Roland MKS80 Power Supply

Introduction This is a feasibility study into a replacement power supply for the Roland MKS80, and it is not a project that AMSynths are aiming to deliver, as Guy Wilkinson owns the rights to the MKS70/JX10 re-engineered power supply which this thinking is based on. Thoughts and ideas are welcomed!

Overview I am fortunate to have a near mint Roland MKS80 (March 1985) and MPG80 (August 1984). This is the second time I have owned one, as I had to sell my first MKS80 in 2001. It is a REV5 with a 110V power supply and therefore requires an external converter, which is all a bit bulky and messy. The transformer in the MKS80 is not tapped for both 240 and 110V operation, so a NOS 240V replacement transformer is needed.

Rather than going down that route I have was inspired by the work done by Guy Wilkinson in replacing the Roland JX10 power supply with a modern switched power supply. Details are here. This approach greatly reduces the heat from the old linear regulators as a bonus, as well as having 240V operation, and is a great long term solution with no capacitors to replace.

Roland are good at designing high quality power supplies with precision rails in the right places for analog circuits. The MKS80 is a completely analog voice circuit, where the MKS70 is using DCO’s and then an analog circuit path. We do not want to fall short of the Roland design specs.

Roland MKS80 PSU

The New JX10 PSU Guy’s design provides +5V and +/-15V rails with low ripple after careful filtering of a set of three Meanwell power converters. Unfortunately this PCB cannot be directly reused in the MKS80 as it requires a +10V precision reference voltage for the VCO/DAC and requires a more power to the 15V rails than a MKS70, even though it has 8 rather than 12 voices and the overall power consumption is similar at 35 W.

The low voltage connections are also in a different location, and the mains filter is on board the power PCB. A new PCB needs to be designed with slightly different Meanwell units to meet the demands of the MKS80. There is an Unregulated +7VDC in the MKS80 to power the MPG80, just like the MKS70 design which also powers the display. Guys design uses a Vigortronix power converter to deliver +9V DC at 0.55A which can be used for the MKS80.

+5V Power
MKS80 Rectifier = 2D4B41 (2A) Main Capacitor = 4700uF 800mA at 8.5VAC
MKS70 Rectifier = 4D4B41 (4A) Main Capacitor = 6800uF
Guys Replacement PSU delivers +5V at 1.5A

+/-15V Power
MKS80 Rectifier = 2B4B41 (2A) Main Capacitor = 3300uF 450mAx2 at 21.5VAC
MKS70 Rectifier = 1B4B1 (1.5A) Main Capacitor = 4700uF
Guys Replacement PSU delivers +/-15V at 600mA x2

Replacement MKS80 Power Supply The MKS80 service manual states that the +5V rail should be within +/-30 mV, which is a tighter tolerance than the IRM Meanwell converter delivers and the actual +4.94 volts measured in Guys design. The +5V does not get used as a reference voltage, but potentially a more accurate converter/solution is needed as AC/DC Converters typically have 2% initial voltage accuracy, and Roland are stating 0.6%.  In the MKS80 the +5V rail is using less power than in the MKS70, so it is possible to retain the Meanwell IRM-10-5 which is rated at 10W (provided we can accept the voltage accuracy).

For the 15V rails the more powerful 20W IRM-20-15, which delivers 33% more current at 800mA per rail, is needed. The 15V rails are at a much lower tolerance of +/-100 and 400 mV, so the Meanwell IRM can easily achieve this accuracy. Other Roland synths require very accurate 15V rails, such as the Juno 60 where the tolerance is +/-10mV which really demands a linear or LDO design.

The +9V supply remains the same as in Guy’s design, using the 5W Vigortronix power converter to deliver +9V DC at 0.55A, more than enough to power the MPG80.

The original Roland +10V power is provided by a 1SZ59 6.5V zener diode, with an Op Amp and trimmer to set this to an accurate +10.00V, the 1SZ59 has a temperature co-efficient of 2ppm/C which is very good! A modern LT1027 precision regulator can deliver the same 2ppm/C. It needs to be buffered with a precision Op Amp on the power supply PCB, although in the voice boards the +10V is further buffered by average performance M5218L Op Amps. The reason to take care here is that this voltage is used in the VCO’s as a reference voltage, as well as the DAC reference voltage.

The rest of the MKS80 replacement power supply can follow Guys design but with a layout and pin out that fits the MKS80 PCB dimensions. The mains power conditioner on the top right of the Roland PCB can be dispensed with and its taking up valuable space at the rear of the MKS80. The power socket needs replacing with a grounded 3 pin IEC socket with power leads going straight to the front panel ON/OFF switch and then running back to the bottom right hand corner of the new PCB.

MKS80 Power Connectors There are 7 low voltage power connectors on the PCB to supply the various MKS80 modules:

  • P1 Analog +15V 10 pin
  • P2 Analog GND 12 pin
  • P3 Analog -15V 8 pin
  • P4 Digital GND 6 pin
  • P5 Digital +5V  6 pin
  • P6 Reference Voltage Analog 2 Pin
  • P7 Unregulated and Digital GND 4 Pin

These need to remain in the same position on the new PCB to ensure the cables and connectors fit.

PCB Design The PCB size can be as large as a rectangular 100 x 200 mm, with the same mounting holes as the original. The removal of the large internal heat sink means there is more space at the rear of the case. I have mocked up a high level design with the 4 converters and the power input at the front corner.

Next step is to do a PCB layout mock up in Eagle CAD to see if the components and the required isolated tracks can fit. A solution to the +5V DC initial voltage accuracy needs to be found as the best AC/DC converter I can find is 1% accurate.

 

Roland IR3R01 Design & Replica

IR3R01 Chip

Introduction The Roland IR3R01 envelope generator chip was used in the early 1980’s in three key synthesizers; the awesome Jupiter 8, the Juno 6/60 and the MKS7. The chip is obsolete and impossible to locate as NOS, which puts these synths at risk of never being repairable in the future.

The chip is a custom design by Roland which although looks similar to a CEM3310 or SSM2055, it is not the same design. This post is about researching how the chip works, building a simulation circuit (especially for the Key Follow Jupiter 8 design) and finally developing an accurate PIC based replica.

Description The IR3R01 generates a positive going 0 to +10V output signal (maybe less on Juno and MKS) with traditional Attach, Decay, Sustain and Release phases. The chip has CV control inputs for the ADSR phases, as well as the usual gate and trigger inputs, and it always runs from +/-15V power with an additional ground pin. The Attack, Decay and Release are driven by 0 to +5V control voltages whilst the sustain control voltage is 0 to +10V. The CEM3310 uses negative control voltages, so is a long way from being a drop in replacement.

The chip also has a Reference Voltage pin as well as a Common pin, and finally a Modulation pin which is used only in the Jupiter 8 to enable the level of the envelope to be controlled by a CV. Here is the pin out:

  1. Modulation Input
  2. Attack Voltage
  3. Decay Voltage
  4. Release Voltage
  5. Sustain Voltage
  6. Common
  7. Reference Voltage
  8. -15V
  9. Ground
  10. Timing Capacitor
  11. Output
  12. Current Input
  13. Gate
  14. Trigger
  15. Retrigger
  16. +15V

Common Pin The input voltage adjusts the timing of the envelope, and a manual trimmer with a Tempco resistor adjusts the attack time maximum, so that all 6 or 8 envelopes have the same timing. On the Jupiter 8 this is set to 6 seconds (although on Page 1 of the service notes its stated as 5 seconds), on the Juno to 3 seconds and MKS 7 to a very short 2.5 seconds. The maximum timing of the Decay and Release is 10 seconds on the Jupiter and 12 seconds on the Juno and MKS. This variation could be due to the increase in Decay and Release CV input resistors from 19k to 22k.

Reference Voltage This is tied to +7.5V (Juno) or +10V (Jupiter) or +5V (MKS). Typically envelope generator chips use this voltage as the highest voltage the attack phase is aiming to, the actual maximum voltage is typically less. Some real world measurements would be useful.

Timing Capacitor This is 68nF (Jupiter and MKS) and 47nF (Juno), this will vary the basic timing of the envelope stages. The manual adjustment of the envelope timing via Common is required because all voices in a polyphonic synth need to have the same envelope times. The adjustment is needed due to variations in the timing capacitor value and variation in chips.

Envelope Oscillator

Modulation Input This is only used in the Jupiter 8 to implement Key Follow, elsewhere the pin is grounded and a different approach is used to implement Key follow in the Juno. In the Jupiter a higher pitch CV shortens envelope time. The IR3R01 circuit is unique with a 28 kHz saw wave being created by a 555 timing chip and sent to all the IR3R01 chips via a buffering Op Amp and a comparator with the Keyboard Follow control voltage.

The IR3R01 receives a 0 to +15V square wave to the modulation pin, which presumably varies in height or frequency to control the envelope speed. It is not yet clear how this works, but it is an important part of the circuit to replicate.

Initial Conclusions The IR3R01 has similarities with the CEM3310 design with a current Input pin and reference voltage, but the control voltages are negative in the CEM3310. The SSM2055 uses positive control voltages and a timing control pin, but no current input. Neither of these chips have the Modulation Pin of the IR3R01. A strong possibility is that the IR3R01 has some basis in the complex Jupiter 4 ADSR which uses oscillators that vary from 10 kHz to 800 kHz, for each of the ADR phases.

The clue is that the ADR front panel sliders on the Jupiter and Juno are non-linear, just like on the Jupiter 4. They have a 4K7 resistor added between pin 1 and 2 to provide a more exponential like CV, rather than getting the microprocessor to do this job. This could be because the ADR voltages in the IR3R01 are controlling the rate of an oscillator, as in the Jupiter 4 design. The modulation pin uses a clocked input to vary the length (not height) of the envelopes, which also suggests a Jupiter 4 ADSR with oscillators. With no IR3R01 datasheet, a more detailed circuit analysis is needed.

Key Follow in JP8

Replicating the IR3R01 There are two uses case of the IR3R01. The simple ADSR in the Juno and MKS where a SMD CEM3310 could be buffered to get a similar “drop-in” response. However a micro controller based design (like Tom Wiltshire’s PIC ADSR here) is going to be less component heavy and potentially could be adapted to the Jupiter 8 use case with the Modulation Pin being used.

By using a digital design the timing will be perfect and the exponential curves can be adjusted if necessary in the data tables, so there is no need for timing adjustment, reference voltage or timing capacitor and these pins can simply be left unconnected. The modulation input needs to be read and the level of the envelope adjusted accordingly. The PIC would need a +5V power supply and most of the input voltages would need reducing (simple resistor dividers?) such as Gate and Trigger, the 10V Sustain CV as well as the 15V Modulation input level.

Whether all this can be done in a 16 pin DIL footprint remains to be seen!

Next Steps I will create a test harness circuit of the Jupiter 8 Key Follow circuit to see what waveform is generated on Pin 1 of the IR3R01 as the Key Follow CV is varied from 0 to +5V.

Replica Design Here are some initial thoughts, I envisage two chips; One for the Jupiter 8 and one for the Juno 6/60, so the envelope timing can be perfectly matched to originals:

  1. Modulation Input – create Jupiter 8 test harness to work out what this does – use a resistor divider to read pulse frequency
  2. Attack Voltage – read directly and map to correct minimum and maximum times
  3. Decay Voltage – read directly and map to correct minimum and maximum times
  4. Release Voltage – read directly and map to correct minimum and maximum times
  5. Sustain Voltage – reduce by 50% using resistor divider
  6. Common – ignore
  7. Reference Voltage – ignore
  8. -15V – use for buffer Op Amp
  9. Ground – use and decouple with +5V
  10. Timing Capacitor – ignore
  11. Output – Op Amp buffer to increase output to 10V
  12. Current Input – ignore or use external resistors with the buffer Op Amp
  13. Gate – reduce to +5V using resistor divider
  14. Trigger – reduce by 50% using resistor divider or ignore and connect internally with Gate
  15. Retrigger – ignore
  16. +15V – use for buffer Op Amp and create +5V power for PIC using regulator

 

Jupiter 4 Voice Card Recap

Overview As part of my Jupiter One project I needed to recap and upgrade the Jupiter 4 Voice Card, ready for fitting to the Controller and Pot PCB’s  that I have designed. Hopefully these notes will help others recapping their own voice cards.

Why Recap? My voice card was made sometime in 1979 and 40 years later the capacitors have deteriorated, losing the original value. This is especially important in the timing capacitors in the envelope generators and the VCO, as well as the power supply bypass. I have used modern capacitors which are higher quality, and specifically upgraded the HPF filter capacitor and the audio capacitor between the VCO and VCF.

Further Changes To fit the Voice Card to my Controller PCB I need to carefully remove the 9 pin connectors as these will be replaced with SIP connectors. I also want to replace all the trimmer pots with cermet types to ensure the tuning of the card is accurate and stable. This includes replacing VR10 which had been swapped out for a potentiometer.

Power Capacitors The 100nF power supply bypass capacitors are C31 and C32 are are replaced with MLCC ceramics. C20 and C29 are 100nF Mylar capacitors (power bypass) and they have been replaced with modern versions that are smaller (this helps fit the Voice Card into 11 mm of clearance). C33 and C34 are 33uF 16V electrolytics mounted on the back of the PCB, and are very important for a stable power supply. I have uprated the voltage to 25V (to provide a safety margin on the 15V power rails) and used 39uF Panasonic FC power capacitors.

Timing Capacitors The envelope generator timing caps C12 and C17 (220pf), C18, C19, C23, C24 (100pf) were all upgraded to MLCC ceramics, along with C2 in the VCO.

VCO Capacitors I have left the two polystyrene timing capacitors alone as they look in good condition. I have replaced C36 and C37 which are kludges by Roland onto the rear of the PCB, they are old ceramic plate capacitors, which I have replaced with modern MLCC type.

Filter Capacitors C10 is the 10uF coupling capacitor between the VCO and HPF which I have upgraded to a nice audio quality capacitor, and at a higher voltage rating. The ceramic disk HPF 470pF capacitor at C11 was upgraded to a precision polypropylene capacitor, which improves the HPF sound. I have chosen to leave the 150pF VCF capacitors as the original disk ceramics until I conduct audio testing. In the AM8104 I use Polypropylene 1% types.

The coupling capacitors going into the VCF and VCA are quite low value 22nF Mylar film, I use much higher 10uF electrolytics in the 8104 JP04 Filter to improve the bass response. I have swapped them out with Vishay MKP 1% types at 68nF. This opens up the bass response of the filter a little, and Roland started using 56nF on the E revision of the PCB.  The capacitor on the CV summing Op Amp at C28 (22pF) has been replaced with MLCC type.

New Trimmer

Trimmers An ideal replacement for the single turn is the Piher PTC10LH10 which has the same pin spacing, however they are difficult to locate at 100k value. I also need a trimmer which can fit within the 11 mm clearance of my Controller PCB, so a smaller trimmer is needed. I have used the Copal CT6 which need Pin 2 bending out to fit the PCB. They are similar to the type used in the Jupiter 8 and are just 7 mm in height. I will also use this trimmer on my Controller PCB.

The VCO multi turn trimmers are easy to replace with Bourns 3006 types which just drop in. I have used real versions rather than cheap copies.

Project Status By early May 2019 the new power supply PCB for the Moog 60HP case has been prototyped and proven. The Controller and Pot PCB’s have been designed, checked and ordered. The panel has also been designed and you can see a mock up picture at the start of this post. More updates to follow!

 

 

 

Roland S10 OS Modification

Only half a display

Introduction I fitted a new OLED display to my Roland S-10 sampler in February 2019, but the new display only shows the first 8 characters, as the original display is configured as 2×8 lines when it actually shows 1×16 lines. The new display is not a drop in replacement. This problem also affects the Roland Alpha Juno 1 and 2 which uses the same display. The solution is to modify the operating system to change the assembler code that writes the characters onto the LCD.

Processor History Roland started off in the late 1970’s using the Intel 8048 processor with inbuilt mask ROM in the Jupiter 4, Pro-Mars, CSQ sequencers and the 184 keyboard. This had a tiny 1k byte of ROM space, so the programs had to be very small and tight! The Juno 6/60 used the 8049 which has double the ROM size, whilst the Jupiter 8 and MSQ700 use the Z80 and external ROM chips. By the mid 80s’ Roland were using the Hitachi 630X processor family alongside the Intel 8051 family. The S-10/MKS100/S220 samplers all use the same Intel 8032 processor with 64k bytes of external ROM to hold the operating system. The UV EPROM chip used was the MBM27C512.

OS Change Process To correct the display readout I need to modify the code in the Roland S-10 Operating System which is written in 8051 assembler. Fortunately this has been done before to alter the MIDI notes off (in both the S-10 and Alpha Juno), so I am treading a proven path. The OS is extracted from the ROM into a binary file, then disassembled. I can then alter the display code to provide the corrected offset on the LCD, and assemble the code back into a binary. This can then be burnt into a new EPROM and put into the S-10 replacing the old OS chip. I am using a modern Micro Chip 27C512 chip with 70 ns read speed, and they cost £1.50 each.

Roland S10 OS Binary The latest OS version from Roland is 2.07 file, and this ROM has already been extracted and is available here. I used this version as it saved a step in the process and meant I could look at the disassembled code quicker, and see whether the project was possible.

OS 2.07 Features I suspect most S-10’s have the latest OS, but if you have an earlier version this is what you are missing:

  • MIDI system exclusive capability.
  • Fix to fast appregiations dropping notes.
  • Fix for the GATE input silencing voices.

Disassembly I tried using a free Windows program from Spice Logic to disassemble the code, but this did not read the hex file successfully. I then used a Windows 10 cloud based disassembler and this gave me both a reference and disassembled file. Looking through the code I could see what looks like an LCD write subroutine at address 73C3 with two sets of mov instructions spaced 40h apart, which matches the spacing of the original LCD data structure. Here is the code:

Address_73C3:
mov a,#0x80
mov Register_68,a
mov P2,a
mov r0,#0x00
movx @r0,a
inc r0
clr c
lcall Address_3AE0
lcall Address_78ED
mov r0,#0x00
mov a,#0xC0
movx @r0,a
inc r0
lcall Address_3AE0
lcall Address_78ED

In theory all that needs to change is the second mov command to use an address of 0x88, so that all the 2nd line display commends have an offset of 8h in addressing rather than 40h. The code would be well written if all the LCD calls use this sub routine. Register 68 looks like the LCD display register. There are around 200  text messages in the S-10 operating system, including messages for the test routine, and an Easter Egg at address DB7E (SUZUKI GSX-R 400, HONDA INTEGRA GSI). This text is longer than 16 characters and was never displayed!

I have mapped the text messages in the code to the address lines, and then back to the LCD write sub routine at 73C3. I have not mapped them all back but I am sufficiently confident to change the code and burn a new EPROM along with a change in OS version to 2.09, as 2.08 is the unofficial modified version for correcting the all MIDI Notes Off.

Assemble and Burn The change is only two bytes in the whole code, so my initial approach is going to be to alter the binary code at the appropriate addresses rather than re-assembling the disassembled code, as this is more straight forward and the as51 assembler may not read the disassembled code correctly. The first stage is to buy some EPROM chips (Micro Chip 27C512) and burn the 2.08 code into one and check the S-10 works with no modifications by me.  I am using my trusty TOPS 2009 ROM burner, but there are more recent versions that do the same job.

The second stage is to modify the binary code in two places (OS version and LCD offset), then burn a new EPROM and see if it corrects the display writing on ALL 200 messages. If there are any exceptions I will go back to my message address map and see how the display is being called for that message that fails. I will post the results soon!

 

 

Roland S10 Sampler Refurb

Introduction In February 2018 I bought a Roland S10 sampler for £50, which looked in reasonable condition on eBay but with the usual QD drive failure. My plan was to re-use the 4 octave keyboard in another project, as S10’s often have had a hard life! However once I had cleaned up the sampler at home, I found it worked well except for the drive, so I decided to keep it as a sampler in my workshop and restore it back to the best condition I could by replacing the worn out components. The sampler may have been bought and used by J-M-J originally, so it seems a good idea to save it.

This post details the refurbishment work that has been done.

History The 12-bit S10 keyboard sampler was introduced in May 1986 at an initial price of £1099, and sold strongly in 1986/7 with over 8000 manufactured before it was phased out with the arrival of the 12-bit S330 in early 1988. In February 1988 the S10 was was being blown out at £699 and the S10 was one of the cheapest ways of getting into sampling, albeit just 4 seconds and with a really unreliable storage drive. The S10 and its rack mount brothers the MKS100 and S220 have fallen out of favor with musicians and can easily be found secondhand for under £100, they are actually quite usable and immediate.

Old LCD Display

LCD to OLED The 16×1 blue back lit LCD had faded significantly (as expected) with the EL reaching end of life and the inverter whining. Take a look at the picture, hard to use a sampler with a faded display with just 16 characters!

This was the first component to be replaced on the S10. I used a modern OLED display which provides a much clearer display and wide viewing angle. There is no need for a contrast adjustment.

The OLED I fitted is the Vishay O016N001AGPP5N0000, which is the correct overall size of 80mm x 36mm and has a green character colour. I bought it from Mouser for £13. It is an easy fit once the main front panel has been removed, and the LCD holder unscrewed. The inverter cabling to the power supply and inverter are removed, and the LCD cable desoldered from the old display and moved to the new display. The contrast cable wire is not required and should be removed (it did go to Pin 3 on the display).

New OLED

The new display works fine but only displays the first 8 characters, just like the Alpha Juno displays upgraded to 16×1 OLED’s. The solution is to hack the operating system which is fortunately on a separate ROM chip. All the 2nd line display commends have to be revised to use 8h in addressing rather than 40h.

I want to keep the OLED, so a firmware hack has to be done! The OS code can be downloaded as a binary and I now need to reverse engineer it. The firmware for the Alpha Juno has been successfully hacked to use a modern OLED, so I will follow the same route and disassemble the binary into the source assembler code. Then make the change , assemble and create a new ROM.

I will post progress on this and the final ROM code.

Old (L) and New (R)

Panel Switches Some of the button switches were a little unresponsive after 33 years, so I bought a complete new set of 30x from Mouser. This is much cheaper at just £5 than using eBay. The Alps part number you need is: SKHHAMA010.  I clipped off the old switches, desoldered the legs and removed the old solder, then soldered in the new switches making sure they are flat against the PCB’s. I removed and cleaned the PCB’s along with button caps. If you need switches for the MKS100 or S220 they are Alps SKHHBSA010, also available from Mouser.

Panel Buttons Two of the buttons were badly damaged, still usable but I have replaced them with NOS items. A bit expensive but worth it.

Keys A couple of the white keys had minor damage, a visible cigarette burn melted into the top surface and a black spot of corrosive paint to mark the upper split point (why?). I was able to tidy these up with a craft knife and some sandpaper, so I did not need to buy replacement keys. All the keys had to be thoroughly cleaned with soapy water, as they were filthy, I use shower glass cleaner.

Two of the keys did not work, so the whole keyboard has to be removed and the contacts cleaned.

Slider Dust Cover The S10 has a dust cover for the 3 slide potentiometers to the left of the keyboard. The dust cover had perished and looked awful sitting in front of the sliders and visible through the plastic bezel. I cleaned the slider and bender plastic bezel after removing it from the sampler, and removed the bend lever (to clean it) along with the PCB. I could then remove the old dust cover, which is a push fit, and copied the outline onto a new one cut out of 2mm neoprene sheet. Quick refit of the parts and… Perfect and it looks much better!

New Lithium Battery

Battery The 3V lithium battery needed replacing after 30+ years, as on power on the Performance Data was lost, and a message came up on the LCD requesting the Bender range be entered. The battery is on the main PCB but tucked under the keyboard, so that had to come out! The old battery measured 3.29V, whilst the new one measured 2.94V, do not be fooled, a 30 year old battery needs to be replaced! A new battery was bought off eBay and soldered in. The old and new battery soldering can be done (carefully) with the PCB still mounted in the casing.

Power Supply All the large and small electrolytic capacitors on the power supply PCB were replaced with new high performance versions with the same pin out and dimensions.

Disk Drive  The disk drive was in good condition with no motor whining but it failed to load QD disks. I took the drive out and it was immediately obvious that the belt had perished. I ordered a new drive belt from eBay (Germany) and in the mean time removed the old belt goo with isopropyl alcohol and cotton buds. The new belt was fitted and the drive worked perfectly after a few issues reading it settled in and worked.

Disks The S10 sampler came with 16 Quick Disks in various conditions but only the 3 newer disks were readable and the rest were thrown away.

S10 Manager I downloaded this software onto a Windows 7 PC and soon had it working well with the S10 (and S220’s), making sample transfer easy. I spent an evening working through the factory library, the JP8  Brass & Strings is the high light!

 

Roland SH-5 Patch Archive

Patch Sheet

Introduction  In November 2018 I was lucky enough to win a VEMIA auction. Not a vintage analog synthesizer, as they are too expensive, but a set of original Roland SH-5 Patch Sheets. 231 patches for this amazing synthesizer, written by hand on official Roland patch sheets between 1977 and 1983.

Each patch carefully documented and grouped into 13 categories from BASS to ANIMALS. There is unfortunately no record of who created this amazing archive of sound, but I guess they bought their SH-5 in 1977, a year after it was launched in Spring 1976. They would have paid around £900 new (around £5k today!), which was well beyond my means at the time as a university student.

Patch Sheet

Patch Sheet Detail These are printed in monotone with a light grey header featuring the Roland logo on the right and Sample Sound Synthesis to the left in caps. Must be the first time Roland used the word Sample, and the contemporary 100M and 100 patch sheets do not use this name.

The patch sheets are nicely printed on a traditional Japanese Shirokuan paper size of approximately 189 x 262 mm. They are single sided and contain two patches, using a graphical representation of the SH-5 front panel.

Patch Index

Patch Archive I have written up all the patches into a spreadsheet, and reproduced the complete set of patches on newly created SH-5 Patch Sheets (designed in Adobe Illustrator). This will be available as a digital download book, along with blank SH-5 Patch Sheets.

I have also translated the patches (and tried them out!) on an AMSynths SH05 Modular version of the SH-5, and these patches are also available as a digital download book. This means musicians using either the original SH-5 or the new SH05 can explore and enjoy over 200 sounds from a very creative individual.

SH-5 Schematics

SH-5 Service Manual The patches came with an original Roland service manual, printed on thin almost clear paper. A higher quality than the photocopied version on the Internet (which is a Roland scan), with fold out large pages for the circuit diagrams.

It is exactly the same document from 30 October 1976, with hand written Japanese annotations. This original manual was stapled together but the staples have been removed, and there is no front cover, which would have been like the SH-3A service manual cover.

Roland S220 12-bit Sampler

Introduction In October 2018 I bought a 31-year old Roland S220 sampler for £70, with no photo of it working this was a bit of a risky purchase! Why buy one in the first place, given it only manages 4.4 seconds of sample time at 30 kHz and 12-bits? Even the disk drive is horribly unreliable and impossible to replace or even find blank QD diskettes for.

Well its that 12-bit lo-fi sample sound that I am after, so I can sample some modular synth sounds and create bass lines and strings. I also want to develop a Quick Disk drive replacement using a microprocessor, which will store 100 sample banks in memory.

Worn Out 16×2 LCD

History The Roland S220 was introduced in September 1987 as an upgraded 16-voice version of the S10/MKS100 but with the same limited 256k byte RAM. It lasted about a year in production before the improved S330 was launched with 756k byte of RAM, so its a relatively rare sampler. The S220 received a new operating system with a lot of improvements over the S10 including 4-part multi-timbrality and up to 4 audio outputs and a good auto loop. There is an external trigger and arpeggiator, so it plays nicely with modular synthesizers.

My S220 I bought a new mains lead and carefully powered it on, after checking there were no internal issues such as damaged power caps. It booted successfully but the LCD screen was extremely worn out and barely visible, so a new green 16 x 2 OLED display from Raystar Electronics (Part Number: REC001602AGPP5N) was bought for £22. You can buy the OLED in different colours, all part numbers start with REC001602A and they are the correct 80mm x 36mm size.

New OLED

New OLED Display It is a simple swap, the new LCD is the same size as the new OLED, and the EL back light power can be disconnected from the power supply. The front panel has to come out to gain access to the LCD, and the interface wiring has to be desoldered from the old LCD and re-soldered to the new LCD without pin 3 connected (no need for contrast).

This is easy to do as the wiring unplugs from the motherboard. The OLED needs careful alignment with the front panel bezel so all the top and bottom of the characters can be seen. The character size of the new display is slightly larger, notice the tight fit of the lettering.

Roland L-109

Sample Disks I also bought a set of new Roland sample diskettes, the L-109 Synth & Organ, as it has a rather nice Jupiter 8 and VP330 sample bank. This loaded successfully on the second attempt after I cleaned the disk head and felt pad. Amazingly the QD drive works, although it is easy to do sample bank transfers over MIDI using the S10 Manager software.

Second S220 A couple of weeks later I bought a second S220, but this was in much worse condition . The electronics seem okay but the QD motor was permanently on and whining loudly, and the drive belt had melted. I did manage to reduce the whining and improve the head tracking but the Motor will not stop when powered on. So this S220 will be the “mule” for the QD drive replacement project.

Maintenance The lithium batteries were replaced in both samplers, they keep power to the 8k bytes of Performance Data held in SRAM. A symptom of the battery failing is that the Bender range setting needs to be input at power on. Don’t believe the voltage measured on the battery, even if its above 3V replace it. The old battery legs can be cut away and the new one soldered in place without removing the PCB.  The power supplies were refreshed with new electrolytic capacitors, especially the 4700uf and 2x 2200uf reservoirs as they have been working for 30 years, along with a lot of cleaning out of dust from the casing!

QD Drive Replacement  Part of the reason to buy a Roland S220 was to design and produce a QD drive replacement. The QD drive uses a single spiral track recorded onto each side of the diskette, and is half like a cassette and half like a 3.5″ floppy drive. It transfers data slowly at 100 kHz and is encoded in MFM format with 64k bytes each side taking about 5 seconds to load. Samples can be loaded and saved but also the disk can be read to provide the sound name and bank setting.

The drive replacement will use a micro-controller to read and write the data, a small OLED display and a SD card for storage. This project will take place in 2Q19 and it has its own dedicated web page here.

SSI2144 and SSM2044 Comparison

Overview In 2017 Sound Semiconductor launched the SSI2144 voltage controlled low pass filter, a re-engineered SSM2044. At AMSynths we have used both chips in our modules; the SSM chip in the AM8044 (2014) and the SSI chip in the AM8144 (2018). We also make adapter PCB’s that convert the different pin out of the SSI2144 to the SSM2044 pin out, this enables them to be a drop in replacement for synthesizers that use the SSM chip.

The SSI chip has a set of minor but worthwhile improvements and specifically the resonance control has on-chip control, and the Q control current for oscillation has been reduced from 425uA to 400uA. The Q control suggested external circuit is also slightly different to the SSM version.

So lets hear and see a comparison between the two chips with the same input signal at 50Hz from an Emu Systems Modular VCO, as I dial up the resonance from zero to maximum.

Original SSM2044 Here is the SSM chip manufactured in March 1987 notice the longer duration before oscillation comes in.

Re-imagined SSI2144 And now the SSI chip from 30 years later, notice the oscillation comes in quicker. In our modules we use the recommended Q external circuits for each chip, and we trim the onset of  oscillation when setting up the modules for customer shipment. So our modules will sound similar. In polyphonic analog synthesizer with a mix of SSI and SSM chips the resonance trimmers will need to be adjusted when a SSM is replaced with an SSI, or better still replace all the 2044’s.

Boss PH-2 into a Jupiter VCF

Introduction  The Boss PH-2 Phaser pedal contains two valuable Roland IR3109 OTA chips. This post explains how they can be carefully removed and used in the AMSynths AM8109 JP8 VCF or the AM8060 JP6 VCF module. The PH-2 was manufactured for nearly 20 years, from the early 1980’s until at least 2000 by Boss (Roland) and is available second-hand from 20 – 90 GBP depending on condition, try not to pay more than 40 GBP.

Customers following these procedures do so at their own risk and there is no guarantee of success made by AMSynths.

Buy a Boss PH-2 We bought this pedal on eBay in November 2014, it is serial number FN21896 and was produced in March 2000.

Stage 1 – Circuit Board Removal
The first stage is to remove the paxolin PCB from the pedal.

  1. Undo the 4x black posidrive screws in the base of the pedal.
  2. Remove the base plate and clear plastic protector.
  3. Pull the PCB out of the case, it is attached via a grey ribbon cable and 10 wires.
  4. Cut the wires and ribbon cable near the PCB and there you have it – the PCB with two IR3109 chips. Nice!!

Stage 2 – IR3109 Chip Removal

  1. The second stage is to extract the IR3109 chips from the PCB. We prefer to cut the chips out before then carefully desoldering the pins.
  2. Using a coping or fret saw cut around the two chips very carefully, you may want to remove some components by clipping them out with cutters, so that you get straight and safe cut lines.
  3. Now remove each IR3109 attached to the PCB as 2 separate sections.
  4. Mount the small PCB section in a pin vice upside down.
  5. Cut through the back of each PCB very carefully between each set of 4 pins. Make sure you don’t cut down into the IR3109.
  6. Heat the solder pads of the first section and using a pair of fine nose pliers gently ease the PCB section off the 4 pins.
  7. DO NOT OVERHEAT the Pins or IR3109.
  8. Alternatively you can use a de-solder pump and not cut the PCB into sections.

Note: We have found the IR3109 to be a robust chip which is not over sensitive to static or heat. But PLEASE take care! They also seem to last many years (decades in fact), it is rare for us to find one that has failed. Of course it’s worth testing your pedal before you extract the chips, so that you know the chips are working.

Stage 3 – Clean Up the IR3109

  1. Once you have removed each IC from the PCB section….
  2. Using de-solder braid remove excess solder from the front and back of each IC pin.
  3. Straighten up the 16 pins with fine pliers.
  4. Optionally use flux remover to brighten up the pins.
  5. Wash the chip in water if you do this procedure, and allow to dry.
  6. You now have a nice clean IR3109 chip!
  7. We usually put them in IC holders to protect the pins, and to enable them to be quickly fitted to a PCB.

Mistakes do happen!

  1. You may break an IC pin on removing the chip or the pin may break because it is fragile.
  2. Don’t worry, they are easy to repair.
  3. Mount the chip in an IC socket and feed a resistor wire into the top of the socket and align to the damaged pin.
  4. Then carefully solder the wire and pin together.

Jupiter One

Dusty Voice Card

Introduction  I picked up a Roland Jupiter 4 voice card on eBay in late 2017 in good condition. In June 2018 I started a project to build a complete single Jupiter voice based around this card. The synth will fit into a Moog 60 HP case (another eBay find) with an aluminium front panel that replicates (most of) the controls of the Jupiter 4.

To create a fully working single synthesizer voice means a lot more electronics are required than just the voice card! The complex module controller card has to be replicated along with a LFO. The arpeggiator has been left off the synth as it requires the 8048 micro-controller, which means no programmabilty or presets either. The JP4 modulation capabilities have been retained and enhanced, with a 5-axis thumb joystick and a revised S&H circuit for the VCF modulation.

Voice Card This is an early 1979 D version of the card with a BA662 based low pass filter. The first job was to clean and dry the card and remove the metal guides. This reveals 4 mounting holes for the top of the PCB. The 4 green connectors are replaced with 9 pin SIP headers, so the card can be mounted to a new motherboard PCB. Some of the ceramic capacitors needed replacing and the trimmers will be replaced with high quality cermet trimmers.

A hack to the card to add a potentiometer to adjust the VCA Gain trim has been removed and a trimmer put back on the PCB. The card is spaced 12 mm from the motherboard. I am hoping the voice works, but I am sure there will be some repairs needed. The card has additional power supply caps mounted on the rear, which I have retained.

The card is a Revision D but with Revision E kludges (power conditioning capacitors, 100 and 2200 pF capacitors around the uA726). The FET Op Amps in the card are LF353 rather than TL082, but they exhibit the same low current at 3.6 mA and 13 V/us slew speed, so equivalent performance.

Mock Up

Module Controller The Jupiter 4 has a dedicated 5th card in the card frame to hold all the circuits that translate and mix the programmable CV voltages for the Voice Card. It also contains the CMOS based envelope clock generators. All these circuits have been put on the Motherboard PCB of the Jupiter One which measure 295 mm x 105 mm.

The module control circuits and Noise circuit has been retained, augmented with the S&H from the Jupiter 8 and revised LFO. Op amps have been upgraded to TL072’s and BA6110 OTA chips replace the BA662’s. This PCB is spaced 11 mm from the panel PCB and it contains the +/- 15V power rails, derived from a switched and precision LDO power supply.

Jupiter LFO The LFO in the Jupiter 4 is on the main board and is almost identical to the one used in the Jupiter 8. The JP4 version uses discrete FET switching and a two chip CMOS 2 to 4 address decoder, this is rather component intensive and a simpler approach is needed. I have taken the same approach as Roland did on the Jupiter 8 LFO and used a 4052 analog switch.

4-Way Switch The Jupiter 4 relies on 4-way slide switches which are difficult to source, but we think we have the right one. It may mean making sub PCB’s to raise them up to the right level, but its the simplest electronic solution. The JP4 micro-controller decodes them the 4-way into 2-bit binary signals, using pull up resistors, for use in the Controller and Voice cards, which we can copy for the Pulse Width Modulation, LFO Wave Form and VCF Keyboard Follow.

Panel PCB The front panel PCB is 295 mm x 105 mm and contains 19 slide potentiometers, 4 rotary potentiometers, 2 rotary switches, 11 slide switches, the joystick and 4x jack sockets. The PCB mounts to the Motherboard via a set of 9 pin SIP sockets and headers. It is spaced 11 mm from the front panel.

Joy Stick

Joy Stick The Jupiter One has a small thumb sized 5-axis Joy Stick mounted on the front panel. This controls the Bend (X axis) and LFO Depth (Y axis) with the integrated push switch acting as Gate On. This provides excellent control over 3 of the synths key parameters. The Jupiter 4 Modulation controls (next to the keyboard) have been retained so that the Bend Amount and LFO can be patched into the VCO, VCF and VCA.

Inputs & Outputs The external connections have been kept to a practical minimum to conserve panel space, so this is not a semi modular. There are inputs for CV and Gate, as well as the X and Y CV’s that over ride the joystick if plugged in (Bend and LFO Depth).  The 6.35 mm analog output jack is mounted on the rear of the 60 HP case, along with the 12V AC power supply input.

Missing Controls The lack of micro controller means there is no arpeggiator or presets or portamento. The Trigger Section of the synth therefore has not been included (like the Pro Mars). The transpose switch has been replaced by having a 4 octave range switch.

Commercial Option I may release the Jupiter One as a commercial project using new Jupiter 4 voice cards based on the BA6110 and a BJT expo generator to replace the ua726. DIY and completed synths may also be available. A two voice version is planned (same 60 HP case) with programmable presets and OLED display.

Copyright AMSynths 2017