(This was originally posted on the X10 Forum.)
We have a solution that has virtually eliminated the smart meter problem at one home in the Duke Cincinnati service area. That particular smart meter was communicating with its data aggregator at 75KHz and 86KHz. Those strong signals were near enough to the 120KHz passband that they were interfering with X10 communication.
Across-line filters in the home, even deep notch filters tuned to the smart meter frequency, were not sufficient in themselves. One thing observed was that the smart meter apparently transmitted an even stronger signal when an attenuator was placed across the line. And there appeared to be more communication attempts, increasing the likelihood of corrupting X10 signals.
My customer found a low-frequency clamp-on ferrite filter made by Intermark to clamp over the incoming service entrance cables to add impedance to the smart meter signal. Initial testing was done with one filter clamped over each hot lead. Along with a filter to provide a low impedance load to the smart meter signal, a pair of the Intermark filters was the first step in restoring acceptable X10 operation.
Since then we have gone through a series of different internal filters and added more Intermark filters to the service entrance cables. The recommended solution is to use 2 or 3 Intermark filters on each of the service entrance cables, and a single shunt filter across the incoming 240VAC. The X10 XPNR is effective for the shunt filter, but a modified version of the XTB-TSS (Tuned Signal Sucker) performed better, particularly at the 86KHz frequency. Since that is closer to the X10 passband, it is more of a problem.
Best performance was actually obtained with an 86KHz notch filter across the 240VAC. However, tuning of a series-resonant notch is very sensitive to powerline inductance. The configuration of the installation being tested had relatively low inductance between the smart meter and where the filters were being installed, so the detuning from line inductance was not too severe. However, the series-resonant notch filter may not be the best general solution because of its sensitivity to detuning from powerline inductance.
Here is a link to the original discussion on the X10 Forum: http://forums.x10.com/index.php?topic=26056.0
Sorry, but we no longer supply the components required for the fix due to the limited availability of the clamp-on inductors.