Long time lurker, occasional SLF, and someone who knows a bit about electronics here, Prada, commenting to Kulverstukas noted:
To my knowledge there is no battery backup for CVR or FDR. It is a long debated issue. Power gone - recording stops. In most cases it means that for pilots there is nothing much to save anymore.
I think this is not
quite true. Modern flight recorders use solid state memory to store data, and the memory chips and associated gubbins will operate from a power supply of 5V or 3.3V. The supply to the recorders from the plane will be either 28VDC or 115AC, so there is a power conversion unit within the recorder. Once the power input stops, there will be a brief period of time during which the recorder electronics will continue to operate, as there are capacitors in the power supply that store charge. It's unlikely to be several seconds, more likely a second or two tops. It may even be necessary for the device to clean up the end of the recording, the equivalent to closing a file on a computer, a process that is initiated by the failure of the power supply.
The other interesting thing is, as I understand it, that the data to the recorder arrives over a bus, upon which many data values are multiplexed. Thus if the bus is severed, all data values from the bus are lost.
This all means that when examining the recording in this supply and/or data lost scenario, there are a number of possibilities.
If the power fails and the data is still available, then data will be recorded right to the bitter end of the recording. The recorder may even record that it has lost power. Or, if the power is on and the data fails, then that failure should be on the recording with nothing after. Or if both power and data fail simultaneously, then to the recorder that is the same as the first scenario, the recorder is still working for a short time, but there is no data input to record.
Thus even though there is no data that is helpful in understanding the flight parameters at the time of the incident, there is a little bit of info as to how the recorder became separated from its data and power, from which it may be possible to make a judgement call as to how the flight recorders separated from the rest of the plane.
There is an assumption here that the recorder remains physically intact through the above failure modes; if the memory container becomes separated from the electronics container then of course recording just stops.