ARP 2500 Keyboard

Overview  The keyboard was designed at an early stage of the ARP 2500 development project in April 1970 and the design was expanded over the next few months as different keyboard variations were designed. For some reason ARP went BIG on the number of variants (18), with almost as many variations as modules! They are all basic pitch generating keyboards with no velocity or aftertouch or pitch bend. They can be single or dual voice, and the keyboard can be split into two, with black and white key reversal on the lower section. Dual keyboards were also available, in the same wooden rectangular cases as the single manual.

  • 3001 One voice
  • 3002 Two voice
  • 3112 Single voice (1 octave) and two voice (4 octaves)
  • 3213 Two single voices and one dual voice (3 octaves)
  • 3221 Two voice (2 octave) and single voice (3 octaves)

The keyboard electronics had an immediate change on the second 2500 made (70-002) with the resistors in the Trigger circuits reduced from 10M to 1M. Much later on the Trigger connection to the bus bar was screened to avoid false triggers.

Block Diagram

Original Design At the core of all the 2500 keyboards is an ARP designed circuit called the 1001 (the first circuit in the 1000 range), which is built on a single PCB with a 25 pin edge connector. The 1001 PCB and circuit was reused in the 3604 keyboard (all three variants) from the early ARP 2600’s, except the 2500 version has the addition of an auxiliary CV output. The circuit has some weaknesses due to the constraints of having only BJT Op Amps available in 1970, which are outlined below.

The keyboard mechanism is the usual Pratt & Read design with 3 buses; Pitch CV, Gate and Trigger. The controls are basic; a Pitch Tuning potentiometer varies pitch upwards by 6 semitones and it can switched ON or OFF, Portamento can be switched ON or OFF and a potentiometer varies the slide time by up to a few seconds, Tone Interval can be switched from a preset 1V/octave or to a variable tuning set by a potentiometer. And finally there is the Auxiliary Output CV which can be switched ON or OFF and the voltage amount varied by a potentiometer.

The Auxiliary Output is used with a two-voice keyboard, to provide the usual duophonic ARP feature, familiar in the Odyssey, with the Aux output connected to VCO2. When set to OFF the Aux potentiometer sets VCO2 to be the same frequency as VCO1. When ON the Aux potentiometer sets VCO2 to be the same frequency as VCO1 when one note is played. When a second note is played VCO2 will play the highest note, with VCO1 playing the lowest.

ARP replaced the 3604 with the 3620 keyboard in 1975 and made a number of improvements, but these changes were not retrofitted back to the 2500 keyboards as the product was essentially End of Life with few sales taking place by then.

8007 module

ARP 8007 The 1001 circuit contains a sub module – the 8007 (or A1) which can be either encapsulated or open. This is the CV Memory for the pitch signal and it contains a dual LM1458 Op Amp, as JFET Op Amps were not available until many years after 1970.

The first Op Amp is buffered by dual matched JFETS and receives the inputs to the CV Memory. The second Op Amp with dual matched JFET’s acts as a second stage and contains the offset trimming, and there is a BJT in the circuit as well.

Encapsulated sub modules seem to have buried the offset trimmer inside the module, whilst the 2500 and 2600 schematics show the trimmer as external and accessible. Early examples of this PCB appear to be of poor quality and contributing to the 1001 shortcomings.

1001 PCB

Shortcomings The 1001 circuit design in the 2500 and 3604 keyboards have a number of shortcomings that are either immediately apparent or appear over time as components age. ARP did a good a job with basic analog circuits in 1970.

A few years later digital keyboard scanning was introduced by E-mu Systems with the 4001 keyboard of mid 1973 which used TTL logic chips. This transformed the analog keyboard into a stable and reliable source of CV’s. The ARP 1001 shortcomings can be addressed by making some simple component upgrades, and one day I might make a replacement AM1001 PCB which would plug straight in!

It is also worth reflecting that ARP was a slow to adopt digital techniques after making partial use of TTL logic in 1970 with the 1050 Mix Sequencer. A digitally scanned keyboard would have to wait until 1978 and the exploitation of CMOS logic until 1976 (1601 Sequencer). The same year ARP attempted the disastrous Centaur/Avatar projects which contained CMOS and TTL logic, but really needed microprocessors and DSP chips to achieve the design objectives.

If only ARP had licensed the Emu Systems digital scanning design for the 3620 keyboard of 1975, and they could have gone down the same road as Oberheim to create a 4-Voice polyphonic keyboard.

ARP SP2 Panel

Trigger Delay This circuit introduces a 15 ms delay into the Trigger signal which provides an important switch debounce circuit to reduce/eliminate false triggering, as the switch on the bus bar is closed. But 15 ms is quite a long time and 10x longer than the minimum attack timing of an envelope. A modern circuit would use a Schmitt trigger IC or a fast micro controller to reduce this delay time to 10 ms or less and avoid any deterioration of the timing as components values vary over time.

ARP moved to using a Trigger signal derived from the Gate signal in the 3620 keyboard (Roland also used this approach), which reduces the delay and means only two bus bars are needed. They used a 4011 CMOS chip which was not cost effective in 1970 and the 2500 project.

CV Buffer This circuit relies on the A1 (or 8007) Hi Z Op Amp sub module, described above, and a 100nF polycarbonate capacitor to hold the pitch voltage between key presses. The A1 requires trimming to set the offset voltage to zero and its prone to downwards drift of pitch as the circuit ages. The 3620 made an improvement with a JFET before the CV hold capacitor to help reduce drift.

A modern solution is to use a low drift and high input impedance FET Op Amp such as the CA3140 or even the LF398 S&H chip. Either of these will produce a big improvement in drift performance, but of course were not available in 1970!

Precision Voltages The pitch CV circuit relies on a 1% 100R resistors in the usual chain underneath the keyboard, with trimmers at the top and bottom and a regulated current supply with no precision reference voltage or metal film resistors. It runs from the standard +15V power rail which is sent from the 1002 power supply via a long cable. There is a signal diode in the constant current circuit but this can easily be improved with a modern precision reference voltage chip. This combined with higher tolerance resistors (0.1%) in the resistor chain, will improve pitch accuracy.

Aux Out The dual voice 2500 keyboards have an auxiliary output which is the highest pitch CV sent through a couple of LM301 Op Amps, and a potentiometer and ON/OFF switch. The LM301’s can be useful upgraded to low drift Op Amps like the LF411, to improve the pitch accuracy of the second voice. The 1001 module came in two configurations (same PCB); the APL-1001-012 which had no Aux Out and the components for this circuit were not fitted and the APL-1001-014 with Aux Out.

AM2500 Keyboard Initially I am using a Roland 181 keyboard to control the 2500 Racks, which is pitch stable and has pitch bending as well as Portamento. A Roland 184 would provide polyphony but it has no triggers. I plan to recreate the E-mu Systems 4003 which has a 5-octave keyboard and CMOS logic based digital scanning. Here is the post.

Copyright AMSynths 2019