VCO design research results
gstopp at fibermux.com
gstopp at fibermux.com
Wed Sep 13 21:47:02 CEST 1995
Hi all,
As a result of all of the VCO experiments I've been doing lately I'm
going to use my current favorite as a "production" design. I've placed
an order for 20 MAT-03FH Matched PNP Pairs at a cost of $8.25 each. I
will re-sell these (minus the few that I keep) to AH and Synth-DIY
members etc. for what I paid plus postage. Lead time is mid-November.
The tempco resistors are being investigated, and like I said before are
not ultra-critical.
The VCO has gone through a few changes and ended up to be exactly what
is shown in the schematic for VCO Option 2 from the Electronotes
"Preferred Circuits Collection", which is a reprint from EN issue 75,
page 11 (the only difference is the sync stuff that I added). Here are
the specs:
VCO Type: Integrator/Schmidt Trigger Tri-Square
Control Voltage Response: 1 volt per octave
CV Input Impedance: 100k ohms for 1 volt per octave
Frequency Range: 0.1 Hz to 57.7 KHz guaranteed, CV input
sweep = -15v to +15v
Power Consumption: +15v @ 31.1 ma, -15v @ 27.8 ma
Waveform Outputs: +/- 5v Sine
+/- 5v Triangle
+/- 5v Sawtooth (rising ramp)
+/- 5v Square
+/- 5v Pulse
Output Impedance for all: 1k ohms
Linear FM Input Impedance: 560k ohms
Sync Input Type: Soft sync to Hard sync depending on sync
signal input level
PWM Control: Initial Pulse Width/Saw Shape pot, PWM in
Additional Features: Symmetrized Ramp Modulation (SRM) on sawtooth
waveform output, controlled by pulse width
control voltage
Trimpots: Volts per octave trim
Saw Shape trim
Sine Symmetry trim
Initial Frequency trim (not on schematic)
Semiconductors used: (3) TL082
(3) CA3080E
(2) LM748
(1) CA3140E
(1) Analog Devices MAT-03
(1) JFET (PN4391, 2N3819, etc.)
Overall this VCO is as close to my original desgin ideal as I think I'm
going to get. All waveforms are absolutely perfect at all frequencies,
except for the sawtooth which gets a little funnuy-looking over 20 KHz.
This is probably due to the 100pf de-spiking cap in the sawtooth converter
as well as the finite rise time of the square wave which is used as the
control signal for the FET in the converter. I don't consider this a
problem since the sawtooth is fine below 20 KHz.
My VCO is built on a piece of Vector prototyping board p/n 45P80-1, which
is a 0.10" grid of 2-sided pads on a board that's 4.5" X 8". I can fit 2
VCOs on this. The construction technique is my usual lead-clipping and wire
soldered rat's nest. Layout is not too critical but I have to mention that
you must de-flux the board once it's built and let it dry completely. If
you leave the flux on you will have potential tracking problems, asymmetric
waveforms at low frequencies, and the VCO will not go down to the lowest
frequency. To de-flux I used the industrial PCB cleaner spray we use here
at work. Isopropyl will probably work but don't use rubbing alcohol because
it has lanolin in it.
If somebody wants to CAD up a circuit board that would be GREAT - I will
probably try eventually but don't hold your breath.
My thoughts on the other VCO designs -
Most sawtooth-based VCOs use NPN transistors in their exponential
converters and therefore tend to gravitate towards the CA3046 transistor
array for the matched pair on pins 1-5. The other transistors in the array
are frequently used in a chip-heater circuit to eliminate the need for a
tempco resistor in the volts/octave prescaler divider. I have found that
the 3046-based VCO's don't go to the extremely low frequencies, which is
okay for audio VCOs but not for LFO-type modulation.
Most of the 3046-based sawtooth VCOs have quite a few components involved
compared to this tri-square VCO. These designs include the ones used in
the Micromoog, Multimoog, Rouge, Prodigy, and Moog 921. Now large
quantities of passive components is okay for production machines but
really a pain in the butt for non-printed-circuit-board assembly so aside
from the superior performance the tri-square VCO is easier to build to
boot. Notice that the Moog 921 is a 3046-based sawtooth design but is has
about five times the number of passive components as the typical design.
The Chroma 3046-based VCO has a low parts count, but I have not had good
luck with this (probably cockpit error). I'll probably look into this one
more in the future but for now I have settled on what I think is the best
all-around design for my home-built systems.
I will make a generic fax that I can send out, of the schematic and the
related text, so Email me with your fax number if you want one sent to
you.
That's it for now,
- Gene
More information about the Synth-diy
mailing list