The title says it all...
If you are about to add a sequencer to your tape based studio, the first thing you need to do is to decide how you are going to synchronise the two together.
The individual tracks of a tape recorder will obviously always stay in sync with each other because they are all recorded on the same physical piece of tape - there is no way for one track to get ahead or lag behind. And, though the tracks within a sequencer are not physically linked in the same way, they are all controlled by the same master crystal clock and so they too will always run together under control of the sequencing software. But problems arise when you want to record part of your song on the tape recorder and another part on the sequencer, because you then need a way to make the sequencer and the tape recorder start together and run at exactly the same speed. It is conceivable that you could start both machines together by hand, but the small fluctuations in tape speed caused by tolerances and electronic drift would certainly cause them to drift out of time before the end of the song.
It is easier to control the speed of a non-mechanical device such as a sequencer than it is to control the speed of a tape recorder, so current synchroniser systems use the tape recorder as the master time keeping device and the sequencer as the slave. The first problem here is to find a reliable way of letting the sequencer know exactly how fast the tape is travelling, and because tape slips slightly as it goes past the capstan, it isn't practical to use the capstan speed as a reference. What we need to know is the actual speed of the tape itself, and this can only be done by recording some kind of time-code onto a spare track.
You can think of the simplest time code as a series of evenly spaced clicks marking out time, and these clicks are generated in the first instance by the sequencer. Usually, the MIDI clock provides the clicks because this is directly related to the tempo of the piece of music recorded onto the sequencer - for a piece of music written in 4/4 time, there are 96 MIDI clocks per bar. The Americans describe MIDI clock as being 24 pulses per quarter note.
A simple synchroniser simply takes this clock, converts it into a form that can be reliably recorded on tape and then this code is recorded onto one of the tracks of an empty reel of tape as you run your sequencer at the right tempo, preferably with a guide rhythm part already programmed in.
If your song has tempo changes in it, these will be faithfully followed by the code so long as they were present in your sequenced guide part at the time the code was put onto tape.
To lock the sequencer to the code on tape, it is only necessary to feed the code track from your multitrack into the synchroniser, the synchroniser's MIDI Out into the sequencer's MIDI In, and to set the sequencer to External MIDI sync. Then, you roll your tape from the top and the sequencer will start with it and remain in time. You can then record new tracks onto the tape by playing along with the sequencer, the only proviso being that you have to start the tape from the beginning every time. The reason for this is that the code only tells the sequencer how fast it should be running, it doesn't tell it whereabouts it should be in the song. In practice, it's easier if you record a rough mix of the sequenced parts onto one track of tape before you start to record new tape tracks. This means you can do your overdubs using only the tape recorder which allows you to do drop-ins or start recording sections anywhere you like on the tape. This way, you only have to take the whole thing from the top when you come to do the final mix. And, of course, you can wipe the guide track and record something else before you come to do the mix. The great beauty of working in this way is that you never have to record your sequenced parts onto tape (apart from the guide track) so you don't lose any quality at all when you come to mix.
Because code systems can be fussy about levels, it is well worth playing the system back in sync immediately after you have recorded the time code and the guide track. By feeding the guide track and the sequencer output into your mixer, you'll be able to hear if there are any timing errors between the two. Some phasing due to very small time delays is to be expected but if you can hear a short echo or flam on any of the drum beats, the system isn't locking properly.
Most synchronisers will read the code quite happily, even if you use the varispeed on the tape recorder, but bear this in mind: the tape tracks will change pitch as you change the tape speed but the sequenced tracks will not.
To get the code to record onto tape reliably, it has to be converted from a square wave clock into bursts of tone and this system of using tones is known as Frequency Shift Keying or FSK for short. Some systems are more fussy than others about the level of code they receive back from tape so you may have to experiment a little with the record and playback levels to get things working properly. Once you've found the best levels, make a note of them for future reference.
You need to record a fairly strong code signal onto tape but not so strong that it crosstalks onto adjacent tracks. Likewise, don't put anything too powerful on the adjacent tracks as they could bleed onto the code track and cause misreading. Drums and bass guitar are to be avoided in this position if at all possible.
You'll probably find that a record level between —10 and —3VU is fine but there is one other problem area - noise reduction. Dolby C doesn't seem to cause any problems, at least I haven't experienced any, but dbx is definitely bad news as far as codes are concerned. If you can't disable the dbx on the channel you wish to record the code onto, I'm afraid that problems are likely.
Once you've got this far, bear in mind that any drop-outs in the code will lead to a loss of timing so don't neglect to clean your heads and also to store your tape carefully. A badly wound tape is asking for trouble because the edge tracks can easily be damaged and time codes are traditionally recorded on the highest numbered track which is, of course, an edge track. The reason for this is that it leaves only one adjacent track to cause crosstalk problems, but on a machine fitted with noise reduction, crosstalk is rarely a problem anyway. In this case, you may be safer recording your code on the second to edge track - for example, on an E16, use track 15. Since I've started doing this with my PPS1, I haven't had any misreading problems (code recorded at around —4dB on the level meters) and crosstalk has never raised its ugly head.
Another point, this time not so much a problem as something to be aware of when you buy a synchroniser: when you come to lock your sequencer to tape, the sequencer's MIDI In is fed from the synchroniser so you can't plug a keyboard into your sequencer to record new parts while listening to the tape tracks unless either your sequencer or your synchroniser has a MIDI Merge facility.
The basic synchroniser just described is relatively inexpensive and usually pretty reliable but this business about having to run the tape from the top every time you want to lock up is a bit of a chore. To get around this, various manufacturers, JL Cooper probably being the first, devised an enhanced FSK system which could record and read MIDI Song Pointer information as part of the code. This will only work with sequencers (or drum machines) that both generate and recognise MIDI Song Pointers but fortunately, most recent sequencers and computer/software sequencing packages do.
What happens is that when you start the tape at some point during the song, the next MIDI song pointer is picked up and sent to the sequencer which then sets itself to the appropriate bar and beat of the song, then starts. Depending on the synchroniser and the sequencer you are using, lock-up can take anything from a fraction of a second to a couple of bars, so it pays to allow a little pre-roll time.
Once the machines are in sync, the system that keeps them there is just like the basic FSK system so small drop-outs can still introduce timing errors. If the drop-outs are serious, then the sequencer will stop and then restart at the correct MIDI Song Pointer position but if you have parts playing when the drop-out occurs, you'll almost certainly hear an adverse effect on the timing till the system sorts itself out again, which could take another couple of bars.
As the code is still a form of FSK, all the precautions that apply to recording a basic time code signal apply here too. The big advantage is that you don't have to start from the beginning of the tape every time and you don't have to worry about tempo changes because these are followed automatically - the system still relies on MIDI clock and MIDI clock is related to tempo.
SMPTE, or 'Society of Motion Picture and Television Engineers' to give it its full title, is an entirely different code from the ones previously discussed. It was never designed for musical applications and carries no tempo information whatsoever - so just what is it and how is it useful?
SMPTE is a time of day code designed for use with film and video formats. When striping the tape, a burst of code is written onto tape once every film or video frame and carries information relating to the time in hours, minutes and seconds as well as individual frames. For example, you might see a SMPTE readout saying 09:45:30:10 which would mean 9 hours, 45 minutes, 30 seconds and 10 frames had elapsed since time 00:00:00:00.
SMPTE is quite a complicated code and uses no fewer than 80 binary bits each time it is written, but to use it for audio sync work, you can remain blissfully ignorant of all that and just treat it as a tool to do a job.
Though SMPTE is described as a time of day code, it doesn't normally relate to the actual time of day - you can set the code to any value you choose before striping your tape with a code track. But some SMPTE equipment doesn't like to run past the midnight hour of 00:00:00:00, their code readers get confused, so it is wise to start some time afterwards. In practice, if you're striping a tape, you might arrange it so that the first song will start at 01:00:00:00, a nice, easy to remember round number, and to allow yourself a little pre-roll time so the system can lock up, you'd start recording the code perhaps 15 seconds before this.
You probably know that the frame rate is different for film and television, and American TV uses a different frame rate from British TV. SMPTE can support four different frame rates, though for music only synchronisation, it doesn't really matter which you choose as long as you stick to it. Most people go for the European or EBU format which is 25fps. American TV uses 30fps while film runs at 24fps.
The fourth format, drop frame, is made necessary by the American NTSC TV system (said by British TV engineers to stand for 'Never Twice the Same Colour') which actually runs at 29.97fps and not exactly at 30fps. Is 30fps were to be used here, the error after one hour would be a little over three and a half seconds (108 frames) which is no good at all. To get around this, the drop frame standard discards or drops two frames at the beginning of each minute except every tenth minute and this loses exactly the right number of frames per hour. Having learned this, you can now push it into the darkest recesses of your mind unless you're planning to do a lot of sound for picture work for American TV.
But we still haven't said what earthly use SMPTE is for locking sequencers to tape recorders - after all, it measures time but it doesn't relate to tempo. The trick is to build a SMPTE generator/reader that can also convert time information into tempo information and then output it in the form of a MIDI clock which your sequencer will understand. And it must be possible to set a start point where the code turns on the sequencer using a MIDI Start command.
It is a fairly simple mathematical feat to convert beats per minute into the correct clock rate, but even here there is a possible pitfall. This conversion is traditionally performed in the SMPTE synchroniser itself, not in the sequencer.
Such conversions inevitably involve the rounding up or down of numbers so that the conversion to tempo may not be entirely accurate - it could be a significant fraction of a beat per minute out - or even more. This doesn't matter if you use the same machine to read and write the code, but if you transfer the tape to another studio that uses a different make of synchroniser, you could find your timing slowly but surely drifting.
If your piece of music has tempo changes in it, it isn't enough to program these into the sequencer because when you're locked up to code, it's the synchroniser that provides the tempo information. What you have to do is to program the value and bar location of these changes into the synchroniser before you can start work and this is rather tedious. Because of these added complexities, you should include as much information with the tape as possible if you're sending it anywhere else to be worked on at a later date. You'll need to specify the make of synchroniser you used, the frame rate you chose, the SMPTE start time, the start tempo and the locations of any tempo changes. This being the case you may ask why use SMPTE at all?
For a start, you must use SMPTE if you are working with video, but even for pure audio work, SMPTE is more reliable and short drop-outs may not cause the problems they do with the simpler systems. Furthermore, you can stripe a whole reel of tape before you start work and then just enter the start points and tempo data for the songs you record as you come to work on them. And, very importantly, SMPTE is an industry standard system which the others are not.
To make life easier, synchronisers are now becoming available that can 'learn' the tempo information from a sequencer and then store it back on disk as part of the song data. This makes the complexity of SMPTE virtually transparent of the user while retaining all its advantages.
Some major sequencing software companies are designing their own dedicated synchronisers which work from within the sequencing software, Steinberg's SMPTE-Track being one example which has been around for a few months now. C-Lab are also taking a similar route with their Unitor which actually synchronises the Atari computer's inner workings directly to SMPTE bypassing MIDI altogether, and this gives virtually instantaneous lock-up plus a tempo accuracy to four decimal places! It also has three merged MIDI Ins which can all be used when the systems are running in sync and two additional MIDI Outs giving effectively 48 MIDI channels.
Another advantage of this type of system, which plugs directly into the computer, is that you don't have to keep repatching your cables whenever you want to stripe the tape and you don't need an additional power supply as the computer's own power supply does the job for you. If your software manufacturer doesn't make a dedicated synchroniser, then there are a few intelligent add-on units like the Southworks Jam Box which will work out most of the tedious stuff for you and save it as part of your sequencer song file.
Lastly, if you intend to synchronise one tape machine to another, this operation also relies on SMPTE, so if you can envisage running two multitracks and a sequencer at a some time in the future, go for SMPTE straight away. You'll need to buy other hardware, of course, to interface the machines and to provide master transport control, and if you're thinking of going this route in the near future, then take a good look at what is on offer as some products overlap into both areas and could save both money and system complexity. Fostex have a wide range of products handling machine to machine sync, SMPTE to MIDI and audio to video sync. And they also have a very informative, free booklet called the 'Fostex Time Code Booklet' which I'm sure Harman will send you if you're interested.
Also on the horizon is the Tascam MIDIizer which will combine machine to machine synchronisation and SMPTE to MIDI conversion in one user-friendly box. Fostex are promising a similar unit to complement their R8 multitrack.
If you never intend to work with video and all you want is a system that will lock your tape machine to your sequencer without making you start from the beginning every run-through, then the MIDI Song Pointer systems are the simplest to use and are relatively inexpensive. You don't have to worry about tempo changes, start times or frame rates - you just plug in and go.
The Tascam MTS-30 is the cheapest I know of but it doesn't allow MIDI Merge so you can't overdub onto your sequencer while it's locked to tape. If you spend a little more, then you can get a JL Cooper PPS1 which does much the same but allows you to Merge at the same time.
Both systems are pretty reliable if you take care with your tape and also ensure the levels are correct, but you might find you have to play about a bit to get the optimum level onto tape. They don't seem so fussy about the level coming off tape. You can use a mixer channel with the EQ set flat to adjust the gain, but my own preference is to use my Boss active volume pedal so that the code never has to go into the desk at all and this must improve crosstalk.
SMPTE too needs to go onto tape at the right level, though there is usually quite a wide range over which the system will run reliably. Even so, I really do think that all manufacturers should fit an output level preset to their code generators so we don't have to bodge up ways of sorting the levels out ourselves - if you have a personal multitracker-type of recorder, you may not have a spare channel or a channel with direct outputs to use for the job and not everyone has an active volume pedal.
If you intend to work with video, then the capability to generate and read SMPTE is a must, but I'd still be inclined to look for one of the newer and more intelligent systems that works out the tempo details for you and stores them with your song data - after all, why make extra work for yourself, coping with the music is tough enough!
As to the near future, MIDI Time Code will offer the same kind of positional information that SMPTE does but being a purpose designed MIDI synchronisation system for musical use, it won't suffer from the compromises that we've had to get used to with existing systems. In theory, the MTC format has been around for well over a year but it is only just now starting to be implemented. In some ways it emulates SMPTE in that it carries precise timing information, but as that information is in a MIDI format, it can be piped directly into your sequencer, or whatever, and the time to tempo conversion carried out there rather than in some external unit. If you're working with SMPTE, you'll still need some form of converter to change SMPTE into MTC but these are relatively inexpensive and JL Cooper's PPS1s have always been equipped to do this by the changing of a simple internal link. MTC also allows for certain cuing information to be conveyed along with the timing so you may in future find that you're using MTC to initiate drop-ins and outs or to spin in bits of music from a sampler - though this will depend on the number of peripheral devices designed to work within the MTC framework. Such devices are likely to have their own internally stored list of actions which they will then implement under MTC command. In fact, all this is leading towards having a central studio control computer which will store all the necessary information to be down loaded into peripheral equipment at switch-on in readiness for the session. Once the music is rolling, all timing will be handled by the one computer giving a better degree of centralised control than we currently enjoy.
But all is not rosy - MIDI Time Code isn't as rock solid in its timing as SMPTE because it has to fit in the data stream along with all the other MIDI information and it may occasionally get delayed by a millisecond or two. This means that for really accurate jobs such as the syncing of one tape machine to another, good old fashioned SMPTE is still the way to go.
How quickly we start to reap the benefits of MTC depends on how confident the manufacturers are that it is what their customers want and, of course, it depends on the system being friendly enough not to further alienate those for whom MIDI is already a four-letter word.
Feature by Paul White
Previous article in this issue:
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!