It's being reported[1 - NYT] that the 107 "turn off time" is inaccurately quoted. It wasn't that the system was turned off at 107, as much as the next expected report at 137 was not received. This leaves a 30 minute window during which the system was disabled. Note also that the last voice communication was at 119, so it could have been disabled after that. We simply do not know.

[1] http://www.nytimes.com/2014/03/18/world/asia/malaysia-airlin...

""turn off time" is inaccurately quoted"

That is part of the problem with any analysis of what could have happened.

We don't know for sure if all the input is correct as reported. The old garbage in garbage out.

We don't know if the background checks (some of which are complete some which various countries haven't even released yet) have issues. Everything is speculation based on fuzzy information.

All we know is what is reported and we don't know how much of it is actually correct. Which is what your comment illustrates ""turn off time" is inaccurately quoted".

Much of the information coming from the Malaysian officials tasked with communication is confused or inaccurate. Naturally it's unclear whether this is a case of "right hand not talking to the left" or broader incompetence, but I think you're right to be conservative in building hypotheses atop this information.

The following is speculation:

If there was an electrical fire maybe it disabled the telemetry system before the last radio comm without them knowing? I don't know all the systems on the plane and what the possibility is of an electrical fire going unnoticed for ~12 minutes. That's my only guess for why the transponder was turned off before last radio comm IF we assume this article/post's point of view is correct.

Couldn't a fire in the electric turn off the telemetry minutes before it gets noticed?

There are alarms for everything under the sun, I don't think it would be possible for a fire to knock off a key system and the pilots not know about it.

In the world of Electrical Engineering ANYTHING can be knocked off without you knowing, because at some stage you just HAVE to assume that given circuitry works. With radio comms, the only way to check if antenna is actually working (a connector is still intact) is to use receiver. That's what commercial radio operators do