So these two are not measuring the same thing, which explains the 3x difference. When we commission a new sensor alongside an old one in the plant, we always follow the same path:1. Bring them to the same quantity. Pass the acceleration of the old node through a 10-1000 Hz band-pass filter, integrate it to velocity and take the RMS. With the ADXL345, a 1-second window and 3200 Hz sampling give a rough but comparable RMS in mm/s. Name the new key vel_rms, and keep sending the old vib for a while.2. Mounting. The transmitter is usually fitted with a stud, while the new node is probably clamped or glued. Adhesive mounting loses a significant amount of value at high frequencies. Attach both to the same bearing housing, in the same direction (radial).3. Chart. Line chart, with two sources in Series: one is the ESP32 node’s vel_rms, the other is the transmitter’s. Choose the same Aggregate (avg) for both. Label the Series rows “Node” and “Transmitter”. While the units differ, you can put one series on the right axis with Value axis: Right, but once they share the same unit, a single axis is more honest. Also draw the ISO 10816 limit (around 4.5 mm/s for this pump) as a Threshold line; it helps when talking to the maintenance company.Settings are here: Chart cards. Watch them side by side for a week; if the difference falls within 10-15%, both are correct.