|Pin 16 on all of the 4051's should be at 5.6 volts.|
|This is the circuit used on each board (upper and lower) to create 5.6 volts. If the diode drifts, the voltage supply will also change.|
|the upper control board|
|one of many suspect mechanical connections|
This particular unit came to me with a disparity between the envelopes on the upper and lower sets of voice cards: the envelopes on the 4 upper voices were shorter than those of the lower 4. When running through the voices in rotary mode, you could hear 4 distinct sets of voices 1-4 with a short envelope, and then the next 4(5-8) would have a longer envelope.
It seemed unlikely that calibration was the answer to this problem, at least in terms of the individual voice cards, because there was such a clear distinction between them in terms of groups of four.
Through some testing and looking at the schematic, I was able to determine that the CV's for the sets of four voice cards came from the "upper control board" for the upper four voice cards, and from the "lower control board" for the lower 4 voice cards. I was able to compare, for example, the decay CV from the lower control board with the decay CV from the upper control board... and I found a disparity of about .05 volts. Although this may not seem to significant, the range in voltages for that particular parameter seems to be .3 volts, so it is clear that .05 volts could have an audible effect. I found that this disparity was involved in all the CV's, the lower board was consistently .05 volts different from the upper.
This difference in voltage could be caused by a disparity in the load (perhaps a short or failing component in the voice cards), or a disparity in the supply current (coming from the upper and lower control boards). To test for this I switched a CV from the upper board and directed it to the lower board. The discrepancy in voltages remained the same respective to the control boards, and I was thus able to conclude that the problem originated from the upper and lower control boards, which generated the CV, and not from problems with the voice boards.
Whenever there are boards connected to boards with plugs and wires, it is useful to clean all the connections. This was the first step I took, as well as cleaning all the socketed IC's. This actually improved the problem, but a discrepancy was still present.
Using the schematic, I could see that those cv's came through op amps, the gain of which was set with 1% resistors. I tested the resistors and could find nothing outside of spec. I swapped op amps between the upper and lower boards, but the problem did not move. Tracing the CV further upstream, I saw that it passed through some resistors, and there were some .01 caps to ground to filter out any ac that may leak into it. I checked these components and swapped caps out, as a leaky cap or a resistor that was out of spec could change the CV... everything was fine, the problem was not there.
I then traced the CV back to a 4051, (demultiplexer), and there was a disparity evident there as well, on the output pin. Swapping the 4051 had no effect.... but at this point I noticed that the VCC on the 4051's (all of them) was 5.6 volts. This was generated through the use of a diode and resistor between the 15 and 5 volt rails... the diode voltage drop of .6 volts was used to drop the 15 volts to 5.6 volts.
There was a separate diode on the upper and lower control boards, and the voltage drop had differed by .05 volts between the two of them... so in essence the lower control board was sending its 4051s a voltage of 5.55 volts and the upper 5.61.
Swapping the diodes (4148's) and the resistors resulted in a much more even VCC for the 4051's on both boards, and resulted in an almost exact match between the CVs.... but the problem was still slightly evident, although the voltage disparity was now .01 volts as opposed to the previous .05 volts. The human ear is a remarkable instrument however, and it was still audible.
Finally, I ran a few soldered jumper wires between the upper and lower control board vcc's... the +/- 15 volts, the 5, the ground, AND the output of the DAC, which supplied the analog voltage to all the 4051's. These actions bypassed all the relevant mechanical connections between boards, made the voltage discrepancy go into the 1000ths instead of 100ths, and resolved the problem! It was only through a combination of these strategies, however, that the problem was repaired. I think the most significant correction came from replacing those drifted diodes with ones I handpicked to have similar voltage drops. The second most effective strategy was soldering jumper wires between the 15, 5, -15 ground and DAC output rails on both boards, bypassing the mechanical connections.
More information about Keyboard and electronic repair can be found at my website, offbeatelectronics.com