ARP 2500 Review (Studio Sound 1972)

In the 1973 Studio Sound magazine, this review ARP 2500 modular system was written by David Kirk. With a discussion on the patchboards and different modules. The full text is below the images.

THE ARP 2500 comes from the same stabIe as the smaller and cheaper 2600 examined in our January issue. There is no resemblance between the two designs, except that they both follow the now weil established principle of voltage contro!. Readers unfamiliar with this are referred to the 2600 field trial and to ‘The Fine Art of Voltage Control’  The most significant difference between one vc synthesiser and another is the means employed for routeing. In the ARP 2500, this is acbieved by switches sliding up or down through 20 discrete positions. These switches are arranged in groups of ten above and below each generating or processing module. The vertical switch positions are numbered 1 to 10 and 11 to 20. Between 10 and 11 are three non calibrated ‘rest’ positions where a switch would normally be placed when not in use. A bus wire connects all points in any one horizont al level together. This starts at a miniature jack socket on the extreme left and terminates at another socket on the extreme right. A sirmlar bus connects the adjacent level contacts to each otber and so on down to level 20. Fig. 1 shows part of an ARP matrix. The arrow pointing downward indicates a signal output, a signal being routed to whichever bus on which switch 1 is set. The upfacing arrow indicates an input, taken in this case from bus 2. Why have we chosen level 2? Merely because it is unoccupied. We migbt equally weil have used level 1, 3 or 4, provided these also were unoccupied. Fig. 2 shows a more realistic ARP patch arrangement in which a sinewave and square- wave from oscillator A (switches 5 and 7 in the leftmatrix) modulate the frequency of oscillator B. A triangle wave from A simultaneously modulates the pulse width of the tone emitted from oscillator B. Two buses are used to separate the frequency and pulse control voltages. The tone produced by B is in this case switcbed to level 4 where it might be routed to a signal processor. Alternatively, it could be coupled through a level 4 miniature jack socket to an external amplifier. Since a ten switch 20 position matrix is incorporated above and below each ARP module, some switches must inevitably be redundant to the system. Not every module requires ten inputs and ten outputs. The 1023 dual oscillator module, for example, has six upper-Ievel inputs, four bottom level inputs, and two bottom level outputs. In other words, it uses 12 of the 20 switches it spans. Since the redundant switches are not even useful in connecting external equipment, one might query the sense of such a routeing system. A typical ARP 2500 might have over 60 redundant switcbes; 25 per cent of its total complement. Add to this the little problem of parallax confusion when looking up or down to a bank of protruding switches, plus the crosstalk between neighbouring buses, and you face the question: why don’t they use pin boards? One answer is that ARP bave chosen the easiest method of ‘custom building’ synthesisers. If so, they have reduced the f1exibility of the instrument by a colossal degree, as the following crude arithmatic will attempt to show: A typical ARP 2500 has 180 inputs or outputs which for tbe sake of argument we will regard as 90 inputs and 90 outputs. Each of these 180 may be switched to any one of 20 positions. The total number of tbeoretically possible combinations is thus 20°, or 40°. If the 90 outputs are arranged down a pin matrix, however, and the 90 inputs across, tbe number of possible combinations more than doubles to 90°. This assumes that the pin matrix is restricted to one pin per row since tbe ARP system is inherently limited to one active point per switch. In practice, a 90 x 90 pin matrix could accept many more than 90 pins: 8,100 to be precise. Which raises beyond comprehension even our 90 combinations. A few other basic points should be covered before we look in detail at each module. Firstly, the photograph does the ARP much less than justice. The 2500 makes considerable use of colour symbolism. For example, a yellow switch may relate to an amber potentiometer, a green switch to a green pot, red to red and white to silver. The redundant switches are coded black. Secondly, if you are ever called up on to move a 2500 from one room to another, resist the temptation to lift it from beneath the keyboard. The main chassis is not bolted to the keyboard module and the two short umbilicals would not prevent a mighty expensive accident if the chassis toppled.  

3001 Keyboard

ARP keyboards are divided into two electrical sections; one with black naturaIs, white sharps, and the other following piano colours. These sections may be coupJed to form a five octave unit or used entirely independently. Your left hand might control the pitch of one oscillator while the right controJs another; a simpJe form of polyphony. More interesting applications come to mind as you grow accustomed to the possibilities of voltage contro!. Simp Ie examples: the left keyboard can be patched to control Joudness, notch filter frequency, vibrator rate, pul se width, or all five characteristics simultaneously. Each keyboard section supplies three items of information: control voltage, gate and trigger. These are prewired to separate buses on the upper level switch groups. The conlrol voltage rises by an appropriate ratio with every semitone shift up the keyboard. Twelve such intervals give a 1 V rise in voltage across one octave. The gate is simply a ramp voltage, a single pulse which holds for as long as a key is pressed. Not to be confused with the trigger, a virtually instantaneous pul se produced whenever a key is pressed but independent of note duration. Gate and trigger have obvious applications in controlling (via an envelope generator) note dynamics. Horizontal panels left and right of the keyboard carry controls governing portamento (gliding tones), absolute tuning and tone interval. Each panel govems the adjacent keyboard section. Fig. 3 shows these two panels. Portamento speed is variabie over a fairly wide range, dependent on the rotary control setting. An adjacent switch overrides this facility. The purpose of other controls on these panels should be obvious.  

1004t oscillator

Fig. 4 shows this, one of several vc oscillators produced for 2500 syntbesisers. Three ingoing arrows at the module top represent frequency control inputs with a fixed sensitivity of 1 V per octave. On the bottom edge of the module, inputs 2 and 3 are also frequency controls. These may be varied in sensitivity to a maximum of 1 V per octave. ‘Input pwm’ controls the mark j space ratio of the pulse output but has no effect on the sine, triangle, square and saw waveforms. Each waveform may be extracted independently, or swi tched into a single output channel (bottom right). Ph ase reversal switches occupy a subpanel. Two of these were incorrect ly wired. Coarse and fine initial frequency controls are incorporated, plus an initial pulse width potentiometer. The ‘enable’ switch disconnects the module from the ARP power supply. Right of this is a low/high frequency range switch. 

1023 dual oscillator

Illustrated in fig. 5, this module is a simplified version of the 1004t. In the same module width it incorporates two entirely independent oscillators, each with concentric coarse and fine frequency controls. No vc pwm in these units, but pulse width presets and facilities for mixing two fm control voltages (2a, 3a and 7b, 8b). 

1045 Oscillator/voice

This unit (fig. 6) can be used as an independent oscillator Oeft hand subpanel) or in conjunction with the built-in enveJope generator, vc amplifier and vc filter. The oscillator suffered from an intermittent fault and in the course of several days grew less inclined to oscillate. 

1047 filter resonator

All ten switches beneath this module are active. One and two (bottom, fig. 7) are audio inputs, mixed by potentiometers bearing the same digits. Output three supplies the ingoing signal treated according to the frequency response curve immediately above switch three: maximum bass falling straight to mllllmum treble. Output four offers a low bass, high mid, low treble format; output seven low bass curving up to high treble; and finally the notch at output eight. Inputs five and six, via potentiometers five and six, accept frequency control voltages. A sinewave at input five, for example, would swing the peak frequency of output four up and down the audio band. A sawtooth control voltage would sweep the peak upwards, cyc1ing instantly back to If. Inputs 9 and 10 permit voltage control of the filter resonance. Initial filter characteristics are set by coarse and fine frequency resonance, notch and final resonance controls. This module appeared rather too sensitive for the rest of the system and often overloaded under conditions in which other synthesisers remained stabie. Slight adjustment to complex patches were of ten upset by the 1047 module overloading. An overload warning light is incorporated to reveal this condition.  1046 quad envelope generator This unit (fig. 8) produces up to four separate envelope waveforms, plus two in antiphase, to control (through a 1006 vc amplifierjlow pass filter module) audio amplitude. It might also be routed to a filter frequency control, oscillator pitch control or some less obvious control input. Attack time, initial decay time, sustain level and final decay time may be adjusted independently in each of the four sub panels. If desired, all four envelopes may be triggered from a single pulse.  

1016 random generator

White noise, pink noise and random ramp voltages are produced by the 1016 (fig. 9). This was the middle unit of the synthesiser supplied for test and sat above and below miniature jackfields rather than switchbanks. The output labelling, conceived for switches, in this case refers to bus levels. Noise appears at levels three and five, random con trol voltages at seven and nine. Each of these four generators  may be switched off when their buses are required for other signaIs. The noise was clean, lacking the grittiness of the generator employed in the smaller ARP 2600. 

1027 sequencer

Fig. 10 illustrates this, a ten state three layer voltage memory. Rows A, Band C show the three layers of potentiometers used in preliminary adjustment. Suppose we wanted to programme a simple melody of ten notes duration. Control voltage output A (tOOd from right, bottom edge) could be routed to the frequency con trol input of an oscillator. If the fust note of our melody is the !owest in pitch, the fust (topmost) sequencer contro! in the A layer may be set to minimum and the oscillator frequency con trol adjusted to the desired pitch. The sequencer at this stage is off (stationary). A green light adjacent to the top row of presets indicates the active controls. We press the ‘step’ button and the first state lamp extinguishes. The lamp below it comes on, and into circuit come the second set of presets. Our next note is tuned by the second from top A layer control, step again, tune the fourth control, and so on to the end of the sequence. If you become confused in the rniddle of tuning a note series, the ‘reset’ button reverts to the beginning of the sequence. When all ten notes are set, the square (illurninating) ‘on’ button sets the sequence ticking away until somebody or something triggers ‘ oif’. The pulse width control adjusts the duration of silence between stages, from discrete notes to a series of non-zeroing ramps. Overall speed is controlled by the pul se repetition frequency knob and the adjacent coarse ‘low/ high’ rate switch. lf we wished to vary the !ength of one or more notes, this could be accomp!ished by employing the ten B-layer presets (middle row) as a source of duration contro! voltages. Output B could be routed to the ‘vc width’ input. Thus the higher a B pot setting, the longer the related A pot note. This leaves layer C free to control an independent melody or the characteristics of another processor. Any external pulse may be applied to the ‘on’ input to trigger a sequence, provided the internal trigger is over-ridden. Similarly a sequence may be halted at any time by applying a pulse to the ‘ off’ input. Here the position gates are useful (outputs Olle to ten, relating to the ten groups of presets). Routing gate ten to the off input halts the sequencer on the tenth note of each run. A slow external pulse oscillator rnight then be used to restart the sequence, The 1027 module occasionally became unstable at high repetition speeds, a fault usually cured by switching off for a few minutes. 

1050 mix sequencer

This unit was a delight to operate and must be one of the smallest 8/2 mixers currently available. Illurninating push-on, push-olf buttons show at a gIance which channels are functioning. Each input has its own level preset, feeding group faders. All ‘on’ channels may be cancelled at the touch of a single ‘exclusive on’ button. The 1050 may also be used as a sequentially controlled analogue gate, channels being activated either singly or in pairs. 

1005 Amplifier

llIustrated in fig. 12, the 1005 incorporates a balanced modulator and voltage controlled amplifier. If equal amplitude sinewaves of frequency fa and fb are routed to audio inputs A and B, the output will comprise the sum and difference tones (fa + fb) (fa – fb). If harmonies are present in the waveform supplied to input A (fa, 2fa, 3fa and so on), the B input remaining a plain fb sinewave, the resultant output signal contains the frequencies (fa + fb), (fa – fb), (2fa + fb), (2fa – fb), (3fa + fb), (3fa – fb), etcetera. Two internal con trol voltage sources may be used to modulate the oscillators feeding A and B. Ratio and initial tuning control are incorporated on the 1005 facia.

1036 sampler/random generator

Like the 1005, this module offers facilities you are unlikely to require until completely familiar with the rest of the 2500 chain. The 1036 is a dual unit with four rotary controls per section labelled ‘dock frequency’ , ‘internal random signal level’, ‘external signal level’, and ‘dock frequency modulation’. If we apply a 1 Hz sinewave to the external signal input (bottom left, fig. 14) and set the dock frequency to 2 Hz, output ‘a’ will deliver two ramp voltages per second, briefly storing the amplitude at each moment the sine is sampled. Noise sampled in the same way produces random control voltage ramps independently of the 1016 random generator. The value of a sample and hold facility dearly exists more in sampling complex repeating waveforms than plain sinewaves. The dock frequency modulation facility makes this a particularly promising module.

There, for what they are worth, are my views of the ARP 2500. Is the system worth E 5400? This can only be sensibly judged by comparison wJth other synthesisers now available. I must point out th at four of the smaller ARP 2600, for example, (totalling E5,400) would form a more flexible basis for an electron ic musie studio than one 2500. You’d miss the sequencer? Then you would still do better with a chain of small synthesisers and a locally made fl ,500 sequencer handling rather more than ten by three states.

Agents’ comment : The 2600 and 2500 are closely related in that they employ the same basic electro nics for sound generation and processing, the principal difference being in the presentation of the ‘package’. Matrix switches are employed in the 2500: (a) To provide a versati le synthesiser system capable of modification to individual requirements. (b) Each in put/output may be switched to any of 40 positions, not 20 as stated. (c) To eliminate the possibility of loss or damage encountered with removable pins or patch cords. Keyboards are available in many configurations. The keyboard rev iewed did in fact have a left-hand single-voice two-octave section and a right-hand two-voice three-octave section . This right-hand section is capable of playing two different notes simultaneously. The filter/resonator is capable of very high Q values (over300) and may obviously become unstable if incorrectly operated. For this reason, an overload indicator is fitted. Should overloading occur, the obvious solution is to reduce the input level. A more delaifed reply is being prepared for our June issue (Ed)   

Are you having synths for breakfast? Leave a reply here...