While working on the XTB-II AGC loop, I discovered it can be difficult to deal with nose generated by compact fluorescent bulbs. A photo of CF noise in the X10 passband can be found at:
This testing was done in my workshop, which is on a non-X10 circuit powered through an ArcFault breaker, and has 4 unfiltered CF ceiling lights. The pulsating seen in the waveform is due to beating between the noise generated from the different bulbs. As the noise pulsates, I measured up to 800mVpp, which lighted up two solid bars on the ESM1.
The reason this noise is difficult to deal with is that it does pulsate. So the point that the AGC samples the noise can be near a null, and the noise can rise to a peak in the middle of the X10 reception window. In fact I can see extra "1"s received when that happens. It results in a reported collision, and a rejected message.
While the AGC in the XTB-II works fine for line transients and relatively constant background noise, it does have a problem with noise that pulsates in this manner. I'm not sure there is a way do deal with it, and that is probably why some CF bulbs cause so many problems.
I thought I would share this info with you.