I am using BAS116 low leakage diodes, one from the signal to gnd (the signal range is 0 to 500mV and cannot go negative) and one from the signal to a certain positive voltage (which is itself robustly clamped so it cannot get "lifted up").
My plan was to create this positive voltage with a forward-biased
1N4001 or some such i.e. +0.6V.In normal operation both will be reverse biased.
The Q is how much of a reverse voltage does this sort of diode need in order to achieve its low leakage?
The data sheet specs it at 75V which is no good.
The reverse bias of the diode between signal and gnd will obviously range from zero to 500mV.
The reverse bias of the diode between signal and the positive voltage will range from 0.6V to 0.1V.
So both of the protection diodes will always be reverse biased but not by much.
Is that a problem?
If say the BAS116 needs 1V reverse to achieve the low leakage I can do that, by having 2x1N4001 (a +1.2V clamp) and by doing something similar on the bottom clamp.
Let me see if I can draw it
--------- +0.6V
D--------- signal (0 to +500mV)
D--------- GND
I know that Vr affects the junction capacitance of a diode but I don't think you need any big Vr to achieve low leakage.
Obviously the effect of any leakage will depend on the signal source impedance; this could be a thermocouple, or a PTAT sensor (1uA/K) on which a 0.1K accuracy (i.e. 100nA) is desired.
I have not been able to find out the Zout of a typical thermocouple but have noticed that most amplifier designs use a very high-Z input (1M plus).