Computer Corner by Bill Kibler

Welcome back to TCJ! I say that only because time is moving on and each
issue is now taking some time to happen. Dave would love to move things
faster than he is, but the rate of money coming into TCJ is rather slow.
When I was editor, I would just dig deep into my pocket and find an
extra thousand or so and keep on moving. Of course I never did get that
money back and had to drop out as editor because of it. Dave however has
no pockets to pull from and thus must print when he finally has the
money.

I would love to help out and have a few times, but my own expenses have
grown and I must make up for three years of losses. That of course
explains why I am working on UNIX systems and doing C. I have almost
recovered from the loss of sleep, free time, and money, only to look
back and wish I could have made it work. Our time as the magazine
supporting collectors is still around the corner, but systems like NT
have people considering the old alternatives.

WINNT

Needing to pay for food and such has caused me to do regular programming
on UNIX platforms. We use Win95 and WinNT platforms to run our tools on.
These tools enable us to talk to the HP-Unix platforms and do our
programming tasks. It gives me a chance to compare and understand better
the advantages of simple and complex systems.

To show how much I have been converted to Unix, my fingers now work very
automatically to move the cursor around in VI, the Unix editor. Vi is a
little simple but very straight forward editor. On almost all platforms
every Unix programmer or administrator must know how to use it. Some
even love it, I am not there yet. It is however some of these little
Unix tools that can give you a good feeling about using that platform.
The tools usually work (hear that M/S), a manual is on-line, and when
you usually have a problem, someone has already created a tools for you.
It also explains why so many Unix tools are ending up on PC's.

Now I have moved my own box at work from NT3.51 to NT4.0. I liked 3.51
better, mostly for it's regular windows setup (like 3.1). The new win95
desktop I find a drag, especially all the layers one must go through to
change something. That of course depends on if you are allowed to change
them at all. They have changed many tool names, removed some, replaced
others. I have tried to add a parallel port hard drive to the box
several times now, all without success. It says it is there, works, is
running, but I am unable to get a drive letter.

Now under DOS I use the fdisk program to find out if I had a drive that
was not setup correctly and thus was the reason it wasn't being seen.
Fdisk is no where to be found in NT, nor can I find a way to test for an
unformatted hard drive. It may be there, they just don't want you to
find it without having gone to one of their special thousand dollar
schools.

At this point I say give me Linux or plain old DOS, even CP/M would be
better than this nonsense. But I have one item to top even this mess.

MMX

Well Intel has their new CPU (the MMX or Pentium Pro, which is more
Media Hype than horsepower upgrade) out and got one in the wings as
well. I was reading an article in EETimes that caught my eye about the
new CPU's. The units have power management built in, not to mention
millions of transistors. This means they can idle back and save power
when possible. These units also are run on anywhere from 1.8 to 3.5
volts depending on CPU speed, current needs, and tested performance at a
given voltage. To do that a 4 or 5 bit signal is supplied to tell the
power supply what voltage to select.

Two things came to mind, one the added complexity and a problem talked
about in the article. I felt the complexity is partly to enable a better
yield on the parts, as some chips were being chucked that fell outside
some of the design parameters. However if the voltage is kept a bit low
or hi these once rejected chips will now work just fine, or at least
they think so. I am not sure it's worth the extra hassles and possible
component costs and failures.

Those cost and problems were of course the main concern. You see the
part can drop current usage to around 300 mills. But hit a key and zap
your up to 12 Amps, yup 12 AMPS! That is a big change for any power
supply, not to mention a whole lot of amps for a single CPU chip. I am
not sure what gets me more, the switching from nothing to full power, or
the fact that full power is 12 amps. You need some pretty big traces to
handle those amps, not to mention the BIG spike such a current change
will cause elsewhere in the system. Talk about a few design nightmares.

The Real World

Reading about all these changes makes me feel good about some of my
other projects. I am still doing some 68HC11 work on the side. It is
mostly C and moving along rather smoothly. I got a good laugh out of an
article in a C support magazine. The writer was talking about doing
embedded C++ and how a reduced set was needed. The funny part was how he
started the article. He mentioned how most embedded products are too
limited for C and especially C++.

This of course didn't stop him from continuing on talking about using
C++ for embedded work. It is true a few organizations are using C and
C++ for embedded systems, it is just those embedded products are very
expensive and usually have DOD stamped all over them. I still have
problems with people who use the same term for million dollar projects
and those in which the whole product will cost less then ten dollars. We
need a new list of terms so people can understand there is a big
difference between these two extremes of the concept.

To me embedded systems generally have a cost of well under $100 and are
not a major design concern or cost. Industrial real time control systems
on the other hand are typically a major part of the cost. Maybe another
idea is one that Dave keeps using, it's embedded if one person can do
the design work, while it is an industrial controller if it takes a full
staff to design and support the product. What terms might you think are
best to separate these two very different fields of work. Send me an
e-mail and I will list the ideas next time.

Product Failure

For those who wonder what has happened with our own CDROM project, I can
say it is moving along slowly. I now have up and running a CCS and a
GodBout S-100 system with 8 inch disks. Between the two I hope to start
going back through all my old disks and some recently acquired disks
looking for items of interests. My main interest is source to BIOS and
utilities. So far I have a good collection, but few have been moved to
my hard drive for cataloging.

Over the years I have heard many stories about users turning on their
old systems only to get nothing. Well it finally happened to me the
other day. I hadn't used several units in about three or four months and
moved them into position to help with the disk cataloging. I tried three
of the Z100's that had worked earlier and one was still connected to the
8 inch drives I was also going to use for copying. Well you can probably
guess that not a one of the units would boot up. A couple had video
problems, one just wouldn't boot no matter what. I finally gave up for
now and dug out an older terminal and used that to test the S-100
systems with.

My idea had been to use a Z-100 for a terminal with the occasional copy
when one of the S-100 systems wouldn't read the 8" disk. The Z-100 also
can copy from CP/M disk to MSDOS disk using one of their own utilities.
Later checking showed the problem to be water damage that eaten away
some of the traces and messed up one of the sockets. Where the water
came from, I have no idea, but what a mess. I now have a better
understanding of the frustration so many of you have told me about.
However don't think using PC's will keep it from happening.

I got a panic call from an old friend who has some knowledge of
computers. His major skill set is two-way radio which he did till he
retired recently. He has played with computers some, but does not yet
have the full grasp of what can go wrong. I have tried to help him
resolve problems with windows which only frustrated him more as we often
hit one wall or another.

This problem had to do with adding a second printer, in this case one of
the newer Epson. When attempting to send data to it, he got back a paper
out message, no matter what. My friend had gotten a second parallel port
card for his PC. When trying that card, nothing work, and thus a phone
call to me. I re-checked all he did and drug the printer home where it
tested just fine.

I went back the next day and tried a few more things. First I checked
the new parallel card and decided all the jumpers needed to be opposite
from what the tech at the store told him. I used MSD and  had a
different, or OK status report on the two printer ports. Now that we
knew the PC tech was wrong and one problems was conflicting port
addresses, we tried the printers again.

After a few minor tests, all started to work properly. The problem it
seems starts with his old parallel port which gives a paper out error on
the Epson no matter what, a mis-configured second board, and not enough
experience to know that most computer store techs have trouble using the
correct end of a screwdriver. I guess at this point I need to do my
normal complaint and say how bad it is and all, but I did that far too
many times already, so I will just let it pass.

Till Next Time

Speaking of passing, I just ran out of space, so hope your trials and
tribulations are minor ones, and keep hacking.

Bill.