The Accu-Vote stopped on eight separate occasions with approximately one orI suspect that there are several things at play here. First some questions. Did the ballot just stop dead or did it pull back and forth a few times before stopping with the ballot jam message? Was there anything on the ballot that would explain a jam such as a fold, tear, sticker or such? Was there a stub still attached to the ballot? Was there anything in the reader? (Hard to tell after the fact).
two inches of the letter sized ballot exiting from the back. In each
situation, the LCD read "Counted Ballot Jammed in Reader". The voted
portion of the ballot had not entered the read area.
My first reaction is that there was a mechanical problem that was causing some ballots to jam in a particular manner. The Accu-Vote should have tried return the ballot, detected the jam, and then tried to clear the jam. The jam clear operation involves reversing the direction of the motor several times. Why the AV reported that the ballot was counted is difficult to answer. The code handling jams is tricky due to the limited information available from the sensors.
I have seen a behaviour something like this on ballots with stubs still attached. The ballot passes through but due to a voting or read error, it is rejected. Upon return, the paper-at-output sensor sees through the perforation of the stub and the Accu-Vote determines that we have seen the end of the returned ballot and then the beginning of a counted ballot being drawn back into the reader. It attempts to separate the ballots and then reports the jam. The same situation is possible when mail-in ballots are opened by automatic letter openers that cut the top of the envelope off. If they slice into a ballot fold, they can leave a tiny slit that can cause the AV to decide that we have two ballots. However, for these cases the ballot should have passed completely through the scanner and encountered problems only on return.
The ballot would go in about half way and recirculate a few times before being returned.Again I would need to know more about the ballot before trying to explain what happened.
Greg, if you can examine one or more of the ballots that exhibited the problem, perhaps you'll find something odd. Otherwise, perhaps you could send us some of these ballots. As for possible sensor problems, Ian and Tab may have additional comments.
Guy Lancaster <firstname.lastname@example.org>
Global Election Systems Inc., Victoria, B.C., Canada