I'm using the MSP430 SD16_A 16 bit Sigma-Delta ADC (page 843) for bipolar differential ADC measurements in a +/- 0.6V range. While the inputs can handle +3.3V, on the negative side the absolute maximum input voltage is -1V. The input seems to be a switched capacitor input with an input impedance of about 200Kohm. The driving opamp, an AD8475, though, can produce +/-4V outputs. I'd like to protect the ADC inputs from overvoltage. I'm considering :
(a) Regular 0.7V forward drop Silicon diodes between each input and Ground (in both directions, or 4 diodes per ADC input pair)
(b) TVS Suppression diodes between each input and ground, though these seem to hard to come by at low voltages.
(c) ESD Suppression Diode structures on each input line, though there isn't a +/-1V rail to clip the signals to.
(c) Some form of 1.8V Zener Diode (a pair in anti-parallel) between A+ and A- to prevent the inputs from going beyond that range. 1.8V seems to be the lowest Zener voltage available. I'm not sure if the AD8475 output common mode voltage will be able to follow the 0V given to it at VOCM when it's outputs are being forced by a Zener Diode, in which case ensuring that the negative voltage remains less than 0.9V becomes difficult. Also, the Zener forward conduction is likely to make this unviable in any case.
Which of these is likely to cause the least distortion to the signal? I expect diode capacitance is likely to mess with bandwidth, but is this likely to ve a problem at, say, 10-100KHz signal bandwidth (but with 1 MSPS 1 bit Sigma-Delta conversions). I'm planning on having and RC anti-aliasing filter in front of this as well. Are there any 'standard' methods to protect ADC inputs of such a low voltage, that preferably don't require the protection voltage to equal the ADC supply rails?