Couple of Prophet 5’s

We have a couple of Prophet 5’s here at the moment, one is a Rev 2 and the other a Rev 3.0. They both have quite different problems as to why they came to us. The Rev 2 passes Autotune perfectly well but then a few oscillators detune rapidly afterwards rendering the synth unusable. Checking with some of the world’s experts on the Rev 2 basically ended up with the idea that it really could be anything. Truly anything, from the power supply, the Sample and Hold systems, the Autotune Multiplexers, the Oscillators themselves, the sockets the chips sit in, the timing capacitors, any of the CMOS switches that do the Polymod functions, the Tuning and Scaling Trimmer pots, the list goes on. Also the problems could be a random mixture of all of the above for different voices. The normal run of events is to try and track down a fault by methodically going through the circuitry and hunting down anomalies, but with the Rev 2 there are no obvious measurable anomalies other than a Tuner telling us things are not right, a few millivolts of any leakage or droop is difficult to monitor in a 10 Volt CV system.

Back in the early days of programmable Polysynths, there was a mindset that they had to be designed to almost mimic their non programmable modular cousins, the 0 to 10 Volt control was the norm, with one Volt per octave being the standard. The Rev 2 Prophet 5 was one and the Oberheim OBX was the other, so all the Sample and Hold circuitry sent out 0 to 10 Volt signals by default, and the circuitry around the Oscillators, Filter and VCA’s were designed to match up with the IC Datasheets which were also marketed as a way to make synth modules with a reduced parts count due to these miracle chips.

With remarkable speed, synth designers realised that it wasn’t necessary or desirable to adhere slavishly to the datasheets, and both Dave Smith and Tom Oberheim redesigned their flagship products to make them more reliable and easier to produce. Those products being the Rev 3 Prophet 5 and the OBX-a. Both designers realised that a 0 to 5 Volt DAC was more feasible and reliable, but also made much of the interface circuitry much simplified, and scaling the resistor values around the expensive Curtis IC’s didn’t change the sound, as they were for the most part current controlled.

We are aware of the anomally that all Analogue synth modules are described as VCO’s, VCF’s and VCA’s but the truth is that they have always been current controlled, voltages from the control interface are converted into currents to work the actual active circuitry, currents charge the capacitor for the Oscillator, control or strangulation of current into the filter dictates the Cutoff Frequency, and so it is with the VCA.


The above image is of a Voice indicator we built for the Rev 2, the tuning drifts made the Sequential technique of putting your finger on the Oscillator chips to see which voice was playing completely futile. Why on earth did no Prophet 5 ever get one of these built in, the Jupiter 8 had them, the OBX-a eventually got them, the Korg Polysix had them. The LED’s are low current 1 mA types and the resistors are 10 kOhm so they don’t load the gate signals to the EG’s. It works well and will be left in the synth. Voice 2 is being triggered here, and the LED’s are much brighter than the photograph shows.

We spent some considerable time with this Rev 2 a while ago, endless scoping and Voltage measurement didn’t get us anywhere, so this time we stocked up on lots of spare parts and just started replacing things.

The logical first step if Autotune doesn’t do it’s job when it thinks that it does is to replace the 4 Multiplexers that handle the Oscillator tuning biases (the error correction voltages) and the oscillator frequencies themselves that are fed to the CPU for comparison. These IC’s are MC14051 chips which are not expensive, the gold plated turned pin sockets we put them into are far more expensive. There was a definite improvement after these were replaced, as the Oscillator drift occured after a few minutes as opposed to seconds, and was not as severe. For some strange reason, it was always Oscillator 1 on a few voices that gave trouble. The next port of call were the MC14016 quad analog switches that handle the waveform switching of Osc 1 and also the Sync, a little bird told us that weird behaviour in these chips can cause detuning even though they are not obviously involved in the frequency control system, so we replaced them all in all 5 voices. There is a current internet wisdom that old CMOS multiplexers and Analog switches will degrade over time, and new ones are much better in a number of ways, such as higher speed of response, lower On resistance and better protection against overloads.

Previously we had gone for the obvious culprits if oscillators won’t stay in tune, bad SSM chips, bad sockets, bad capacitors, bad Sample and hold capacitors… All of these possibilities were tested and components replaced with no improvement. Replacing the CMOS was giving us far better results but not perfection yet.

The main issue for us with Prophet 5’s is that they seem to die a little more while you work on them, changing chips and other components (usually one at a time) requires a great many switch off and then subsequent switch on’s, which on a weirdly ill machine may require hundreds of power cycling events, and as soon as you have finally isolated a problem another one appears, as in when switched on the synth is dead, no lights, no autotune, nothing ! So while you were trying to resolve an analog issue, a whole bunch of the CPU section was waiting to commit Hari-Kiri, and as soon as you have succeeded with the first, the latter kicks in, leading to another very time consuming round of Digital signal analysis to find out what chip has died now. Only one P5 has been exempt from this in our experience (and yes Andrew it’s yours!).

On a seriously sick P5, we will by default change all the Tantalum decoupling capacitors on the power systems for low ESR electrolytics, they are on all the PCB’s and there are around 25 of them. Old Tant’s can go short at any time and besides causing one of the Voltage Rails to self protect, which in theory causes no damage (although the synth won’t power up) the endless Internet jury is still out on the possible side effects of bad power sequencing on the exotic SSM or CEM chips on the voice cards, and potentially the multiplexers also. There are Tantalum caps in other parts of the system but as they have resistors in series with them and many are in the Audio path they are very unlikely to go short and they have not been problematic and thus never been touched, the Audio artifacts attributed to Tantalum’s in Audio systems may be a contributing factor to the Prophet sound, so we leave them alone, and if by chance one were to fail we would replace it with another Tantalum capacitor.

On this particular Rev 2, the object of the rebuild ( it was originally meant to be just a repair, if only it was that simple) was just to get the voices back in tune but even as this process was ongoing, Dying Prophet Disease started to rear its head, one of the SSM2050 EG chips died, and swapping this with a known good one confirmed this, and several switch on’s later a second Envelope died although this wasn’t the chip, it was a ceramic decoupling cap surrounding the chip. After the EG’s were sorted and we thought normality had returned to proceedings, now the autotune tuned everything a Semitone high, leading to the possibility of a stuck bit in the DAC on the CPU board, although it could be almost any part of the master Oscillator tuning section on the CPU board. We have the chips and we have the test gear, but any way you slice it this is going to be very time consuming.