice coil would overheat from continuous DC voltage).

So, what do you do in digital audio to avoid this? You don't start measuring from the "flat land" below (i.e. absolute silence), but you go a bit up the hill/mountain (aka turn down your gain or "trim" the signal), so you can reach and describe the peak of the mountain, without cutting it "flat" on the top (via hitting the bad and absolute digital ceiling).

So, the bit depth is essentially how much "distance" you have available to describe your hill/mountain/sine-wave. And that "distance" how far your sine-wave can travel up and down that you can describe digitally, is your dynamic range.

This means that you don't have to climb as "far up the hill" (aka into audible and wanted audio territory) to start describing the way up the hill and the top, but you can start further down the hill and cover more distance. Ideally, you'd be able to start all the way down from flat land (i.e. total silence).

Any time you have to "climb up the hill" a bit where there's already audio present, you essentially have quieter audio information disappear in background noise (e.g. as is created by the number of steps available we talked about earlier... the fewer steps, the more audible "quantization noise" that creeps into the picture from the "flat land" side of this description. Not counting the noise floor created by devices like microphones, etc., which often is higher/louder than the digital noise floor and available dynamic range... at least at 24-bit).

In other word, a higher bit depth of 24-bit allows you to capture QUIETER signals before they disappear in the (digital) noise floor (after adjusting your gain to make sure you never-ever-ever-ever have the hill be higher/louder than the dynamic range you have available, since it would cut it "flat" on the top, and make your "speaker" stand still at maximum extension, which is deadly/causes clicks/ugly noise. That's why you should NEVER allow the clip LED turn on on digital meters. It usually means that at least 2 samples in a row, where hitting maximum and create a "flat top" - and that's even worse than the hair style, hahaha).

OK... that was quite the ramble, and I think I repeated myself a bunch of times in the process. But quite honestly, I don't want to go back and re-read what I just typed to clean it up. I hope that analogy is still kind of useful, and maybe makes it easier to picture how digital audio works, by comparing a simple analog sine wave, with speaker movements and hills/mountains and steps up and down (sample rate), while considering the maximum available distance (bit depth) you can "record" digitally, as an analogy.

And to come back to the original question... in ADAT or S/PDIF or other digital signals that contain audio data, the word clock is derived from the frequency of the incoming samples (44.1 kHz/48kHz) instead of being a "separate" square wave, as transmitted over a BNC word clock cable.
Only BNC word clock connections have to deal with 75 Ohm termination in that sense. When syncing off the word clock signal that is derived from digital audio connections that also carry the audio information, no such separate termination is available/necessary.

So... sure... if you connect a BNC cable from the D8B/Apogee Clock card (BNC WC-Master) directly to the BNC on the HD24 (BNC-WC Slave), and IF the HD24 is internally terminated or allows you to switch that on, that's how to set that up right.
If you want to set the HD24 to sync as word clock slave from the signal coming into it via the ADAT optical ports, you can do that, too. In that case, you don't need to connect the BNC word clock cables at all and don't have to think about termination.

Again, sorry about the ramble. I hope it's still kinda useful.

[csp]

Y-my-R,

That was absolutely brilliant and so VERY clearly put and exampled.

As I have explained/stated numerous times, I grew up, trained, lived and worked (65+) in basically the analogue world (a lot in valves !!!) and as such digital is really another language to me, so I honestly believe that for the first time since I have been using digital equipment I have understood the full process of word clock and word clocking.

You really should try to publish somewhere what you have just written and if not, you should definitely be a teacher in the general digital area.

I know that I do and I hope that others who are part of this d8b forum appreciate as I have/do all of the effort that you put into helping everyone and --- "Again, sorry about the ramble. I hope it's still kinda useful" IT SURE WAS USEFUL AND WITHOUT THE RAMBLE IT WOULD NOT HAVE BEEN AS CLEAR !!!!!!!

David