ARP 2500 Rack Modular



ARP 2500 – this is what I wanted to replicate!

ARP 2500

Overview The ARP 2500 modular synthesizer was developed as a direct competitor to Bob’s Moog Modular, which had made a massive impact in 1968-69 after “Switched on Bach” launched the analog synthesizer into the public imagination. Alan Pearlman founded the Tonus company in 1968 with $200,000 and in the autumn of 1970 launched the ARP 2500. The Massachusetts company implemented three innovations over the existing Moog and Buchla designs:

  • Stable oscillators with a wide frequency range.
  • Matrix Patching System, with no need for patch cords.
  • Multi-voice keyboards.

David Friend presented an overview of the ARP 2500 to the AES Convention in October 1970 and ARP took immediate market leadership from Moog by implementing an improved VCO design. This was made possible by implementing a stable and accurate exponential converter, which is at the heart of the VCO and VCF cores. These converters were encapsulated into small sub modules to improve temperature stability, and to stop other manufacturers copying the ideas.

Alan Pearlman also introduced Op Amps into his designs and used the Teledyne 1339 Hi-Fi chip, as well as the newly available LM301. The 2500 modules make extensive use of matched transistor pairs, and Op Amps are used in CV summing and final audio stages. The ARP 2500 did not use the new OTA CA3080 chip (1969), and matched transistors are used to build precision VCA components, showing the precision Nexus discrete Op Amp heritage.

ARP Brochure

Digital Technology The 1027 and 1050 modules make use of early digital technology, and probably the first time in synthesizer history.  Both of these modules use the original 74xx TTL logic chips in plastic packaging that were released in 1966 and  common place in the late 1960’s. The 1027 uses a ceramic package Burroughs Nixie chip as a decade counter and was probably the first mass produced digitally based sequencer (before the EMS AKS May 1971). The strange choice of a Burroughs chip is probably due to cost, and that the TTL SN7490 decade chip was not launched until 1970.

The Burroughs decade chip gave the 1027 an unusual 10-step sequence rather than the more usual 8 or 16 steps, although with some patching the sequencer can be 8. The clocks and control circuits in the 2500 are all analog, with no use of the new CMOS chips that were introduced in 1968 by RCA. Just discrete transistor flip flops and relaxation oscillators built from uni-junction transistors – the design of which dates make to the 1950’s.

ARP 4001

Analog Technology The major analog electronics innovation in the ARP 2500 design was the use of high quality exponential converters in the VCO and VCF circuits to produce accurate and temperature stable sounds. Alan Pearlman used a circuit he had created in 1965 and patented in 1968 at his previous company Nexus; the Temperature Compensated Logarithmic Amplifier. This circuit was developed into two exponential generators; the 4001 and 4002.

Each small PCB was populated with up to 5x transistors and a 100R Tempco resistor, and then placed in a potting box which was then filled with epoxy for temperature stability. The core of the circuit was either a matched NPN transistor pair marked in red paint in the 4001 or a matched PNP transistor pair marked in green paint in the 4002. I have recreated the 4001 and 4002 as potted sub modules 30mm square with Linear Systems matched transistor pairs, they work very well and use SMD Tempco resistors from Panasonic.

The 1047 UAF also use potted sub modules for the current mirrors that form the OTA’s in this state variable filter. The 4005 had both NPN and PNP variants, and I have recreated these using the same 30mm square potting boxes and temperature conductive epoxy. the use of discrete matched transistors makes the filter far superior in sound to those based on the 3080.

Teledyne 1339 Op Amp

Op Amps ARP initially used high quality Teledyne 1339 Op Amps, which are pin compatible with the contemporary 741 Op Amp but a completely new and improved design with a fast slew rate at 34V/uS and a low offset of 2mV. They were officially launched in May 1971 by Teledyne Philbrick at US$4 from the same location as ARP, Boston USA. Alan Pearlman set up Nexus Research after working at PhilBrick, both companies being acquired in 1966/67 by Teledyne. The sale of Nexus helped start up ARP and some of the technology (exponential amplifiers, 4 quadrant multipliers, encapsulated modules) made its way into the 2500.

The 1339 proved unreliable with possible lock up problems as the output range is only guaranteed to +/-10V, which could be exceeded in an ARP 2500 module. They were replaced by the slower and more tolerant LM301 Op Amp, and the equivalent high quality MC1439 from Motorola in “safe” positions – such as the 1006 buffer amp and 1004 sine wave buffer amp. I have retained the LM301’s in some modules to be historically accurate, but increasingly as I went through the build I upgraded to modern Op Amps to get closer to the original intent of the fast and low offset 1339.

For low offset I have used LT1013 Op Amps, and for fast slew I used either TLE2071 or 411’s. For example in the 1047 I have used TLE2071’s in the filter core, OPA134 in the audio buffer and LT1012’s in the CV summers.

HP Red 5mm LED

LED technology The ARP 2500 pre-dated the availability of low cost LED’s by a couple of years, which is why the were not used in the original modules. 12V incandescent lights were used instead, either within the Switchcraft “white” push buttons in the 1005, 1050 and 1027 or behind coloured lens in the 1027 Sequencer. Red LED’s had started to appear as 7 segment displays in calculators during the late 1960’s.

However it wasn’t until 1971 that HP introduced the 5082-4404 red LED in the now familiar 5mm T1 3/4 package, and then green LED’s eventually followed a few years later. The last ARP module to go into production was the 1026 Preset Voltages and that does look like it used LED’s in the promotional material of 1974. I have used 12V lamps in the Switchcraft switches but 5mm red or green LED’s for panel lamps with either Fresnel or plain lens. The LED’s reduce power consumption and are faster for sequence movements.

Technology Cascade The ARP 2500 pioneered new analog synthesizer designs but it was a high price tag/low volume product aimed at Universities and the “Moog” record market. Pearlman spent the next 3 years reworking the technology into a complete range of synthesizers at lower prices that appealed to a much wider range of musicians. The ARP 2600 (1971) and Odyssey(1972) can trace their circuits back to 2500 modules, such as the Moog copy filter (4012 module) and the uni-junction transistor clocks. This cascade of technology proved to be very profitable and ARP rose to become the largest synth company by the mid 1970’s, perhaps milking the 2500 analog designs for too many products and years.

Large 150 mm PCB’s

The 2500 Rack  The idea of building a replica ARP 2500 as a set of modules became possible in the spring of 2004 when MusicParts in the USA published the original schematics, complete with the original construction and part listings. Whilst simple schematics are freely available on the Internet, these official ARP documents have far more detail and make replication easier. I decided early on that replicating the matrix switches was not practical due to availability and cost, and that a patched approach was going to be a better and cheaper solution.

I bought the full set of ARP documentation and over the next 10 years gradually designed the PCB’s using Eagle CAD and sourced the tricky parts that were needed. I made a lot of effort in the Summer of 2015 to progress the project with prototypes and production modules, and then again in the Spring of 2019 to dig into the detail of the more complex modules and get them operational. I have also added some of the modules that ARP announced in 1970 but never got round to manufacturing, as the modular market declined from the mid 1970’s.

Open Hardware The AM2500 project was released as Open Hardware on 11 April 2019 with a dedicated repository of documentation and design files at AMSynths on GitHub.

Mechanical Design I set out to design and build a close replica of the 2500 as a set of 5U modules, MOTM style and 3.5″ wide. Most of the modules use an aluminium sub assembly which bolts onto the front panel and provides space and mounting of the large 150 x 150 mm PCB’s. Some modules used pot brackets to fix the PCB directly to the panel, making for a lighter construction.

The matrix panels are very expensive and hard to locate, with a full set running into £1000’s. I have used 3.5mm jack sockets and patch cables, which I think are a better idea as there is no cross talk. I have kept as close as possible to the panel layout, and used similar high quality PRV6 can potentiometers and coloured ARP knobs. Some modules use the 270X potentiometer which are cheaper and come in 1M log values. I also ran out of coloured ARP style knobs after the first 10x modules, so some knobs are simply silver with a line rather than dot. I do plan to get colour knobs remade.

1033 Inside

Electronic Design I have kept close to the original 1970 design but with a few careful component improvements. Some chips are obsolete and needed modern replacements, but all of the transistors are still available or have modern equivalents. Some modules I have used carbon composite resistors like the original (in the 1004-2600 VCO’s), but sometimes I have upgraded to metal film (1036 Dual S&H). Ceramic capacitors are a mix of plate and modern MLCC, and I have used modern smaller polypropylene capacitors to replace the hard to locate polycarbonate capacitors. Mylar capacitors have been retained in the VCO’s and in one of the 1047 UAF’s.

I have also made various Op Amp choices in the AM2500 modules; most of the time keeping to the LM301 in the schematics but sometimes using modern LT1013 low off set Op Amps in the critical positions like the VCO, and fast TLE2071’s. As the project developed I have tended to go back to replicating the intention of the original designs rather than using the compromised slow LM301’s, thereby making more careful choices. My 2500 is not a slavish copy, but hopefully its how the original in 1969 was suppose to sound, with no cross talk and precision components.

Design Characteristics The AMSynths 2500 modules are designed with these characteristics:

  • 5U high panels, MOTM style, 3.5″ wide.
  • +/-15V power supply.
  • 2mm Aluminium front panel designs that are similar to the original.
  • 3.5mm jacks mounted to the front panel, no Cherry matrix switches.
  • Original Switchcraft push buttons with incandescent bulbs.
  • High quality 1/8″ shaft potentiometers (PRV6 and 270X).
  • Large format 150 x 150 mm PCB’s with the original component layout.
  • Wide PCB traces – 50 mil power and 40 mil signal.
  • 1/4W 5% carbon composition resistors as per the original.
  • 1% Metal film resistors where specified by ARP.
  • Ceramic disk capacitors.
  • TO99 “can” Op Amps such as LM301.
  • Encapsulated sub modules, like the original.

Build & Outcome This has been a long project, a total of 15 years! Progress in the early years of 2005 – 2010 was limited to schematic production in Eagle CAD and locating rare parts including a set of ARP 2500 knobs. Faster progress was made in 2015 to 2016 with an initial set of 16 modules completed or nearly so. But a busy day job reduced progress for the next 3 years until I retired in 2019 and had the necessary time to complete the project. The cost of the potentiometers and knobs have increased to the point where I doubt I could afford to do this again! Each module costs over £200, so the total project cost is near to £5000.

There are four 6U Oak Angled Cabinets:

  • Rack A – 3x 1004-2600 VCO’s, 1006 Filter Amp, 1033 Dual DADSR
  • Rack B – 1036 Dual S&H, 1005 Mod Amp, 1047 UAF, 1033 Dual DADSR, 1050 Mix Sequencer
  • Rack C – 3x 1045 Synth Voices and 1027 Sequencer.
  • Rack D – 2x 1004-2500 VCO, 1016 Dual Noise, 1047 UAF, 1050 Mix Sequencer

ARP 1035 Module

The “Lost” Modules ARP announced a further 14 module around 1970 but never launched them into production, only the 1035 has ever been seen in the first 2500 system. I have selected some of the more interesting modules to replicate, click on the links.

  • 1013 Octave Fixed Filter Bank
  • 1015 Half Octave Fixed Filter Bank
  • 1019 Oscilloscope
  • 1024 Minisequencer
  • 1025 Dual Reverberation Module
  • 1028 Slave Module
  • 1029 Frequency Shifter
  • 1031 Multipurpose Filter
  • 1032 Mixer/Distributor
  • 1035 Triple Modulator – prototype produced
  • 1037 Dual Voltage Controlled Envelope Generators
  • 1038 Dual Preset Envelope Generator
  • 1040 Noisillator (VCO & Noise Source)
  • 1042 Triple VCA

Commercial Potential I have no intention of building and selling lots of replica 2500’s! I will leave that to other companies with deeper pockets and a bigger risk appetite. However I may build some modules in 5U for sale, and I may cascade some of the design down into Euro Rack. Please get in touch if you are interested.

Copyright AMSynths 2017