[evlatests] More on 3bit-8bit polarization leakage and stability

[Michael Rupen]

>  There are two significant exceptions here -- ea06 and 
> ea23 both showed interesting differences between the two days.  ea23 might be 
> explained by the change of hardware (sampler board?) made last week.


Mike replaced the data path 0 delay module for ea23-A1 between the two runs,
because it was giving parity errors (suggesting a memory problem) and gave 
some odd variability in some 8-bit tests (see an earlier Rick report).
I would hope this would only improve ea23's performance; and in any event
this should affect only ea23-A1 (C1 does not use this particular delay
module, and A2C2 use a different Station Board entirely).

So far as I know, nothing interesting was done to ea06 between the two
observations.

Do the data show problems on A2C2 as well as A1C1?  Are ea06 and ea23 as
stable as the other antennas, within a given observation (i.e., on
timescales of hours)?

Note that we have a total of 7 antennas currently outfitted with 3-bit
samplers.
   ea06 and ea23 are reported not to be stable over days
   ea07 has very high noise and other oddities
   ea10, ea14, ea18, ea26 are said to be fine (apart from occasional
     complaints about ea10's sensitivity in Vivek's reports)
So 3 of 7 examples have problems of one sort or another.  Not sure how to
interpret that.

              -- Michael