Re: Restoring a NorthStar Horizon, problems with SRAM board

On Jun 1, 8:28 am, Dave.Dunfi...@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
(Dave Dunfield) wrote:
I am trying to determine if the computer is reliable or not. Do you
have any suggestions on what I can run from a NS DOS 5.1DQ disk to
verify the computer is working reliably? I have run various monitor
programs and BASIC and all seem to work as far as I can tell. The
system stays up for extended periods but I do not have any way to
really gauge whether it is working well or not. Also, I have many
disks which came with it but I do not know which ones are blank or
not. I can do LI commands from the DOS prompt but that does not work
on all disks. I think these disks are a collage of many systems disks
including a NorthStar Horizon and I am not sure what is in there.

The N* monitor has a memory test command - One version of the monitor
can be configured to run stand-alone - use that one, and test the blocks
of RAM above and below the block occupied by the monitor for an
extended period of time (each) - This will give you a pretty good indication
of the RAM.


Hi Dave and Herb,

I have run the Test Memory program from the monitor. So, it has
succeeded although I had to break it up into 4K blocks since it went
rather slow. Also I had to do it from various monitors since I kept
stomping on my memory map. (ie, you can't run memory test on $0000-
$1000 if you use M0000, etc. Also, NS DOS keeps a lot of its IO
routines in the $2000 region which gets stomped on too. M5700 comes
in very handy for those last bits of memory. I did run it for all the
4K memory blocks from $0000 through $DFFF which is the extent of my
SRAM memory.

There is also a DT utility on disk - you will need a scratch/blank disk,
and it will test the disks read/write - of both of these tests pass, your
system is in pretty good shape.


I ran the Disk Test program twice for each drive and they both passed
with no problems. So I am feeling pretty good about the hardware.

Although I had to lift the machine up for some reason and then it
would not boot the next time. So I removed all the S-100 cards and
reseated them and the problem went away.

The disks that do not accept 'LI' - Have you tried both UPPER and
lower case? - N* requires upper case, and most I/O routines do the
translation, but some do not - It's also remarkably easy to corrupt
the command table entry for "LI" before saving a customized version
of DOS - do any of the other commands work on these disks?

If you get no response at all, it could be that the I/O routines are
customized for a different input device...


This problem seems to have resolved itself. I think the drives were
just depressed from being ignored for 25 years and now they know
someone loves them again they are coming back to life. :-)

I guess it was just creakiness. I did clean the drive head rails and
used a little sewing machine oil on them. Should I lubricate the head
positioning disk as well? What other things should be lubricated?

All the monitors on the NorthStar boot disk have been customized to
use the Flashwriter for video output and the parallel keyboard for
input. All except the M5700 monitor which uses the first serial port
for IO. After I review NST.ASM, which appears fully correct for the
default case my machine is, I assemble the NST program using ASMZ80
per instructions and that goes fine. Then I launch the NST program
with default parameters "NST UH NST" which also goes fine.

Just keep track of the memory location used by the monitor and
the NST program - this is covered in my information. Depending
on which N* monitor you use, you may have to move the data
transfer buffer.


Dave, your HORIZON simulator, NST, and NSI are just about the most
fantastically wonderful programs I have ever used. Seriously, your
software is near miraculous. Without your software for easing the
transition, restoring the NorthStar Horizon would be a LOT harder. I
am also able to make backups of the disks which came with the computer
and independently view them offline using the simulator. Very cool.
I am very impressed.

I am running full screen on my laptop and when I start the M5700
monitor on the NorthStar, I see "MONITOR 5.1" come up on the laptop
screen and a ">" prompt. I can type return and simple commands to the
monitor and it seems to work.
However, here is where the problem occurs: When I type DS 2D00 at the
">" prompt per instructions, I get the prompt for the first value and
I press F1 to send the NST file. Then things start going wrong
because some of the characters are just dropped and the NST file is
never completely uploaded.
Some hex numbers come across fine, others only a single character,
some none at all. Usually, it gets several lines down and stalls out
the entry program. Pressing return at prompt regains control and
returns me to a ">" prompt.

I expect the problem is that you have modified I/O drivers in the
Horizon monitor.


My NorthStar Horizon has a video card and parallel keyboard so all the
IO routines are customized. However, it does boot and function with
the "default" NSDOS50D.NSI boot disk image using the standard terminal
settings on the first serial port.

I did have a minor suggestion for the NST program which may help its
robustness when dealing with old floppies with occasional sector
errors. Make two NST programs. Keep the first one the way it is
exactly it is small and easily transferred. Then add second advanced
NST program which allows for robust dealing with sector errors. I
suggest four options: BEST, which just reads what is there and makes a
best effort to read the sector and passes legal but possibly incorrect
data, SKIP, which just passes along default values for the sector and
ignores what is there, RETRY, which just just rereads the sector which
deals with some of the "weak bits" that sometimes read right if you
try often enough, and ABORT which does what it does now and errors out
the disk image transfer.

Also, I noticed that sometimes the last byte of the NST program does
not get uploaded. I just type in B6 at the end of the upload and it
works fine.

You may need to poke around with the monitor and determine what the
N* serial port is being set to - it should work fine of they match.

You could try the paremter: U=0'-'~S or U=1'='~S

To maximiize transfer speed, NST allows for characters being buffered
in the uarts on RX/TX - the default U=2'='~S allows for two outstanding
characters (one in each direction). The one suggested above allows for
0 or 1 oustanding characters, so the monitor will send a character and
wait for the echo before sending the next character.

You can also use my HDM80 with appropriate U= parameters. In this
case you would need to download the full "DOS + client" image for
NST (assuming you started the system from the HDM80 ROM).


I do not know if I made a mistake or what but when I tried this again
later, NST does work with U=2"="~S but not U=1"="~S or U=0"="~S. It
works reliably now so I am happy.

I read the instructions again but maybe I am not understanding what is
happening here. I tried varying the communication parameters but only
C=1:81N9600 seems to work. C=1:71E9600 also works but has the same
characteristics. Changing baud rate does not work for other than
9600. Is there a way to specify hardware flow control of the serial
port? Does it affect serial communications?
When I press F1 at the regular monitor prompt (not in the
instructions, I know, just an experiment) I see the string of hex
numbers being sent but they seem awful slow for 9600 bps.

Win2k often takes 100ms to report a recived character to the DOS
box ... this slows down single-character at a time transfers
considerably - It could also be that NST is waiting for the echo
and not getting it, or getting out of sync from the monitor '?'
error response and timeing out to recover.

I tried the W= parameter as it sounded promising that maybe Win2K was
interfering with the serial port somehow but I do not know what a more
reasonable value would be if 18 is the default. Changing the value to
a larger number does not seem to have an effect. I do not know why
and I don't really understand so I leave it alone.

It should be fine under W2K - I've seen XP take *10 seconds* to report
a character received by the physical uart up to the DOS box ... 2K is
usually on the order of a few 10s of milliseconds.

What does seem to solve the problem is the U= parameter. The default
of 2 is apparently not working on my computer so I tried U=0"="~S but
that did not work either. However, when I tried U=1"="~S it seems to
work. Then I was able to get the NST program to upload the NST stub
program on the NorthStar just as I would expect. Once that occurred,
I just followed the directions and did a "JP 2D00" and the monitor
prompt disappeared. I am not sure why this worked but I suspect it
has to do with my old and slow laptop running Win2K and having less
than ideal serial port drivers.

Could be. Best to do this stuff under real DOS or Win9X (which is
real DOS) - The further you go up Microsloths OS "progression",
he worse the DOS box works, especially for serial comms.

It all looks good so I press F10 to exit the upload terminal and run
the NST command locally on the laptop with "NST RD MASTER". The
computer is busy for a while and the NorthStar is reading sectors and
the laptop is recording them as you would expect. After a while, it
finishes and there is a MASTER.NSI file in my C:\HORIZON directory.
With the new master boot image recorded, I load the HORIZON simulator,
mount the NSDOS50D disk in drive 1 and MASTER in drive 2. Start the
simulator and do a LI command at the DOS prompt. I can see the
simulators boot disk directory. Do a "LI 2" command at the DOS prompt
and I see the MASTER boot disk directory I just recorded. In addition,
I was able to start the M5700 monitor in the simulator and it appears
to run fine.

Good.

By the way, I was also able to put some disks in the drive 2 of the
NorthStar and get them to give a directory listing. However, there
are other disks which when I type "LI 2" on the NorthStar NSDOS
prompt, I just get garbage. I do not know if this is another
operating system like CP/M or is it a single density floppy disk or
just a corrupted disk. There are many disks like this and with rather
mysterious contents.

If is is single-density, it should still show a directory (DD DOS handles
both SD and DD disks) - most likely it's CP/M, or possibly another OS
(UCSD for example).


All the disks I have found so far are double density.

Do any of these boot? (ie: does another OS come up) - keep in
mind that they may boot to the serial console or your flashwriter
depending on how they are configured.


Yes, I have found a boot disk for NS DOS 5.1DQ and one for Lifeboat CP/
M. Both have been customized for my IO routines. I will let you know
if I come across anything more interesting. I have about 80 disks to
sort through and I have no idea what is on them.

If this does not get you anywhere, read some of the disks into
images. Then you can boot CP/M on my simulator and see if you
can read the directories (indicating that they are CP/M disks) -
if that does not work, try hex-editing the image files and look for
ASCII strings that might identity the content.


Yes, that seems to work just fine. The simulator has been invaluable
in diagnosing what is on the disks.

Again, sorry for the long message but I would like to keep you and
anyone else interested up to date on my progress on restoring the
NorthStar Horizon. There are still many things to do on this
wonderful old computer but tonight has been a good night so that is
all for now.

Excellent - Glad to know it's alive and well!


Thanks to you guys!

PS, one profoundly ironic thing I noticed is my NorthStar takes about
5-10 seconds to boot from power off to DOS prompt. My laptop running
Win2K and easily 2 orders of magnitude greater processing power, takes
about 10 minutes to boot. This drives me nuts. How is it that
computers actually boot SLOWER as time progresses? :-)

"Progress"...

Dave

--
Dunfield Development Services http://www.dunfield.com
Low cost software development tools for embedded systems
Software/firmware development services Fax:613-256-5821

THANKS AGAIN! Have a great weekend!

Andrew Lynch

.

Relevant Pages

  • Re: Restoring a NorthStar Horizon, problems with SRAM board
    ... I can do LI commands from the DOS prompt but that does not work ... on all disks. ... The N* monitor has a memory test command - One version of the monitor ... I assemble the NST program using ASMZ80 ...
    (comp.sys.northstar)
  • Re: NS DOS utilities and blank disks
    ... it appears to be a loadable NS DOS utility but I do not find ... DOS and CP/M boot disks. ... for end users except occasionally to verify your boot disk is still at ... routines, it would not help very much. ...
    (comp.sys.northstar)
  • Re: Found at Yard Sale
    ... Bill H www.ts1000.us ... Also from same yardsale, CGI monitor, and about 400 5 1/4 floppy disks, ... Ms dos 3.3 original disks for Everix Step computers ... Dbase III Plus original install disks ...
    (comp.sys.ibm.pc.classic)
  • Re: Control C to reload?
    ... After a few disks I remembers that with CPM 86 you have to do ... to do the same in Dos, including the Dos that Win 95 runs at. ... Bill H www.ts1000.us ... Without this support an OS like DOS or Windows has to assume that the diskette has been swapped with another diskette between operations, and thus the OS can't cache any data. ...
    (comp.sys.ibm.pc.classic)
  • Re: "Failed SMART usage Attribute: 9 Power_On_Hours."?
    ... # list the devices that they wish to monitor. ... # Alternative setting to report more useful raw temperature in syslog. ... # become common with SATA disks in SAS and FC environments. ...
    (comp.os.linux.hardware)