Lexicon PCM60 Refurb

Overview Back in 1980’s I could only dream of a Lexicon reverb, slim expensive 1U rack FX’s that lived in Pro recording studios. My first FX unit was (IIRC) an ART reverb which was quickly replaced with an Alesis Quadraverb, which served me well for 10 years in the 1990’s. During 2000/2010 I downsized the studio and had no need for FX, even a TC3000 only lasted a year. New synths around this time tended to have in built FX.

In 2017 I built a new studio space with room for some outboard. I was fortunate enough to buy both a PCM70 and PCM80 from a couple of old recording studios that were closing down, one of them in France. These are the back bone of my reverb setup, along with an old Yamaha REV7, which I first saw in a recording studio in Reigate in the mid 80’s. Gated Reverb – ha!

Whilst the PCM70 and 80 have some great reverbs, they are more useful during mix down, rather than for tracking synths. They stay hooked up the the analog console as outboard. For tracking synths I prefer older Roland SDE delays which sound like refined tape delays, a refurbished Quadraverb or my newly acquired PCM60 reverb.

Lexicon PCM60 The advantage of the 60 is you get two reverb algorithms derived from the legendary Lexicon 224 (Room and Plate) with just 4 time delay settings and on/off low and high EQ. No presets, everything is real time button selection, no deep spaces to get lost in – just rooms. The 60 also has an effects send and return which means you can patch in a delay or EG to further change the sounds.

It took me a few years to find one, and its not a perfect example by any means. But I finally have a PCM60 which I use with my “Big Moog”, its patched into the 984 Matrix Mixer, whilst the PCM60 effects loop goes of to a SDE1000 Digital Delay. I prefer to print sounds and FX on the way into my DAW, rather than in post production.

Refurbishment Outside the PCM60 needed a clean, and there is writing underneath the buttons where someone has recorded their patch settings. Unfortunately this permanent ink will not come off and is embedded in the now faded blue part of the panel. I will have to live with this, unless I can find a better panel in the future.

Electronically the PCM60 works well with only two issues; The input level pot is worn and jumps to maximum causing a lot of distortion as the input stage is overloaded. Fortunately the pot is still available from Mouser, so I replaced it. The second issue is that one of the signal level indicator LED’s is not working. I suspect this is a problem with the LM3915 driver chip which is socketed, so easy for me to replace.

The Lexicon components are high quality and the PCB layout is excellent. The power supply capacitors date back to 1984, so I have replaced them with new Panasonic high quality versions. I am pleased with the sound and noise floor, so I haven’t made any Op Amp changes which would be; FET input and output S&H to AD823,  Bipolar input/output to LM4562.

Outcomes The PCM60 is cheaper and rarer than the well loved PCM70, but it is very well engineered and the two sets of sound are very nice. There is a set of version 2 ROM’s that replace the Plate reverb with an Inverse Room, which was popular back in the 1980’s, but a cliché now. The V1.0 ROM’s are 24-pin 2732’s and the V2.0 is 28-pin 2764, they can be kludged together and manually switched but the Inverse Room is not worth the bother and cost.

 

Roland SDE1000 Repair

Roland SDE1000

Overview I bought a broken Roland SDE1000 Digital Delay in November 2020 for just £40, but with the known fault of it not powering on. I already have a mint and boxed SDE1000 which I find to be a fantastic delay unit for using with analog mono synths. The plan is to repair this one and use it with my “Big Moog”, instead of the Revox A77 tape delays that Klaus used.

This particular SDE1000 was manufactured in March 1985. This was two years before I rekindled my interest in synths and a secondhand ARP Odyssey Mk3.  Although the Roland was a lot cheaper than studio FX at the time, it was an expensive luxury for me at the time.

The Roland SDE-1000 was one of the Japanese company’s first effects unit, launched in July 1983 at £399 alongside the more powerful “studio version” SDE3000 which was double the price. The delay unit is quick & simple to use, with a very straightforward front panel & interface.  Sound quality is surprisingly good with a smooth, very analogue sound. This is in part due to the companding circuits to get the analog data into just 12-bits.

Maximum delay time is 375ms in standard mode (750 ms in x2 mode) and 605/1210 ms using the x1.5 rear panel control.  The sound of the SDE in X2 mode is quite reminiscent of a slow tape delay, and increasing feedback results in a gradually decaying, dulling repeat – again, like tape.

Delay time is displayed on a 4-digit blue fluorescent display with an Up – Down rocker switch to alter delay settings. The LFO has speed & depth controls, which can give deep chorusing effects or a gently shifting delay with phasing.  There are only 4 memory presets and no MIDI but the SDE1000 was a sales success with over 7,000 made, and it remained popular before Alesis entered the market with custom DSP chips.

SDE1000 Advert

Technology The SDE1000 is an early digital delay which uses a Gate Array chip as the main controller rather than a DSP chip, and a 12-bit R2R DAC rather than a dedicated DAC chip. The microprocessor is the familiar 8049 which Roland used in many products during the early 1980’s. The analog signal is compressed into a 12 bit data word with three 64k bit RAM chips used to store the digital data.

Roland used good quality NE5532 Op Amps in the output circuits and a dual transistor input buffer

Changes The SDE100 went through a number of circuit changes in 1983 to improve the headroom and HF response. They are documented in the service manual. This particular unit in from 1985 so it has these changes implemented and a V3 PCB which is not mentioned in the service manual.

The Repair This SDE1000 is in reasonably good external condition and a sound interior, which has possibly had a small amount of repair work (like the rear pot for time adjustment). It is a robust and reliable design, so I am not expecting chip failure but a power supply problem.

The power supply provides many different voltage rails;

  • +/-15V rails for the analog circuits using a discrete voltage regulator
  • +5V for the digital chips using a 7805
  • +12V rail (7812) for the front panel LED level indicators
  • +12V for the DAC voltage reference
  • +20V for the LCD driver chips
  • +1.7V for the LCD itself

Any part of these circuits could be where the short circuit is, so this is going to take some time! I checked the X2 safety capacitor on the mains side of the transformer and it was ok but I have replaced it. I disconnected the power connectors until I was left with the +1.7V rail which was the source of a short circuit.

I disconnected the LCD and switch PCB from the main PCB to eliminate it as a source of a short, however there was an additional short in the 15V rails. In the end I replaced the W02 regulators and power diodes, all the power supply capacitors, power transistors and power capacitors. Even though none tested as failed, this complete overhaul solved the problem and the SDE1000 powered up perfectly and works a treat.

Sounds The original factory SDE1000 presets, which can be overwritten, are:

  • Long Delay: 750ms with feedback and light modulation
  • Doubler: 30ms delay with feedback and no modulation
  • Chorus: 50ms of delay with light modulation and feedback
  • Flanger: 15ms of delay with deep modulation

The sound quality does deteriorate when in x2 mode with a limited bandwidth of just 8kHz. So I keep the delay in x1 mode with the rear trimmer set to 1x or 1.3x.

Outcomes A nice warm digital delay line for £40 and another £40 on component replacements, that will maintain the SDE1000 for another 35 years! I replaced the original old battery while I had the delay apart and I will re-calibrate using the notes in the service manual before it finally goes in the studio. The slight change in power supply voltage rails may not be exactly as it came in the factory back in 1985.

Behringer 904A rewinds to 1967

Moog 904A Caps

The 904A Low Pass Filter The Moog low pass filter module dates back to 1966 and is probably the reason you want a Moog modular! The Behringer VC LPF module is a faithful copy and uses THD polyester capacitors in the filter stages, with smaller values to ensure they fit onto the PCB.

It sounds very nice, but lets make it sound fantastic and rewind back to 1967 and fit the same filter capacitors that Moog used in the IIIP.

Rewind to 1967 To get to the 1967 Moog 904A is tricky, as the larger capacitor values are much harder to locate, and are large in size – if the correct polypropylene types are used. These caps won’t drop into the existing Behringer PCB space, so the original capacitors need to be removed and a new daughter board fitted to the rear.

I developed the daughter board to use a combination of 1.5uF, 390nF and 100nF hand matched polypropylene capacitors. It plugs into where the old Behringer capacitors were on the main PCB, using PCB pins and a right angle 10-pin power socket replaces the original. The PCB also supports the AMSynths 904C, as it puts the jack socket inputs and outputs onto a rear mounted Molex connector.

Behringer 904A

Outcomes Testing the upgraded filter correctly shows that the frequency cutoff is lower due to the larger filter capacitors and the frequency cutoff is easily adjusted with the RANGE trimmer as there are holes in the new PCB so that both trimmers can be adjusted.

Listening to the audio output and measuring with a frequency analyzer reveals that the filter resonance quality has improved, with resonance down to a lower 40Hz and a 3rd order harmonic peak appearing in addition to the 1st and 2nd orders. This is as a direct result of using matched polypropylene caps.

These are subtle rather than dramatic improvements but certainly worthwhile, in terms of filter resonance and the overall quality of sound. It is now the sort of filter that you want to use all the time! It is worth noting that the 1967 version of the filter did not self oscillate, and it was only later that the circuit was modified. I have retained the self oscillation as its a key part of the Moog sound.

Costs The PCB, caps and parts cost a total of about £30 and it takes a couple of hours to carefully desolder the Behringer caps and install the new PCB. This is an easy project for the competent SDIYer. Matching is done with a capacitance meter and 10x of each value need to be bought to get four closely matched.

SCI 700 Programmer

SCI 700 Mark 1

Overview  Back in 2002/3 I bought and restored a couple of SCI 700 Programmers, and here is the “restored” blog page from my old web site. The Sequential Circuits Model 700 Programmer is a rather useful way of adding 64 patch memories to an analog synthesizer. In November 2002 I bought one from the USA that need some repair work.

The front panel had corroded quite badly and it was externally in a poor condition. So I decided to do a full electronic restoration and rebuild it into a 5U 19″ panel so it could sit with the AM Modular cabinets. The picture on the right shows the Model 700 as it arrived from the USA. In April 2003 I acquired a second Programmer 700 (a Mark 1), this required a lot less work to get it fully operational. I sold both of these in the mid 2000’s.

Overview  Back in 1976 Sequential Circuits was a small music technology company based in San Jose, California and run by Dave Smith from his garage. After successfully designing and launching a digital sequencer (Model 800) Dave and new recruit John Bowden (x-Moog) went on to create the Model 700 Programmer. The new product provided rudimentary patch memories for analog mono-synths like the Mini-Moog and ARP 2600. The table top programmer had 64 memories of 3 control voltages to drive the VCO’s and two DADSR envelopes to drive the VCF and VCA of the partner synthesizer. In 1977 this was a major step forward, with only Oberheim providing patch memories on its 4 and 8 voice synthesizers.

Mark 1  Dave launched the Model 700 in 1977 and sold one a week, mainly in the USA. Here is a Mark 1 model, all the knobs are large, there are no CV trimmers on the top left hand panel and the patch switches are engraved with numbers. The record switch is part of the right hand lower toggle switch.

SCI 700 Mark 2

Mark 2 In 1979 the Mark 2 model was released with front panel V/octave trimmers for the 3 control voltages, and a single multi-way socket for a single cable connection to the partner synthesizer. The Mark 2 has revised PCB’s and a slightly different circuit design. The original CA3080 and matched transistor design was updated to use SSM2050 and SSM2020 chips, along with TL072 Op Amp’s.

The high quality sealed cermet pots were also replaced with standard carbon pots. All these changes meant the Model 700 could be manufactured for less cost and Sequential Circuits went on to sell over 200 Programmers, and it was still in the catalog in March 1981.

The Model 700 has a place in synthesizer history and it provided some funding and technical R&D for the legendary Prophet 5. Below is a Mark 2 version with Prophet 5 knobs, and lettering above the tactile switches. The 7 segment display is larger and the patch Record facility has a red button rather than being on the right hand lower toggle switch.

Rick’s SCI 700

Description The Model 700 has 3 independent control voltages for controlling the pitch of the partner synthesizers VCO’s. Each pitch is controlled by a rotary potentiometer, and a single external control voltage from a keyboard can be optionally added into each pitch with 3 toggle switches. The pitches are quantized to semitone values.

Two envelopes can be programmed with the usual ADSR and an initial delay – very Dave Rossum! The envelopes are triggered from gate or trigger inputs, usually from the external keyboard. The envelopes are designed to drive the VCA and VCF of the partner synthesizer directly. Each envelope has its own VCA so that the envelope volume can be controlled from the 700. There is an additional offset control for envelope 2, so that the initial VCF cut-off can be set too.

Memories Once a patch has been set up with the programmer you can store it into one of 64 memory locations, selected by 8 switches (Programs) and a rotary switch (Banks). A red Record button is partially sunken into the control panel, which means it is not accidentally pressed. Patches can be recalled or the programmer can simply by used as a live set of controls.

Tricks  The 700 has a few neat tricks up its sleeve. The 8 Programs within any Bank can be incremented electronically from a footswitch or a LFO or keyboard. Using an external LFO you can create a traditional 3 control voltage x 8 step sequencer with different VCF/VCA sounds for each step. The number of Programs that are incremented can be adjusted from a rotary front panel control from 0 – 8. When set to 3 the Model 700 will switch Programs as 1-2-3-1-2-3-etc.

Technology The Model 700 is based around CMOS logic chips, there is no micro-processor as the new Zilog Z80 was too expensive to use in 1976. Dave couldn’t use ADC chips either, so he emulated one with discrete digital and analog chips. The control voltage resolution is 6 -bits, which for the VCO controls equates to 64 discrete voltages, 5 octaves from 0 – 5V. However the envelopes also get the same resolution, which is a bit more of a limitation.

Inside there are two neat and well laid out PCB’s stacked on top of each other. The first board contains 16 pots which are scanned and stored as 6-bit digital words into 768 bytes of SRAM storage, which is backed up by a 3V lithium battery. This data is used to drive the 3 control voltages, and to control two DADSR envelopes based on SSM2050’s, along with two VCA’s built from SSM2020’s. This design is very similar to the Dual Transient Generator made by E-mu Systems.

SCI 700 Panel

Restoration Part 1 The first step was to see if the Model 700 would power on, and to do a smoke test! A quick internal visual inspection showed no burn outs, so I powered it on and was greeted by a set of LED’s that worked perfectly but no control voltages or memories. Plus the tale tell smell of burning – a tantalum capacitor was on its way out!
A detailed inspection revealed:

  • The +5V rail was okay – the regulator was replaced as a precaution
  • The +15/-15V rails were dead – both regulators were replaced
  • The comparator reference voltage had failed – thanks to the blown tantalum capacitor. I replaced this with an electrolytic and replaced the scarred 47 ohm resistor on the inbound connection to the LM723 regulator chip – which I also replaced as a precaution.
  • The lithium battery was dead – so it was replaced with a Varta 3V lithium 2/3AA
  • The transformer was in very good condition – so it was left alone.
  • Some IC’s had previously been replaced, as there were some new IC sockets. I added sockets when replacing chips, as Dave Smith had only put sockets in for the SRAM memory chips.

An order for new regulator chips was sent out, and whilst I waited for the parts to arrive I replaced the tantalum capacitors with electrolytic capacitors, and completed the following upgrades:

  • The electrolytic power regulator capacitors were replaced with high temperature versions – they are smaller too!
  • All ceramic disk capacitors were replaced with new dipped versions
  • The rusty old pots were replaced with Bourns sealed conductive versions
  • New E-mu control knobs in solid aluminum replaced the beaten up originals
  • New slide switches replaced the worn out Switchcraft switches which had started to have a redundant center position, as well as lots of surface rust.

Restoration Part 2 With the new power regulators in place and the rails working again, it was possible to power up the 700 and see what else had failed. Quite a list!:

  • Voltage 1 had gone permanently -12V, this was fixed by replacing IC4 (LM348 Op Amp)
  • Voltage 2 was fine.
  • Voltage 3 was correct but modulated at about 20Hz, once again replacing a LM348 Op Amp at IC3 sorted the problem.
  • Envelope 1 was at 0V – the SSM2050 and SSM2020 checked out okay, so it was clear that IC3 was the problem. When replaced the envelop worked perfectly.
  • Envelope 2 was fine, although the release time was over 30 seconds! I will add a trimming control, as in the SSM2050 application notes.
  • The patch buttons didn’t work consistently, so I replaced almost all of the CMOS logic around the scan and clock logic, using new IC sockets. The key problem turned out to be IC14 – it needed an exact replacement MC14163. Near equivalents gave erratic responses.
  • The 12 position rotary switches have been replaced with new plastic ones. The original shafts were damaged when I removed the control knobs, and they were rather clunky.

April 28th 2003 (5 months after purchase!) and everything was working perfectly again.

New Panel And finally a nice new 3mm aluminum front panel, 5U high, 19″ across in aluminum with black lettering. This replicates the old panel exactly, so that the PCB’s can be mounted directly to the panel. The rear jack sockets for inputs and outputs have been moved to the right hand side and are 3.5mm jacks (my studio analog synth standard). The new front panel was designed during May – July 2003, with many checks against a paper print out of the front panel. The panel was finally ordered on 8th August.

Restoration – Again The second Model 700 I bought in April 2003 required a lot less work, I’ve added a lithium battery and replaced all the decoupling and PSU capacitors. It is a Mark 1 model, which means its based around lots of CA3080’s, with no SSM chips, very different to the Mark 2.

I have kept the original red LED’s and 7 segment display, replaced the potentiometers with high quality Spectrol units (there is not enough space for the deeper Bourns 91A). The control knobs were replaced as well. They stood proud of the front panel by 1/4″ thanks to the factory not cutting the pot shafts down to size! Overall the unit worked very well as bought, which is a good job as it cost £300.

AMSynths 700 Well its now 2020 and the time is right to reproduce the 700 as a EuroRack module. My “Big Moog” has a set of 3x 921B VCO’s and a PPG303 filter that are ideal to connect up to a new 700 programmer with 64 patches. The availability of CEM3310 clones makes a reproduction 700 viable and the patches (and notes) can be driven by the 960 sequencers, so the timbre of each note changes. The 700 quantizes the VCO notes into semitones, so this is a perfect match for a 960.

A&H Spectrum Mixer Refurb 3

Second Master/Group Section

Overview During March and April 2020 I had more time to complete the refurbishment and was lucky to find a set of two 8 channel Spectrum input channels with panels and frame on eBay. This enables me to refurbish a full set of channels outside the mixer, as well as having a full set of spare pot and switch buttons. The panels were more rusted than my original, so I am only transferring the channel cards in.

In October I located a Master/Group section (made 30 April 1991) with rather nice wood ends, some different coloured pot caps and a rusty front panel. This enables me to restore a set of master and group cards whilst keeping the working Spectrum in the Studio. A previous owner had started to modernise the Left and Right master channels with nice Nichicon caps, OPA2134’s and unusual Roederstein EKU Bakelite Bipolar Caps. Unfortunately cheap IC sockets were used and the Op Amps move about in them, so these will be replaced with turned pin versions.

Master Section Refurb The two master channels have the highest level of noise in the mixer even with the groups and channels off. The L and R signals are buffered by a 5532 Op Amps and then it is TL072’s downstream from there on FX Sends. The 5532 buffer was replaced with bi-polar OPA1612 Op Amps and the FX return TL072 with OPA1642. Both the new chips are SMD and are fitted with PCB adapters and 100nF SMD power decoupling capacitors.

The OPA1642 was chosen due to its low noise (5nV) and low power consumption at 28% higher than the TL072. It also has low THD and reasonable offset at 1mV. The 5532 is also used in the five balanced output drivers where it is replaced by a new PCB and THAT 1646 chips. The usual electrolytic recap was completed at the same time and the slate oscillator fixed. The ten Op Amp upgrade cost was around £45.

This was the original plan, but the second mixer had already been upgraded with OPA2134 and OPA2604 so I will try with the cards initially on the Left and Right channels.

Original EBOS Board

Master Balanced Outputs The Master section has five balanced outputs that use a separate EBOS PCB with Main Left & Right, Mono and Aux 3 and 4 outputs. It is an important part of the mixer and I want to ensure it has high performance.

The circuit uses NE5532 Op Amps and a trimmer to get the offset correct. Whilst this design was fine in the 1990’s, the rest of the PCB needs updating with new caps, 1% resistors and new cermet trimmers. The PCB is single sided and poor quality, with traces breaking, not at all like the main PCB’s. So rather than restore the old PCB, I have designed a new PCB that uses modern THAT1646 chips which provide an uplift in performance and reliability.

The new PCB has RFI suppression added to each circuit, with 100pF caps and inductors on the outputs. The THAT chips reduce the power consumption from 40mA to 30mA, which isn’t much, but any power reduction is worth having! The slew rate and THD also improve, so we should hear a better sound.

Group Refurb The eight group circuits have a high noise floor and a couple of the groups are not working correctly. The design has five sections:

  • Group bus summer
  • FX Sends for the lower and upper channels
  • EQ for the lower and upper channels

The bus summing is done with NE5532’s which have been replaced with bi-polar OPA1612 Op Amps. The FX and EQ originally used TL072 Op Amps which I have replaced with JFET OPA1642’s which reduced noise, and improve slew and THD. The upgrade cost is £22 per Group, with 10 IC’s replaced per card.

Channel Refurb The channels are a lower priority for upgrading due to the reduced cost/benefit, and the priority is to get all 16 channels working. Each channel has 4x TL072 Op Amps and a NE5532 as the combined Mic/Line pre-amplifier. Replacing all the Op Amps across 16 channels would cost £350, and I have to be careful not to increase power consumption. Upgrading the TAPE inputs to balanced is a priority.

With more channel cards available from buying another Spectrum I can experiment with different op amps.

PC Noise My Spectrum lives above a Windows PC used for audio recording, which puts out a lot of high frequency junk onto the mains which the mixer picks up rather loudly. This has been sorted out, but moving the mixer onto a clean power supply did not make much impact. I have added an external mains EMI filter to the PC along with ferrite rings on the PC and Mixer power cables. I also have an EMI internal filter for the RPS power supply and I will check the mixer frame ground is sound.

 

 

Prophet 600 Refurb – Part 1

P600 as bought

Overview In June 2020 I rescued a rather forlorn and broken Prophet 600 for way too much money, but I decided it was worth the time and effort to get it back it to perfect working order. Although financially it does not make sense, as I will probably spend more on it than its currently worth, once I have fixed the P600, I plan to implement a number of upgrades.

  • Gligli OS upgrade
  • New power supply
  • Stereo chorus and voice panning
  • New Fatar keybed
  • New walnut end cheeks

The Prophet 600 was announced at NAMM 1983 and shipped a month earlier in December 1982 with the first MIDI implementation ever (thanks Dave!). It was a reasonably successful product with 6000 (approx) sold until mid 1985, against the heavy competition from the DX7. The innovative use of digital envelope generators and a digital LFO pushed the capability of the original Z80 microprocessor, the slow envelopes and obviously stepped pots put customers off. The budget nature of the P600 was a bit too apparent!

CPU board

Part 1: Getting The P600 Working The first stage is to get P600 tested and working again with no modifications, although dusty the circuits look to be in good condition and better than I expected. I downloaded the Technical Service manual from July 1983, as I am sure this is going to be a constant companion for the next few weeks! My Prophet 600, serial number 1406, arrived at the workshop on 12th June and I logged these initial visible faults:

  • Large number (11) of keys missing
  • The keybed is in very poor condition, needs to be replaced
  • Slightly damaged Mylar control panel
  • No control knobs
  • One pot missing
  • One pot with spindle cut
  • OS ROM missing, all other chips in place
  • All cabling intact but disconnected
  • Pitch and Mod wheels stiff
  • Battery disconnected

The front panel was cleaned and found to be a very good condition with only one visible chip down to the metal which will be carefully infilled with black paint. The rear and underside of the casing have not survived as well and there are numerous areas of black paint missing. The white silk screen lettering has survived, so it needs a black repaint feathered into the original. The rubber feet are in place but the wooden end panels will be replaced. All the intact pots rotate ok and need a clean, and the toggle switches work fine.

Power On Whilst a set of OS (Version 8) and Diagnostic ROM’s arrive from Germany, I can test some basics; is the power correct and the microprocessor being clocked. There are no visible signs of power failure, such as blown caps or regulator heat damage. The first step is to power on the transformer at the correct 240V but disconnected from the CPU PCB. I then checked the AC output voltage rails which should be 18 and 36 VAC. Measurement shows all ok at 22 and 44 VAC with slight transformer whine.

Onto powering up the CPU and Voice PCB’s (they are hard wired together) and checking the analog and digital DC rails. I left the keyboard, bender and panel PCB’s disconnected at this stage. The power rails measured good (+4.97V, +15.07V, -4.96V,-14.95V) and there is a 4 MHz clock at pin 6 of the Z80 chip. I ordered new OS ROM and Diagnostic ROM chips from a German eBay seller and they arrived very quickly.

Initial Repairs Two new potentiometers were ordered from Wine Country in the US and the wooden sides measured, so that new American walnut sides could be made by MintCase. The piano hinge for the front panel is missing the six machine screws that attach it to the rear casing. I ordered a pack of ten M3 x 8mm black posidrive machine screws, and they were sometimes a loose fit requiring a M3 nut to be fitted as well. More progress in Part II to follow, when we try and boot up the P600.

 

 

 

OB-Xpander Replica

OB-Xpander

Overview The Oberheim OB-Xpander was a prototype analog synth created around 1982, and not to be confused with the later Oberheim Xpander which although similar is significantly different. There is very little information about the first Xpander, just a picture in Mark Vails Vintage Synth book (volume 1) and some text in the A-Z of Analog Synthesizers. From this limited information I have concluded that the OB-Xpander was based on the OB-8 and developed in the Summer/Autumn of 1982 (because the panel graphics reflect an early OB-8 with Page 2).

Features Here is a probable feature list:

  • Part of the Oberheim System alongside OB-XA, DMX and DSX
  • Four independent “multi-timbral” analog voices, using one OB-8 voice card
  • Four independent LFO’s in the  modulation section
  • Keyboard controls, Split/Layer, Unison have been removed
  • Probably the same OB-8 CPU board with some minor modifications
  • Four sets of CV/Gate inputs as well as the Oberheim Computer Interface
  • Stereo and Mono Outputs with preset voice panning
  • Cassette storage for patches is retained
  • No MIDI, as this did not arrive at Oberheim until Spring of 1983
  • Page 2 included, as the Modulation has the additional white silk screening

Modulation Section The Modulation section needs to be per voice with 4 separate LFO’s, to ensure each voice has different sounding patches. The OB-XA has two analog LFO’s for the Lower and Upper voice banks, whilst the OB-8 has two digital LFO’s with the extra sawtooth waveforms that are screen printed on the OB-Xpander panel. So could the OB-8 OS have been extended to generate 4 LFO’s in the OB-Xpander? The limitation is in the Z80 processor speed, but the keyboard scanning workload has been removed, CV/gates added and the envelopes are analog.

The OB-Xpander has a switch in the lower corner of the modulation section, which the OB-XA does not have and was used by the OB-8 as VOLUME MOD. However the labeling is shorter, so maybe this was just VOL.

Master Section This has been reduced down to one column of controls;

  • MASTER VOLUME potentiometer
  • AUTO switch, labelled as per OB-XA
  • MASTER TUNE potentiometer

Control Section This remains as one column but has one change;

  • PROGRAM VOLUME potentiometer sets Voice Volume
  • PORTAMENTO potentiometer
  • OSC2 DETUNE potentiometer and red LED

Oscillator Section The VCO controls are the same as the OB-8 and OB-XA and the panel lettering verifies the OB-Xpander is based on the OB-8 voice card, as it has the lettering underneath the Pulse and Saw switches for Triangle. This is how early OB-8’s were lettered before the Page 2 functions were added on the silk screen.

4 Voices on CPU board

Keyboard Section This has been renamed to three words which are too blurred to read, maybe PROGRAM VOICE SELECT. There are switches to select each voice for programming and patch save/recall. with a fifth switch (assumed) to call up a four layer program (one of 12 available, as in Double mode). The cassette Play and Check functions are retained with the corresponding red LED. The OB-8 VOICE board actually has 4 voice select signals from the CPU board (at A13, A15, A17, A19) but only two are used for Lower and Upper voicing across two PCB’s. This confirms that the OB-8 is the basis of the Xpander and that it would have used 4 voice boards, each with only one voice populated.

Programmer Section The patch selection remains the same as the OB-8 with the familiar row of A-D and 1-8 switches, with 120 patches and a WRITE/RECORD switch at the far right hand side, labelled as per the OB-XA.

Prototype Only The OB-Xpander never made it past prototype stage, probably due to being only 4 voices, and the limitations of the Z80 processor running 4 independent digital LFO’s. Oberheim reworked the idea in early 1984 as the Xpander, with 6 independent voices, lots more digital control, MIDI and CV/gate. The processing power was increased to cope with more features and the Xpander has stood the test of time and is one of the greatest analog synths ever made.

Case Dimensions Peter Forrest states this as; 90cm x 30cm x 12.7cm, the same depth and height as the Oberheim DSX and with a weight of 9Kg, which is half the weight of an OB-8. The Xpander case does not look deep enough for a set of OB-XA voice cards to fit in.

Licio Comisso OB-8 VOICE card

Replica It is possible to recreate the OB-Xpander? It would make a very nice sounding 4-voice multi-timbral synth. All the CEM chips are readily available (CEM and AS), and both the VOICE and CPU boards could be recreated from the schematics. Licio Comisso has already successfully cloned the OB-8 VOICE and CPU boards to recreate the OB-8, but when I contacted him he only had OB-XA boards available.

A rather wide 30″ case would need to be fabricated, with a silk screened front panel. Using smaller width switches might mean the case width could be useful reduced. The power supply would be mounted off the CPU board, and improved.

OB-8 OS Modification I initially considering changing the OB-8 Z80 source code to support the Xpander. The largest change is to increase the existing two digital LFO’s to four, which is no easy task as there is no source code and its quite a large operating system to reverse engineer (16 KB), however this has been done before. The rest of the changes required are smaller. I disassembled the source code but it is a lot of work and time to get the changes made.

In addition I need to find a way to retain the calibration capabilities which use the Bender Board and 1st octave of the keyboard, probably by using a set of miniature slide switches mounted internally. This means retaining the keyboard scanning.

Custom Gligli OS A possibly easier approach is to modify the P600 Gligli code to add the extra digital LFO’s, and at the same time removing the digital ADSR’s, cassette interface, and keyboard scanning. This would also give a strong MIDI capability, but the user interface and MIDI Sysex/cc needs changing to match the Xpander, and then there is all the Page 2 functionality to consider, and the loss of the Oberheim calibration software. The advantage is that the Gligli code is open source and written in C.

Conclusion Whilst an exact replica of the Xpander is possible, the limitation of 4 voices is a bit too limiting, given the amount of time and cost needed to build a replica. My preference is to build the OB-Xpander but as a desktop OB-8 with 8 voices, Page 2 and MIDI. This means no OS changes are needed and the CPU board can have MIDI built in. I will also remove the Noise circuit and add in two new filter modes (12dB HP and 6dB LP).

 

 

EEH DS 500 Sequencer

EEH DS 500

Overview This is a possible project to recreate a very rare and unusual early digital sequencer the EEH DS 500. Only a handful were manufactured in 1981/82 in Germany and used by Tangerine Dream and Klaus Schulze, rather briefly as they quickly adopted MIDI sequencers in 1983.

I came across the sequencer in May 2020 and the web pages by  Till Kooper here, which has bote the schematics and a German user manual. I have translated and improved the user manual into English and its here.

EEH DS 500 Opeartions Manual (English)

Looking at the schematics and after locating the OS EPROM binary I realised I could recreate the DS 500 completely. Whilst the sequencer is hard to program and has only 500 notes, it does have loops within loops, and makes an interesting project during the 2020 Covid-19 lockdown.

Technology Whilst the technology used in the DS 500 is 40 years old, the IC’s are still available at a low cost expect the DAC and unusual SN4929 combined NAND and Inverter CMOS chip, which are harder to locate and more expensive.

  • NEC D780C (Z80 microprocessor) £4.50
  • Intel 2716 EPROM (2KB)
  • Two NEC D444C SRAM 450ns £7.00
  • Intel P8225A £3.50
  • Ferranti ZN426E DAC £20+
  • 74LS48 and 74LS145 £3 each
  • SN4929N
  • CMOS 4016, 4030, 4011 , low cost
  • LM324 replace with low drift OP2277 Op Amps
  • DL304 7-segment displays  £7.50

Switches & Buttons The original Rafi PPG style switches and buttoms are no longer manufactured. The closest match is from Marquadt and the 6425 series of momentary switches. These have 16mm sqaure caps in different coloures, lettering and with space for a LED. After lots of research I gave up on finding every numeric key cap (7 was not in stock) and went with the correct light grey, dark grey and black cap colours with a red start/stop button. I will label the numerics and button descriptions onto the front panel.

Here is the BOM:

  • 2x Switch Body Red LED –  6425.4111
  • 4x Switch Body Yellow LED –  6425.4121
  • 4x Switch Body Green LED –  6425.4131
  • 10x Switch Body no LED –  6425.1101
  • 2x Red button with LED – 826.000.071
  • 8x Dark Grey button with LED – 826.000.021
  • 10x Light Grey buttons without LED – 826.000.031

EEH CM-4

Project Status PCB design stage as at 16 May 2020.

EEH CM4 This is a much more powerful digital sequencer that was used by Tangerne Dream in the Poland concerts, with 4-tracks and 16 steps. Five units were made around 1983. The CM4 lives in an 8U 19″ rack panel and has the following features:

  • 4x CV and Gate Outputs
  • Clock in and out
  • Clock divider switch
  • MIDI out
  • 40 memories (+ panel memory)
  • Patterns can be linked to songs.
  • Per track: display, end of sequence programming (reset value), note / pause ratio.
  • Quantization of notes.
  • Per step: potentiometer for programming the CV/gate values, switch with LED for switching the gates on and off.

EEH CM4 Close Up

It was based on the same Z80 technology used in the DS 500, but with more RAM (6 KB) and additional displays and front panel key pad controls. A coupld of 16-way multiplexers and two 8225 Programmable Periperal Interface chips are needed for all the I/O. A different DAC was used maybe a Burr Brown PCM54. The connections are at the base of the front panel, where you can see the CV and Gate outputs as well as MIDI, which presumably is just MIDI clock out, maybe clock in.

The usual PPG style Rafi push button switches with inbuilt LED’s are visible throughout, a total of 32. Unfortunately the CM4 must have been just as difficult to programme via a numeric key pad as the DS 500 and soon obsoleted by MIDI computers and seqeuncers. Tangerine Dream sold there two probably in 1984 and as least one still exists today.

Information and images courtesy of Mirko Luethge August 2002 and Carsten Werner 2005.

Roland MC500 Refurb

MC-500 As Bought

Overview I bought an original 1986 MC-500 in May 2020 for a nice low price but sold as a repair job. It turns out to be in very good condition and the issue was a damaged OS floppy diskette, which refused to boot and gave I/O ERROR 3. Once I had powered it up with a brand new DS/DD diskette with Super MRC-500 software it works perfectly. The LCD had faded after 35 years and needs an OLED replacement and the diskete drive is rather loud and clunky but I will keep it until it fails.

Old LCD Display

The MC-500 was way out my budget in 1986 at £999 and I bought an Atari 520STFM (£299) in 1987 and used it with MasterTracks Pro for many years until it was struck by lighting one night and exploded.

I had added a 20MB hard disk to the Atari to enable a large collection of songs to be built up. The price of hardware sequencers has of course fallen significantly, so I can now experience and use this approach in the Garden Studio.

New OLED Display

OLED Display This is an easy plug and play replacement using the Newhaven 2×20 NHD-0220D2W-ABJ, which is a nice blue on black OLED. The EL backlight connections are not needed and can be removed along with the EL transformer, which is on the PSU PCB.

My trannsformer has a slight wine so I removed R1 and R2 from the PCB to cut the power to the transformer, which avoided the need to take the PCB out. The display is hand wired to the 14-way cabling rather than using a plug and socket. I removed the old LCD from its metal frame and then carefully peeled the metal frame from the display bezel which has some doubled sided tape.

Diskette Drive The drive works ok and uses 3.5″ DS/DD diskettes which are a bit expensive to buy these days. The sensible approach is to install a Gotek drive but I rather like using floppy diskettes as I can write stuff on the label about the song. So whilst the drive still works I will keep it in place. I also have a MC-300 and MC-50, so I can swap songs between them.

Power Supply Recap I decided against a recap of the power supply, as the main electroytic caps were in good condition and it meant quite a bit of dismantling and desoldering cables. In the future I will do a full power supply refurb along with the regulators, but for now no changes, If you do want to replace them us high quality Panasonic EEU-FC’s with the same diameter and height:

  • 4700uF 16V – EEU-FC1C472SB
  • 2200uF 25V – EEU-FC1E222SB

Power Supply

Technology The MC-500 is based around an enhanced Z80 micro processor manufactured by Hitachi as the HD64180 with integrated memory management and on chip peripherals. It was launched in 1985 and initially Roland used the DIP64 pin version the R0, before moving onto the 80-pin SMD R1 version for the MC-500 Mark II launched in January 1988. The R0 DIP version can only address a maximum of 512KB, with just 256KB fitted in the MC-500 using 8x 32KB RAM chips.

The MC-500 Mk II uses the R1 micro processor which can address 1 MB of RAM, with 756 KB fitted, and the subsequent MC-50 and MC-50 Mk II also use the same microprocessor giving the this micro composer a life span of nearly ten years. The MC-50 versions use onboard memory for the OS to live in, so its faster to use and boot, but not as classic in style!

Memory Upgrade My MC-500 was manufactured in July 1986 only a few months after the launch in January, and it uses the original PCB layout with 256 KB of RAM and the R0 version of the HD64180 which has 19 addresss lines. This limitation means the early MC-500’s can not be easily upgraded to 786 KB of memory. The processor and RAM need swapping out and the extra address line added. Roland did offer the OM-500 upgrade, which must be a main PCB swap, and the resulting sequencer is called a MC-500B by Roland..

MRB-500 Software

Software Options & Versions In 1986 the MC-500 was launched with an initail software release (MRC-500) which was upgraded with more detailed features as Super MRC in 1988 along with a wider range of software to enable SysEx storage, chained songs for live performance, MIDI file conversion and a set of rthymn tracks. Here are the various software releases, click on the titles to download the .OUT file:

MC-500 like new

I use the MRB-500 to store MKS-80 Patch Banks as it perfroms the correct handshake and MRM-500 to transfer MIDI files to Cubase and back.

Recreating the Software The MRP and MRB software cannot be downloaded from Roland and the original boxed software is expensive at £50 – 100, when you can find it secondhand. Fortunately 20 years ago R. Kevin Grannum archived the files onto his web site in .DAT format using obscure uCopy software. The web page exists but the links are all broken, but with Wayback Machine I found the files and downloaded them.

With some careful hex editing and byte level compaisons I was able to trim the start of the files and add extra bytes at the end to make them into binary 720 KB files in .OUT format that SDISK (or WDISK) can write to a 3.5″ DS/DD floppy diskette. The diskette must be pre-formatted by the MC-300/500/50.

E-mu Systems AVI

Emu Systems AVI

Overview The Emulator I was released before the advent of MIDI in 1983. The MIDI specification was created by Sequential Circuits and Yamaha, and E-mu Systems were not able to add their influence. They were after a RS422 serial protocol – which is much faster than MIDI. This is why early Emulators and Emax’s had the RS422 serial interface – which allowed connection to Mac computers. Dave Rossum (quite rightly) thought this was a better and faster interface than MIDI. However MIDI won the day, and the faster RS422 standard never took hold.

So although the Emulator I eventually gained a JLCooper MIDI external interface, for the first few years owners were after a CV/Gate interface to enable the Emulator to be driven from other keyboards and most importantly from the new microprocessor digital sequencers – such as the Roland MC-4 and MC-8. Thus was born the Analog Voltage Interface, model number 6040.

Tomita The original Analog Voltage Interface was probably commissioned by Tomita for use in his 1981 Album of Orchestral music, with the Emulator playing Timpani from a MC-4B or MC-8. The Steve Miller Band used an E1 and this interface on their early 1980’s albums. Vince Clarke used one in 1983, probably with his famous MC-4B.

Blue Box The Analog Voltage Interface is a large metal blue box with a sloping front. There are 8 sets of CV and Gate jack inputs, located right on the front panel, one for each of the eight voice channels of the Emulator I. Below these inputs are eight toggle switches which control the mode of each channel. The three modes are:

  • Test – manually switches the channel CV into the E1
  • Internal – switches the test CV off
  • External – external gates control the E1

The Blue box links up to the Emulator I via a standard RS232 serial cable. The socket is round the back of the AVI. The voltage per octave can be trimmed via a trimmer which can be accessed through a small hole in the AVI front panel. The AVI must be used with the correct Emulator OS software, or samples can be damaged.

AVI OS Software

AVI Software The special Emulator OS version EP8SA.0308 for the AVI box has some additional functionality.

  • Mono Mode – monophonic mode on channel 1
  • Disable/Enable – turns AVI control on and off
  • Gate Test LED – GET SEQ LED shows key on/off
  • Scan Rate Control – speeds up the response time to CV’s

Technical The AVI has a Z80 micro-processor running at 5 MHz with a 2716 EPROM for the simple operating system. The control voltages are read by an ADC0809 8-bit analog to digital converter. The AVI software behaves as if it were the internal sequencer, so this functionality is disabled in the Emulator. The keyboard is still active, but its best to switch it off using the Scan Rate Control, as the response to CV/gates improves by 10 ms to 3 – 4 ms.

Success? Well this box is pretty rare, probably around 50 or so were produced from 1981 – 83. We know of only three in existence today and I owned one in 2000 along with four Emulator I’s. When I spoke with Dave Rossum in January 2001, he commented that it was one of the more unusual and less successful items that E-mu Systems made in the 1980’s. However low demand for a CV/Gate interface, is understandable, especially once MIDI took off in 1983.

Emu Systems AVI Rear

Connections The AVI is connected to the Emulator via a RS232 serial interface and cable. The Emulator acts as the DCE and has a DB25 female socket. The AVI acts as the DTE and has a male socket. The communication protocol is very basic with only TX, RX and GND signals being used by the AVI, although the Emulator has a more complete RS232 implementation. The RS232 socket is wired to the Emulator digital board via a ribbon cable, and it enters the board via a DIL header called IC41. Therefore a standard serial cable wired as follows is required. Only the lower 3 connections need to be made.

E1 DB25 Cable has Male Plug E1 IC41 pin AVI DB25 Cable has Female Socket
6 DCE Ready 7
20? DTE Ready 11
5 RTS 12
4 CTS 13
3 TX 14 RX 3
2 RX 15 TX 2
7 GND 10 GND 7
Copyright AMSynths 2019