Sampler guts

One nice thing I did during the otherwise mostly unmemorable previous year of 2023 was to perform further needed repairs and general maintenance on my ancient and beloved SCI Prophet 2000 sampling keyboard, serial # 00064, as it nears the end of its fourth decade in operation. Over the years I have replaced voltage regulators and other power supply components, along with countless electrolytic and tantalum capacitors that had either already failed or were likely to do so given their age. I have had to replace a dead Curtis chip on an occasion. I have also replaced the brittle grey ribbon cable that served as the wiring harness connecting the digital and analog circuit boards with sturdier, flexible wires connected individually from terminal to terminal, since the original solder joints had begun to crack. So far this has worked astonishingly well, although early on I had to go back over and track down a couple of loose connections.

 

 

This time I discovered I had to replace the little coupling capacitor between the VCF and VCA portions of the Curtis Electromusic 3379 chip for Voice 1. Each one of the eight voices of this 12-bit sampler uses a single CEM3379 for the analog signal processing of that voice channel. A quirk of that chip's design, allowing for maximum design flexibility, is that all the functions on the chip - voltage controlled filter, voltage controlled amplifier and voltage controlled panpot - are available independently with separate inputs and outputs for each function. On the analog board of the Prophet 2000 the VCF outputs are wired directly into the input pins of their corresponding VCAs through small capacitors that "AC-couple" the filter to the input of the VCA. Basically, this means that any direct current (DC) bias present in the output of the filters will not be sent through to the amplifier inputs. A failed capacitor at this location results in a noticeable loss of signal level and perceived fullness for that voice channel compared to others that are working normally.

 

 

After removing the part I suspected, it was immediately obvious that it was indeed a bad capacitor, confirmed by electrical testing. Wary of introducing subtle differences between the sonics of the voice channels, I decided it would be best to replace all 8 coupling capacitors with identical parts at the same time rather than just replace the faulty cap in Voice 1. After all, the remaining 7 were just as ancient and likely to fail before too much longer! I also replaced the remaining electrolytic caps associated with the analog voice channels which, as it turns out, form part of the filter CV bias and scaling circuits. So now the venerable Prophet has a new lease on life in the filter / VCA section and is sounding crisper and punchier than ever!!

 

 

 

As always, there is still room for improvement: I have yet to definitively diagnose and fix a random, very occasional sample playback and / or CV triggering glitch that occurs with certain envelope settings and (usually) many voices being triggered at once. The fact that it seems to happen with some envelope settings and does not appear to be associated with any one voice channel suggests that it might be a bad component in the control voltage (CV) sample-and-hold section, such as the LF356 op-amp at the output of the CV DAC that feeds the demultiplexing circuitry, or an associated component on that part of the board. Alternatively it could be a diode, a data latch or some other part associated with an individual voice channel that only acts up intermittently. The strange thing is that certain patches and settings don't seem to be affected by this glitch at all, whereas with others the error starts to crop up almost immediately. Fortunately the technical manual for this remarkable instrument is extensive and incredibly informative, so with a bit of luck and tenacity I think I'll eventually be able to figure it out!

(PS - sorry for my awful pictures. I promise I'll get a decent camera someday!)