2nd board of freq shifter finished

[Haible Juergen]

For all those who are interested:

Yesterday I finished board #2 (of 5) of my frequency shifter.
It contains the fixed frequency oscillator (24kHz), sine shaper,
all pass filter for 90 degree shift at 24kHz, and two 4-quadrant
multipliers for the quadrature BFO operation.

A few remarks (don't ask for shematics now - I will release them
when the whole thing is finished):

Temperature stability:
Much more important than on ordinary VCOs. If you want a frequency shift
of 1Hz, you get this by holding one oscillator at 24000 Hz, and a second
one at 24001 Hz - not easy!
So I want the fixed oscillator to have the same circuit as the VCO, so
that remaining temperature dependencies of components cancel out.
(Bode/Moog did this, but the Electronotes version does not.)
As nothing ever will cancel 100%, a drift free design is still
necessary.
I used a 3080 as current reversing switch, and *two* 3080's in cascade
as hysteretic switch (schmitt trigger). The 3080 is a good choice for 
a fast, precise schmitt trigger, as it has an exactly defined output
swing
(unlike an opamp), and is quite fast. But, alas, it has rather low gain,
so the temperature dependance of its input differential pair will change
the thresholds. This is cured by using 2 otas in cascade, and a positive
feedback loop around both of them.
And, of course, the bias currents for all otas are set by precicion
current
sources, not just resistors to vcc or gnd.

Multipliers:
I wanted precise and affordable components. So I used RC4200 chips.
They are 1-quadrant devices, but as they are temperature compensated,
a 4-quadrant multiplier can be made together with a resistor network.
(see Barry Klein's Book, or the 4200 data sheet)
Care must be taken to balance offset currents as well as offset
voltages,
so you have to choose the right resistor values at the offset inputs of
the
4200. (I mention this because it's sub-optimal in Barry's book, but then
again his aim was a simple ringmodulator.)

I have tested the circuit as far as I could at this state of the
project.
It works, but I can't say anything about temperature stabillity yet.
When the 2nd oscillator is built, I can tell if the above was worth it
all ore not.

BTW: This is my first project where replacing a 741 with a TL071
does *not* work !!
The Bode design uses a 741 as buffer in the HF-oscillator. (Yes,
it works fine at 20 or even 24 kHz, if you keep the amplitude low!)
I wanted to use a TL071, but it latches up on power-on when the
input voltage is too low (but still larger than the negative supply).
I tried several things like clamping with diodes etc., but it all had
unpleasant side effects, and so I finally tried a good old MC1741.
Works fine.

JH.