Searching \ for 'Placing blame on IBM (RA4 inversion)' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=placing+blame+ibm
Search entire site for: 'Placing blame on IBM (RA4 inversion)'.

Truncated match.
PICList Thread
'Placing blame on IBM (RA4 inversion)'
1996\12\21@184217 by Marv

flavicon
face
As part of several comments on the PIC 16C84 RA4 pin the following appeared:

 "Now if only IBM had kept all the input pins on the printer ports the same
polarity, I'd be walking on Cloud 9. Wake me up when the ride is over!"

 Do not blame IBM for that one.  That was Centronics' doing... and they are
long since gone.

And later in the fray:

 "... **BUT** some parallel ports, most notably Toshiba laptops, do NOT
support this."

 Blame Toshiba for that one! Try using Interlink (formerly Laplink) with
one of those Turkey's and look like a fool.

 Even a bad standard, is preferable to no standard.

Happy Holidays,
 Marv

1996\12\22@054312 by John Payson

picon face
>
> As part of several comments on the PIC 16C84 RA4 pin the following appeared:
>
>   "... **BUT** some parallel ports, most notably Toshiba laptops, do NOT
> support this."
>
>   Blame Toshiba for that one! Try using Interlink (formerly Laplink) with
> one of those Turkey's and look like a fool.

Do you (or does anyone) have any idea whether that wierdness is only a
problem with Toshiba laptops, or if it is more widespread?  I have no
real qualm with the /Strobe wire not being open-collector (if it's open
collector it will pose a bottleneck when trying to send data quickly to
a printer).  My qualm is that the other control outputs, even though they
are open-collector, can't be read back; a read to the control port will
show the latched value rather than what's actually on the pin!

[btw, even for the strobe wire, they should have done something analagous
to the 87C51 pullup design: make it so that when the pin has a low->high
transition it will be pulled up hard, briefly, and so that it's pulled up
weakly the rest of the time.]

1996\12\22@102657 by Martin J. Maney

flavicon
face
On Sun, 22 Dec 1996, John Payson wrote:
> Do you (or does anyone) have any idea whether that wierdness is only a
> problem with Toshiba laptops, or if it is more widespread?  I have no
> real qualm with the /Strobe wire not being open-collector (if it's open
> collector it will pose a bottleneck when trying to send data quickly to

This proves not to be the case.  That interface was designed in the early
days of TTL, and is designed to have a pull-up that is by modern standards
an awfully small value - I recall 220 ohms or 330 ohms as being common for
plain-old-TTL pullups.  And in fact, the Centronics interface was spec'd
with setup and strobe (minimum) pulse width on the order of a couple
microseconds not for the sake of the logic gates but to allow for a
reasonably long cable.

> [btw, even for the strobe wire, they should have done something analagous
> to the 87C51 pullup design: make it so that when the pin has a low->high
> transition it will be pulled up hard, briefly, and so that it's pulled up
> weakly the rest of the time.]

That would have been interesting, but I'm afraid you're forgetting how old
the design is.  They used an earlier design paradigm's "hard" pull-up
rather than one that hadn't been invented yet.  :-)

1996\12\22@145249 by Bob Blick

flavicon
face
>>   Blame Toshiba for that one! Try using Interlink (formerly Laplink) with
>> one of those Turkey's and look like a fool.
>
>Do you (or does anyone) have any idea whether that wierdness is only a
>problem with Toshiba laptops, or if it is more widespread?  I have no

I think it might be one or two models that gave toshiba such a bad name. I
have a T2400 that has always acted just like a real computer, even with me
bit-banging the printer ports in Basic and assembly.

If you want to start a cursing-fest, mention the Compaq Contura Aero, that
one has some "special" things going on with its parallel port.

Cheers, Bob

1996\12\22@172209 by John Payson

picon face
> On Sun, 22 Dec 1996, John Payson wrote:
> > Do you (or does anyone) have any idea whether that wierdness is only a
> > problem with Toshiba laptops, or if it is more widespread?  I have no
> > real qualm with the /Strobe wire not being open-collector (if it's open
> > collector it will pose a bottleneck when trying to send data quickly to
>
> This proves not to be the case.  That interface was designed in the early
> days of TTL, and is designed to have a pull-up that is by modern standards
> an awfully small value - I recall 220 ohms or 330 ohms as being common for
> plain-old-TTL pullups.  And in fact, the Centronics interface was spec'd
> with setup and strobe (minimum) pulse width on the order of a couple
> microseconds not for the sake of the logic gates but to allow for a
> reasonably long cable.

220 ohms pullup... with 4.4 volts across it (pulling down to 0.6volts) that's
20 mils.  Plausible I guess, but even so a PIC should be able to sink that
with no problem.  On the Toshiba, though, the pullup on that thing was REAL
STIFF--even a PIC output couldn't pull it very far (now THAT's stiff).

> > [btw, even for the strobe wire, they should have done something analagous
> > to the 87C51 pullup design: make it so that when the pin has a low->high
> > transition it will be pulled up hard, briefly, and so that it's pulled up
> > weakly the rest of the time.]
>
> That would have been interesting, but I'm afraid you're forgetting how old
> the design is.  They used an earlier design paradigm's "hard" pull-up
> rather than one that hadn't been invented yet.  :-)

The Toshiba's pullup on the strobe wire is, at least from my perception, even
stiffer than a 220 ohm resistor; as for changing paradigm, there really aren't
any definitive electrical specifications for the old parallel port and design-
ing a circuit so that a rising edge on "strobe" would switch a 50 ohm pullup
for 2us followed by leaving on a 4.7K (or even 330ohm) pullup would have been
quite in line with the specifications.

Oh well...

1996\12\22@231732 by Martin J. Maney

flavicon
face
On Sun, 22 Dec 1996, John Payson wrote:

> 220 ohms pullup... with 4.4 volts across it (pulling down to 0.6volts) that's
> 20 mils.  Plausible I guess, but even so a PIC should be able to sink that
> with no problem.  On the Toshiba, though, the pullup on that thing was REAL
> STIFF--even a PIC output couldn't pull it very far (now THAT's stiff).

Oops - I wasn't talking about the Toshiba, about the quirks of which I am
quite happy not to have to know <grin>; rather I was addressing the
issue of switching speed with a pullup.  For a moderate length of
twisted-pair wiring, which was what the old Centronics spec called for,
the use of pull-up instead of some active current source was not even
close to being the limiting factor.

> The Toshiba's pullup on the strobe wire is, at least from my perception, even
> stiffer than a 220 ohm resistor; as for changing paradigm, there really aren't
> any definitive electrical specifications for the old parallel port and design-
> ing a circuit so that a rising edge on "strobe" would switch a 50 ohm pullup
> for 2us followed by leaving on a 4.7K (or even 330ohm) pullup would have been
> quite in line with the specifications.

There may not be any electrical specifications for the IBM parallel port,
but that's because they were (nominally, at least) designing to the
relatively ancient Centronics interface.  I have no idea if there was ever
a formal standard sanctioned by a national standards body, but there
certainly were published specs for that interface.  Granted that no one
has likely felt the need to publish them in the last fifteen or twenty
years...

And yeah, it was an earlier paradigm.  Think for a moment how much
additional circuitry it takes to add that dynamic pullup when you're
building it out of SSI TTL gates and discrete components.  Back then, if
you needed X mA of pullup you simply designed that in as a static load, as
they did.  It would be unfair to say that they didn't care how much power
it took; rather it was just that this wasn't an unusual expenditure of
power for the time.

Not that I would want to go back to those design rules... pocket
calculators that could heat good-sized rooms and such.  :-(  :-)

1996\12\23@033420 by John Payson

picon face
> On Sun, 22 Dec 1996, John Payson wrote:
> > 220 ohms pullup... with 4.4 volts across it (pulling down to 0.6volts)
that's
> > 20 mils.  Plausible I guess, but even so a PIC should be able to sink that
> > with no problem.  On the Toshiba, though, the pullup on that thing was REAL
> > STIFF--even a PIC output couldn't pull it very far (now THAT's stiff).
>
> Oops - I wasn't talking about the Toshiba, about the quirks of which I am
> quite happy not to have to know <grin>; rather I was addressing the
> issue of switching speed with a pullup.  For a moderate length of
> twisted-pair wiring, which was what the old Centronics spec called for,
> the use of pull-up instead of some active current source was not even
> close to being the limiting factor.

When Centronics designed the interface which would become the standard, not
only was the pull-up speed not the limiting factor, but communications PERIOD
was not a limiting factor.  Simply put--on the printers of the time, a data
transfer rate of 1,000 bytes/second was more than adequate (even a typical
dot matrix printer outputting graphics could only print about 500 columns of
dots per second).  Actually, being limitted to 1,000 bytes/second as some
computers were did pose something of a limitation since many printers of the
time could not simultaneously print and receive data; even a small (2K or so)
buffer between the computer and printer could improve graphics throughput
quite noticeably.

[slight editing of my >> quoted text to improve line lengths]
{Quote hidden}

Certainly there were specifications for the Centronics interface, but since
sending a thousand or so bytes of data down a cable isn't particularly hard
many parameters either were not specified rigidly or else were often imple-
mented contrary to the specification.

> And yeah, it was an earlier paradigm.  Think for a moment how much
> additional circuitry it takes to add that dynamic pullup when you're
> building it out of SSI TTL gates and discrete components.  Back then, if
> you needed X mA of pullup you simply designed that in as a static load, as
> they did.  It would be unfair to say that they didn't care how much power
> it took; rather it was just that this wasn't an unusual expenditure of
> power for the time.

My point was that the Toshiba laptop has substituted for the older paradigm
(put a fairly stiff pullup on the strobe wire) a new paradigm (put a really
incredibly solid active pullup on the strobe wire) which was a departure
from the old one but which may have been desired to improve transfer speeds
(220 ohms is fine for sending 1,000 bytes/second or even 25,000 bytes/second
but may not be good enough when sending 1,000,000 bytes/second).  My point
was simply that it would have been less of a departure from the old paradigm
of they'd made it so that the pullup was mega-stiff only briefly when the
line made a transition and was otherwise only moderately stiff.

> Not that I would want to go back to those design rules... pocket
> calculators that could heat good-sized rooms and such.  :-(  :-)

Heh... though some of the equipment then was pretty impressive.  For example,
there are times when I have a lot of source code to print out and I'm waiting
for the HP LJ/4p's to print (4ppm) when I've wished I had one of those nice
old fashioned chain printers next-door (I don't think I'd want one in _MY_
office...)

1996\12\23@085000 by timetech

flavicon
face
Marv wrote:

>   "... **BUT** some parallel ports, most notably Toshiba laptops, do NOT
> support this."
>
>   Blame Toshiba for that one! Try using Interlink (formerly Laplink) with
> one of those Turkey's and look like a fool.

Hmmm? I did a quick check of my 3600CT & 650CT, and Laplink still works,
as it does on every other Toshiba machine I have access to. In fact, my
only purchase of Laplink is by buying it preinstalled on a Toshiba
portable...

-- Tom Rogers

1996\12\23@085003 by timetech

flavicon
face
The standard of the time was to use open collector drivers for ttl level
signals. The pull up at the far end was typically between 220 and 1000
ohms. There was a world of machines that used this basic topology in
everything from external connections to fast backplane busses. There
were several terminating schemes for bidirectional busses, including
split termination with a 220/330 combination.

The IBM pc design team (the Boca Raton group) apparently misunderstood
the use of open collector logic. They chose totem pole active high drive
as their standard for irq lines, despite the fact that the universal
method was/is to use active low open collector lines. This fact alone
has caused a lot more problems in pc use than the printer port
definition.

-- Tom Rogers

1996\12\23@101014 by Michael Mullen

picon face
I think the choice of the IBM printer port were determined by costs.
I just checked the PC (not XT) reference manual, and the mochrome/
printer adapter card uses a single 7405 for all the device control lines.
These lines come from a latch which is cleared by master reset.
Might have cost another chip to change it.

A more interesting question is why the port was not made bidirectional.
All the gates are there, and if an unused latched output had been
connected to the output enable on the driver -- a no-cost change --
we would have had an 8 bit bidirectional port with control lines.  Slow
(because of the caps on the data lines) but much better than what
we have been stuck with since then.

Mike Mullen

1996\12\23@151028 by Martin J. Maney

flavicon
face
On Mon, 23 Dec 1996, John Payson wrote:

> was simply that it would have been less of a departure from the old paradigm
> of they'd made it so that the pullup was mega-stiff only briefly when the
> line made a transition and was otherwise only moderately stiff.

Somehow out of all that I had gotten the impression you were wishing that
they'd specified that sort of drive 'way back when - no wonder this whole
conversation has been confusing to me!

> Heh... though some of the equipment then was pretty impressive.  For example,
> there are times when I have a lot of source code to print out and I'm waiting
> for the HP LJ/4p's to print (4ppm) when I've wished I had one of those nice
> old fashioned chain printers next-door (I don't think I'd want one in _MY_
> office...)

<shudder>  Next door is too close.  Third building over is about right,
and by then it has to be a pretty large printout for this to be a win
unless you consider the other benefits of getting up and walking away
from the 'puter once in a while.

Hey, Merry Christmas, all!

1996\12\23@201452 by John Payson

picon face
> I think the choice of the IBM printer port were determined by costs.
> I just checked the PC (not XT) reference manual, and the mochrome/
> printer adapter card uses a single 7405 for all the device control lines.
> These lines come from a latch which is cleared by master reset.
> Might have cost another chip to change it.
>
> A more interesting question is why the port was not made bidirectional.
> All the gates are there, and if an unused latched output had been
> connected to the output enable on the driver -- a no-cost change --
> we would have had an 8 bit bidirectional port with control lines.  Slow
> (because of the caps on the data lines) but much better than what
> we have been stuck with since then.

Actually, if the control lines are used bidirectionally the current printer
port design really isn't all that bad; in fact, in some cases it may be
better to use the port as two eight-bit paths and one control line than to
use the data bits bidirectionally.  Unfortunately, the control signals are
not bidirectional on the Toshiba Satellites and I'm a bit concerned about
whether they may have that problem on other machines as well.

More... (looser matching)
- Last day of these posts
- In 1996 , 1997 only
- Today
- New search...