While your at it – can we
get an engineer to take a look at why these memory cards are going bad. We would hardly have an issue in this Florida
recount if not for failed memory cards. -----Original Message----- Take it to the top. Say it needs to
be done. If they want sales to do it we will. -----
Original Message -----
To: support@gesn.com Sent: Sunday, November 12,
2000 3:09 PM Subject: RE:
Central Count Reader Testing Fair
enough. I agree. But my plate is full. I believe that is the
same for everybody else as well. Prioritizing aside (because
everything is a priority), who is going to do it? For years, I've
advocated for a testing department that is structured between development and
support. Perhaps now, these resources will be created. Until
then... -----Original
Message----- All further
testing should be done in McKinney. Now. Not by sales and field
support at customer sites 6 months from now when we're running other
elections. -----
Original Message -----
To: support@gesn.com Sent: Sunday, November 12,
2000 1:09 PM Subject: RE:
Central Count Reader Testing BTW:
I'm not suggesting this as a fix but just as a test. The results could
lead us into a permanent fix. Ian -----Original
Message----- Try
running the AccuVote on battery power but leave the AccuFeed on AC power.
Running the AccuVote on battery power will cause the ballot speed in the reader
to slow. The difference in the speed between the AccuFeed and the
AccuVote should be irrelevant. See if
this decrease in ballot speed gives you lower rejects for these errors. Ian -----Original
Message----- One
additional factor to note on the central count lucid reader issues. I
will be testing this during the next week. There are some ballots that
continue to get "No Ender" or "Calibration", depending on
feed orientation, after being fed 6 or 7 times, but will then finally go
through. I believe what we are seeing here is that these ballots appears
to have cuts that are inside the cut mark at the top and below the timing marks
at the bottom. This shift causes too little time for the AccuVote to
calibrate or to see the first set of marks as its fed through. This means
that we've got several factors combining together. I believe we can work
around these by knowing what to do both procedurally and in the manufacturing
to identify central count units from precinct units. Factors appear to
be: · reader throat
thickness being too tight for folded ballots in many cases · folds themselves
being too thick for throat thickness, requiring smoothing · variability in
some accuVote readers themselves, ie. motor speed, reader sensitivity · cut marks on
ballots affecting "timing" and reading of critical ballot marks -----
Original Message ----- From: Steve
Knecht To: support@gesn.com Sent: Tuesday, November 07,
2000 1:51 AM Subject:
Central Count Reader Testing The following
information may be useful in determining future course of action for central
count. The following test was run. I tried five
Lucid readers without shimming (.007 width of ballot paper) using 50% ballots
that had been "smoothed" and 50% not smoothed. Smoothing means
that someone takes a smooth blunt object and presses the crease made by the
fold. I then shimmed 5 units. This is what I noticed: · A significant
improvement when shimmed, I assume allowing for the timing and the fold
dragging on the way thru the machine, affecting how the machine tracks the
timing marks thru the machine. · The shimmed units
were somewhat variable. That is one machine of the five had consistently
higher number of "no ender marks", "calibration errors"
etc. with or without smoothing. · The
"smoothing" process greatly affects the ballots going thru the units
smoothly. The folks in Santa Barbara have accepted this process and seem
ok with it. The "low staffing" sites, such as Humboldt, bristle
at the thought of having to manually smooth out the fold on each ballot.
But they may not have a choice. Conclusion:
Variability in Lucid readers can be significant. Folds are a definite
contributing factor to thruput. Shimming for thickness of folds helps in
central count. |