28 Dec
2002
28 Dec
'02
6:25 p.m.
Sellam Ismail <foo(a)siconic.com> wrote:
Whether this will work depends on how punch card
encoding works. From
what I can tell so far, there will only ever be one of the 1-9 numbered
rows punched per column, but there can be any combination of the 0, 11,
and 12 rows. Is this correct?
No. Symbol characters do not follow this rule, for example the period
character '.' is encoded as punches in rows 12, 3, and 8.
You should assume that any combination of rows may be punched. As a
general rule, I think you'll find that for cards containing printable
characters, no more than 3 punches are made in any column, but I don't
think you can assume this for cards containing binary data.
Doug Jones has a page that you should look at:
<http://www.cs.uiowa.edu/~jones/cards/codes.html>
...but I can't get there right now, so you might need to look in
Google's cache (long URL w/wrap):
<http://www.google.com/search?q=cache:AeJS4EmmG4gC:www.cs.uiowa.edu/~jones/cards/codes.html>
-Frank McConnell
28 Dec
28 Dec
8:32 p.m.
On 28 Dec 2002, Frank McConnell wrote:
No. Symbol characters do not follow this rule, for
example the period
character '.' is encoded as punches in rows 12, 3, and 8.
You should assume that any combination of rows may be punched. As a
general rule, I think you'll find that for cards containing printable
characters, no more than 3 punches are made in any column, but I don't
think you can assume this for cards containing binary data.
That's what I feared. I'll have to come up with an elegant way to decode
the data.
Doug Jones has a page that you should look at:
<http://www.cs.uiowa.edu/~jones/cards/codes.html>
...but I can't get there right now, so you might need to look in
Google's cache (long URL w/wrap):
<http://www.google.com/search?q=cache:AeJS4EmmG4gC:www.cs.uiowa.edu/~jones/cards/codes.html>
Just what I was looking for! Thanks!
Sellam Ismail Vintage Computer Festival
------------------------------------------------------------------------------
International Man of Intrigue and Danger http://www.vintage.org
* Old computing resources for business and academia at www.VintageTech.com *
8:45 p.m.
On Sat, 28 Dec 2002, Sellam Ismail wrote:
That's what I feared. I'll have to come up
with an elegant way to decode
the data.
I did this once with ASCII paper tape: I made a loop that was triggered
by the 'ready' bit and then sequentially tested each bit in the frame of
7, setting corresponding bits in a register, a kind of stupid software
UART. The problem is to time the loop to be done right as the data bits
are latched, and before the next 'ready' trigger... but the 6522 has
hardware timers, no?
Once you've latched the card column data, you can store it in a
super-frame that corresponds to the card itself - a 'column frame' and a
'unit record' frame.
Then decode at your leisure.
Of course I'm a hardware geek - it is more than likely a far more elegant
scheme or algorithm will be offered by Those Who Know These Things.
Cheers
John
29 Dec
29 Dec
6:08 a.m.
On Sun, 29 Dec 2002, John Lawson wrote:
I did this once with ASCII paper tape: I made a loop
that was triggered
by the 'ready' bit and then sequentially tested each bit in the frame of
7, setting corresponding bits in a register, a kind of stupid software
UART. The problem is to time the loop to be done right as the data bits
are latched, and before the next 'ready' trigger...
The way I'm thinking to do it is to first determine what grouping the
character being read is in. It is either going to be in the 12, 11, or 0
"zone", or in the extended characters, the 8-3 zone. And then there are
the extended IBM character sets (i.e. EBCDIC) where you have 8-x zones
where x can be any number. What these represent is the 8 and x columns
being punched together as a sort of "meta" code (kinda like a control key
on a keyboard).
Then, there will be 1-9 possible characters in each zone. After
determining the zone, finding the correct character will be a matter of a
simple table lookup. The table will contain the ASCII character code of
that character, and then off that goes over the serial port to the
receiving machine (or buffered in some manner).
Doug Jone's article on punch card codes helped a lot. Doug Jones ROCKS!
but the 6522 has hardware timers, no?
Jes.
Once you've latched the card column data, you can
store it in a
super-frame that corresponds to the card itself - a 'column frame' and a
'unit record' frame.
I'll probably collect a whole card's worth of data before sending it out
the serial port. I'll have to see if this will be possible given the
speed of the reader (200 cards a minute). It should be possible to spit
it out in between card feeds at 9600bps:
80 characters x 11 bits = 880 bits / 9600 bps = .092 seconds. Should be
no problem. I'll see what the manual says about timing between card
reads.
...or I might be able to spit out characters as they are read, queueing
them in software if necessary. I think that's how I will go about it.
Sellam Ismail Vintage Computer Festival
------------------------------------------------------------------------------
International Man of Intrigue and Danger http://www.vintage.org
* Old computing resources for business and academia at www.VintageTech.com *
7:43 a.m.
On Sun, 29 Dec 2002, Sellam Ismail wrote:
The way I'm thinking to do it is to first
determine what grouping the
character being read is in. It is either going to be in the 12, 11, or 0
"zone", or in the extended characters, the 8-3 zone. And then there are
the extended IBM character sets (i.e. EBCDIC) where you have 8-x zones
where x can be any number. What these represent is the 8 and x columns
being punched together as a sort of "meta" code (kinda like a control key
on a keyboard).
And I forgot to mention the really funky codes that combine two of the 12,
11, 0 zones and multiple rows to form special characters (like 12-0-1-8-9
for NUL in standard IBM EBCDIC).
Then, there will be 1-9 possible characters in each
zone. After
determining the zone, finding the correct character will be a matter of a
simple table lookup. The table will contain the ASCII character code of
that character, and then off that goes over the serial port to the
receiving machine (or buffered in some manner).
The idea is to create a character set template for each known set that can
be loaded before each batch to be read. The template will include the
exceptions like where multiple rows are punched for meta characters. Then
it will be a simple matter to translate the code to the appropriate ASCII
equivalent.
Sellam Ismail Vintage Computer Festival
------------------------------------------------------------------------------
International Man of Intrigue and Danger http://www.vintage.org
* Old computing resources for business and academia at www.VintageTech.com *
8:06 a.m.
On Sun, 29 Dec 2002, Sellam Ismail wrote:
And I forgot to mention the really funky codes that
combine two of the 12,
11, 0 zones and multiple rows to form special characters (like 12-0-1-8-9
for NUL in standard IBM EBCDIC).
And don't forget the silliness of punching graphic images on cards. I was
showing punch cards to my OS class, including portapunch (right after a
certain Florida election). I wanted to show them some of the variations,
such as round-hole and other sizes (there are some great stories about
some of those!), and the only example of the half sized card that I had
was punched with holes making a picture of a hand giving the finger.
Why not just read two columns, each of which is twelve bits, pass that as
three bytes of raw binary, and then decode in the host machine?
--
Grumpy Ol' Fred cisin(a)xenosoft.com
10:39 a.m.
Fred Cisin (XenoSoft) wrote:
On Sun, 29 Dec 2002, Sellam Ismail wrote:
And I forgot to mention the really funky codes
that combine two of the 12,
11, 0 zones and multiple rows to form special characters (like 12-0-1-8-9
for NUL in standard IBM EBCDIC).
And don't forget the silliness of punching graphic images on cards. I was
showing punch cards to my OS class, including portapunch (right after a
certain Florida election). I wanted to show them some of the variations,
such as round-hole and other sizes (there are some great stories about
some of those!), and the only example of the half sized card that I had
was punched with holes making a picture of a hand giving the finger. Why not just read two columns, each of which is twelve
bits, pass that as
three bytes of raw binary, and then decode in the host machine?
I would read all 80 columns 16 bits each.
Remember often a blank column could be read
as a leading zero. One place to look for card
to ascii? encoding is old data books on mask
programable roms as that was a common use for
them.
Ben.
10:51 a.m.
> Why not just read two columns, each of which is
twelve bits, pass that as
> three bytes of raw binary, and then decode in the host machine?
On Sun, 29 Dec 2002, Ben Franchuk wrote:
I would read all 80 columns
Of course. If there is a data communications bottleneck, trailing blanks
could be inferred, rather than explicitly transmitted from the reader to
the host machine, but ALL columns must be read.
16 bits each.
Why 16 bits? There are 12 physical bits present. (Y,X,0-9)
Or is that just for the convenience of doing it in 16 bit machines?
Remember often a blank column could be read
as a leading zero. One place to look for card
to ascii? encoding is old data books on mask
programable roms as that was a common use for
them.
Decoding is the only part that remains truly trivial (in software). Even
a hardware based decoding could be done fairly easily with a ROM and using
the 12 bits to find the appropriate value in an array of 4096 possible
outputs.
1:33 p.m.
On Sun, 29 Dec 2002, Fred Cisin (XenoSoft) wrote:
And don't forget the silliness of punching graphic
images on cards. I was
showing punch cards to my OS class, including portapunch (right after a
certain Florida election). I wanted to show them some of the variations,
such as round-hole and other sizes (there are some great stories about
some of those!), and the only example of the half sized card that I had
was punched with holes making a picture of a hand giving the finger.
I won't make an effort to read those cards ;)
Why not just read two columns, each of which is twelve
bits, pass that
as three bytes of raw binary, and then decode in the host machine?
That's a pretty good idea. But I kinda prefer programming on the Apple
//e and like the idea of having it do the translation so that it spits
out raw ASCII. Plus it's much funner ;)
Sellam Ismail Vintage Computer Festival
------------------------------------------------------------------------------
International Man of Intrigue and Danger http://www.vintage.org
* Old computing resources for business and academia at www.VintageTech.com *
7826
days inactive
7827
days old
8 comments
5 participants
participants (5)
-
bfranchuk＠jetnet.ab.ca -
cisin＠xenosoft.com -
fmc＠reanimators.org -
foo＠siconic.com -
jpl15＠panix.com