Jupiter One is a ProMars

Prototype PCB’s

Overview Back in 2018 I bought an original Jupiter 4 Voice Card with the plan to turn it into a complete single voice synthesizer in a Moog 60 HP case.  In 2020 I got as far as a prototype (see photo of the Motherboard and Panel PCB’s) but with some problems, which resulted in the project being shelved. In 2022 I realised I could use the voice card to make a ProMars (without patches) by adding a second VCO. In May I redesigned the front panel and controller PCB’s to overcome the issues and to use the new AS662 chip.

There are four PCB’s in the project:

• Original JP-4 Voice Card
• New Controller PCB
• New Panel PCB
• New VCO-2 PCB

Block Diagram

The designs are based on the Jupiter 4/ProMars but simplified as there is no patch storage. This means the VCA’s that control CV’s from stored patches can be removed and replaced with the direct connection of slide potentiometers. The PCB’s are interlinked with 9-pin SIP connectors and 15V DC power goes into the Panel PCB and is then distributed to the other PCB’s. The Panel PCB attaches to the 2mm aluminum front panel with the jack sockets and rotary controls. Only one countersunk mounting is needed on the top right. The panel attaches into a Moog 60HP case with a new internal 15V Meanwell power supply, which runs off an external AC adapter.

Controller PCB In June the PCB’s arrived and I started by building up the new Controller PCB which has the ADSR clocks, LFO and LFO Delay, Noise and HPF controller. The voltage controlled  clock section of the LFO is from the JP-4 with LFO RATE controlled by a +5V slide pot. The wave shapers are from the JP-8, as is the waveform selection using a 4052 demux chip. I have also added the Jupiter 8 S&H circuit, so the Jupiter 4 feature of the S&H going into the filter can be retained without building a second CMOS LFO.

The 4052 has a maximum power supply range of 20V. In the JP-8 this is +13V and -5V, which is a bit inconvenient! I have used +8V and -8V linear regulators which keeps it under the 20V limit, and provides sufficient headroom for the +/-5V LFO waveforms, which are amplified by x2 after the 4052.

The first test was to check the LFO was working, along with the Noise and LFO Delay circuits. These tests were done standalone without any other PCB’s,  and it checks the AS662 circuits are working ok. The six ADSR clocks were also tested to check they were the correct ranges; 10 kHz to 800 kHz for the ATTACK and x3 this for Decay and Release. The LFO needed a few adjustments to adjust the JP-8 timings to match the JP-4.

• LFO Rate increased from 0.05Hz – 40Hz to 0.1Hz – 80Hz
• LFO Delay increased from 4s to 10s

Panel PCB The next build was to populate the Panel PCB, which holds all the pots and switches, a Gate Boost circuit, LFO, VCO and VCF controllers and output stage. The controllers have all been simplified, as there is no patch storage, and an external keyboard will do the pitch bending. The LFO controller is a new design which enables an external MOD CV (e.g. velocity or aftertouch) to control the level of the LFO, which can then be sent onto the VCO, VCF and VCA via a manual Depth potentiometer.

The LFO controller was the first circuit to build and test

The VCO pulse width circuit has been modified to use a continuous slider rather than a 4-position switch, and to add the VCF envelope as a PW modulation  source. The LFO waveform selection is on a rotary switch, and VCF key follow is also a slider rather than a 4 -way switch.

Voice Card Mods The original JP-4 Voice card requires some minor modifications, which is connected via new 9-pin SIP sockets to the Controller PCB. The key follow circuit has been removed and the key follow CV from a front panel slider is patched directly into the VCF via the KF0 signal. The KF1 signal is used to send the VCF envelope output to the front panel PCB, where it is a new source for PW modulation. VCO-2 also has to be connected into the voice card VCF input.

A trace has to be cut and two new wires connect it to VCO-2, so that the VCO-1 signal is sent to the VCO-2 PCB and mixed into the VCO-2 signal by an Op Amp and sent back to the input of the VCF. In previous posts I have described the component improvements I have already made to the JP-4 Voice Card,

VCO-2 PCB In the original ProMars there is a shortened PCB in the card frame for the second VCO, which is identical to the Voice Card VCO but obviously does not have the ADSR’s, VCF or VCA. I have recreated this PCB which mounts to the left hand side of the panel PCB. I populated a prototype in June 2022 and noticed that the VCO has two +15V power rails.

One for the analog side of the VCO and a second (called +15V*) which is decoupled and supplies the digital side of the VCO. I missed this feature out in the AM8400 VCO and there is no clock bleed from the CMOS or 555 timer into the audio. The split power rail was introduced in March 1979 and all ProMars have this implemented.

I will implement the split power rails in the production VCO-2 PCB’s and a SMD SSM2212 as the THD part is obsolete. The decoupling resistor for the +15V* rail  is labelled R121 but there is already a R121 in the original schematic which should have been R12 but Roland messed up the numbering of resistors! It is R120 in the AMSynths PCB.

My design of VCO-2 is slightly different to the AM8400 JPO-4 VCO, where I used precision voltage rails for the expo generator and implemented a HF trimmer. I have kept very close to the original Roland design by replacing only the UA726 with a SSM2210/SSM2212 and R4 is now a 1K Tempco. There is no HF trimmer. It will be interesting to compare the two VCO designs.

Outcomes In June 2022 the prototype VCO-2 PCB was populated, corrected and worked perfectly. There is a typo in the original schematics, R39 (the pull up resistor for the VCO-2 off/on JFET) looks like 1k but its actually 47k. The 1k resistor got very hot and pulled lots of current! The CONTROLLER PCB was fully populated and the PANEL PCB gradually built and each section tested in turn.

By the end of June most of the circuits were working well except for:

• Some minor trace errors needed correcting
• Attack Clock interfering with the LFO speed, needs a PCB trace rerouting
• LFO bleed into VCO via power supply, improved regulation needed
• Envelope clocks bleeding into power supply (not audible), improved regulation needed

A third PCB run is needed to correct the LFO bleed problem and to add LDO’s for each of the main sections (LFO, VCO-1, VCO-2 and ENV GEN clocks).

I also took the opportunity to implement a fix to the LFO bleeding into the voice card output(s) via the LPF modulation signal. These two signals are right next to each other and in the JP-4 the traces run in parallel. I am unlikely to encounter the same amount of problem, but its worth doing. The post VCA 47k resistor at R97 on the voice card is shorted and moved to the input of the audio output summing Op Amp (IC30B). In my PROMARS this is an easy change with a 0R resistor at R97 and a new 47k resistor on the PANEL PCB.