Guys,
I followed this thread towards the end of last year and being intrigued I went ahead and built this as per spec - only recently been able to test it for calibration (had cloud, rain, cloud, more rain, for what seems months!). So far seems its operation may be a bit unpredictable, as described below, but I need to do more analysis to clarify that!
Not sure if anyone will still be following this thread after all this time, but this is my update so far for anyone who may still be interested ....
Basically, the system worked first time - brilliant. What I have noticed, however, is how sensitive the MLX90614 sensor actually is! I have it mounted on top of my 'cloud sensor' enclosure pointing vertically straight upwards as required, this enclosure is in turn is mounted on the side of a portable box I use to contain power supply, mini pc, router, etc. which I normally place adjacent to the 'scope + mount when I hook everything up.
This evening was clear (the first decent clear night for ages) so decided to set just the box up outside to assess the 'night time, clear sky' readings (no mount/'scope). This was positioned about 3 metres from the rear of my house on patio, with the cloud sensor box located on the side facing towards the house - I tend to do it that way, so the mini-pc would also face the house in that orientation, ensuring good wi-fi reception to my router.
To start with, it seemed the sky reading was somewhat 'warmer' temperature (~ -13 degrees C or so) and, with the ambient reading ~ +8.9 degrees C or so, the Ave. Diff. was thereby wobbling around the cut-off value, causing periodic unsafe conditions to be registered even thought the sky was perfectly clear. By chance, I then turned the box around by 90 degrees so that the sensor was on the 'away from house' side and immediately the sky temperature dropped dramatically to around ~ -20 degrees or lower! This allowed the Ave. Diff. to change quite soon to a higher value, meaning it was more above the 'safe' limit than before ....
In other words, I can only presume that the sensor must have been susceptible to the radiated heat coming from my house (?), and turning it away from the house has allowed it to read the sky more accurately? I don't know, only presuming here ....
In any event, a daytime 'cloudy' measurement I managed to do last week resulted in about the same Ave. Diff. readings as I'm currently reading this evening after I turned the box around! So, I'm not totally sure what this means - is the technology or the theory not as accurate or reliable as it could/should be? Should we therefore expect to take these readings with a pinch of salt, not to be relied upon completely? To be fair, I haven't yet had an opportunity to do some night time, cloudy (and dry!) readings for comparison yet, so I guess that might be the proper comparison test against this evening's results.
If anyone's interested, I'll try to report back here with more info if/when I get it!
Stay safe,
Richard