We, MIC Metering took on a project last year, measuring the liquid level of septic tanks and establishing wireless data networks. The depth of septic tanks is 1.2 meters, with diameters ranging from 60cm for larger ones to 30cm for smaller ones. The on-site environment is complex, and the liquid level of septic tanks is uneven. At the beginning, we conducted on-site testing of wireless NBiot ultrasonic sensors with single 120KHz piezoelectric ceramic transducer. We divided the testing into two batches, one with 10 units and the other with 15 units. The first batch was tested for 3 months, and the second batch was tested for 2 months. The test results are good, and the data accurately reflects the liquid level height of the septic tank with an error of 1-2 centimeters, without any false alarms or omissions. But later on, due to insufficient customer budget, we switched to a cheaper wireless ultrasonic sensor with dual 40KHz transducers for measurement, and the problem arose.

When we conducted testing experiments in the company, we used a large garbage bin filled with garbage for simulation testing. We randomly selected 10 40KHz NBiot wireless ultrasonic sensors for 3 days of testing, and the data reporting frequency was 5 minutes. The data was true and accurate, with no false alarms or omissions. But after arriving at the scene, the data reporting frequency was changed to every 2 hours, and the data would significantly jump to once every 2-3 days. Considering that the emission angle of the 40KHz dual probe ultrasonic sensor is relatively large (we have made physical adjustments to the angle, reducing it from 60 ° to 25 ° for the first time), there may be diffuse reflection or sound waves hitting the wall of the septic tank, causing data jumping. Therefore, we have made physical structural adjustments to further reduce the emission angle to 15 °. We simulated the height and depth of the septic tank in the company and conducted a simulated garbage bin test, and the data was accurate and stable. However, during on-site testing, there was still a problem of data jumping. Environmental factors such as interference sources or temperature changes can be excluded, as well as issues such as unstable installation, short sampling time, and insufficient filtering processing. Later, due to the inability to solve this problem, we missed out on this project. Until now, this question still lingers in my mind. I would like to share it with everyone and seek guidance. Thank you!