Jasen Betts wrote in a message to Matt Mc_Carthy:

 RJT>> Checked the battery,  it's a common coin cell,  CR2032,  and 
 RJT>> measures 0.5v! Never saw one get that low...  Pulled one off
 RJT>> another board that was laying around here that I can't use for
 RJT>> other reasons (keyboard connector's in a strange place),  and
 RJT>> checked that -- 2.8v -- but it gives me the same result.

 MM> That 2.8V on the Lithium batteries is just as good as the .5V,
 MM> means _dead_ to get anything less than 3.0V.  That 2.8V reading
 MM> comes from stray bits of materials within the cell, there's _no_
 MM> current associated with the voltage reading
 MM> .........................

 JB> typically a moving coil voltmeter puts a stronger draw on the
 JB> battery than does the cmos. but yeah ISTR them being fussy about
 JB> getting 3.0 motorola seems to think theirs need atleast 3V to
 JB> function but notes they can keep the time with only 2.2

Yeah,  I've seen a number of cases where a CMOS battery will maintain
settings but won't keep the clock going,  too.  That 3.0 is a data sheet
spec,  not necessarily one that always holds in the real world...

---