Each month we review the latest Electro-Music Equipment — from synthesisers to sound reproduction and effects! E&MM's special in-depth reviews look at what's new in the world of commercial music — a vital updating for both electronics designers and musicians.

PAIA are a small company in Oklahoma City, USA, producing a wide range of music synthesis equipment, and the 8700 Computer/Controller represents the main product that really distinguishes them from other synthesiser manufacturers. The guiding force behind PAIA is John Simonton and it is his design philosophy that makes PAIA so interesting.

Basically, this is to make available a comprehensive range of synthesiser modules with all the necessary interfacing hardware for the minimum possible price, and to provide computer control over all voltage-controllable functions. A complete system (the 4700/J) in kit form, consisting of 3 VCOs, 2 ADSRs, 2 VCFs, 2 VCAs, control oscillator/noise source, various interface units, a digitally-encoded keyboard and the computer/controller, sells in the USA for around $800. This system is capable of various modes of polyphonic operation (though obviously the number of voices is limited to the number of available VCOs) and, to a certain extent, offers multi-track capability, the 'recorded' music then being dumped onto tape.

The individual modules range in price from $35 for a kit VCO to $26 for an ADSR and display a no-frills attitude to design. The modules show considerable design ingenuity, but I do feel that more modern circuits based on, say, Curtis Electromusic Specialties chips could perhaps improve some performance parameters and certainly offer additional facilities. For some reason apparent to only themselves, PAIA have opted for linear response modules (like Yamaha) rather than the linearly-controlled exponential modules common to ARP, Moog and Oberheim. So, real problems arise if one starts interfacing PAIA modules with the linearly-controlled systems common in this country. Linear to exponential converters are one expensive solution, but a preferable alternative is to use the interesting part of the system, the 8700 computer/controller, with one's own favourite linearly-controlled modules.

Computer/Controller Functions

The block diagram set out below should illustrate how the Computer/Controller is interfaced between a Keyboard and synthesiser modules (Figure 1). The 6503 CPU (same software compatability as the standard 6502 but with fewer address lines) is used as what might be described as a 'fairly dedicated' processor (less dedicated than those in some washing machines, more dedicated than the average home computer), i.e., solely for making music rather than indulging in pan-galactic fantasies or working out how much you'll be putting into the Chancellor's pocket!

Pitch information is derived from an encoder that scans a 7x7 matrixed keyboard for depressed keys. The beauty of this type of encoder, in comparison to the customary 'brute-force' diode encoder, is that polyphonic information is also encoded.

Program entry is accomplished by loading from tape, operating a touch-triggered keypad, or by accessing a PROM. The cassette interface provides visual (LED and display) and auditory (bleep tone) indication of successful loading and dumping, and unusually also offers the facility of motion control. Software for the cassette interface is derived from firmware, a PROM located on the main 8700 board. This format appears gratifyingly tolerant of speed variations and/or distortion, and caused no problems other than those derived from user error. The keypad didn't get quite such high marks, however, for though a 'bleep' provides feedback of data entry, the passive action of a capacitively-operated touch switch makes for easy mistakes when entering a long program, or when aiming a finger at a crucial control function during the middle of a bit of demonic keyboard work. On the plus side, though, this type of keypad doesn't wear out, and, as with PAIA's Programmable Drum Set, there's no doubt that you do digitally adapt to it.

Moving to the other side of the CPU, the data output, PAIA protocols use 6 bits of an 8-bit word to specify an analogue parameter (control voltage) while the other 2 bits are flags (flag 1 is used as a 'trigger', and flag 2 as a general-purpose control bit which the Quad Addressable Sample-and-Hold recognises as a glide control bit).

One digital to analogue conversion later, the newly-generated control voltage is applied to another peripheral, the QuASH (Quad Addressable Sample and Hold), which via some address lines from the CPU turns one CV (Figure 2) into multiple control parameters for multiple VCOs, VCFs, or whatever (Figure 3). The other half of the 4052 multiplexes Flag 1 giving us four triggers as well as the four CVs from each QuASH. With two QuASH's we have a system reminiscent of the Roland MC8 Micro Composer (which offers microprocessor control of eight analogue voices), but at a fraction of the cost. However, to get the maximum benefit of all this, we do need a couple of VCOs (to give a 'fat' sound) for each voice, not to mention sundry ADSRs, VCFs and VCAs.

Undoubtably the answer to this alarming escalation of expense is to interface the 8700 directly with a bank of digital VCOs. In theory, digital VCOs should make VCFs and VCAs redundant since a waveform's harmonic structure and overall amplitude can be controlled in realtime. As practice is inclined to be eons away from initial theory, General Instruments' AY-3-8910 may be a sensible interim step towards a super-amazing waveform generator.

Software Options

Meanwhile, and returning to things nearer home, i.e., standard analogue synthesiser modules, it's time to look at PAIA's software options for the 8700. Over the three years since the 8700 was introduced, PAIA have produced a variety of software ranging from the sublime to the less than amazing. Table 1 gives the current options with PAIA's selling points and my 'test comments'.

To construct a program, PAIA start by running a LOOK subroutine which, by scanning the keyboard, generates KTABLE, a list of notes held down. Another sub-routine, NOTEOUT, takes care of note output by reading the sequential entries from NTABLE, a list of notes to be outputted, and causing the D/A converter to turn key data into CVs which are assigned to the first, second, third or fourth S/H sections of the QuASH.

In between KTABLE and NTABLE comes a middle program which determines the actual personality of the program. Two straightforward possibilities for a middle program are shown in Figure 4.

Let's examine two of the software options to see what it is about the middle program that makes the 8700 more than just a pretty face. Firstly, MUS-1.0: this consists of POLY 1.0, a polyphonic allocation algorithm that assigns up to 16 depressed notes to respective channels of QuASH; INIT, an initialization routine that serves to set variables and buffer areas; and TRGN, a routine that serves as a software transient generator, a cunning substitute for the old analogue ADSR. TRGN responds to a note that has just been triggered by producing an ADSR voltage transient that is assigned to even number QuASH channels, whilst pitch setting voltages (CVs) are assigned to odd number QuASH channels. The value of this software transient generator is more apparent if we examine an ADSR transient like the following: This sort of transient starts out as a non-percussive kind of swell with a percussive 'kick' added at the last instant before the transition to the Decay and Sustain cycles. It's also possible to defeat portions of the ADSR cycle by, for instance, going from the middle of the Attack period to the Release state without including the intermediate Decay and Sustain cycles. This exciting type of flexibility just isn't feasible with traditional ADSR's and gives you that characteristic super-sharp digital sound much in favour with rock producers.

The second program I'm particularly enthusiastic about is SEGUE 1.0. PAIA's aim is to introduce an entire family of sequencer software, ranging from a humble monophonic program to an ultra-sophisticated system capable of complex multi-tracking. At present, only SEGUE 1.0, the 'universal monotonic sequencer', is available, although the 4-channel version, POLY SEGUE, is in the proverbial pipeline.

SEGUE 1.0 seems to stand way above the rest of the monophonic digital sequencer crowd. The Electronic Dream Plant 'Spider' and Roland's CSQ100 have a reasonable note capacity (126 and 168 notes, respectively), but the options available for adjusting sequences (or, as some would say, for 'musicalizing' a sequencer), either in terms of note length or key, are nowhere near as good as with SEGUE 1.0. In the case of the 'Spider', transpositions can be effected during playback of a sequence by touching pads marked '2nd', '3rd', '4th' or'5th', which transposes the sequence downwards accordingly. This is adequate in some situations, but for instance to transpose by a 7th, it's necessary to touch the '5th' and '2nd' pads together, and this doesn't exactly make for easy operation! It's interesting to note that this is exactly the same principle as that used in EMS's Synthi DK1 keyboard, and that's almost ten years old.

With SEGUE 1.0, once a line has been played into the memory, the sequence can be transposed by depressing any note on the keyboard and, if necessary, this information can be entered into the memory as a separate transposition sequence. What's more, the entire note and transposition sequences can be dumped onto tape and recalled ad lib. With 1K RAM, the sequence length isn't exactly limitless, and, if the POLY SEGUE program were available, you'd be lucky to squash a 4-part jingle into the available memory space.

You can hear some music using this system on the E&MM demonstration cassette No. 1.

If PAIA could amalgmate a polyphonic version of SEGUE 1.0 with a routine offering software transient generators, expand memory space so that multitracking is really feasible, and add digital waveform generators, then they'd definitely be on to a winning system!