Conventional wisdom is that in a ring modulator — a real ring modulator, one with a diode ring and transformers — diodes with low forward voltage are preferred. Matthew Skala writes (PDF):
in a ring modulator, higher forward voltages mean higher signal levels are necessary and there will be more tendency for the carrier to bleed into the output
and Ken Stone:
Traditionally these diodes should be germanium, though Schottky diodes such as the BAT48 can be used. Even 1N4148 silicon diodes will work, though with greater distortion due to the larger voltage drop.
But is that the truth and the whole truth? I’ve done some LTSpice simulations and I’m not so sure. I started with a simulation file from here and played with the diodes, including using 1N914 (silicon junction, higher Vfwd), 1N5817 (Schottky, lower Vfwd), and a simple diode model,
.model Didl D(Ron=1 Roff=100G Vfwd=0) and varying Vfwd.
Here are outputs using 1N914 with sine wave signal input amplitude ±0.5 V (left, blue) and ±1.5 V (right, red):
You can see increasing the signal amplitude much above ±0.5 V results in distortion, a flattening of the peaks. With the Schottky:
Here even at ±0.5 V the shape is flattened, and at ±1.5 V it is very flattened, with an output amplitude hardly larger than with the smaller input.
Here are results using the above Didl model. I apologize for the fact they’re presented in a different and harder to see format. The cyan line — actually it’s four coincident lines with cyan on top — is the output with ±0.5 V input and Vfwd from 0.01 V to 0.7 V. The other lines are with ±1.5 V input and the same Vfwd range, with magenta being the smallest Vfwd and dark blue the largest.
So here it seems the ±0.5 V signal results in a Vfwd-independent output, while the larger signal does show larger and larger flattening with smaller and smaller Vfwd.
Based on the above, it seems smaller Vfwd gives you more distortion and smaller amplitude! Now, I’m not an expert on any of this and there could be something fundamentally wrong with these simulations. I don’t have any extra transformers so can’t do breadboard tests right now. So don’t take this as definitive debunking of the assertions of people who know more about electronics than I do! But maybe think about the possibility low Vfwd is not unequivocally good.