Home -> Magazines -> Issues -> Articles in this issue -> View
When Is Sync | |
Article from One Two Testing, August 1985 |
the science of synchronisation explained
Having recently bought a television after five year's abstinence, I find myself a target for all manner of low humour and constant interrogation as to which mindless soap/quiz I have become addicted. The fact that the set was bought for its RGB computer input capability does nothing to allay the scorn of my detractors.
But to the point, a not too distant edition of 'Whistle Test' (nice to see John Noakes still getting work) featured the Eurythmics recording in Paris. Now, the media have got a reputation for getting technicalities muddled and it's true that I was slightly Ruddled at the time, but Dave and Annie's recording technique appeared to start by laying down a guitar and vocal track 'live'. This done, Dave explained that they then bung on a drum machine, a sequenced bass line, more guitar etc and additional vocals to complete the track. Fair enough, but one moment's thought will throw out the question of how do you synchronise a drum machine to the temporal wanderings of excitable earthlets?
In time-honoured tradition, a sync signal should have been recorded on a spare track while the performance was rendered to a guide rhythm. This would, of course, dictate a constant tempo and prohibit any spontaneity. The Eurythmics' scheme is obviously preferable, but how can it be done?
Now I can't lay claim to inside knowledge but one possibility would be to use a new gizzmo from Roland by the name of an SBX80. This new toy is described as a programmable rhythm controller/sync box, and by all accounts it's jolly clever. In essence, a button can be tapped in time with a replayed 'live' performance, and a sync track generated to suit. The SBX80 does a lot more besides but as this isn't supposed to be a review, let's leave it for a while and explore this syncing business in a little more detail.
Of necessity, any discussion about synchronisation rapidly becomes horological with all sorts of talk concerning clocks. Nothing mind-bending to comprehend here, a clock signal is physically akin to the square wave output from a voltage-controlled oscillator (VCO — remember those?). Whereas the output from an oscillator is destined to reach your ears in some modified form, a clock signal is a controlling stream of prompts telling its receiver to move on by one slot (smallest unit of quantisation). That slot may be the next note in the case of a sequencer or the next hemi-demi-semi-quaver beat in the case of a drum machine.
Clock signals are like currency in that every manufacturer has a different exchange rate of pulses to the crotchet beat. Standardisation is so lacking that even good old crotchets get called quarter-notes (qn) outside of the U.K. Provided that the number of sync pulses per crotchet (qn) match at both ends of the connecting cord, two machines should run with one tempo. Anyone who's tried this out across various manufacturers' products will know that interconnection is never that straightforward. Features such as Roland's Start/Stop control thwart attempts at interfacing even when clock rates are matched.
The number of clock pulses per crotchet has to be exactly divisible by both three and four so that triplets (three notes played in the time of two) can be correctly scaled. For example, in a 24ppc (pulse per crotchet) system, a quaver (eighth note) is 'worth' 12 and a triplet quaver equates with 8 'ticks'. A quick bit of mental arithmetic gives three quaver triplets equals one crotchet (3x8 = 24).
Let's now slip back to this business of putting a sync track on tape. The easiest way is to take the clock oscillator output and squirt it directly onto the tape. This is the way that Linn do it, and a quick sum will give the frequency to be recorded for a 120 beat per minute song as: (120/60) x 48 = 96Hz. In case you're wondering where the 48 came from, it's the number of clock pulses per crotchet (qn) for the Linn. OK, 96Hz is well within the hi-fi capabilities of a tape deck, but with slower tempi the clock signal is no longer recorded as a square wave but becomes a sloping-topped mess as a result of differentiation by coupling capacitors.
The clock can be recovered by clever circuitry, but as this is inside the Linn and the machine doesn't provide a clock out when syncing to tape, it presents difficulties in linking with additional sequencers. Another problem involved with recording the neat clock frequency is that phase inversion can occur between record and playback. The result is that the synchronised signal can be half a clock pulse out.
To overcome these difficulties, frequency shift keying (fsk) is used which translates the mark and space phases of the clock's square wave into two different audio frequency tones. These tones alternate at the clock frequency and, however slow the clock, the recorder only has to handle audio tones, usually an octave apart, which may be chosen to optimise crosstalk to adjacent tape tracks. Phase no longer becomes a problem, but as the primary vendors of fsk tape sync — Roland and Oberheim — use different tones and clock rates there is no compatibility at the tape level.
As we're discussing encoding for tape storage, it's worth describing how data such as rhythm patterns and computer programs may be handled. First, the data needs to be converted from the computer's internal 'byte-wide' format into a single stream of ones and noughts (serial). This task is performed by a specialised chip known in general terms as a UART (Universal Asynchronous Receiver/Transmitter). It is this same chip type which sits at either end of a MIDI link. The data, having been serialised, is passed to further circuitry which, in the case of the CUTS standard, records a logic '1' as eight cycles of 2400Hz and logic '0' as four cycles of 1200Hz. Incoming data at a rate of 300 bits per second (Baud) can be encoded in this way. Pushing the circuit to the limit allows 1200 Baud which is the rate used on the BBC micro. As will be explained later, there is an alternative for even higher rates.
But back to syncing. How can clock rates be multiplied and divided to drive 'incompatible' equipment? Both can be achieved if we're talking powers of two. A binary divider chip costs about fifty pence but somehow, when attractively cased and promoted, its cost goes through the roof. Multiplication is not so easy and unless a microprocessor is used to interpolate the given input, a trusty analogue phase-locked-loop (PLL) has to be used. PLLs consist of a voltage-controlled oscillator whose control input is derived from the comparison of its own output with the frequency of the signal to be locked onto. Ordinarily, the PLL mimics the input one for one, but if a binary divider is used to halve the output of the VCO before the comparator, the confused circuit will lock at twice the given frequency. Needless to say, this is a gross over-simplification and the PLL needs time to capture a signal and settle down.
So far, so good, but how does MIDI fit into all this? The MIDI spec was thoughtful enough to include System Real Time data using a clock rate of 24 transmissions per crotchet (qn). We no longer have a square wave to record or encode but are given the rather nebulous concept of data patterns corresponding to F8 hexadecimal thundering down a cable at 31.25 thousand bits per second. This is the Timing Clock In Play signal which, to confuse further, gives way to a Measure End clock at the end of every bar.
To convert MIDI to clock pulse and tape formats we could use a box with KMS-30 emblazoned on the top. This impressive and cheap unit from Korg allows synchronisation from and to any of the above mentioned schemes. Not only that but it takes care of 24/48 pulse conversion as well. Of course you can't please all of the people all of the time and owners of Oberheim (96ppc), Fairlight (192ppc) and PPG (64ppc) equipment will have to resort to the esoteric delights of Garfield Electronics' Doctor Click 2 to solve their particular problems. Serves 'em right for being up-market trendies.
As synchronisation becomes more matter of fact, there is an increasing reluctance during rehearsal and recording to have to start each sequence from the beginning. Up until recently, there was no option but with Roland's SBX80, 'dropping in' now becomes a reality. To achieve this flexibility, each bar must be identifiable and a taped SMPTE time code is used as a reference. This time code is well known in video work and is also used when slaving 24-track tape machines together. An individual bar within a sequence may be addressed by MIDI's System Common Measure Select and as the SBX80 knows exactly where the tape is by reading the time code, the sequencer can be fired into action, bang on cue, wherever the tape is started.
There's not much space to say much more about SMPTE here except to tie in one issue raised earlier. This particular time code shoves 80 bits of data onto tape for each frame and there are 25 frames per second in Europe and 30fps in the USA. Pencils out again chaps, that corresponds to 2000 and 2400 Baud respectively. Now how do you get that on tape? The rate is far too fast for fsk methods and so Manchester Bi phase Modulation is used which encodes data by phase changes within a constant frequency clock. And with that last nugget I must go. Dallas is on in a few minutes!
Thanks to Kevin at Tune Inn, Catford, for the loan of the hardware to turn the theory into practice.
15 Fab Software Tips |
Interfacing The Past (Part 1) |
Troubleshooting with the Friend Chip SRC AT - SMPTE/MIDI Processor |
That Syncing Feeling (Part 1) |
Technically Speaking |
Adrift On An MTC - MIDI Time Code |
![]() Using Timecodes - An Introduction To Timecode Synchronisation (Part 1) |
Short Circuit - Time Machine Revisited |
SMPTE Uncovered |
MIDI Matters - Song Position Pointers (Part 1) |
Everything but the Kitchen... (Part 1) |
Getting into Video (Part 1) |
Browse by Topic:
Feature
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!
New issues that have been donated or scanned for us this month.
All donations and support are gratefully appreciated - thank you.
Do you have any of these magazine issues?
If so, and you can donate, lend or scan them to help complete our archive, please get in touch via the Contribute page - thanks!