At 09:39 AM 3/25/98 -0600, you wrote:
>David Wollmann <dwollmann_at_ibmhelp.com> wrote:
>>I'm working with
>>Hollerith cards on my system right now with an optical card scanner. Data
>>storage capacity measured by bytes/pound anyone?
>
>Can you tell us more about what you're doing? Is this a commercial
>card reader, or something you built yourself? I was daydreaming...
This is a Chattsworth optical Hollerith card reader designed for reading
test and survey cards. It's specially modified to be able to handle cards
without timing marks, and I suspect the surface behind the read area is
specially coated so the punches are visible to the scanner as something
like "pencil marks." It feeds cards lengthwise from a 150-card hopper with
a rubber roller and an actuated plate that drops the stack on the roller to
feed each card.
It communicates asynch, top speed 9600 bps (ouch) and offers the option to
interpret the punches as standard Hollerith, or output them in a special
2-byte binary format with bit 5 in each byte high for error checking and
control character filtering. The customer told me these were standard
Hollerith, but of course they're not--they're a custom blend of Hollerith
and indicator columns that store survey results. Writing the C to translate
the output has been a lot of fun. Since the xlate table was so large (lots
of holes in it), I created a human-readable text file, then a C program to
convert that to the source for the array.
The machine is slow and not well documented. I'm stuck running it on a DOS
box, and don't have a good async lib for DOS, so I have to read the cards
using ASCII file xfer from an old version of Procomm. Next time around I'm
going to hook the reader up to a Linux box and port the code so I can xlate
on the fly and handle jams more gracefully.
The big problem is, these cards have been in storage somewhere in Spain
since 1964 (close to the water I suspect) and rubber-banded together. Most
of the rubber bands have melted and then petrified, so I have to scrape the
remains from the cards with my handy Opinel blade before I can feed them
(the key slot is very tight). As you can imagine, the cards also have quite
a bit of curl which causes a lot of jams.
I suspect any other cards we may get from other projects will be in just as
bad a condition or worse, so I think we would have been much better off if
we had been able to kludge an old IBM reader. That's probably going to be
my next project--hooking a real reader to an Intel Linux machine.
Does anyone have any CE docs for IBM card readers?
BTW, there is a web site with some good background on Hollerith cards. It
lists several of the character sets and some historical information:
http://www.cs.uiowa.edu/~jones/cards/index.html
There is also a Univac page that provides some good historical information
on punched cards IIRC, but I can't find the shortcut right now.
>- John
>Jefferson Computer Museum <http://www.threedee.com/jcm>
>
>
--
David Wollmann |
dwollmann_at_ibmhelp.com | Support for legacy IBM products.
DST ibmhelp.com Technical Support | Data, document and file conversion for IBM
http://www.ibmhelp.com/ | legacy file and media formats.
Received on Wed Mar 25 1998 - 13:07:27 GMT