DaveB is looking into it, but DTT isn't ramping down excitations using noise injection. This is something we've been seeing, but usually it just causes saturations. I think it *may* have caused a lockloss a week or so ago, but it definitely killed a lock today.
I attach an ndscope showing that the excitation just suddenly stops, as well as 2 screenshots showing my excitation and measurement settings. You can see in the Measurements tab that I've got a rampdown time set for the excitation.
For now, since these are noise injections, I can use awggui for the injection, which doesn't seem to have these same ramp down problems.
I append these screenshots to FRS ticket 11367.
This could be a repetition of a problem two years ago where the output filter rings for a long time such that at the time the excitation is turned off the output amplitude is still non-zero.
We could only come up with a work-around, using diag to manually control the output gain stage of the awgtpman excitation channel to ramp the output down to zero over a ramp time.
Here is the wiki page describing the work around:
Here are the original alogs from 2016:
I was able to reproduce the problem on a PEM-MY test channel (attached plot)
This current problem is covered by FRS FRS-11367
I think awggui does ramp down correctly (or at least someone should check) so a good option is for people to just use that for their excitations until dtt can be really properly fixed.