There is a mistake in your last eqn. The inverting state is a unity gain, the gain is -Rf/Rin = -1k/1k = -1. Yet you introduce an Ra factor in your last eqn, where the last stage should only be -v(antilog)Here are the following equations that I computed awhile ago
logarithmic:
Vo1 = -Vtln(V1/Is*R1)
Vo2 = -Vtln(V2/Is*R2)
summing:
Vo' = Vtln(V1*V2/Is^2*R^2)
where R = R1*R2 since these two resistors are equal. I evaluated them as one giving us R^2.
anti-log:
Vo = -Rf(V1*V2/Is*R^2)
inverting:
Vo = -Rf/Ra (-Rf*V1*V2/Is*R^2)