[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: Report Failure Error on 1.94w

The memory cards (32KB) work fine initially (for printing reports and counting ballots), but at some point (that point in time varies) they fail to allow the printing of reports, displaying the message "REPORT FAILURE / PRESS ANY BUTTON".  Once you get this message, (even if the card previously printed reports) the memory card will not print on any AccuVote again.  It will keep repeating the error message above when you attempt to print zero totals or results totals, in pre-election or in election mode. I got a call from Bill Vanderburg (Adkins).  He has experienced this 7 times in 5 different accounts in Mossouri.  Back in October/November last year and as recent as today. He has experienced this with Absentee memory cards downloaded from VTS 1.92 and with Absentee and Precinct Memory Cards downloaded from GEMS 1.11.9. If you duplicate the card, the error follows the duplicate.
  All of this indicates that the downloaded Accu-Basic program (the .abo file) is getting corrupted.  This data is not covered by a checksum and therefore the corruption is not reported as such.  Whether the corruption is firmware related or not is hard to tell definitively.  As with all memory card errors, I recommend recording:
 - the AV serial number
 - the firmware release
 - the time and date
 - what the AV was doing when the error was detected?
 - did the AV crash or do anything "odd" just before the corruption was detected?
 - if the error occurred during counting, record how the last ballot counted was voted (note that this is not the ballot being read when the corruption was reported)

  A few random corruptions are unfortunately normal and we chalk these up to static, battery problems, electrical noise, dirty connections, and of course the infamous cosmic rays.  If particular AVs are corrupting memory cards, then there could be power supply, noisy motor, or other internal problems and the units should be serviced.  If the corruptions are happening during particular types of operations on random AVs, then it could be a firmware problem which we need to track down.

  So, are these 7 corruptions random?  The Accu-Basic text can represent roughly a quarter of the data on a memory card which would suggest that there should be a few times this number of other corruptions.  If this is the case then we are not likely going to find anything firmware related.

He has saved the most recent failed memory card.  If anybody needs to examine that card please contact me in McKinney.
  Analysis of the memory card(s) may or may not help to diagnose what caused the error.  If you want to send it to me, please let me know what .abo file they downloaded so I can compare it to what's on the affected card.  Otherwise I would be more interested in answers to the questions above.

  Memory card corruptions are a sensitive issue.  I would like to supply error report forms with every AV and have them used to report all errors.  Then we could do statistical analysis and get a better understanding of what is really happening.  Unfortunately we need to keep quiet about these errors so this has not been an option.  It's frustrating for all of us.