Temperature is one of the variables that matters most in hydroponic cultivation, since it shapes how much nutrient dissolves, how much oxygen the solution holds, and how fast roots take up what they need. Most Internet of Things (IoT) hydroponic setups read it with inexpensive digital sensors, because those sensors are cheap enough to deploy at scale. What is far less clear is how well such sensors hold up over weeks of continuous use in the warm, damp air around a nutrient channel. This paper reports two things. We first describe a low-cost monitoring system built around a nutrient film technique (NFT) rig, in which two temperature sensors sit in the nutrient channel and a third measures the surrounding air, while every reading is timestamped, filtered, and stored at the edge and the cloud is used only to visualize and archive. We then turn the redundant sensors into a way of studying drift directly. Over 40 days of hourly logging, that is 960 readings per sensor, the primary sensor pulls away from its neighbor by a mean absolute deviation of about 0.82 degrees Celsius and peaks near 1.47 degrees Celsius, and the gap only becomes obvious after the first ten to fifteen days. The short-term scatter widens along with the bias. We read these numbers as a caution for closed-loop control and, more usefully, as a baseline that later calibration work can be tested against.
Copyrights © 2026