Chris Whitten wrote:
I did a test to determine how horrible my VCO's were stacking up as far as
base stability (not temperature cycling mind you). I tested mine against
the Blacet, A. Systems RS-95, PCO and an old Steiner-Parker VCO which isn't
the greatest as far as features, but does have the purest sinewaves I've
ever heard in my life.
Mine did fair better than the PCO, A. Sys and Steiner-Parker, and just a
hair below the Blacet (by about 2 hertz at 1k nominal). But this isn't why
I'm bringing this up. While I didn't have the 258 here during that test, I
did measure the one I did have a couple of months before and the 258 had
less drift than the PCO when both were running over an extended period of
time with the exact same test conditions.
The 'you really don't need to read this unless you're spellbound by what I'm
talking about' department:
Without going into full-bore VCO 101 - their core is really nothing more
than as self cycling AR generator. Some have equal attack and decay (thus a
triangle wave) like the Buchla, Wiard, Aries and Plan B does. Others have
all attach and no decay (thus a saw) like the Serge and many many other VCO
do. Stuck in that feedback loop is what's called an expediential converter.
An expo converter forces current into the VCO's core loop which in turn
increases or decreases it's oscillating frequency. That's how VCOs work.
After the core, the signal is fed through a series of waveshapers to derive
the other waveforms, but that's also not important here.
What is important is the expo converter, it's main component being two
transistors which are linked together to form what's called a current mirror
in which a varying voltage going into them is converted to the varying
current that's forced into the core loop. These two trannies are VERY
sensitive to outside temperature fluctuations and this is where the
instability comes from in VCO designs. While there ain't no way to get rid
of it outside of going (ack!) digital, you CAN compensate for it in analog
VCOs by either doing your best to normalize the operating temp these
trannies are exposed to or by introducing temperature compensating circuitry
such as a tempco resistor that (somewhat) counteracts the current drift.
Neither of these two methods are fail safe, which is why Technosaurus was
full of crap when they stated their analog VCOs didn't drift.
Let's discuss the first method. By intentionally heating the two
transistors up they become less effected by outside temperature variations.
This is what Serge and Buchla (and Moog for that matter) did, yet they did
it in different ways. The Buchla 258 and the Minimoog used a matched tranny
pair called a UA726 which is sealed in a metal can and has an internal
heater in it - it's designed into the part. You use an external resistor to
determine the internal operating temperature at the chip inside. The UA726
is designed to be heated up to normalize the current drift The package is
so thermally isolated you can hardly feel the rise in temperature inside if
you touch the metal can. - this is also intentional: if you can't feel the
heat coming out, it stands to reason you can't feel the cold coming in,
either.
Serge used another part called a 3046, which is not self heating. A 3046
consists of a bunch of isolated transistors in a single DIP package, two of
which are configured in exactly the way required to create a current mirror.
Since this part is monolithic (meaning all of the transistors are on a
single chip inside) what Serge and Analogue Systems did is configure the
unused trannies in this package to heat up, thus raising the temp of the
chip inside to attempt to do what the UA726 did by design. The big
difference is however the 3046 is not as well insulated as the 726 is, so
more heat dissipates out of the package which increases it's vulnerability
to the thermal conditions outside.
> You're also assuming the Serge VCO is more stable than a Buchla VCO.Chris speaks the truth here and here's why:
> In my experience (in use), they are just about the same.
I did a test to determine how horrible my VCO's were stacking up as far as
base stability (not temperature cycling mind you). I tested mine against
the Blacet, A. Systems RS-95, PCO and an old Steiner-Parker VCO which isn't
the greatest as far as features, but does have the purest sinewaves I've
ever heard in my life.
Mine did fair better than the PCO, A. Sys and Steiner-Parker, and just a
hair below the Blacet (by about 2 hertz at 1k nominal). But this isn't why
I'm bringing this up. While I didn't have the 258 here during that test, I
did measure the one I did have a couple of months before and the 258 had
less drift than the PCO when both were running over an extended period of
time with the exact same test conditions.
The 'you really don't need to read this unless you're spellbound by what I'm
talking about' department:
Without going into full-bore VCO 101 - their core is really nothing more
than as self cycling AR generator. Some have equal attack and decay (thus a
triangle wave) like the Buchla, Wiard, Aries and Plan B does. Others have
all attach and no decay (thus a saw) like the Serge and many many other VCO
do. Stuck in that feedback loop is what's called an expediential converter.
An expo converter forces current into the VCO's core loop which in turn
increases or decreases it's oscillating frequency. That's how VCOs work.
After the core, the signal is fed through a series of waveshapers to derive
the other waveforms, but that's also not important here.
What is important is the expo converter, it's main component being two
transistors which are linked together to form what's called a current mirror
in which a varying voltage going into them is converted to the varying
current that's forced into the core loop. These two trannies are VERY
sensitive to outside temperature fluctuations and this is where the
instability comes from in VCO designs. While there ain't no way to get rid
of it outside of going (ack!) digital, you CAN compensate for it in analog
VCOs by either doing your best to normalize the operating temp these
trannies are exposed to or by introducing temperature compensating circuitry
such as a tempco resistor that (somewhat) counteracts the current drift.
Neither of these two methods are fail safe, which is why Technosaurus was
full of crap when they stated their analog VCOs didn't drift.
Let's discuss the first method. By intentionally heating the two
transistors up they become less effected by outside temperature variations.
This is what Serge and Buchla (and Moog for that matter) did, yet they did
it in different ways. The Buchla 258 and the Minimoog used a matched tranny
pair called a UA726 which is sealed in a metal can and has an internal
heater in it - it's designed into the part. You use an external resistor to
determine the internal operating temperature at the chip inside. The UA726
is designed to be heated up to normalize the current drift The package is
so thermally isolated you can hardly feel the rise in temperature inside if
you touch the metal can. - this is also intentional: if you can't feel the
heat coming out, it stands to reason you can't feel the cold coming in,
either.
Serge used another part called a 3046, which is not self heating. A 3046
consists of a bunch of isolated transistors in a single DIP package, two of
which are configured in exactly the way required to create a current mirror.
Since this part is monolithic (meaning all of the transistors are on a
single chip inside) what Serge and Analogue Systems did is configure the
unused trannies in this package to heat up, thus raising the temp of the
chip inside to attempt to do what the UA726 did by design. The big
difference is however the 3046 is not as well insulated as the 726 is, so
more heat dissipates out of the package which increases it's vulnerability
to the thermal conditions outside.