From nwilson@morgan.com Mon Dec 5 13:38:58 1994 Date: Mon, 5 Dec 1994 15:20:46 -0500 From: Ab Wilson To: electronic music mailing list Subject: Re: TG 77 tips Michael Dvorkin writes: > Can anyone give me any tips on programming. I am not that much of a > FM expert.... It seems that all I ever do these days is give people tips on how to use the TG77. I ought to make my own FAQ. Before I get into FM I want to point out how to get good sounds out of the filters (it took me ages to figure this out). The TG has got great filters but they are a bit of a pain (to say the least) to program. First thing: each element of in a TG actually has two filters in series - a 2 pole low pass, and a 2 pole switchable high pass/low pass. This setup gives a great deal of flexibility in creating sounds but also has a lot of (cunningly interdependent) parameters to program. Despite all this flexibility, most of the time (if I guess correctly) you're going to want to make it sound like a 303 (or some other trendy plastic box) in which case you're going to want to configure the filters as a resonant 4 pole low pass. To do this just set both filters to low pass (this is the only mode in which the resonance has any effect) and (an important and) give them the same settings for everything (including the envelopes). If you don't do this the resonance won't be half as pronounced. (Note the resonance control works for both filters in tandem - I don't know why this is but I suspect it's something to do with digital filter algorithms and feedback loops.) Try this: set both the filters to low pass set both their frequencies to about 200Hz, then turn up the resonance until you get self oscilation. Now change the frequency of one filter. For small changes you'll still get the oscilation, although less intense. But change the frequency by more than a couple of notches and the oscilation dies. (This control actually becomes very useful when you know what you're doing since it allows you to vary the width of the resonant peak and generally mess around with the shape of the filter slope.) If you now set the frequency of the other filter the oscilation will come back. I wish Yamaha had added some form of parameter link function because this setup makes setting the filter that little bit less intuitive - instead of just moving one control you have to go through an iterative process involving each of the filter frequencies. So tip number one - get into the habit of using filter copy whenever you change settings of one of the filters (especially the envelope settings). The second thing that gave me grief (until I worked out what the manual was trying to say) was how the hell do I get long filter sweeps. The confusion here is two fold: Firstly the TG has no free running LFOs. Every time you hit a note the LFO starts again from whatever initial phase you set. So you can't program fast arpeggios with slow filter sweeps over the top (although there are hacks - see below). IMHO this is the single biggest flaw of the TG. I said some very rude words when I discovered this. On top of that the TG's filter modulation routing is bizarre and badly documented. If you look on the filter cutoff page on the far right you'll see a column that probably has ENV written in it. This is the filter modulation routing. Each filter can have it's own routing. There are three choices of routing which are confusing, badly named and non-orthogonal. From memory these are: o LFO - the filter is modulated by the LFO and also in realtime from whatever midi CC is assigned to filter cutoff. It's envelope is turned off. o ENV - the filter is modulated by it's envelope. o VA-ENV - the filter is modulated by it's envelope and by the level of the midi CC assigned to filter cutoff sampled at key-on. ie no envelope plus LFO mode. I've been reading up on how digital envelopes are implemented and this actually makes sense to me now. Also since the LFOs restart on key-on anyway it's not a great loss. Any road up, the first one of these isn't that useful unless you want fast wobbly sounds or you want to control the filter in real time. I haven't figured out why Yamaha bothered with the second mode at all (since if you don't want midi filter cutoff control you can always turn it of from the common page). The third mode however lets you make squelches with the envelopes and slowly sweep the filter using midi CCs (probably from a sequencer). This is how I usually go about making slow sweeping bleepy noises. So tip number two for getting analogue imitations out of the TG is use VA-ENV (I might have the spelling wrong) mode for filter mod. A couple of other quick tips before I finish with the filters. Unlike every other synth in the known universe the filters in an AFM element of the TG come after the VCA (well there aren't really any VCAs but conceptually in FM the carriers act as VCAs). So if you make an AFM sound with filter oscilation it will go on and on for ever unless you do something to stop it with the filter envelopes. Alternatively you could use an AWM element. Also the TG's envelopes are linear were as nice envelopes are exponential. You can compensate for this by using the extra steps to aproximate an exponential curve - hassel but worth it. I've just realised that this message is too long as it is so I'll have to leave the FM tips for another day. BTW there is a mailing list for Yamaha SY users: sy-list@chorus.fr Could be better - too much stuff on SY85s and sample dumps - but useful for information. Ab. From nwilson@morgan.com Tue Dec 6 09:06:40 1994 Date: Tue, 6 Dec 1994 05:43:31 -0500 From: Ab Wilson To: Herbert Janssen Cc: electronica@andrew.cmu.edu Subject: Re: TG 77 tips Herbert Janssen writes: > ENV: modulates the filter cutoff by the envelope and sets the > initial frequency at key-on. > The velocity changes the ***cutoff level*** ! > VA-ENV: modulates the filter cutoff by the envelope and sets the > initial frequency at key-on. > The velocity changes the ***envelope rate*** (I think it is > only the attack rate) ! I'll have to check up on this when I get home but that sounds like it makes sense. > I definitely know that the frequency in ENV mode can be controlled via > CC because I do that all the time. It's definately the case with the TG-77 that you can only use CCs to control the filter in realtime if you're not using envelope modulation. If you use envelope mod then the CC level is sampled at key-on. (In practise for short bleepy sounds this isn't a big limitation.) Is this true of the 99? The more I think about this the more I want to go home a read the manual again. > BTW, the 99 has one tiny feature the 77 lacks: you can "link" the > settings for both filters. It also has better FX routing if I recall doesn't it? It sounds to me like the software in SY-99 was result of people's comment's on the faults of the 77. Now if only they'd made a rack version of the 99 with all the outputs. > > So you can't program fast arpeggios with slow filter > > sweeps over the top (although there are hacks - see below). > > Did I miss something or did you skip this "below" ? > I'd love to hear how to do that. I mentioned using sequenced midi-CCs to simulate free running LFOs but that's such an obvious hack that you probably didn't see it as such. The other way (if you don't mind loosing envelope control) is to put the voice in mono mode and play legato. This is particularly useful for long evolving FM sounds. It's when you start doing things like this that conventional software sequencers (I use Cubase) really start to piss you off. Only a single loop for the whole song - so much for running an LFO with a period of 3 bars over a 4 bar loop. Cannot play legato over a loop point - so much for sounds that evolve forever and ever (on the TG at least). Send `reset all controllers' at loop start without any means of turning it off - I always wanted a horrible glich at the beginning of all my loops. What I want is a package that just does what I tell it. Anyone know much about Max? Ab. PS I'll post some basic tips on FM when the boss isn't looking over my shoulder. From nwilson@morgan.com Tue Dec 6 14:31:54 1994 Date: Tue, 6 Dec 1994 15:36:20 -0500 From: Ab Wilson To: Michael Dvorkin Cc: electronica Subject: Re: TG 77 tips I write: > > FM is IMHO pretty good for bass sounds (well FM bass sounds at least). Michael Dvorkin writes: > Cool! I've never though that FM was good for that. I wanted it for evolving > pads. And some purely wierd sounds. But I guess I can use it for much more. > Interesting!.... I suppose I'll leave the pads part to my WaveStation.... It's pretty damn amazing for evolving pads too - in one AFM element you have six loopable envelopes and two LFOs. You can make sounds that go on and on for hours. Hmm, maybe it's time for some general FM tips. Ok, first tip is take it easy. This holds for everything - the way you go about programming a voice in general, operator output levels, the way you use the envelopes - everything. The most important of these is your overall approach to programming. Start off really simple and only get more complicated as you know what you're doing. Set up two operators in an modulator/carrier pair and turn all the rest off (you'll be doing that quite a bit). This gives use something like this: op2 -> op1 For now set both ops to 1:1 frequency ratio. Go to the output page and select the subpage that gives you all the operator levels as a bar graph. Set the output of the carrier (op1) to full and now increase the level of op2 (the modulator). This will give you a feel for the general workings of FM - ie the higher the level of modulator (the modulation ratio), the brighter the sound. This is almost but not quite like changing the cutoff frequency of a lowpass filter. If you turn the level up too far you'll get nasty noises which start off sounding distorted and end up at full modulation ratio as gritty digital white noise. This leads to the second point - take it easy with modulator output levels unless you want all you're voices to sound like finger nails scaping down a blackboard. So now you should have two ops stacked up making some sort of agreeable sound. There are a number of directions you can go in from here. Firstly you can go to the operator page and play with the frequency ratio of the modulator. Briefly, the higher the frequency ratio, the more spaced out the partials in the resulting waveform will be. At a ratio of 1:1 you get all the harmonics, at higher ratios some get missed out. I don't know the exact figures here but I think a modulation ratio of about 70% and a frequency ratio of 1:1 gives you a waveform with the same harmonic content as a sawwave, using a frequency ratio of 1:2 should give you something like a square wave. (Using feedback you can get a sawwave using a single operator.) That covers the two basic dimensions of a static, two operator FM sounds. The fun begins when you start adding modulation and extra operators. Next go to the envelope page. I find when programming the TG I spend the most time playing with envelope settings. You can spend a very long time on this. To get a feel for what's going on select the envelope for op2 (the modulator). Select the rates page and move the cursor to attack rate one. Set the rate to just less than full speed. The sound should now have a nasty spikey attack to it. This puzzled me for a long time - why would slowing down the attack make a spike at the begining of the sound? Well I'm not sure about this (ie I haven't heard it from a Yamaha designer) but IMHO the reason is to do with the sampling rate of the control sources in the TG. Basically the control rates are a lot slower than they could be. So if you set a very fast (but not instant) rate over a big range you'll end up with the envelopes moving in a series of big steps. Each step imposes an impulse function on the output which results in the spike of noise that you hear. For a very long time I thought this was just what happens when you set fast FM envelopes and there was nothing you could do about it. In fact this is none problem. If you set op2's attack rate really slow and play a key you'll hear why. What you should hear is something like this: You press the key and hear a sine wave then after quite a long time you start hearing some overtones which eventually build up to the sound you had originally. Ok we all know how envelopes work, the point I'm making is that over most of the range the effect of the modulator is minimal and in the context of a complex voice you won't be able to hear it. So now go to the level page and set the initial level for op2 to just less (say 10%) than maximum. (A very important feature of Yammy FM envelopes is that they don't have to start at zero.) When you press a key now you should get the slow sweep that you were expecting last time. Now if you go back to the rate page you can set a faster rate without distortion. These two tips combine to make the third point I started with - go easy with the envelopes. You don't need a very big change in a modulator level to have big effect on the output so long as the change is over an interesting region. Do not under estimate the importance and value of this advice. In fact it is the single most important piece of advice I can give when it comes to FM. It took me nearly a year before I was making sensible FM sounds because of this. (I had great fun making mad/horrible noises for a long time before though.) There are loads of other things I haven't covered - non-integer modulation ratios, fixed frequency operators, the pitch envelope, dynamics, RCM etc. However most of these are a case of suck it and see and using your imagination. I'm going home now. Ab.