Jump to content

Word Clock


Recommended Posts

This is a spill over from another discussion but I was wondering where folks are taking their word clock from?

I usually make my mixer, in this case an 01v96, the master and lock all the other machines, MOUT 828 MK2 to it as slaves.

MLan is breaking this style of build as it requires the computer attached to be the word clock master.

Here I am experimenting w/ Boomcorder via mLan and it doesn't seem to me that my powerbook would be the best source for sync but it is required to be the master in the world of mLan. Hopefully this will change.

Darren Brisker has posted about running TC into a MOTU traveler and then having word sync/line up to that so that TC is bascially generating the master.

My perception was that TC and video sync were pertinent to syncing to picture and word clock was a different and unrelated  syncing clock used only to get all the digital audio participants to lock together.

I've never locked word to anything but an A/D or a Big Ben type word clock that wasn't tied at all to TC. Never got any calls of some very long rolls but that doesn't mean that drift didn't happen.

I'd love to hear how folks are setting things up and what did and didn't work.

Scott Harber

Link to comment
Share on other sites

  • Replies 52
  • Created
  • Last Reply

Top Posters In This Topic

Hello Scott,

I believe there are several versions of the ambient LockIt, some producing only the TC. Others producing a black burst, tri-level sync and/or word clock as well.

So the audio folk would use the one generating both TC and word clock, while a HD-video camera operator would use the TC with tri-level sync version.

But, I've only read the manual of these things, not actual experience :-(

Take

Link to comment
Share on other sites

Currently clocking Traveler from Denecke SB1 TC (or HHB TCDAT) to input 8, or (in the studio) from word clock made in the Tascam IF88AE and shared with the DA88s and the Traveler.  With the DAT I can save the input and use the HHBs word clock out to the Traveler.  Looking to see the new Ambient box I've heard of that can make TC and clock etc.  HDP2 also clocks from incoming TC, or internal.  The clocking of either seems ok on internal as well if the takes are minutes long, not tens of minutes. Clocking from the Mac internal clock is generally not a good idea.

Philip Perkins

Link to comment
Share on other sites

Currently clocking Traveler from Denecke SB1 TC (or HHB TCDAT) to input 8, or (in the studio) from word clock made in the Tascam IF88AE and shared with the DA88s and the Traveler.  With the DAT I can save the input and use the HHBs word clock out to the Traveler.  Looking to see the new Ambient box I've heard of that can make TC and clock etc.  HDP2 also clocks from incoming TC, or internal.  The clocking of either seems ok on internal as well if the takes are minutes long, not tens of minutes. Clocking from the Mac internal clock is generally not a good idea.

Philip Perkins

I can definitely get word out of a lockit box but I've never heard of jamming it w/ tc and then having it output word in sync w/ the TC. I've only used it to output video sync in various styles.

Is it helpful to jam TC and using the word out? Are there advantages?

I've always run word as a seperate entity but I was just curious if there was some positive aspect to doing so.

I can get word from many different sources, usually my mixer, there's no shortage of options there but my concerns are w/ word and TC syncing and their drift from each other if there is any.

Scott Harber

Link to comment
Share on other sites

I can get word from many different sources, usually my mixer, there's no shortage of options... Scott Harber

There are many devices that generate timecode, some are built in to recorders, some are dedicated generators, some are built in to our slates and take over generating the timecode for the slate once the slate has been jammed.  As we have all experienced, not all timecode generators are created equal.  Some are much more accurate than others.

The same holds true for the clocks in your various digital devices.  While all of these devices can output word clock to your various other devices and provide you with a Master Clock source, not all of them are created equal.  Some devices provide a much more accurate word clock source than others.

So... while, yes there are no shortage of options, it is still quite important to choose the best one.

Best,

Darren

Link to comment
Share on other sites

  • 3 months later...

Darren Brisker has posted about running TC into a MOTU traveler and then having word sync/line up to that so that TC is bascially generating the master.

My perception was that TC and video sync were pertinent to syncing to picture and word clock was a different and unrelated  syncing clock used only to get all the digital audio participants to lock together.

I've never locked word to anything but an A/D or a Big Ben type word clock that wasn't tied at all to TC. Never got any calls of some very long rolls but that doesn't mean that drift didn't happen.

I would also love to hear a detailed explanation of the part timecode plays in word clock relationships between devices, or even if time code being generated by a word clock "master" has effect on the word clock itself.  Since most of the NL recorders these days do not actually roll "time code" as a "track" or sync pulse, but simply put a timestamp on the .WAV file that TC-intelligent software can then create a TC signal from, I wonder if time code continues to be relevant to clocking speed and issues.  If anyone can shed light on this, it'd be appreciated!

Regards,

Noah Timan

Link to comment
Share on other sites

Hello Noah,

I believe that most NL edit software and Boom Recorder count the number of samples to calculate the current timecode.

So it is very important to have the sample rate of the audio interface to be very accurate to have the calculations yield an exact time. Well, I am lying, the sample rate need not to be exact in the real world, but it needs to be in complete sync with the camera; as here as well a film-edit applications count how many frames has passed to calculate the time.

So if you don't want your audio to drift compared to the camera, one would need both a word clock and a timecode signal to be generated from the same device (or two devices that are calibrated to the same speed).

Film cameras may lock to a timecode signal, but video cameras often need a blackburst signal or tri-level sync in combination with a timecode signal. Some audio interfaces like from MOTU can also lock their sample rate to a timecode signal without a word clock.

But all of this is not needed when you just do a feature film the takes for dialogue is often too short to notice any drifting, because at the start of each take both the audio recorder and camera uses the current timecode at that moment.

But if you are recording concerts or beauty pageants, where you need to record audio continuously for many hours and have multiple cameras this becomes an issue.

Link to comment
Share on other sites

I Use a MOTU midi Time Piece AV as a Master Word Clock, from there I distribute Word sync to my Yamaha O2r and other digital devices like a DAT machine, digital Sub-mixer, Digital Mic Pre-amps, Digital multi band compressor, and computer Audio Interface. I find this setup to be the most stable as opposed to have the Mixer to act as a Word Clock Master.

Just my 2 pesos...

Link to comment
Share on other sites

Hi Take,

Thanks for the info.  I *do* understand what word clock is and why it needs to be employed -- what I'm fuzzy on is how the timecode application of a device specifically affects the clock speed, if it does at all.  Yes, different machines connected differently all need to have the same exact idea of 48kHz or whatever the elected sample rate is -- in short, to be running at the same "speed".  Now, does the TC generator in one of the machines in the chain affect that sample rate, or is it simply a marking system that will apply regardless of the variance of the digital clock speed? 

nvt

Hello Noah,

I believe that most NL edit software and Boom Recorder count the number of samples to calculate the current timecode.

So it is very important to have the sample rate of the audio interface to be very accurate to have the calculations yield an exact time. Well, I am lying, the sample rate need not to be exact in the real world, but it needs to be in complete sync with the camera; as here as well a film-edit applications count how many frames has passed to calculate the time.

So if you don't want your audio to drift compared to the camera, one would need both a word clock and a timecode signal to be generated from the same device (or two devices that are calibrated to the same speed).

Film cameras may lock to a timecode signal, but video cameras often need a blackburst signal or tri-level sync in combination with a timecode signal. Some audio interfaces like from MOTU can also lock their sample rate to a timecode signal without a word clock.

But all of this is not needed when you just do a feature film the takes for dialogue is often too short to notice any drifting, because at the start of each take both the audio recorder and camera uses the current timecode at that moment.

But if you are recording concerts or beauty pageants, where you need to record audio continuously for many hours and have multiple cameras this becomes an issue.

Link to comment
Share on other sites

On the two recorders I'm using now: Tascam P2 and Metacorder w/ Traveler, the TC only affects the clock speed of the recorder if I tell it to.  On the P2 I can tell it to calculate its internal word clock from incoming TC, and the Traveler can do the same via FireWire Console by connecting the TC to one of the Traveler's audio inputs.  If I don't do these actions then the recorders will time-stamp the files correctly, but the sample rate will be whatever the machine's internal clock is, not that of whatever is generating the TC.

Philip Perkins

Hi Take,

Thanks for the info.  I *do* understand what word clock is and why it needs to be employed -- what I'm fuzzy on is how the timecode application of a device specifically affects the clock speed, if it does at all.  Yes, different machines connected differently all need to have the same exact idea of 48kHz or whatever the elected sample rate is -- in short, to be running at the same "speed".  Now, does the TC generator in one of the machines in the chain affect that sample rate, or is it simply a marking system that will apply regardless of the variance of the digital clock speed? 

nvt

Link to comment
Share on other sites

Hi All,

I'll revisit this and try to explain it as best I can.

First of all, in the digital world, the better your clock (lowest PPM and least jitter), the better your audio quality will be.  Poor quality clocking results in reduced fidelity at best and a totally unusable signal at worse.  Somewhere in the middle are pops and clicks.  Very poor clocking will result in distortion, or a totally unuseable signal.  So it is very important to use the best clock available to you as your Master Clock.  In the music (read commonly available interfaces) world, most decent clocks spec out at around 50-100ppm.  Many are much less than that.  Really high end interfaces may spec out at 5-10ppm.  As we all know, we have now become fortunate enough that our t/c clocks are now spec'ed as low as 1ppm (Denecke) or 0.1ppm (Ambient).  As well, most timecode clocks are TXCO rated whereas Wordclocks generally are not.  In summary, our timecode clocks are generally FAR more accurate and temperature stable than most word clock Master Clocks.  For this reason, if you have a method of slaving your Master Word Clock to your timecode, you should do it.  You will have a much more stable word clock for doing so.

The second reason for slaving your word clock to your timecode does indeed have to do with sync.

As we know, in the non-linear world linear timecode does not exist.  Timecode is derived from a stamp at the beginning of the file and then *samples* are counted from there to extrapolate what the timecode should be at any given point in the timeline of the audio file.  So, it stands to reason that, if your timecode clock (0.1ppm?) and your word clock (100ppm?) are not *exactly* in sync because they have drifted apart, the *extrapolated* timecode for a given position in an audio file's timeline and the *actual* timecode for that same position will not be the same.  If my math is correct, even a spec of 100ppm could allow for a slip of up to 1 frame every 6 mins.  This will cause problems with the editing and especially the re-conforming process.  Locking your word clock to your timecode clock will 100% ensure that the extrapolated timecode and the linear timecode (such as what the slates are seeing, etc) will *always* remain locked together.  This is referred to as "Sample Accurate Timecode" or "Sample Accurate Sync".

I hope that this will help shed a little more light on this subject.

-Darren

Link to comment
Share on other sites

So if you don't want your audio to drift compared to the camera, one would need both a word clock and a timecode signal to be generated from the same device

This can be done using a 744T, or the new Ambient 501, as both of these devices emit *both* wordclock and t/c.  Or any other device that emits both t/c AND wordclock.  There are probably other recorders capable of this as well.  Or, as Take mentioned, it can be a device such as the MOTU Traveler that is capable of *syncing it's word clock to incoming timecode*.  The Traveler would then go onto be used as the Master Word Clock for the rest of the devices in the chain.

-Darren

Link to comment
Share on other sites

On the two recorders I'm using now: Tascam P2 and Metacorder w/ Traveler, the TC only affects the clock speed of the recorder if I tell it to.

And how, pray tell, does it do that?

  On the P2 I can tell it to calculate its internal word clock from incoming TC,

What I'm curious about is specifically how it calculates its internal word clock from incoming TC.  I'm not clear on the relationship between frame rate and sample rate in this context of one redefining the other and how (and with what tools) these manipulations take place.

Thanks!

NVT

Link to comment
Share on other sites

Thanks, Darren, for shedding a little light. 

Locking your word clock to your timecode clock 

Here I'm a bit foggy...are you suggesting this action is to take a word clock signal out from the machine that also happens to be generating code and bus it to word clock in on the second recorder, or are you suggesting that one literally take a SMPTE timecode signal and bus it to a word clock in signal?  I never tried the latter, but I don't suspect it would work. 

What I'm really trying to understand, beyond the gozintas and gozoutas, is the relationship between samples per second and frames per second and how the TC signal can affect the sampling rate.  I understand how the sampling rate and time code frame rate can create relationships and offsets, both synchronous and non-synchronous, but I am unclear on how one clock causes another to actively behave differently.  It's an issue that we never had to worry about in most of our old linear workflows.  Anyone who's got their head wrapped around this, your advice is most welcome!

Regards,

Noah Timan

Link to comment
Share on other sites

Thanks, Darren, for shedding a little light. 

Here I'm a bit foggy...are you suggesting this action is to take a word clock signal out from the machine that also happens to be generating code and bus it to word clock in on the second recorder, or are you suggesting that one literally take a SMPTE timecode signal and bus it to a word clock in signal?  I never tried the latter, but I don't suspect it would work. 

What I'm really trying to understand, beyond the gozintas and gozoutas, is the relationship between samples per second and frames per second and how the TC signal can affect the sampling rate.  I understand how the sampling rate and time code frame rate can create relationships and offsets, both synchronous and non-synchronous, but I am unclear on how one clock causes another to actively behave differently.  It's an issue that we never had to worry about in most of our old linear workflows.  Anyone who's got their head wrapped around this, your advice is most welcome!

Regards,

Noah Timan

A analog to digital convertor has to look at some clock source to decide its sample rate.  Very accurate clocks--good enough for hours long takes etc. are still quite expensive.  On some inexpensive recorders one way to make them be more accurate sync wise is to allow their internal clocks to look at incoming timecode, and derive a clock from that.  That way a single connection (as from a video camera) can provide both a positional reference (the TC, the file stamp @ the start of recording) and a rate-of-speed reference (the sample rate).  Both of these are necessary to both mark a recording and make sure that it is running at the same speed as the master.  This is a lot less of an issue in digital recording than it was in the analog world (where it was a very big deal).  With even decent clocks one can get away with most film-length takes ( a few minutes) without worrying about clock rate.  But for longer takes there can be real drift if the clocks of all the convertors (ie in the video camera and in the recorder, or etc) aren't locked together somehow.  TC is one way, video sync (or trilevel sync if HD video) is another. 

Philip Perkins

Link to comment
Share on other sites

Here I'm a bit foggy...are you suggesting this action is to take a word clock signal out from the machine that also happens to be generating code and bus it to word clock in on the second recorder, or are you suggesting that one literally take a SMPTE timecode signal and bus it to a word clock in signal?  I never tried the latter, but I don't suspect it would work. 

Noah Timan

I am suggesting that you use the "machine" that generates your t/c as your Master Clock with regards to wordclock.  If this machine happens to also have a wordclock output, you may use that, Digital equipment can also be clocked via the actual digital audio signal as well.  So, if you are feeding your second recorder via AES, SPDIF, or ADAT Optical, it will be able to lock to these as well; it is not necessary to use the actual wordclock BNC connection.  In fact, locking to the ADAT or AES signal can often be the preferred method.  It all depends on your exact setup with regards to signal flow.  If you are feeding both of your hardware (744, cantar, etc) recorders from an analog board and both of your recorders have built in timecode generators, it is probably not necessary at all to lock them together via anything other than timecode as they will probably already have their timecode and wordclocks tied together.  You could also lock them via sending wordclock from the unit generating the t/c to the b/u recorder.  If you are also using a digital mixer, everything changes as that now also has to have its wordclock synced as well.  There is no simple answer as it completely depends on your unique combination of equipment and signal flow.  If you tell us what you have, we may be able to offer suggestions as to how to clock.

Thus far *my* clocking arrangement has been:

GR-1 supplies timecode.  Timecode is fed into an analog input of MOTU Traveler.  Traveler is told to clock to the incoming t/c via software that's on the Traveler (If you look at the photo of my cart, I believe you will see that the front panel of the Traveler shows that it is using SMPTE as a clock source).  01V96 console is then slaved to the Traveler via clocking signal contained within ADAT optical signal that I use for "Tape" returns from the MOTU going into the 01V96.  Any backup recorder, if required, will have its clock slaved via the sync information that is contained within the AES or SPDIF digital audio that will be sent to it from the mixing console or the Traveler.

You are correct in your assumption that you can not simply feed SMPTE t/c into a wordclock input.

-Darren

Link to comment
Share on other sites

On some inexpensive recorders one way to make them be more accurate sync wise is to allow their internal clocks to look at incoming timecode, and derive a clock from that. 

Right.  Now how, specifically, do they do that?  That's what I'm after...sorry to be repetitive.  How does the TC signal control the sample rate, one way or another?  You are suggesting the SMPTE signal specifically alters the speed at which the A/D digitizes.  I'm curious as to how that process takes place.  Sorry if I'm being a bonehead and overlooking something obvious, but this is what is mysterious.  I understand multiple cross reference points for later, but I don't necessarily understand how a time code mark that is disconnected to the sampling rate -- and simply making a stamp in the chunk as opposed to being an active part of the recording process -- controls and furthermore *maintains* the speed of digitization.

nvt

Link to comment
Share on other sites

  If you tell us what you have, we may be able to offer suggestions as to how to clock.

Thanks.  I apologize as it seems that I have not been clear.  I'm not looking for instructions on how to interconnect my equipment, but rather a greater understanding of the specific process by which SMPTE TC affects sampling rate (without specific regard to this or that recorder or TC generator or whatever).  It seems to me that at least in theory, the best master clock would be the most accurate digital clock available, which might or might not be the one built into an all-in-one portable recorder designed to do many functions.  There are numerous permutations and combinations of gear in all of our many different workflows, which is why I don't think a discussion of “take a BNC cable and plug it here on this box” doesn't really help us out in this regard - more interesting is a theoretical understanding that we can apply to many different individual scenarios. 

Specifically, I'm curious about the notion put forth that the recorder generating time code should be the master clock first and foremost, even if a more accurate A/D exists somewhere else in the chain.  If I can gain a specific understanding of how the TC is processing, affecting, and changing the digital sampling process, I can better understand why it might make sense to potentially slave a superior A/D to an inferior one (if the inferior one is the TC recorder).  Since the digital clocking speed can have a potentially marked effect on fidelity, it seems important to ensure that we are always making our master the best clock.

Regards,

Noah Timan

Link to comment
Share on other sites

Thanks.  I apologize as it seems that I have not been clear.  I'm not looking for instructions on how to interconnect my equipment, but rather a greater understanding of the specific process by which SMPTE TC affects sampling rate (without specific regard to this or that recorder or TC generator or whatever).  It seems to me that at least in theory, the best master clock would be the most accurate digital clock available, which might or might not be the one built into an all-in-one portable recorder designed to do many functions.  There are numerous permutations and combinations of gear in all of our many different workflows, which is why I don't think a discussion of “take a BNC cable and plug it here on this box” doesn't really help us out in this regard - more interesting is a theoretical understanding that we can apply to many different individual scenarios. 

Specifically, I'm curious about the notion put forth that the recorder generating time code should be the master clock first and foremost, even if a more accurate A/D exists somewhere else in the chain.  If I can gain a specific understanding of how the TC is processing, affecting, and changing the digital sampling process, I can better understand why it might make sense to potentially slave a superior A/D to an inferior one (if the inferior one is the TC recorder).  Since the digital clocking speed can have a potentially marked effect on fidelity, it seems important to ensure that we are always making our master the best clock.

Regards,

Noah Timan

True, but in the video world we have to follow what camera/picture is doing  electronically, or we all have to follow the same clock (ie SBT or Lockit).  I'm not anywhere near enough of an engineer to tell you exactly how a recorder can look @ incoming TC and derive a clock ref from it, ostensibly by counting, multiplying and dividing, but they do do this.  I have no doubt that there are more accurate clock sources around than the TC coming from a video camera or deck, but I need to follow them so that's what I use.

Philip Perkins

Link to comment
Share on other sites

Specifically, I'm curious about the notion put forth that the recorder generating time code should be the master clock first and foremost, even if a more accurate A/D exists somewhere else in the chain.

Noah Timan

Noah,

There are probably many factors involved in what makes an A/D "more accurate".  For the purposes of this thread let's assume that by more accurate we are only referring to a more accurate clock.

I'm not sure that the idea that the timecode generator should be the master clock first and foremost is a completely accurate way of describing this scenario.  There are two separate theories at play here.  The 1st is that it is not a bad idea to have your timecode clock and your system sync clock tied together for the purposes of Sample accurate timecode.  The 2nd is that, yes, you should use your best clock for your system sync.  Most probably you will find that it just so happens that your timecode generator IS your most accurate clock.  How much difference will this actually make to quality of your sound?  Hard to say.  I suppose the improvement will be inversely proportional to the quality (ppm/jitter) of the wordclock in your current system.

I don't believe that you will find any A/D's with clocks that are more accurate than those that are in timecode generators.  As previously mentioned, you will probably not find ANY A/D's with .1ppm TXCO clocks.

Some may not consider any of this to be absolutely critical.  I suppose one could just consider this the optimal way of doing things.  In any case, this may not be possible with all combinations of equipment.  As mentioned, it does require either a t/c generator (whether built into a hardware recorder or not) that can also output some sort of clocking signal that the rest of your digital gear can accept.  Also in the case of hardware recorders such as the 7 series, Cantar, Fostex, Deva, etc, this is probably al ready occuring internally anyway.  These machines are often the only digital device on one's cart and therefore are already being used as both the A/D and the Master T/C generator.

-Darren

Link to comment
Share on other sites

The 1st is that it is not a bad idea to have your timecode clock and your system sync clock tied together for the purposes of Sample accurate timecode. 

I don't believe that you will find any A/D's with clocks that are more accurate than those that are in timecode generators.  As previously mentioned, you will probably not find ANY A/D's with .1ppm TXCO clocks.

It seems inconclusive that the TC is going to be 'sample accurate' or tied into the A/D converter circuitry at all, even if one machine controls both functions.  It seems from your post that you are suggesting that the TXCO clock controlling the TC generator and the digital clock controlling the sample rate are one and the same, which is not my understanding.  Or maybe you're suggesting that the two separate clocks, carrying out different functions in the machine, are talking to one another.  If this is the case, it'd be good to learn how exactly they are doing so.  But first we must conclusively prove that this is in fact happening before finding out *how* it happens.

Most probably you will find that it just so happens that your timecode generator IS your most accurate clock. 

That depends upon one's setup, of which, as we all know, there are many variations.

I suppose the improvement will be inversely proportional to the quality (ppm/jitter) of the wordclock in your current system.

That's my understanding, yes.

These machines are often the only digital device on one's cart

That seems to be occurring less and less in this day and age.  I know a lot of folks using multiple digital recorders, Yamaha digital mixers, and other outboard gear.  Your posted workflow is a perfect example.

Regards,

Noah Timan

nvt

Link to comment
Share on other sites

It seems inconclusive that the TC is going to be 'sample accurate' or tied into the A/D converter circuitry at all, even if one machine controls both functions.  It seems from your post that you are suggesting that the TXCO clock controlling the TC generator and the digital clock controlling the sample rate are one and the same, which is not my understanding.  Or maybe you're suggesting that the two separate clocks, carrying out different functions in the machine, are talking to one another.  If this is the case, it'd be good to learn how exactly they are doing so.  But first we must conclusively prove that this is in fact happening before finding out *how* it happens.

Hi again Noah,

Without evidence to support the contrary, I would have to say that you are right in that it is not a proven fact that the machines that we commonly use that do output both wordclock and timecode do indeed have both of these clocks tied together in one way or another.  That being said, I'd be both shocked and extremely disappointed in the manufacturers if they are not.  This especially holds true for devices such as the new Ambient 501.  I'm not certain what the point of this device would be at all if its wordclock and t/c are not locked together.

All of that being said, perhaps the next stop for you may be the manufacturers themselves.

Best!

Darren

Link to comment
Share on other sites

Without evidence to support the contrary, I would have to say that you are right in that it is not a proven fact that the machines that we commonly use that do output both wordclock and timecode do indeed have both of these clocks tied together in one way or another.  That being said, I'd be both shocked and extremely disappointed in the manufacturers if they are not. 

Hey Darren,

I'm not sure it's something that we need to be mad at the manufacturers for -- it may well be that although both clocks are called "clock" and both control a certain frequency of events, they are not really intercompatible devices.  My understanding is that they are two separate mechanisms with very different purposes and functionality -- in other words, calculating minor variances in a set number of samples per second (and making sure all the devices in a chain are locked into having the exact same variances in the exact same places) is not part and parcel of to marking a second by a set number of frames.  I'm sure if the Ambient controller has a word clock output and doesn't handle audio, this means that there is a way to interrelate the two, but we add a layer of complexity to this when we consider that unlike with a TC Nagra or other device, the time code is no longer an integral part of the recording with NLR, but rather simply a stamp in the file header that is used only for calculation for displaying values on subsequent *playback*. 

Hopefully we will learn more soon.  Your suggestion to contact the manufacturers is a good one, especially if no one with certain knowledge is able to chime in here.

Regards,

Noah Timan

Link to comment
Share on other sites

Hello everyone,

Someone asked me to comment on how word clock and timecode signals can change the sample rate. So here I am doing a brain dump, I hope I will be coherent.

Every AD converter uses an oscillator, the early AD converters 16 bit/48 kHz oscillate once for each sample. For each pulse it measures the voltage on the input and converts it into a PCM value of 16 bits. You could switch between the internal oscillator, or the word clock driving the AD converter directly.

These days we want 24 bit samples, but simply measuring the voltage doesn't work anymore, 24 bit AD converters are usually delta-sigma (or sigma-delta) converters (or hybrids between the old and delta-sigma). A delta-sigma doesn't directly measure the input voltage, but does this indirectly by measuring the time for a bucket to fill. It needs a very high frequency oscillator for this, mostly in the high Mhz range. You can no longer directly drive these AD converters with a word clock, we come to that later.

The precision of these delta-sigma AD converters is directly related to the stability of the oscillator. With stability I mean that each pulse of the oscillator takes exactly the same amount of time, low jitter. If the oscillator jitters, the noise floor of the AD converter raises. Crystal oscillators by itself are not stable enough to reach the full 24 bit of precision, therefore they make them temperature controlled (putting the oscillator in a small oven with a thermostat), and may contain multiple crystals, or whatever.

The frequency of the oscillator, of course, controls the sample rate. But the frequency of the oscillator is much higher than the sample rate. The frequency of the oscillator can be tuned, this can be done by a PLL (Phase Locked Loop). The frequency of the oscillator is divided to the sample rate frequency and then compared to the incoming word clock, the frequency of the oscillator is then tuned up or down, until the two are locked together. This is a pretty fluid process, so if the word clock has jitter this is averaged away and does not increase the noise floor of the AD converter, but it could add a lower frequency modulation.

The same PLL trick can also be done using the timecode signal, which has a frequency itself. The speed of the motor of an analogue tape deck is controlled in a similar way.

The timecode signal also contains the current time stamp and is used to say when a sample is recorded. An BWF audio file only shows the time stamp for the first sample in the audio file. Which is why it is important to run the timecode at the same speed as the sample rate; either by locking the sample rate to the timecode signal, or by locking the sample rate to a word clock that runs at the same speed of the timecode signal.

Some house clocks can deliver a word clock, a timecode signal, tri-level sync and black burst. The simplest way to design one of these would be to start with a stable high frequency oscillator to make a black burst signal, then you would use frequency dividers to create the lower frequency word clock, tri-level sync and a timecode signal from this. I've seen many house clocks that have a black burst input to generate the other three, you can also get separate black burst generators.

Cheers,

    Take

Link to comment
Share on other sites

Take, thanks so much for sharing your info here with us.

This quote gets to the bottom of our unanswered query here:

The same PLL trick can also be done using the timecode signal, which has a frequency itself. The speed of the motor of an analogue tape deck is controlled in a similar way.

The timecode signal also contains the current time stamp and is used to say when a sample is recorded. An BWF audio file only shows the time stamp for the first sample in the audio file. Which is why it is important to run the timecode at the same speed as the sample rate; either by locking the sample rate to the timecode signal, or by locking the sample rate to a word clock that runs at the same speed of the timecode signal.

What I am wondering specifically is how one locks the sample rate to the frequency timecode signal.  Are these two devices running off the same clock/oscillator in a time code recorder that also processes the A/D?  If not, or if one is using separate devices, how do these two separate clocks "lock"?  Does the TC chase the sample rate speed or vice versa?  How does the word clock integrate between the oscillator you mention utilized by the A/D and the TC rate that is presumably calculated on its own clock?  I understand the basic function of the devices, but I have been unable to learn exactly how they interrelate, and any possible effect on the A/D and its function when the frequency of the frame rate is leading the sample rate.

Thanks for any further illumination you may be able to provide!

Regards,

Noah Timan

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...