|Re: Errors reading with a HP 82153A Barcode Optical Wand |
Message #19 Posted by Luiz C. Vieira (Brazil) on 4 Apr 2012, 12:49 p.m.,
in response to message #1 by aurelio
As we all (many of us...) know, the HP41 system itself comes from a time were memory was kind of a luxury. Although ROM was not that much expensive as RAM, addressing was also tricky, and because the HP41 uses a serial addressing mode, large amount of memory would also take larger amount of time to read/write (sorry if I am teaching priests how to pray...).
Card readers, tape drivers and the wand as well use a, say, heterogeneous, 'variable' media as a place to store information. Although reliable under certain considerations, in all three cases there is no way to make sure the system will provide a constant read/write speed while either storing/retrieving information. Neither the card reader nor the tape driver use step motors, so small yet acceptable speed variations while reading data should be corrected by software, and error-correcting software tends to grow in size faster than they grow in recovery capabilities. Our HP41 system is not big enough for providing larger data recovery routines, so the best ratio between efficiency and system use is supposed to be achieved. And we all know that storing a program in magnetic card or in magnetic tape does not ensure success in any future retrieval.
Storing data coded in bars gives us a far more stable representation WHEN we have a high-quality printing system. I have tested scanned images from bar codes and I always use 400dpi or more when scanning original printout. After that, I reduce colors to two (B&W) when printing so there is no 'in between' gray, shaded area. It is either space or black printing. The wand SW has a limited error correcting algorithm for reading bars mostly based on speed variations, i.e., because we cannot ensure our arm provides constant speed while reading, the 'size' of the bars is not constant (time measured when reading spaces and printed bars), and there are two types of bars. The small error-correcting portion of the reading routine constantly adjusts input data so the best ratio possible is achieved, so if there are both variations in speed while reading AND variations in the size of the printed bars, I guess the error correcting routines will fail.
I have a Brazilian magazine form the 80's with very interesting information about the HP wand system. I'll locate it and find the reference to the original English article and post about it later.