Yes - a fair point. Even a 1k resistor in series with the signal will limit the current to 600uA, assuming the signal is clamped to +5.6V and the ADC has a +5V rail.
Currently I have 1k resistors in there, but they are before the clamps
- to limit the current in the diodes. The reason for this is because the product will have some terminals with +24V on them so some monkey installer might connect this +24V to the ADC input, and I don't want it to blow up.
The 1k resistors would limit the current in the BAS116s to about 20mA which is 400mW dissipation in the 1k resistor. Hmmm, I might make it
2k2. The Zin of the ADC (the 16-bit one) is 100M which is a 22ppm error, bit the Zin is relatively constant and we will be doing factory cal with the correction stored in an EEPROM so to a first order approximation this should be OK.The 22-bit ADC is a different part of the circuit and I can protect it differently, with something more clever. But it looks like even the
2xBAS116 will be fine leakage-wise.Actually there is a bit more detail: Of the four ADC inputs, I am using a clamping rail of +1.2V (two series diodes forward biased) for three of them since the max signal will be about 500mV, and a clamping rail of 2.7V for the fourth input which will see a max signal of 2.2V.
So at the upper end there isn't a problem because I will not be going anywhere near the ADC's VCC rail.
The problem occurs at the bottom end, for negative inputs, but the product won't have any negative voltages which could be mis-connected. So it is just static...