I added a sound decoder to a Farish class 57 locomotive this weekend. And it included a new feature I've not had on the railway before: a "keep alive" capacitor (a 470uF tantalum capacitor to provide some local energy storage if there is a poor track contact).
The inrush current into that capacitor could be substantial. After putting it on the track, we found that we'd get occasional "track shorts" reported by PM42 and worse at the fiddle yard end.
I used Loconet-Checker to change the current thresholds on all the PM42 up one notch (from 3A to 4.5A on two, and from 4.5A to 6A on the others, noting that I have two different build standards with different sets of thresholds). That seems to have solved the problem.