The Cellet cellphone charger arrived today, and I tested it on a "private" circuit with a XTB-IIR. This is clearly a very nasty device for X10 communication. I made my measurements at the XTB-IIR input bandpass filter before the active gain stage. That is a convenient place because all 60Hz has been removed. There is some "in band" gain due to the filter, so my numbers are probably about double what would be seen directly on the powerline.
When loaded to about half its rating, I measured bursts of 125KHz exceeding1Vpp immediately after the zero crossing. That is exactly where X10 signals are transmitted. Under load, the bursts are constant. Unloaded, the bursts are random, and can appear at other points in the waveform too. The frequency remains around 125KHz.
The XTB-IIR AGC totally rejected the noise from the unit under load, but occasional "1" bits were detected when the unit was unloaded. I never saw the XTB-IIR respond to the noise as a valid X10 signal. However, it is interesting that the ESM1 gave a solid a "Good" indication when the unit was under load.
I can certainly see how this charger will block X10 communication if the X10 modules do not incorporate AGC. I'll do some more testing, and will add a report on the X10 Troubleshooting page when I get the chance.