In the GPX series detectors the gate drive to the input gets is directly driven from the ADG333 via a BC857 directly to the gates, my take on this is that the designer tried to get as much gate saturation as possible lowering the conduction resistance of the fets. To get lower input noise and less gate capacitance to ground it is advisable to cut the track from the emitter of the 857 and add a smd 10 ohm resistor in series. This value can be played with but take it too high and the Fets may get hot due to resistive losses. On small gold tests I gained near 10% extra depth performance on tiny gold of 0.01 gram. Also a reduction in 1/f noise possibly by reducing saturated carrier electron storage but one would think that the long time duration between transitional states would negate this effect. Less capacitance = more initial signal input. As i do not design mosfets it is a guess why the noise performance is better, I just observe the effects and take advantage of the circuit improvements. Maybe optical or galvanic isolation is the next thing to test. So far the modded GPX detectors work much better than the SDC2300 and the GPZ7000 on most gold sizes subject to ground conditions.
