DIY parts

[terry michaels]

> Thanks for taking the time to answer the questions about the gate
> capacitance. I think I have a better handle on what is going on now. 

> I found a good web site for FET info -- belonging to the
> Temic/Siliconix/Vishay people -- and they had the spec sheets I was
> looking for. I was surprised to see they still have the 2N4859 in their
> product list! Looking at the 2N4392 family I noticed that the 2N4391
> seems like it might be an alternative to the 2N4859. It has only a bit
> more resistance, and much less leakage and faster switching. 

Hi Ian

Thanks for the info on the Temic FETs, I'll definately check it out..  I
have their data book here, but I completely forgot about it.

One FET I'm going to investigate is the J310.  It has reasonable on
resistance and gate capacitance.   It is commonly used for RF circuits, and
is available from a number of sources.

> As far as the output impedance of the 3130 ... have you ever thought of
> not using an active integrator? I notice that the MOOG schematics I
> found on the web have the cap just attached between a voltage reference
> and the exponential current source. So the ramp starts at Vref and ramps
> down to zero, where a comparitor fires the discharge circuit.  

I chose the integrator type of design for the following reasons.  First,
the input to the integrator is within a few millivolts of ground potential
at all times.  This allows you to run a ground ring around everything on
the input node: the CA3030 pin 2, one end of the timing cap, the source
lead of the FET, and the collector of the exponential transistor in the
CA3046.  The ground ring eliminates leakage from elsewhere on the PCB into
the integrator, and is highly recommended.  Another advantage is the base
to collector of the exponential transistor is within a few millivolts of 0
volts.  This greatly reduces leakage in the transistor.  All of those
things extend the lower range of the oscillator greatly, with no compromise
to the upper range.  Also the sawtooth ramps up from 0 volts.  The ramp is
perfectly straight. 

The disadvanges of the Moog approach are:  The base to collector voltage of
the exponential transistor is never 0, so transistor leakage current is
much greater.  Also, the transistor has a collector inpedance that is less
than infinity, which means the output current, and the current gain of the
transistor, varies at different points of the sawtooth, which results in a
sawtooth that's not perfectly straight.  It's also tricky to get one end of
the sawtooth to reach zero, so greater level shifting is needed. 

Either design gives good performance, neither is a truly perfect design. 
The choice between the two is ultimately a personal one.  This is true for
nearly anything in an analog synth, there are usually many ways of
designing any one module.

Terry