I simply don't have any real experience evaluating SM surge suppressors (under that name anyway).
Before I go, I'll cite from Wikipedia
Experimental results show that most surge energies occur at under 100 Joules, so exceeding the SM design parameters is unlikely, but it provides no contingency should rare events induce energies that exceed it. SM suppressors do present a theoretical fire risk should the absorbed energy exceed design limits of the dielectric material of the components. In practice, surge energy is also limited via arc-over to ground during lightning strikes, leaving a surge remnant that often does not exceed a theoretical maximum (such as 6000 V at 3000 A with a modeled shape of 8 x 20 microsecond waveform specified by IEEE/ANSI C62.41).
SM suppression focuses its protective philosophy on a power supply input, but offers nothing to protect against surges appearing between the input of an SM device and data lines, such as antennae, telephone or LAN connections, or multiple such devices cascaded and linked to the primary devices. In this design philosophy, such events are already protected against by the SM device before the power supply. The limitation of such filter approaches has been examined. SM low-pass filters are generally not suitable for data communications circuits, because they would also block high-speed data signals from getting through.
In comparison to devices relying on components that operate only briefly and do not normally conduct electricity (such as MOVs or GDTs), SM devices tend to be bulkier and heavier than those simpler spike shunting components. The initial costs of SM filters are higher, typically 130 USD and up, but a long service life can be expected if they are used properly. In-field installation costs can be higher, since SM devices are installed in series with the power feed, requiring the feed to be cut and reconnected.
