Your statement "I still need to prevent greater than 5volts across the pot" is probably far too strong of a constraint. Where's the datasheet for your digital pot? Since you haven't provided one, I'll use this one as an example (first one that I found):
MCP4017 datasheetIf you look at what the
actual requirements are, in the absolute maximum ratings section:
Voltage on VDD with respect to VSS ..... -0.6V to +7.0V
Voltage on SCL, and SDA with respect to VSS..... -0.6V to 12.5V
Voltage on all other pins (A, W, and B) with respect to VSS ..... -0.3V to VDD + 0.3V
Input clamp current (VI < 0, VI > VDD, VI > VPP ON HV pins) ........... ±20 mA
Output clamp current (VO < 0 or VO > VDD) ....................................... ±20 mA
That shows you that not only is a voltage of up to VDD+0.3V (i.e., 5.3V) OK, but it also describes the limits of the included clamp diodes. So you don't need to worry at all about the voltage spec; as long as you keep the shunt currents below 20mA, you'll be fine.
For instance, the 100 ohm resistor I mentioned earlier will provide protection for voltages up to 7V (20mA * 100 ohm + 5V). If you use a 1k resistor, then suddenly that goes up to protection up to 25V. This is all without using any extra diodes.
If you want to use silicon protection diodes (to avoid leaky Schottky diodes), then it's pointless to put them directly in parallel with the dpot pins because the dpot clamp diodes will die before the Silicon diodes even start conducting. If, instead, you place the silicon diodes before the protection resistor, then you can create an extremely robust solution indeed with quite small resistances.
There's basically no configuration in which Schottky diodes make sense, because if you put them directly on the dpot pins, you can't guarantee that the diode will actually start conducting before the clamp diodes in the IC, and if you put them before the resistor, there's no point in them being Schottky because even a normal 0.7V drop diode will work perfectly well and will mess with your signal less.
The point is, this is all simulatable and calculatable using the datasheet figures; there's nothing magical or different about the job of protecting inputs.