Beyond joking and sweating, these sensors (speed of sound) are noticeably affected by heat. With careful mount to fixed distance you could use it as thermometer. 
I definitely see a shelf in the graph when the coffee machine turns onā¦but given the measurement interval at that point is 10 seconds, Iād have expected to see a slope or a curve.
The 20us pulse has definitely improved things. At first it was 100% successfulā¦but itās like proving something doesnāt happen. I do get some logs that it has failed to start measurement. Iām currently fiddling with timeout to see if that improves things. I know this sensor has a significant (at these timescales) delay before the result is returned. With the ātimeout: 8mā it brings the timeout to 46647 Āµs as reported in the logs. That should be longer than the 42ms Iād measured with an oscilloscope so weāll see if that helps or if Iām barking up the wrong tree! It does specifically log that it fails to START measurement but itās always possible itās a collision with the previous result or mis-logged or something. Worth a shot anyway. So far itās logging all goodā¦but as I say, itās waiting for something to not happen 