Have you had any luck deriving the equations?
I think I can see where they made their mistake. Just deriving the
expression for the gain from the Vin- node for example, use the Y-Delta
transformation on the R1,R2,R3 triad. Temporarily ground the Vcc- node and
then the Tee network of R1,R2,R3 will become a Pi network, with a resistor
connected from the Vin- node to ground (which can be ignored because it is
just a load on the voltage source input and has no effect on gain), a
series resistor from Vin- to the - input of the opamp (this is the resistor
which sets gain), and a resistor from the - input of the opamp to ground
(which also will have no effect on the gain from the Vin- input). The
inputs to the + side of the opamp are also grounded for this calculation.
The standard gain expression for an inverting amp configuration with a
feedback resistor, Rf, and an input resistor, Ri, is Av = -Rf/Ri. In this
circuit Rf is R4, and using the Y-Delta transformation, we have
Ri = (R1*R2 + R1*R3 + R2*R3)/R2. This gives a gain from the Vin- node of:
-R2*R4
---------------------
R1*R2 + R1*R3 + R2*R3
Now if you left out the first term in the denominator, you would have:
-R2*R4
-------------
R1*R3 + R2*R3
which could be factored as:
R2 ( R4 )
------- * (- -- )
R1 + R2 ( R3 )
which is what they have, and which is incorrect. A similar error is
present in most of their expressions.