Articles Home

Features

Reviews

 

Get In Sync

© Dan Phillips

This article first appeared in Electronic Musician magazine in 1998.

Synchronization - getting two or more devices to record and/or play back together - is one of the fundamental parts of making a studio work. Sync issues are similar for both analog and digital devices - in fact, syncing digital devices together is often simpler than syncing their analog counterparts. There are some different methods in the digital synchronization toolkit, however, which may be unfamiliar at first. In this article, we'll look at the basic concepts of sync, and how they apply in the digital domain.

Start sync and continuous sync

To being with, it's important to realize there are two separate components to synchronization: starting at the same time, and proceeding at the same speed as time continues. For the sake of convenience, let's call the first "start sync," and the second "continuous sync." In the analog world, these are frequently two facets of a single synchronization method, such as SMPTE. With digital devices, on the other hand, they may be handled separately.

Analog Sync

Let's start with the analog case first. With analog tape, both start and continuous sync are generally handled by SMPTE timecode. For instance, to sync two analog 24-track tape machines together, you would stripe both tapes with SMPTE, and connect the two machines together using an analog tape synchronizer.

The synchronizer listens to SMPTE from both machines, and has direct control of the tape machine's motors, as well as simple transport control. When you press play on the master machine, the synchronizer shuttles and varispeeds the tapes until both machines are playing back from the same point in the music - for instance, at the beginning of the second chorus. Once this happens (and it can sometimes take a while), start sync is achieved.

If that was all that the synchronizer did, however, it wouldn't be enough; the two tapes would start together, and then gradually drift apart. This is because the two machines' motors will run at slightly different speeds, even though both are ostensibly moving the tape at exactly 30 inches per second.

To solve this problem, the synchronizer listens to the two SMPTE streams throughout playback. If necessary, it makes continuous adjustments to the tape speed to keep the two machines playing together.

Analog to Digital sync

Syncing an analog system to a digital system - either tape or DAW - is conceptually the same as syncing two analog machines. The two systems need to start at the same point, and they need to agree on the playback speed.

As with analog machines, timecode (such as SMPTE or MTC) is still used for communicating the start point. Continuous sync, however, is done differently.

One might imagine controlling the speed of a digital tape machine's transport, just as with analog machines (although that wouldn't be completely correct). But DAW's don't have any tape motors to control, so there has to be another way to set the playback speed.

Enter word clock (fanfare, please, maestro).

A digital audio recording consists of a long stream of individual samples; you might think of each sample as representing a very small area on the surface of an analog tape. Just as the analog tape moves across the tape heads at a constant speed (such as 30 inches per second), the samples of the digital audio recording flow by at the sample rate (such as 44.1kHz, or 48 kHz).

The rate of this flow is controlled by the word clock, which sets the precise sample rate of the digital system. Each time the clock ticks, the system sends and/or receives another sample. If the device's sample rate is 48 kHz, for instance, the clock will tick once each 1/48,000 of a second. The word clock therefore controls the "tape speed" of a digital system.

Even on common digital tape systems, in fact, it is the word clock that controls the speed of the physical tape transport; the tape speed is adjusted to match the word clock.

Word clock can be generated internally, or received from an external source. Most digital audio connection formats, including S/PDIF, AES/EBU, TDIF, and ADAT optical, include the word clock signal. Word clock can also be transmitted separately, without any audio data.

So, to sync a digital system to an analog system, one needs to translate the analog tape speed into the digital word clock. A number of synchronizers (such as the Mark of the Unicorn MTP-AV, Opcode Studio 64 XTC, and the Digidesign Universal Slave Driver) do exactly this; they read the SMPTE or MTC signal from the analog tape, and use the rate of the timecode to generate a word clock signal.

For instance, if SMPTE is running at 30 frames per second, and the audio sample rate is 48 kHz, each SMPTE frame would correspond to 1600 ticks of the word clock. This ensures that when the SMPTE from the analog tape runs faster or slower, the word clock - and thus the "tape speed" of the digital audio system - speeds up or slows down accordingly.

Some digital audio software and hardware can do this internally, without actually changing the audio hardware's word clock rate. Instead, they measure the incoming SMPTE/MTC stream, varispeed the audio in software, and then send out the processed audio at a constant sample rate.

Digital to Digital

As in the all-analog and analog/digital hybrid cases above, all-digital setups still use timecode for start sync. Depending on the equipment, they may use either the familiar MTC/SMPTE timecodes or new proprietary, sample-accurate timecodes (as discussed below).

Continuous sync, happily, is much easier with all-digital setups. You no longer need to directly control physical tape speeds, or varispeed word clock based on SMPTE/MTC. Instead, you can simply directly connect the word clocks of all the devices, so that they all play and record at the same exact sample rate - and thus all have the same "tape speed."

Now, you might be thinking that with the precision of digital equipment, 48kHz on one machine should be the same as 48kHz on another, so that word clock connections would not be necessary. Unfortunately, this isn't the case. Just as analog tape decks don't exactly agree on what 30 inches per second means, digital devices all have slightly different interpretations of sample rates. When two machines are both set to 48kHz, for example, one might really be 47.998 kHz, while the other is at 48.001 kHz.

The solution is a master-slave setup, very similar in concept to timecode configurations. The master sets the sample rate, and the slaves ignore their internal clocks, using the master's clock instead. A properly set up word clock arrangement will ensure perfect, driftless continuous sync between all connected devices.

 

 

 

 

Digital to Digital (continued)

You can think of each device's word clock as a mechanical gear, with every tooth in the gear representing a single sample. Spinning by themselves, they can move at any rate they like. But when two or more gears are fitted together, they move in lock-step precision. Every time one gear advances by a single tooth, the others turn exactly one tooth as well - no more, no less.

When the word clocks of digital audio devices are synchronized, they act in the same way. Each time the master plays a single sample, the slaves play one as well. If two devices are each playing the same digital audio file, they will take exactly the same amount of time to play that file - accurate to the sample.

It's important to note that word clock synchronization is critical not only for continuous sync, but for accurate transmission of digital audio data under any circumstances. Without proper word clock sync, you'll be plagued by audio artifacts, including pops, clicks, and distortion.

As a side note, standard word clock signals "tick" once per sample. In contrast, Digidesign systems use a "superclock", which "ticks" 256 times per sample. The two are not directly compatible, although some synchronizers offer both options. Some may even offer a single set of word clock I/O, switchable between the two, so make sure that you've selected the correct option for your gear!

Setting up word clock synchronization

Generally, setting up a studio's word clock synchronization involves two separate steps. First, the word clocks must be physically connected. This can often be done simply by making a digital audio connection from the master to the slave, such as S/PDIF out to S/PDIF in, ADAT optical out to ADAT optical in, and so on.

Sometimes, it may also be necessary to use dedicated word clock cables, such as with complex setups, or with devices which have word clock I/O but not digital audio I/O. In larger systems, one can also mix and match formats as required by the I/O available on each device.

Second, the slave devices must be set up to use the word clock from the master device. In digital audio software/hardware combinations, this is referred to as "sync source," "audio clock source," or other similar terms. On hardware devices, there may be an obvious front-panel selection (such as Clock Source on the ADAT XT), or a hidden key combination (as with original ADATs). Some older or simpler devices, such as DATs, may simply switch automatically as digital I/O is enabled or disabled, which can limit one's options; fortunately, most new devices are more flexible.

If your digital audio software supports software-based continuous sync, you should disable this feature; the word clock connection will take care of this in hardware. In some cases, in fact, software-based continuous sync can interfere with hardware-based sync, causing erroneous audio artifacts or even total sync failure. Different software will call this parameter different names; in Deck II, for instance, you should enable Trigger Sync, while in Cakewalk, you should set the Audio Options>Advanced SMPTE/MTC Sync to Freewheel.

The final frontier: sample-accurate timecode

So, with a proper word clock setup in an all-digital system, it's easy to achieve perfect, sample-accurate continuous synchronization. The accuracy of the start point, on the other hand, depends upon the type of timecode being used.

MTC, for instance, offers resolution of 1/4 frame at 30 frames per second, or about 1/120 of a second - about 400 samples at 48kHz. SMPTE may offer greater resolution in some cases. However, if you're using a digital audio program, chances are that SMPTE must be converted to MTC before reaching the software - which means that you're still operating at MTC resolution.

ADAT and DA-88 timecodes, in contrast, are based on the word clock. Each tick of the clock is also a tick of the timecode, in a convergence of timecode and "tape speed" which brings us full circle from the dual-function SMPTE of all-analog sync. Both systems offer single-sample accuracy for start sync, with a resolution of 1/48,000 of a second at 48kHz; hence, the term "sample-accurate sync."

Several combinations of computer-based audio software and hardware also offer direct support for sample-accurate timecode. Due to the slight indeterminacies of current computer operating systems, the sync may be perfect, or there may be a sample or so of variation. Even with a bit of slack, however, this is easily 100 times more accurate than MTC.

Resolution and results

So, what does that increased start-point accuracy get you? The most compelling use is in computer editing of audio tracks from a digital tape recorder. With sample-accurate sync, the audio can be transferred into the computer and then laid back to tape with extreme precision.

For example, try recording a track twice from an MDM into digital audio software, once using sample-accurate sync, and the other time using MTC. Then, play back both the computer and MDM tracks simultaneously, again using sample-accurate sync for playback of the first, and MTC for playback of the second. You'll notice that the sample-accurate tracks just sound louder, with little or no phasing, while the MTC tracks phase noticeably.

Similarly, with MTC, tracks transferred in different record passes may have noticeable differences in phase. For best results when transferring phase-correlated audio (such as a drum kit recorded with multiple mics), it's best to transfer all tracks in a single pass.

Sample-accurate timecode has certainly upped the resolution ante. Remember, though, that almost every CD on your shelves was recorded using SMPTE/MTC for synchronization (if the project used sync at all). They sound just fine, don't they? So, if SMPTE/MTC sync is what you have available, don't sweat it.

Also remember that, regardless of start-point resolution, a proper word clock setup will always ensure drift-free continuous sync. So, watch those word clocks, and be safe sync'ers out there!

 

Dan Phillips is a singer, songwriter, and producer, and is part of the team at Korg Research and Development.