Jeff Liebermann wrote:
The only time I've seen an incorrect
display was when I was building a WWVB emulator and spraying garbage
data everywhe
http://www.instructables.com/id/WWVB-radio-time-signal-generator-for-ATTINY45-or-A/?ALLSTEPS
However, I've never seen a random erroneous date or time.

Nice project. Now I think of someone trying to increase its range with a power stage and going way too far...

This reminds me of these projects:
http://www.erikyyy.de/tempest/
http://bellard.org/dvbt/

I guess something similar could be done to generate a DCF77/WWVB signal.

The algorithm could also check for a reasonable deviation against
the current setting.

I don't think so. Once it gets a valid time to display, it turns off
the receiver to save battery power. No need to decode more than one
or maybe two frames.

Oh, I think I didn't explain it right. I meant an additional check for valid data would be to verify that the received time falls within a reasonable window around the current time. If the clock knows that now is 1:00am +/- 2 minutes it makes no sense to receive 7:42am and take that as valid time.

So the algorithm would be: if there was never a sync before or the user set the time manually = belive whatever time is received, in this case decoding of several minutes can be done as an extra check).
Otherwise (there was a valid sync before and clock time was not changed manually) check the received time is within a few minutes of clock time.
That would help save battery since decoding a single minute would be enough for safe daily sync.