Hooked up the voltmeter to the signal output and found there wasn't very
much interference from hazy sunlight directly on the sensor.
I believe the sensor lens would have to be
aimed directly at the sun,
before any big changes in output would be seen.
Off-angle sun light and direct light from my 1W LED flashlight had very little effect,
when compared to the large changes seen when operating in Bounce mode. (using a white paper reflector).
Notice the IR glow of the LED.. My camera can see IR!!
The no-reflect voltage of 4.5v is a really good TTL '1'. (Even covered, it never when up to 5V. My power supply might be a tad low).
From my white-paper measurements, (simulated frost) it seems like 0.4 to 0.5" from the HX fins would be prefect for detecting frost at TTL levels.
It sure looks like no Analog measurement would be needed.
But, I would use a Schmitt trigger TTL input chip, to provide noise immunity.
10pcs 74HC14 7414 74HC14N Hex Inverter with Schmitt Trigger DIP 14 | eBay
IMHO, water droplets hanging off the lens could cause a false alarm. So, a rain shield would be a good idea.
~~~
If ice forms on the lens pair, it's very likely that IR light will be coupled over
to the P-transistor and give you a defrost alarm.
Due to the direction of air flow, past the sensor and into the HX fins,
Sensor icing could be a problem. The sensor might still be signalling
ice on the HX,
after a good defrost.
Or, maybe the LED will warm up the sensor body and melt off the ice..?.
For sure, Real world testing will be needed to tweak the setup..
The sensor might need it's own little defrost heater (100 ohm 1/4w resistor?).