Automatic Gainsay wrote:But in those three, one of which having been very professionally maintained, none of the 901s held pitch for very long at all, and none of them had scaling that stayed in pitch beyond two octaves.
Several things going on there, one depends on the revision of 901 you are using.
First, those 901's do not tune up like a standard oscillator many are accustom to today. You need to tune two areas of the oscillator. Pitch and scale. First tune them with zero volts going into them, usually a low C note. (Tuning with nothing plugged into the CV inputs.) Much like you would any other oscillator. Next you need to tune the scale or intonation. Several ways to do this. The original 950/951 keyboards had a scale adjust. But using a precise 1 volt per octave output like from a MIDI to CV you would need to send two of those CV's to the 901 driver with one attenuated. A simple patch really. So send it 5 volts straight with the one patch cord. Should be a five octave higher C. If it is a little flat you need to add in a tiny bit of the other attenuated CV to bring it into tune. (a zero to 10 setting of about 0.2 or so) If sending 5 volts the C is a little sharp disconnect the straight CV and patch the attenuated one until in tune. (a setting of about 9.8 or so) The important thing here if they go slightly out of tune due to temperature/humidity changes you first tune the scale as the zero volt low note will rarely be what is going out of tune. It is usually the scale that changes on those. And adjusting the incomming scale will not change the low note it is tuned to.
So once you figure out how to handle the tuning it is not that difficult. Similar to if you tuned an open guitar string to E, and if you have 24 frets finding the high E note is sharp or flat. If you start retuning the machine head you will throw the lower notes out and never properly tune it. You have to set the intonation/scale on those 901's too and have an easy way to adjust/compensate if needed. If I am playing for a while and I find the 901's are slightliy flat or sharp, a slight tweek of the scale attenuator is all that is needed.
The drifting at a slight breeze on a 901 is probably related to the revision. Early 901's up to as late as winter of 1967 had those large germanium "top hat" diodes in the 901 drivers. These would be the one's Carlos used on SOB and many early systems had. I'm sure Bob Moog was aware of the drifting and made changes to the later revision 901's to replace the diodes with a pair of CA3019 doide arrays. An 8 legged can that had several matched diodes inside. This must have increased the stability greatly as the 901's I have here, and probably most made from 1968 on, have these inside. My 1969 901 oscillators while as temperature sensitive as any 70's analog oscillator, won't go out of tune just because someone walks by them, or if there is a breeze. In a studio environment they are rock solid once warmed up 10 minutes, like any other vintage oscillator design.
I do not doubt all those stories of needing to retune those early revision 901's or the breeze being all that was needed to make them go out. My experience with germanum diode matching by measuring the voltage drop is that any slight temp change will make the readings go all over the place. To match a set of those old diodes I needed a constant breeze from a fan across the diodes and had to handle them with pliars. Even hand temperature was enough to make the readings unstable. So, if your 901's are next to impossible to keep in tune, they may be early revision ones. That Norlander system had many of those and they made many of that revision 901 up into late 1967.
Here is the different diodes on the 901 driver schematic. Early individual ones on left and CA3019 on the right.
Long explanation, but there has to be a reason these 901's have the poor reputation of staying in tune. Bob did get it right on the later ones and thanks to musicians like Carlos insisting on better stability, revisions were made.
There is more to the 901 quirkyness too. For instance the tuning is not quite a linear line, but has some bumps in it. So even if both high C and low C are in tune, the middle C could be slightly out. (Or any one note in between) A bump in the tuning range and using the fixed voltage on the 901A driver and retuning the 901B frequency, you can move the bump around if needed. Not a problem, but sometimes those bumps need to be moved around if it falls in the wrong place. Rare but it happens.
The other thing that could limit the range is they will c**p out if you send more than 6.5 or so more volts to them. That is what the "Fixed Control Voltage " on the 901A rotary switch is for. To keep the 901's is their optimum 0 to 6 volt range by offsetting or adding voltage. So for instance, if a five octave keyboard is controlling the 901's with 0 to 5 volts and you are already adding in 3 volts via the 901A driver, about half way up the keyboard the 901's will hit a wall and not get any higher note. This is probably due to the voltage rails on the Moog system of +12/-6 volts. Many of those adder circuits mirror the voltages and it can't go below the -6 volt rail even if the output is then inverted to +6.
That's my take on the 901's.
Get the ones with CA3019's and you should have reasonably stable oscillators. Tune both pitch and scale as the scale may drift. And don't send too much voltage. I can get 6 octaves in tune no problem, 7 octaves where the low bass might not be exactly perfect but acceptable.