Sync Or Swim

(main text from adambaby.com April 2010, updates added July 2010, May, August 2011, and April 2013)

Synchronising music hardware to software in the 21st century

If you are a music producer that works entirely ITB ("in the box") then what follows will probably have little interest for you. If like me you enjoy running your old (or new) drum machines, hardware sequencers, auto-arrangers and synth arpeggiators while building a track on your DAW (digital audio workstation i.e. computer sequencer) you probably will have run into problems getting them all to play nice together.

Sync problems have been around as long as recording, so it's no surprise that there's many different standards used. Since the nineteen eighties MIDI has been one of the commonest methods in studios that don't need video/SMPTE sync, however it is infamously sloppy, due to its serial nature, and the fact that it is generated by busy CPUs that don't prioritise timing duties.

So you may have wondered if there is a practical way to avoid "midi slop" and lock your hardware and software instruments together rhythmically. There are some options.

Innerclock Systems have just released a one-box solution, called the Sync Lock, reviewed on this page.

MOTU have Volta, an ingenious method of using a software plug-in to generate control voltages as audio. Similar to this is Silent Way, a suite of plug-ins. Both of these are aimed at modular analogue synthesizers, but the CVs they generate can be used for clock/sync purposes. Another neat option if you use old gear is Sync Unit DC, which generates din sync 24 from audio. It has some limitations. They all require an audio interface with certain electrical characteristics. I have yet to try them.

Or, if you have access to certain old drum machines and sequencers, you could try synthesizing an audio-level sync signal known as "FSK" in your DAW and sending this sync signal from an audio output to compatible hardware, as explained below. Once you've got it working, you can then do cool things such as get your 808 or 606 to shuffle and swing. You'll need a unit that reads FSK, also known as "tape sync", and puts out din sync and/or midi (models listed). You may also need some kind of a midi-to-trigger converter (options discussed).

Synthesizing FSK

"Tape sync" is the term some companies used for the recording and playback of a sync tone, known as FSK, or "frequency shift keying". The tone is a square-ish wave that shifts between two frequencies continually, high-low-high-low-high-low etc in equal periods. Each high-low shift represents a "pulse". For Roland gear, the high frequency is specified at 2400 hz and the low is 1200 hz, and they alternate at a rate of 24 pulses per beat (as in beats-per-minute). Musically speaking, whenever you double the frequency of any note, the note becomes an octave higher. So FSK is simply two notes alternating an octave apart, continually. The faster the octave pulse plays, the faster the tempo.

It turns out that the frequencies specified roughly correlate to midi notes D6 (note number 86, about 1174 hertz) and D7 (note number 98, about 2349 hertz).

             
Above: the FSK pulse pattern at 24 pulses per beat (quarter-note). Note the quantise grid value in Live at the lower right corner, which needs to be in a triplet division.

I've provided some midi files using these values, at the Roland rate. (Oberheim, Linn, and Korg used different rates).

These files were looped, playing a simple synth plug-in set to a monophonic square wave, unfiltered, with a gate-type amp envelope. By sending the audio to a dedicated output (it could be a headphone out, say) connected to the TAPE SYNC IN jack of an MSQ-700 and experimenting with the levels, eventually I got the external hardware to start recognising clock info, and midi and din sync clock started being transmitted. (You could also use a SBX-10, TR-707, TR-909, MC-202 etc, see below for quirks of various models).

OK, so my clock info is now locked to my DAW, regardless of tempo shifts. So far so good, now, how do I start and stop the external units?

For FSK-synced units to start at the same place each time, they need to be put into a "ready" mode by hitting Start (either on the front panel or by footswitch). The next high-low pulse that comes along will cause the unit to send a start message, whether it be midi and/or din sync 24. Luckily, most Roland rhythm units (and many others) have a footswitch for start-stop.

The ideal method would, like the clock generation, tie the start message to an audio pulse, for the best resolution and consistency. I found by sending a single "D7" pulse at the start of the bar, from the headphone output of my MOTU 828mkii to the start/stop footswitch jack of the MSQ-700, I could trigger the start mode. It requires a fairly healthy volume level, hence the use of the headphone output. If this trick doesn't work, it's time for Plan B.

Slightly less-than-ideal method: If you have a midi-to-CV converter, you can use a midi message to trigger the start-stop footswitch input. The converter may need to be able to trigger a short-to-ground (S-trig) or a negative going voltage change. If you don't have a converter that does this, there are lots of options around for any midi-to-X type of job you need to do, for example check out the kits at Highly Liquid.

Because there will be a certain lag between the midi message and the generation of the trigger, you'll need to experiment with negative track delays for that midi channel, which most DAWs let you do these days.

Above: a Live template for starting a new track, with a 4/4 kick pattern for adjusting the start delay, currently set at minus 85 milliseconds

I found I needed to send the start message 80 -100 milliseconds before the audio started on the DAW. Also it seemed to vary slightly with tempo so I made a note of the different values and made up a set of templates projects for different tempos. Having a simple 4/4 kick drum pattern recorded as part of my DAW template allows me to quickly fine-tune this with the hardware.

Using the audio pulse for Start/stop is virtually instantaneous, and doesn't require significant track delay, nor does it differ at different tempos.

In practice, my Live template has a track at the top for the one-shot start-stop message, next to a midi track loaded with a soft synth, playing the FSK loop, next to an audio track containing a 4/4 kick drum loop for alignment of the start delay. The FSK clip sits on every scene, the Start message, just where it's needed.

In Use

Having all my midi, din sync 24, and tape sync devices start up and lock to my DAW is a wonderful thing (I've waited far too long for it!). But there are also some musical applications available by controlling the clock pulses. Check out the midi files. After a lot of trial and error nudging the pulses around I came up with some useful variations that give varying amounts of shuffle and tempo shift. There's probably more to find.

Audio demo: here several Roland models (with a x0x in their name) are playing together controlled as described. Gradually, FSK clips with increasing shuffle amount are launched, then a half-time clip, then half-time shuffle, then back to straight time.

If you intend to play around with the timing of these files, just remember to leave no gaps between the notes, but to stretch out the lower note D6 (the pilot tone) to fill up any space. Dropping the tone out altogether tends to really mess up the sync.



Above: the pulse pattern for a shuffle rhythm. The first 6 pulses are straight on the quantise grid, the second 6 are dragged back a little, with the last pulse of the second group being shortened to allow the first pulse of the next group to be back on the grid again.




Above: the pulse pattern for "12 off, 12 on" which can give a half-time shuffle feel.



Above: the pulse pattern for "2 off, 4 on"

A word on the midi files for the clock pulses. Not all will have their loop points set correctly when dragging into your DAW - make sure they are looped end to end. For some reason I found that with the half-time (e.g. "6 off, 6 on") and three-quarter time ("3 off, 6 on") files, it worked best if the file started with the lower, pilot tone (i.e. note D6) part of the pattern, rather than the high pulse part. Your mileage may vary. Also, not all the files work for all tempos or rhythms!

Midi files

FSK24ppqn this is the regular sync pattern

SHUFFLE_1

SHUFFLE_2

SHUFFLE_3 increasing amounts of shuffle or swing applied

3OFF3ON half tempo

6OFF6ON half tempo, a different feel depending on how busy the rhythm is

12OFF12ON half tempo shuffle

2OFF4ON three-quarter tempo

3OFF6ON three-quarter tempo, different feel

6OFF12ON three-quarter tempo, ditto



If you use Ableton Live as your DAW and you are just after the regular type of shuffle rhythms, using the global shuffle in Live is easier than using a specific clock clip, but remember to set the main FSK sync clip to "Swing 16" for it to work.

Overall, the system works well when using the audio start message, but when using the midi message I could notice slight differences in feel between start-ups. Maybe one time out of seven would be just a fraction out, requiring a re-start. As I mentioned, the amount of track delay that was needed on the start track was different for different tempos. Conceivably this could also vary depending on how busy the midi interface was at the start point. Of course, it is this very inconsistency with midi transmission that we are trying to avoid! You could perhaps get around this by some kind of audio-to-trigger plug-in or circuit, such as the trigger plug from Silent Way.


Update April 2013:

Ok, so I have found there is a much more reliable and repeatable way of starting the hardware: when you send the start pulse don't start the FSK pattern straight away, send a bar (or more if you want) of just the pilot tone, followed by the FSK pattern. In Live, you can do this automatically by using a "Follow Action" in the pilot tone clip, so that it plays for however long you want, then the next clip starts. Yes, the trade-off is waiting for a bar to lock your hardware up, but it is such a reliable and solid method that it's worth the wait. Just pretend you're back in the good old days of syncing to tape!



If faffing about with obsolete second-hand gear and midi-CV converters doesn't appeal to you, there is a new, compact, high-resolution device that works even better. Enter...




The Innerclock Systems Sync Lock

What got me fired up to find a better sync method for my studio was discovering the Innerclock web page a couple of years ago and carrying out the "litmus tests" described there. The tests confirmed things were a bit of a mess. What I had been talking myself out of for years was there in the numbers - I could deny it no longer! Unfortunately, their proposed solution, the Sync Lock, was, at that time, not yet finished. After a long wait, finally, it's here!

David from Innerclock has come up with a brilliantly simple one-box solution to this problem. No second-hand hardware, midi-to-CV converter, or particular audio interface model needed either. The principle is probably fairly similar to what I ended up doing with FSK above, only with much finer resolution and recurring precision. An audio track on its own dedicated output provides clock info which is converted by the external box into 1) midi clock, 2) din sync 24, and 3) a 5 volt trigger, sixteen to the bar, giving you a variety of sync options. A second dedicated audio track provides the start/stop info for the unit.

What you get with the box (which by the way is a robust, compact and techy-cool design in black and chrome) is a footswitch, a dual-function connector cable, a PSU, an operating manual and a CD. The CD contains an audio file of the pulse needed to drive the inputs, and also a plug-in called "Sync Gen" in AU and WIN VST. You have the option of generating the clock via the plug-in, or from an audio file from the sample provided. Innerclock don't guarantee the plug-in will work with all DAWs, hence the audio sample, which will. Obviously, you need to dedicate two mono audio outputs from your interface to the Sync Lock. I hung a cheap Behringer ADA 8000 off the optical ADAT connection on my MOTU 828 mk2 interface, which did the trick (leaving room for another 3 Sync Locks!).



Above: the red LED on the Sync Lock shows the stopped state. Also at this end of the unit is the midi out, the dual socket breakout cable for the trigger (black) and footswitch (red) connections, and the din sync out.

Connections


Power is 9 volts, around 200mA, via a barrel connector, negative tip (In other words, your regular Roland/Boss guitar pedal power plug). You may need both midi and din cables to connect the box to your external devices. The dual function cable is a tip-ring-sleeve (TRS) standard instrument plug, that is broken out to two mono line sockets. One of these is an output for the voltage triggers, the other is an input for the footswitch. A little clunky perhaps, but given the compact size of the unit, the need to minimise the number of jacks is understandable. The footswitch is necessary to reset the unit (i.e. force a stop message) in the case of the DAW being stopped mid-sequence, when a start-stop pulse has not been triggered. I found this really handy (if that isn't a contradiction when discussing a foot switch).

Setting Up the DAW


Setting up the audio files for the clock track and the start-stop track is walked through in the manual, and requires a bit of copy-and-paste of the click sample (a wav file is provided) to construct a one-bar loop, for the clock output, and a 1-bar clip with a single click at the very start, which is not looped usually (although it can be - for stutter starts!). Once done, saving the project with these files as a template means you don't have to bother with it again, in theory. However, I had some problems using the audio sample as the clock track.

As you've seen, I'm using Ableton Live as my DAW (v.5.2.2). If Live was set to a certain tempo, say 120 bpm, when I put together and consolidated my 1-bar loop from the click wav file, then it would run quite smoothly at that tempo and say 8 to 10 bpm either side of that. Outside of that range, the Sync Lock would lose sync with the DAW. I put together 1-bar loops at different tempos, and again, they worked fine only within a certain range of the original tempo. Changing the warp mode from "Beats 1/16" to "Beats 1/32" just allowed a slightly wider range of variation. In addition, I found the click loop didn't respond to Ableton Live's "Global Groove" (shuffle) control as it should. Something happened, but I don't think you'd call it shuffle! Changing the clip's own swing or warp settings seemed to have no effect here.

In practical use, where radical tempo changes aren't needed, the click sample loop worked fine. The sync was consistently rock-solid and unvarying between my midi clocked devices, the din sync clocked ones, and the loops running in Live. Fantastic!

So, what happens if you need rollercoaster tempo changes in your track, or would like a little or a lot of swing to help the groove?

This is where the Sync Gen plug-in comes in, providing another dimension of control over the rhythm. Currently there are WIN VST and MAC AU versions - no MAC VST yet. Sync-Gen provides the clock impulses instead of the click loop. The Start-Stop clip is used as before. The plug-in's GUI shows a representation of the pulse train as a horizontal strip of short vertical bars, a bit like an 808's strip of LED switches. Clicking on these toggles them on and off. By playing with this while the pattern is running, you can drastically or subtly change the feel of the rhythm.




A "randomise" button mutes a random pattern of these pulses each time you hit it, which usually gives a lurching feel a bit like a drunken sailor! See the demo below. There is a dial for shuffle, from 0 to 100%, and various controls for adjusting how the plug-in reads the pattern of pulses. Multiple patterns can be saved in the plug-in.

I would have liked to have seen these patterns available as midi clips (see update!), which would then allow you to build a library of "feels", and be able to launch them at any time. You may be able to achieve this with plug-in automation however.

Stutter


There are creative possibilities using the start-stop control too. Due to it's high resolution, stutter starts are a breeze, especially in Live, where you can set a clip to trigger on repeat for as long as it's held down. The trick to this is making sure you have the right number of clicks to end up in a RUN state, not a STOP state! Playing with this revealed some curious behaviour when shuffle had been dialled in. It sounded as if the plug-in's sync pattern had been offset, so that almost an "inverse" shuffle was playing. Normally, a shuffle or swing feel is obtained by delaying beat subdivisions just ahead of the downbeat. After stutter starts, or occasionally after radical tempo changes, the delayed pulses occurred with a different relation to the downbeat, causing a feel that wasn't exactly swing, but was nonetheless quite interesting, especially since all the machines stayed in lockstep!

Audio demo: a bunch of old Roland gear synced by one Sync Lock, starting out in straight time, slowly increasing the shuffle percentage, and then a couple of stutter starts. Note the curious feel after some of the stutters - fixed by a re-start message, eventually! Then, a smooth tempo increase, until the randomise button is hit at 1:23, cue drunken lurching, all pulled back into line at once with a new pattern from the Sync Gen plug-in.

Audio demo: Check that swing! From straight time to swing in a few bars, then a synth arpeggiator chimes in, swinging, triggered also by the Sync Lock.

Voltage trigger


The voltage trigger is positive 5 volts, 20msec pulse, sixteen to the bar. I tested it with my modified TR-77, which runs at 4 ppqn, that is, sixteenth note pulses. It worked well. And, the sixteenths swing in response to the Sync Gen shuffle control, so you could for example "swing" your synth arpeggiators to the groove. If you have more than one destination for the single voltage trigger provided (e.g. the VCA on your modular, and the Arp clock on your Juno-60), usually all it needs is a Y-splitter cable which you can buy or make yourself.

Conclusion


Despite a few hiccups in some applications, the Sync Lock works very well. The sync was so solid that the tuning of the phasing caused by simultaneous kick drums didn't vary from bar to bar!! I was gobsmacked! Overall, I think the Sync Lock is a winner because it uniquely solves a long-standing problem of synchronisation in the modern studio, while bringing with it new creative possibilities. Check out the Sync Lock page for more info.


Update July 2010

Reverting to my old sequencer (Cubase VST 5 on the Mac OS 9.2), I was forced recently to take another look at the options for Sync Lock users that don't have access to the Sync Gen plug-in for whatever reason. In doing so, I came up with a rather neat answer to my stated wish above that the sync patterns could be in the form of midi files which would allow groove customising.

I hit upon the idea of using Propellerheads Recycle program to analyse the groove of the output of the Sync Gen plug in, and exporting that as a REX file into Cubase, which would then output audio pulses to the Sync Lock. It worked perfectly! And with an added bonus... getting Recycle to also export a MID file of its analysis, you can then bring that MID file into Cubase, and assign a REX player (e.g. such as Phatmatic) to play the REX file under control of the MID file, thus giving you complete control on a pulse-by-pulse basis of the groove of your external machines. For example. you can nudge groups of pulses around to give you a shuffle that increases towards the end of the bar. Or, increase the shuffle just in the fourth bar of a group, or in the breakdown, say. Or, push the snare on beat 2, and delay it on beat 4, etc... You get the idea - lots of creative possibilities! To give you a graphic shot of how this might look in a midi file editor, see below...

Above: Midi file for control of a REX player, as seen in Cubase VST midi editor. One bar is shown. The first half of the bar shows pulses in straight time, the second half shows pulses in swing or shuffle time. Note the elastic, sinuous shape of the line of notes in the second half.

Having midi note control over the sync pulses works just like the FSK system we started with - grouping or nudging the notes around gives similar feels. For example, having a group of twelve pulses muted, followed by an active group of twelve gives a half-tempo shuffle feel. So the midi files I've provided will work here also. Just delete the "D6" notes, then transpose the "D7" notes down to a lower octave so your REX player will see them. And of course you can use this technique in any DAW where you can use a REX player as a plug-in, although you may need to adjust the audio settings so that the click slices aren't mangled too much by the player.

Below I have provided some REX files of the Sync Gen plug-in. These can then be easily tweaked for your own purposes. Note the REX files were created in OS 9.2 on Recycle 1.6. To get modern players to recognise them just replace the ".aif" with".rex".

Sync Gen Swing 00 (no swing)

Sync Gen Swing 10 %

Sync Gen Swing 20 %

Sync Gen Swing 30%

Sync Gen Swing 40 %

Sync Gen Swing 50 %Sync Gen Swing 60 %

Sync Gen Swing 70 %Sync Gen Swing 80 %

Sync Gen Swing 90 %

Sync Gen Swing 100 %



Update August 2011

Innerclock have been very busy of late. The Sync Lock is getting a minor hardware upgrade, to accomodate a new software front-end. It's called the Sync Gen II, and the plug-in is AU and VST compatible. More info here, and available soon.



"Smart FSK" April 2013

Timing sync in this studio has been tight and trouble-free since the arrival of the Innerclock Sync Lock/Sync Gen system. The Sync Gen takes care of the clock duties, the FSK track drives a Roland hardware sequencer (MC-50mkII) which takes care of midi notes etc, leaving Ableton Live to do it's magic with audio, avoiding its midi weaknesses, yet remaining the master conductor of the whole show.
While this is fine for loop-based song creation, wouldn't it be nice when working on a more linear arrangement, e.g. intro/verse/chorus etc, to have the hardware chase the song position of the DAW? With just a click on a locator point, you could jump immmediately to the area you are working on, and all your hardware would jump to that point too.
Well, that's just what "Smart FSK" allows you to do. At least, if your hardware recognizes it, it does. Smart FSK provides the usual timing clock pulses of FSK, however, embedded in those pulses is an additional code. This code represents a midi mesage known as Song Position Pointer (SPP). It is updated every sixteenth note, and is basically a counter that tells the hardware how many sixteenth notes have passed since the start of the song. When a machine recognizes a SPP, it updates it's own beat count and then begins to count clock pulses from that point, ignoring any further SPPs.

I could find no details of the Smart FSK specification on the internet, so to try to understand how the extra info was embedded in the FSK pulses, I analyzed a recording of Smart FSK from my MC-50 sequencer. I noticed that the high frequency pulses were different to that of regular FSK. In FSK they are all of equal duration, in Smart FSK there are short ones and longer ones, but their leading edges are all still regularly spaced as in FSK.
My understanding of this was that it could be binary code. However, I had no way of knowing how Roland actually implemented this code in the pulse-generating/pulse-reading hardware in the MC-50, or how it related to the actual byte format of the midi SPP.
So I transcribed the code from a few different bar numbers on to paper as a series of dots and dashes. Since a sixteenth note at midi resolution (24 PPQN) is counted by 6 pulses, I wrote out the pulses in groups of six. Certain patterns began to emerge, which helped speed up this fairly tedious process. I made an eyeball estimate of the mark:space ratio of the short pulses, then wrote a midi file with the notes in the pattern of the code.

Essentially, it worked!
Start Live playing an arrangement, and in under 2 seconds the hardware is locked on to the song position and playing in sync!

There are a few quirks I've noticed. Very occasionally it will get bar numbers confused - quickly fixed with a stop/re-start. Also, the MC-50 seems to prefer to start seeing the FSK from somewhere later than the start of the bar, rather than from beat one. So by putting my Locate points at beat 2 of the first of a two bar segment, it will be locked on by the end of the second bar. There are more idiosyncrasies that I think may be due to the very short midi notes used - extreme amounts of swing don't work, and neither do tempos below about 80 bpm. I hope to find ways around these problems.

Now of course, it would be a really tedious process to generate a unique midi file for each bar. What I have done for the present, until someone invents a plug-in that can do this, is take a few 2 bar segments from between bar 1 and bar 177, spaced out fairly evenly so as to be useful in most arrangements, for jumping to a point closest to where I'm working. The workflow is quick, the computer keys control playback, and there is NO MOUSING NEEDED!

Smart FSK Sync template2 download

Here is an Ableton Live set created on version 5.2, that contains the midi information I use. Switch to the Arrange page. You will need to have a simple soft synth loaded into the FSK track, programmed to have instant attack and release, using a monophonic squarewave with no filtering, no velocity or modulation. I used the freeware TAL Elek7ro II VST synth. Send the audio to a spare mono output plugged in to your hardware's "Tape Sync In" jack, and fiddle with the levels until you find the sweet spot.

I'd love to make the info available as a simple midi file, however I find that the ultra-short notes never seem to survive the translation.

What Is DIN Sync?

(borrowed partly from the Jambox 4+ manual, copyright ©1987 Southworth Music Systems)

Before the midi standard (1983) there were a number of sync types for sequencers and drum machines. The commonest was TTL 5 volt sync. This type of sync is a series of 5 volt pulses sent over a wire. The change in voltage from zero to five volts represents a clock pulse. Tempo is calculated from the time between clock pulses. A musical beat, i.e. a quarter note, is divided into a certain number of clock pulses. The number of clock pulses per quarter note (ppqn) varied between manufacturers. DIN sync is just a type of 5 volt TTL sync that uses a five pin DIN connector. It can be 24 ppqn (Roland) or 48 ppqn (Korg) 96, 192, or even 384 ppqn.

DIN TTL sync is usually sent using two wires, both of which send zero to five volt transitions. One wire sends clock pulse transitions constantly, regardless of whether the machines are running or stopped. The other wire is the start-stop line. A voltage of zero indicates stop mode, a voltage of five volts indicates play mode. The third wire used is for ground

                                   
Like midi, DIN sync uses a 5 pin 180 degree DIN connector. However, it doesn't use the same pins. Pins on male connectors are numbered (from right to left, viewed from outside of the connector, with the 5 pins upwards, and facing them): 1–4–2–5–3. Holes on female connectors/jacks are also numbered 1-4-2-5-3, but from left to right (facing the holes). Pin 2, the centre one, is ground in DIN sync and midi (although ground is only connected at midi outputs, not inputs). Midi uses pins 4 and 5, DIN sync uses 1 and 3.

You need to know this if you're going to make up special cables that tap off these voltages for different purposes (say, to send to a non-standard machine), or if you want to send your sync signals to different parts of your studio. DIN sync suffers minimal loss of signal when you connect multiple devices, so it's easy to make up a passive distributor for the clock and run/stop voltages, essentially you are just taking multiple taps from each wire. It's a voltage, not a serial stream of data like midi, so its instantaneous, there is no lag. My homemade DIN distribution box is seen here next to the Sync Lock.

TECH NOTE:

Florian Anwander of the Synth DIY list recently brought the following details of the DIN sync spec to attention, concerning the requirement of tempo clock pulses to be reset for 9 ms when a "Start" pulse is received. As explained on page 5 of the Roland TR808 service manual:-





(copyright ©Roland Corp)



Sync quirks of various models

Roland were big on tape sync. It featured in many of their products in the seventies, eighties and early nineties, starting with their first sequencer, the MC-8 in 1977, right up to the MC-50 mark II in 1992. In machines such as the TR707, SBX-10, TR-909 or the MSQ-700, both midi clock and din sync 24 are generated when tape sync is received, which makes them very handy for linking midi gear with pre-midi systems. Some machines exhibit a very annoying behaviour - when the tape sync pulse stops being received, they automatically drop out of tape mode, requiring a manual reset before they will sync again. Both the 707 and the 909 do this, so if you're considering using these as your sync hub, have them positioned handy to your work area!

The TR-909 may also sync poorly if slaved to midi clock due to a bug in the software. Machines with V. 1 ROMs can "lose" a midi clock under some circumstances. This can be fixed by upgrading the ROM chip to a later version.

Other manufacturers that used FSK in some form were Linn, Sequential, Yamaha, Korg and Oberheim.


Above: detail from CR5000 manual showing connections at rear panel (copyright ©Roland Corp)

Some preset machines have no apparent sync options, but often will sport a start-stop footswitch jack. A midi-to-trigger circuit of some kind will usually work here (as mentioned in the FSK section), with the midi messages at the start and end of a one or two bar loop, forcing a re-start to keep the sync. Obviously you'll need to play with the tempo a bit, but great results can be had. You're not restricted to playing the presets as they are programmed either, by messing with the tempo and syncopating the re-start messages!

Roland

MC-8 - CV and gate sequencer, din sync, tape sync

MC-4 - ditto

MC-202 - 2 track CV/gate sequencer with built in monophonic synth, din sync, tape sync, no midi



MSQ-700 - infuriating (!) early midi and DCB sequencer, much more useful now as a sync interface between tape, din and midi

MSQ-100 - midi sequencer, midi and din sync, no tape sync

MC-300, 500 series - midi sequencers with tape sync, no din sync

MC-50 - midi sequencer, used "tape sync 2" or "smart FSK", which was FSK with song position information encoded in it. Could also read the original tape sync signals, say, from an MSQ-700.

TR-808, 606 - din sync 24 ppqn

CR-5000 - preset drum machine with analogue voices. On first inspection it seems there is no sync option, but this machine features a start-stop and a re-start jack for a footswitch, which can be triggered by a converter as explained above. Other remote control inputs are for Intro/Fill and Register, for changing patterns on the fly. Also a voltage trigger output (5 volts, 10 msec) that occurs on the eighths, sixteenths, or in time with the Accent pulses. A trick with these machines is to plug the trigger out into the Register jack, to switch between two patterns of your choice creating a hybrid rhythm, makes up a bit for the lack of user programs.

CR-8000 - presets as well as user programs, big brother of the 5000, has din sync, as well as the remote control footswitch jacks and trigger mentioned above.

TR-909 - analogue and digital drums, see bug mentioned above. Needs manual reset to tape sync mode. Only has din sync in, not out.

TR-707 and 727- digital drums, needs manual reset to tape sync. Generates both din sync and midi clock when syncing to tape, generates midi clock when syncing to din, but alas does not generate din clock when syncing to midi! Generates all three signals when used as master.


Boss Dr Rhythm DR-55 - analogue voices, start/stop jack, trigger voltages out

Boss Dr Rhythm DR-110 - analogue drums, no sync, Accent trigger voltage output

TR-626 - digital drums, FSK, midi

MPU-401 - early midi computer interface, with tape sync

SBX-10 - good all round sync box, FSK, din sync at 24, 48, 96 ppqn, and midi

SBX-80 - professional studio sync box, SMPTE, no FSK, din sync and midi.

R-8 Rhythm machine - reads FSK and outputs midi clock, no din sync

R-70 Rhythm machine - Smart FSK (Tape Sync II), midi



Korg

KR-55 - preset drum machine, analogue voices, no sync, only start-stop footswitch input

KMS-30 - Korgs most useful sync device, midi, din sync as 24 or 48 ppqn, tape sync

KPR-77 - analogue drum voices, din sync 48 ppqn

DDM-110 and 220 - digital drum voices, din sync 48 ppqn



Yamaha




YMC-10 - tape sync to midi converter

RX-11 - digital drums, midi, no FSK, can be synced from external 5 volt pulses at the "cassette in" minijack, with a selectable timebase 24, 48, 96 ppqn

QX-1, QX-3, QX-5 - midi sequencers, tape sync

EMU

Drumulator - FSK, din sync 24 48 and 96 ppqn, serial data, and, later, midi

SP12 - midi, SMPTE

SP1200 - ditto

Others

Southworth Music Systems Jambox 4+ - a brilliant, early Mac interface that featured SMPTE, MTC, midi clock, din (TTL) sync at 24, 48, 96, 192, and 384 ppqn (set by internal jumpers) and could also follow an audio click track!

Linn LM1, Linndrum - external clock, FSK on some versions, the Linn 9000 had midi and SMPTE

Kawai Q80 - midi sequencer with FSK



My background with sync issues

After learning the basics of multi-track recording on a cassette portastudio, I graduated to a Roland MC-50 sequencer synchronised to a Fostex R8 reel-to-reel tape machine. The sync system was called by Roland "tape sync", and involved recording a pulse generated by the MC-50 onto one track of the tape. You needed to be sure in advance what tempo and just how long you wanted the track, there was no changing it later. When the tape played back, the MC-50 would pick up the pulse and run in tempo with it, allowing you to record vocals and guitars etc while the midi parts played "live" in sync with the tape.
It was very reliable - I can't remember it ever failing!

In the late nineties I moved to a computer DAW (Cubase) and hard disk recording. Immediately I noticed problems when synchronising my hardware drum machines and pattern generators to my software sequencer. If they started well, the timing would tend to slip, or, from the start there would be a lag in the machines that often progressed into chaos. Eventually I worked out a system using Kenton MCV converters (to convert midi to DIN sync) and a dedicated midi output whereby the clocked devices all received the same clock source at once. I found my assortment of Roland "digit-zero-digit" boxes ran much tighter off the older voltage clock protocol rather than midi clock. Things improved even more with my moving over to using Ableton Live as my main DAW. Ableton LIve has the ability to offset the start of midi clock coming from the interface against the start time of the audio tracks.

However it was always a compromise, and it wasn't until about two years ago when someone alerted me to the Innerclock website, that I realised how much. The "litmus test" described there demonstrated that the timing info in my studio was all over the shop! Dismayed, I actually went back to using hardware for a while, and using Live as a simple recorder to capture the performance. These tracks exhibited a noticeably better feel.
Innerclock then announced the imminent release of their "Sync Lock" - designed to sync your midi or DIN sync rhythm devices to your DAW. At last! I thought, this is what I've been waiting for! But something was niggling at me. What did this system remind me of? Of course, it sounded a bit like good old "tape sync".

The Sync Lock was some way off being finished, so I started taking another look at tape sync. I guess I'd always assumed that because it was older technology than midi, it would be inferior. But think about it - midi is a serial stream of data. If anything happens at either the transmitting end or the receiving end where the CPU has to attend to something else, then there is a delay. It might be microseconds, it might be milliseconds, but the delays will add up, and not in any musical way. There is clearly an advantage to be gained by keeping the pathway for the timing info clear of the midi data stream. And that's what tape sync does.

My experiments began by recording a segment of tape sync from my MSQ-700, editing it in Ableton, and looping it as a clip, then playing this channel back to the MSQ. It didn't work very well at all, after a couple of bars things clearly fell apart. I messed around with edit and loop points with only marginal improvements. It wasn't until I looked at just what FSK was, exactly, that I realised I didn't need to stuff around getting the sample to sit right, I could synthesize it directly, allowing freedom for tempo changes and shuffle etc, (See: Synthesizing FSK, above).

Comments