PPG Wave 2.3 and Waveterm
A special report on the latest versions of these successful German instruments. Review by David Ellis.
With its distinctive blue livery, the PPG Wave 2.2 has become an instantly recognisable synthetic friend of the modern electronic musician. More than that, its designers' dogged determination to push the fusion of digital and analogue synthesis techniques to ever greater heights has meant that its sounds are just as much a fingerprint. David Ellis takes a look at the latest products from this German manufacturer, the new Wave 2.3 and updated Waveterm.
The Wave 2.3 takes the development of PPG's total music system concept one stage further, and, at the same time, puts right some of the problematic areas of the 2.2, particularly the alpha-numeric display and some of the sequencing functions. It also adds some valuable features of its own, such as the ability to play eight different sounds from the keyboard or with the sequencer.
Development and testing of the Wave 2.3 was actually completed just before the Frankfurt Musik Messe, so only two or three systems have actually made it into the UK so far, and that's reflected in the fact that the documentation for the 2.3 supplied for review comprised a 2.2 manual and a few extra 2.3-specific sheets. Needless to say, however, a proper 2.3 manual is in the offing.
Outwardly, things are much the same apart, that is, from the newly-incorporated, almost inevitable MIDI in, out, and thru sockets on the rear panel. Inside, there are a couple of PCBs for the Analogue panel and Digital Display Select panel, and a motherboard complete with four plug-in cards. First, there's the I/O board, with 6809 processor, EPROMs, program RAM, sequencer clock, and ADC for reading the front panel controls; second, the so-called 'PROZ' board, with lots of RAM and CMOS chips (about which little has been divulged!); and third and fourth, the two OF4 voice cards, each of which provides four sound channels via 12-bit AD7548 DAOs (as compared to 8-bit AD558s in the Wave 2.2), SSM2044 VCFs, and CEM3360 dual VCAs.
The basic plan of action according to the PPG principle of synthesis is to use multiple waveform tables as the equivalent of analogue VCOs, and then pass these through DACs to VCFs and VCAs controlled in the usual way by CVs under micro control. In fact, the waveform table approach permits an incredible range of oscillation possibilities, because not only can you use the full whack of 64 waveforms available in each sound program, but it's also possible to select much shorter waveform segments, or even just one of the 64. In fact, the last segments of each wavetable also contain sawtooth, square, triangle, and pulse waves just in case you want to use more traditional raw synthesis techniques.
As with the 2.2, waveforms are supplied in 16K of EPROM (2 x 2764), in what PPG somewhat confusingly call 'wavetables'. Now, you or I have got accustomed to thinking of a wavetable as a series of bytes representative of one cycle of a waveform. PPG, on the other hand, think of their wavetable as a set of waveforms. So, in their parlance, there are 30 wavetables in that 16K of EPROM space, each of which contain 64 waveforms, and these, in turn, comprise 128 8-bit values. Now, a simple bit of mathematics would suggest that these 1,920 waveforms need around 228K of storage, so there's obviously a clever trick going on in PPG's Hamburg headquarters if they've succeeded in squashing them down into just 16K. I'd guess that what's happening is some form of harmonic encoding, so that when a particular sound program is yanked from EPROM, a fast Fourier is performed to actually put the right 128 8-bit values into the 2.3's sound program workspace. That's just academic conjecture and is neither here nor there as far as what the Wave 2.3 actually sounds like!
The Wave 2.3 keeps the 2 x 40 character LCD display of the 2.2, but adds (very significantly!) some back-lighting so that you're not perpetually hunting for a torch. As with the 2.2, this display is made to work for its living by means of the Display Select panel, which enables the display of parameter values for all the different areas of the instrument. In fact, the Wave 2.3 has 3,456 control functions, so it's understandable that not all of them are displayed at once. Mind you, I look forward to the day when instruments like this use the 8-line, 80-column displays that are now being used on some portable micros (the Gavilan, for instance). It's certainly a little on the boring side having to hop around from one set of display parameters to another.
On powering up the Wave 2.3, various system checks are carried out, and a set of wavetables is loaded into the workspace RAM ready for action on the part of the user and processor. As with any programmable synth that comes equipped with factory presets, it's temptation beyond endurance not to try them out, and, in the case of the Wave 2.3, there's a lot of them about...
As with the 2.2, the sound programs efficiently cover the gamut of what's possible with the PPG synthesis approach, ranging from the utterly sublime to the equally ridiculous. There's certainly plenty of variations on metallophone-type percussion and plucked instruments (at which the 2.2 and 2.3 excel), but there are also some glorious waveform sequencing effects, including '05' (a chorus of amphetamine cats, perhaps?), '16' (a detuned choir with gut reinforcement), and a host of others. As a bonus, the sound programs supplied with the 2.3 also give you the opportunity of indulging in that favourite pastime of 'Spot the Tangerine Dream preset'.
In fact, the 87 sound programs are set up in 20 'combiprograms' (a feature new for the 2.3), which comprise the wavetables and parameter values for two Groups of sounds (a and b), including data for setting up split points. Anyhow, the first LCD display that greets the user is shown in Figure 1(a).
The top display line indicates the number of the combiprogram (CP), the nature of each of the eight sound programs in the eight banks (BK) of Group a, a global detuning parameter (DET), the keyboard mode (KBM) - ranging from eight-note polyphonic with two 'oscillators' per voice to monophonic with sixteen, the split points (SPL), the position of the split points (KEY), and the data transfer function (DTF) for storing or loading sounds, sequences, or whatever. Repeated jabbing of the left and right arrow keys on the keypad sends a cursor across the lines of the display so that particular parameters can be altered. It'd be nice if those arrow keys had auto-repeat, but I suppose you can't have everything. Sending the cursor past 'DTF' switches to an alternative display, illustrated in Figure 1(b).
This is equivalent to the main display of the 2.2 and, in the case of the 2.3, provides the option for putting the 2.3 into the 2.2 mode (ie. with just a single split point capability).
One of the most impressive features of the 2.3 is the enormous range of modulation options available. Aside from the obvious use of different envelopes to control the 24dB/octave VCF and VCA, plus the LFO functions, it's also possible to control the accessing of waveforms out of a Bank's wavetable with keyboard pitch, velocity, or after-touch. This is achieved by going to the Digital display, and toggling one or other parameter (see Figure 1(c)).
Here, the display indicates sources and destinations. The former group includes K (keyboard), M (modulation generator), T (touch sensor), V (velocity), and B (bender), while the latter has W (waveforms), F (filter cut-off), L (loudness), and M (modulation intensity).
Thus, assigning a '1' to KW (keyboard wave control) varies the manner in which a wave-table is spread over the keyboard; doing the same to the TW (touch wave control) parameter results in digital keyboard digging going further into the wavetable (a very satisfying effect!); and MW (modulation wave control) switches the wave control over to the modulation wheel.
However, the really dramatic effects are created by using the ADSR Envelope 1 to determine the ways and means of waveform sequencing through the contents of a current bank's wavetable. In addition, the second or sub-oscillator of each voice can be treated in a similar fashion with the AD Envelope 3 - independent of the main oscillator. So, not only could you have sequencing upwards through the 64 waveforms going to one oscillator, but also sequencing downwards through another set of 64 waveforms going to the other - an affect that's utterly mind-boggling!
Each sequencer channel can be used with one of the eight sound-program Banks. However, because of the more flexible voice card management on the 2.3, several channels can also use the same Bank. Again, the sequencer has its own display lines selected from the appropriate button on the Display Select panel, and a sample display is shown in Figure 1(d).
SEQ sets the sequence or arpeggio mode, LOOPS determines the length of the sequence or arpeggio, RECM sets the record mode. TMC determines the quantisation applied to the note events, SP sets the speed, RUN starts and stops the sequencer, and CH indicates the status of each channel (0 = normal, 1 = record, and 3 = off).
There are eight individual recording Channels handling up to a total of 1,000 events, so if you're using a multi-sample program such as the 12-bit drum sounds supplied on one of the Waveterm disks, you preset the sequencer for 4 bars, go to Channel 1 and prime that for recording, run the sequencer to hear the 4- beat count-in of the metronome, and then start playing the bass drum pattern. Then, by switching channel 2 to record mode, you can overdub the snare, followed by the toms, hi-hats et al on the other Channels.
However, sequence recording doesn't necessarily have to stop with the note events, their Banks and their split points. On playback, eight of the pots on the Analog panel are assigned to the eight sequencer Channels. By assigning a value other than 0-3 to the Channels (the bottom line of the Sequencer display), the pots can be set to update the pitch (values of 4 or 5), loudness (6), filter cutoff (7), waveform sequencing (8), or filter attenuation (9). Playing back the sequence with one or more of the Channels assigned to these pots will result in any knob twiddles being stored in the sequencer memory, so that, next time around, those parameter updates will be included in the sequence playback. In effect, this feature transforms the humble polyphonic sequencer into something like a computer mix-down facility.
The 2.3 software is certainly a great improvement over the 2.2 as far as sequencing is concerned. The main new attractions are that you can set up a sequence in advance provided you know the number of events (bars) you require, and that you can also go straight into polyphonic record mode. Also, if you have a combiprogram with, say, five sounds across the keyboard, it's now possible to set one Channel only of the sequencer to record, play from the different split sections into that Channel, and then on playback you'll have not only the relevant note values but also the corresponding sounds with their split points.
The other point to bear in mind is that the sequencer also records dynamics of performance, and this is particularly useful with the Processor Keyboard. In fact, the Processor Keyboard includes a slot for its own multi-sampled sounds (grand piano, choir, or drums, for instance) and will interface directly with the 2.3 without requiring the facilities of the waveterm.
The Wave 2.3's back panel certainly has its fair share of interfacing sockets, and includes, from left to right: Cassette, Phones, Stereo outputs, Sustain (for a foot pedal), CV in (for controlling a single voice), Trig in (ditto), Trig out, Program (a DIP switch to adjust the output clock rate of the sequencer), Rhythm (5-pin DIN to control drum machines), Communications Bus (to the Waveterm), MIDI in, thru, and out, and last, but far from least, eight separate channel outputs. All pretty impressive stuff.
Excellent though the Wave 2.3 on its own maybe, it's the Waveterm that's responsible for transforming it into a complete computer music system. This is really the controlling heart of the PPG empire, comprising as it does all the ins, outs, and processing of a quality 8-bit micro, a 12" green VDU, dual 5.25" disk drives, and an audio signal processor for sampling sounds.
In fact, the Waveterm has been through a rather bumpy ride as far as development is concerned, starting off in 1982 with a unit using a single 8" drive and one row of function keys below the screen, and ending with the present 1984 version using dual 5.25" drives and two rows of function keys. However, the insides are basically the same, with a single Eltec board holding the ubiquitous 6809, 64K, RAM, double-density disk controller, and video controller.
The quality of the display is really very high, with both excellent resolution and absolute stability. Indeed, additional software is actually available for using the Waveterm as a word processor! To do this you'd obviously need a QWERTY keyboard, so PPG provide a socket on the back of the unit for connecting any standard ASCII-encoded unit. Also, for anyone making extensive use of the non-real-time entry side of the system, a proper alphanumeric keyboard would be a useful addition. However, that's not to say that the standard Waveterm arrangement of one row of numeric keys and an upper row of software-defined function keys doesn't work well: in fact, it seems very user-friendly, and the non-typing musician may find this approach easier to get on with in the short term.
So what does the Waveterm enable you to do? Well, there are essentially two sides to the unit:
1 Creating sounds. Either by sound sampling or additive synthesis, or a combination of the two, with or without analogue modification.
2 Composing. Either with real-time sequences from the 2.3 or wholly in non-real time with the Event Generator.
To get to these various operations, you have to access one of the five current display pages.
Life in the Waveterm scheme of things starts with Page 0, which provides a 'systems check and communications management', but don't let that put you off! The left-hand column tells the user which components are connected to the three communications sockets on the back of the Waveterm. Apart from the Wave 2.3 which was connected in this case, other options might be the Expansion Voice Unit or the Processor Keyboard, and there are also other items to follow, such as a trigger box for use with the Event Generator (Page 5).
Page 0 also includes a keyboard map with the eight memory banks indicated above it. There's no immediate control over defining the sound until something is done to allocate a sound to one or more of the banks. So, for instance, the display shown of Page 0 indicates that five sounds have been allocated to banks 0-4, and the markers on the keyboard indicate the split points. These sounds actually represent sampled transient (hence the 'T' prefix) sounds of a string orchestra (an ad hoc one, judging by the sound of it), an opera chorus (shades of Tallis' Spem in Alium), and a fuzz guitar (much admired by Trevor Horn) taken from one of the demo disks provided with the Waveterm, but they could just as well have been your own sampled via Page 3. However, whereas these and other PPG-provided samples have 12-bit resolution, the Waveterm only provides the means for 8-bit sampling - hence PPG's current involvement in preparing a library of high quality samples.
Once these sounds have been loaded up from disk, they're then all playable from the 2.3 keyboard, and, what's more, the last sound played can be modified in real time by manipulating the 2.3's controls. For instance, an ADSR envelope (1) can be assigned to controlling waveform sequencing, the start point of the sample to the 'Waves-Osc' control, and the other envelope to the VCF and VCA.
The other point about the interface between the 2.3 and the Waveterm is that modifications made to a sampled sound - a change in the overall envelope or an alteration or two to one or more waveform segments, for instance - can then be stored back on the user disk as an update to the original sound, and this will be loaded with the original sound automatically next time around.
Page 1 provides the means to 'Compute a Wave'. There's a mass of waveforms already on the system disk, and they can be called up simply by using the 'Get' softkey. This page's Fourier synthesis operates with up to 32 harmonics, each with relative amplitudes from 0-63. Going to Page 2 then allows you to assemble previously constructed and stored waveforms into PPG's idiosyncratic 'wave-tables'.
This is the actual sound sampling facility. The sampling rate is variable from about 50kHz to 4kHz into 16K or RAM. Various facilities are provided for examining a sample, including the zoom facility which allows you to select the frame size of the sound snapshot from as much as the full 16K to as little as a single 128-byte set of values, and then display it. In addition, previously stored sounds can be recalled from disk for further analysis, and alternative start/stop points of the sample can be set up to make up new sounds. Although at present sampling is only 8-bit, PPG are working on building up a 12-bit sample sound library.
A recent software addition to Page 3 also allows two different transient sampled sounds to be 'merged' together (interpolated, in other words), and this also extends to using reversed segments of samples. So all in all, this is a page to watch out for...
The area that many sampling systems tend to gloss over is what happens to the sound when it's played miles away from the original sample pitch. Real instruments do not simply increase or decrease their harmonic constituents in a blind arithmetic fashion according to the pitch played. Instead, subtle harmonic changes occur, so that by the time you've got to a high G on a violin, the harmonic constituents of the sound are vastly different to the open G string. If you don't take that into account when a sample is playing away from home, you end up with the whole gamut of Suzuki-type larger and smaller violins, or, with the human voice, the old Chipmunk syndrome. Page 4 provides the Waveterm user with the option of constructing a resonance curve that'll determine which frequencies are emphasised and which ones are attenuated over the range of the 2.3's keyboard, theoretically eliminating many of these problems.
Page 5 is the Event Generator, and this can work in creating sequences either via Event Tables, or by using sample sequences played in real time and downloaded from the 2.3 at a later stage - as with the Space Elegie demo supplied with the review system. In fact, this was actually created for 16 tracks, using the Expansion Voice Unit in addition to the Wave 2.3, and also makes use of a special multisample set of sounds, including an authentic tambourine, a Queen Mary-type fog horn, a thumb-dislocating slapped bass, the Berlin Philharmonic at the end of their tether, the inevitable strings, fuzz guitar, and a banjo-picker's delight.
Using this demo piece as an example, the 70 or so separate sequences are chained together with the 'Play' command, so that pressing the softkey marked 'Execute' causes the Waveterm to read the corresponding sequences from disk and transfer them to whatever workspace memory is available in the 2.3. When 'ready to start' appears on the screen, the piece can then be run by keying 'Run' on the Digital Display Select panel and whatever play mode is required.
For instance, entering '1' on the numeric keypad instructs the 2.3 to change sound programs wherever they've been included in the sequence. In this case, however, where the same program is used throughout each part, '5' is more appropriate, as this ensures that the same sample is consistently played for each part. Though this demo didn't actually make use of the facilities, the Waveterm Event Generator also enables each simple sequence to change program, tempo, tuning, and dynamics wherever and whenever the user feels like it.
The point about the Event Generator is that the number of note events is virtually unlimited. For example, you can write a Play command with constantly changing sequences and then exchange disks with new sets of sequences at relevant points during the composition. That's taking an extreme situation, though, because in reality the 0.5Mbyte disks provide ample sequence space for most musical mortals without recourse to Hal Chamberlin musical chairs. What the Waveterm actually does is to load up segments of sequences from disk and transfer these to the 2.3 as notes are played by the latter - a virtual memory situation, in fact.
Although downloading of a simple sequence from the 2.3 to Waveterm also provides the means for editing real-time note events, you can also work in non-real time by starting note event input directly from the 'Edit' mode of Page 5. Up to 16 tracks are available, and bars are displayed one at a time on the VDU. Parameter values that can be programmed for each note event include 'Time' (where a note begins), 'Gate' (the length of anote), 'Oct' (the register of a note), 'Sem' (the name of a note), 'Bank' (the sound program), 'Updat' (the update parameter - dynamics, filter cut-off, wavetable start, or whatever - applied to a note event by manipulating the 2.3's controls), and 'Ch' (the sequencer channel for the corresponding event). The numbers and columns approach to non-real time note entry isn't always the quickest way of going about creative pursuits, but the Waveterm's software is reasonably kind in this respect - particularly by virtue of the 'Copy' facility that enables bars or events to be copied from one part of the score to another with or without updates of transposition, program change, and so on.
There's no denying the impressive quality of the combination of Page 5 and 2.3. Still, it'd be nice to see PPG turning their attention to a more MCL-orientated from of non-real time entry.
There's no doubt in my mind that the Wave 2.3 is one of the most versatile digital synthesisers around. In fact, it could be seen as the ideal foil to the recipient FM synthesis bandwagon. The only reservations I have about the sound quality lie with the limitations of 8-bit waveforms, but PPG's move towards 12-bit resolution is rapidly casting these aside.
Some of the preset waveform sequencing programs are very dramatic, others less so, and a few demonstrated slightly annoying glitches as the sequencing switched from one waveform to another. That's obviously hard to avoid given that waveform segments can't always be relied upon to start and stop bang on zero amplitude at the right point in a sample, but since PPG's own factory presets also do that, it makes me think that a little more effort in this direction might not go amiss. However, as we go to press it seems that the company are in the process of curing these glitching problems. Another point worthy of attention is that the output is a mite noisy for a synthesiser of this calibre. It does sound rather as if some digital noise is leaking to the output without the intervention of the VCA. Still, that's a minor point, really.
Obviously, there are hardware differences between the 2.2 and 2.3 which help to account for its improved performance. At present, people using the 2.2 with the Waveterm are only able to work with 8-bit resolution, and are limited to a maximum of two samples played across the keyboard. However, there will be an update available for their machines which will provide the means of assigning eight samples at once, though still only at 8-bit resolution. At least, that's how things are at present, though there are plans in the pipeline to ensure that 2.2 owners and their machines are brought right up to scratch, so to speak.
What can't be denied is that the potential combination of the Wave 2.3, Waveterm, Expansion Voice Unit, and Processor Keyboard, for something in the region of £12,000, would make for a quite superb computer music system that easily overshadows its competitors, and the fact that the system can be acquired bit-by-bit only makes it more attractive. Indeed, taking just the Wave 2.3 and Waveterm together (for a VAT-inclusive total of £8,585), you've got a system that's equally suitable for stage, studio, or use as a compositional tool, and does just about everything you might want with a great deal of panache.
VAT-inclusive prices for the Wave 2.3 and Waveterm are £3,995 and £4,590 respectively. PPG are now handling their own UK distribution, and their agent is to be found at (Contact Details). Alternatively, contact PPG-Vertrieb Duren KG, (Contact Details).
Gear in this article:
Review by David Ellis
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!