[sdiy] [SDIY]Improving Moog 901A

[Terry Michaels]

Message text written by Magnus Danielson
>> Paul Schreiber's recent modifications for the 921A&B makes me wonder
whether
> anyone has done something similar for the 901A & B? 

Naturally you could do things, but last time I looked carefully aty
the 901 schems I was not overly impressed by the solution. They form
an interesting insight to the early attempts. Otherwise the
recommendation would be to switch to 921's ;)

But still, if we meditated long enought over the schematics, we surely
could come up with ideas of improvements.

Cheers,
Magnus<

Hi Magnus:

Unlike the 921A/921B, which can be improved by swapping in better
components, the drift problems with the 901A/901B is primarily due to the
design concept.  Updating it would require major reworking of the PCB to
accommodate many new components in an entirely new circuit configuration.  

Here are some examples.  The CV summer is an operational amplifier made up
of discrete transistors. The performance of this op amp doesn't even come
close to what is available today with monolithic parts.  You could remove
all of those parts and hard wire in a monolithic opamp, but it would be a
messy type of update, not like just dropping in one new part.

The exponential conversion is done by exploiting the V/I characteristics of
silicon diodes.  Later 901's used monolithic diodes packaged together
(CA3019), the earlier 901's used two diode strings each consisting of 4
silicon rectifiers!   Needless to say, the early version had essentially no
thermal coupling between diodes, and neither version had a tempco resistor
to compensate for the +3300 PPM scale factor change.

The exponential current needed to keep the diode strings in balance was
derived by another discrete opamp, driving a PNP transistor with a resistor
from the emitter to the positive supply, with the current taken from the
collector.  The voltage required to produce the exponential current for the
diode string was the actual voltage sent out to the 901B's in the system.  
In each 901B, that voltage drives a PNP transistor with a resistor from
emitter to the positive supply, creating a current source that
theoretically duplicates the current driving the 901A diode string. 
Because these PNP transistors are physically in different modules, there is
no thermal coupling between them, so you have another major source of
thermal drift.

The best way to do a circuit design upgrade would probably consist of
reworking the 901A to be strictly a control voltage summer (similar to the
921A), and send the summed voltage to the 901B's.  Each 901B would have a
conventional exponential converter using a matched pair of PNP's, they
would produce the exponential current that drives the UJT oscillator core.

Instead of hacking up vintage Moog 901's it is probably better to relegate
them to LFO duties, where their poor frequency accuracy isn't an issue.  

Or you can make really BIG money by selling them on Ebay.

Terry Michaels