• Please review our updated Terms and Rules here

Found a rare Novell Data Systems "Nexus I" computer. Need help with diagnosis.

R

redruM69

Experienced Member
Joined
Jul 18, 2007
Messages
64
Location
Salt Lake City, Utah
Found a rare Novell Data Systems "Nexus I" computer. Need help with diagnosis.

Hey guys. Just found this beast today! I picked it up from a former Novell employee about 10 miles away from the original headquarters in Orem, UT. I believe it is quite rare, and found extremely little information online. Serial number is 544.



I have opened the monitor unit up, and found it has 2 identical motherboards, Z80 CPU's, but with different ROMs and RAM amounts. One board appears to be the "main" CPU board, controlling IO, the network, and the disk controller (remote in the disk unit). It then has a 2nd motherboard, that seems to be controlling the terminal BIOS, and has a ribbon leading to the video card. The video card has a F6800 on it.

After some tinkering, I did manage to get it to boot from the 5MB HDD. It boots to a CP/M prompt, and still has dbase installed.
However, the screen output is quite corrupt. Take a look:



It begins scrolling continuous underscores as soon as it passes the power on checks. These underscores insert between each character displayed on BIOS messages, before the HDD boots. I have already tried disconnecting the keyboard. Once the HDD boots, the hyphens stop, but still has corrupted characters and carriage returns. I have also swapped the ROM's in the two CPU boards, and swapped them around. It wont boot the disk bios anymore (lack of RAM?), but after it passes the power on checks, it still scrolls underscores.

Here's a video of the boot sequence:


Does anybody have any resources or info of any kind on this oddity?

Thanks in advance!
 
Last edited:
Nice find!

Two things spring to mind.

  • Bad memory - if you have access to an ICE, connect it to the main CPU and run a memory test.
  • A problem with the character generator, if it has one.

It's interesting that you seem to be getting corruption at end of line. I wonder if the screen handling code is corrupt? All depends on how the screen interpretation of control characters is implemented.

Open it up and take some good pictures of the board(s), then we can see what ICs it is using.
 
Tonight I will dig deeper, and start mapping out the PCB's This machine also functions as a networked terminal. The way the underscores scroll like they do make me suspect the terminal portion is going wacky. If I can isolate that portion and shut it down, it may start behaving better.
 
JonB said:
Open it up and take some good pictures of the board(s), then we can see what ICs it is using.
Click to expand...

Here's some more pictures: http://imgur.com/a/MRpb1

I was wrong about the proccessors. Only 1 Z80 per main board. The other chips are Z8470 DARTs. The video card has a F6800 series CPU.
I have cleaned the corrosion from the video card. Made no change.

I have a hunch the corrupted characters and the scrolling underscores are 2 different issues. The previous owner showed an attempt to boot a couple of years ago: https://www.youtube.com/watch?v=xnm1DbUmkRY
It had some corrupted characters at that time, but no scrolling underscores. The repeated lines in which he thought it needed a boot disk was actually an error that I received also at first. After aligning the HDD head a little bit, it gets past that and boots to the corrupted CP/M prompt that I showed in my first post.
 
Small update. Cleaning the corrosion on the video board did not help. Power supply has been tested under load, and gives good voltages.
I will try piggybacking RAM chips soon.
 
It's neat isn't it? I expected the logic boards to be in the drive box..

Can you run it with boards removed? Reason I ask is you will want to use a logic probe or scope at some point to measure signals.

Hmm, try to run without the serial board connected. Maybe these underscores are coming from there, then stop after the board is initialised at boot?
 
redruM69 said:
Hey guys. Just found this beast today!
Click to expand...

If you can borrow one, get an eprom programmer and dump all of the eproms.
Also try to get one of Dave Gesswein's MFM disk emulators from someone long enough to image the hard disk.

It is likely there is bit rot in the eprom images, and you said you had to fiddle with the drive to get it to boot, so
the OS image is at risk.
 
Hey that's an awesome find! I was always fond of Novell products, this is the earliest one I've ever seen. Is there any more information you can get out of the person you got it from?
 
JonB said:
It's neat isn't it? I expected the logic boards to be in the drive box..

Can you run it with boards removed? Reason I ask is you will want to use a logic probe or scope at some point to measure signals.

Hmm, try to run without the serial board connected. Maybe these underscores are coming from there, then stop after the board is initialised at boot?
Click to expand...

Scrolling persists without the serial board attached. It also persists when the main CPU board is removed. It stops if the secondary CPU board is removed. The issue lies between the 2nd CPU board and video board.



Al Kossow said:
If you can borrow one, get an eprom programmer and dump all of the eproms.
Also try to get one of Dave Gesswein's MFM disk emulators from someone long enough to image the hard disk.

It is likely there is bit rot in the eprom images, and you said you had to fiddle with the drive to get it to boot, so
the OS image is at risk.
Click to expand...

I have a willem, and I plan on dumping the ROMs soon. Getting this drive imaged may prove tricky. I do have a 286 with an MFM card in it. I should be able to grab an image using that.

new_castle_j said:
Hey that's an awesome find! I was always fond of Novell products, this is the earliest one I've ever seen. Is there any more information you can get out of the person you got it from?
Click to expand...

He worked for Novell for 24 years, as a Support Engineer, Senior Manager, and "Search Architect". He also had a huge collection of Novell historical artifacts. Mostly shwag. You can browse his collection here: https://wiki.microfocus.com/index.php/Novell_Museum. He said he found this computer on eBay a few years back, and that it belonged to a trucking company. They paid $24k for this machine back in the day!
 
Al Kossow said:
No, it will require this
http://www.pdp8.net/mfm/mfm.shtml
because you don't know what the sector format is.
Click to expand...

Maybe I am not familiar enough with MFM. I dont quite understand. Couldn't the drive image be dumped in full regardless of sector size, and later figured out? How would using this emulator be different then imaging in a PC with MFM controller?
 
Another update. I have been attempting to contact Bruce Damer, curator @ Digibarn museum. Apparently he has one of these machines. If I could borrow the two suspect boards from him, it would drastically ease diagnosis. I could compare each board on my scope and figure whats dead. So far he's been pretty unresponsive though.
 
If this thing boots CP/M (albeit with lousy display), you might be able to export the image of the hard disk out the serial port (if you can get that to work) with a bit of code.

The most extreme case of this I ever had was a machine with no functioning floppies, but CP/M on a hard drive, but no workable serial ports. I figured out how to get a keyboard LED to blink under keyboard control, so I taped a phototransistor over the LED and bit-banged the data out of the machine. It took more than a day, but the machine stayed up.

I don't think I'd be desperate enough to do that today--I've gotten lazy in my Medicare years.
 
redruM69 said:
How would using this emulator be different then imaging in a PC with MFM controller?
Click to expand...
The low-level sector format on MFM and RLL drives differs slightly between every single hard disk controller model, even from the same vendor.

On a PC, if you have to replace a hard disk controller card, it must be EXACTLY the same model, otherwise it will usually appear completely blank. Before using a hard drive with a different model of controller card, the drive must be freshly low-level formated, wiping all the data.

And that it just between IBM PC style controllers. Fat chance (FAT12 chance?) reading an MFM hard drive from a non IBM-PC.

Later interfaces such as SCSI and IDE eliminate this annoyance by putting the controller logic on the drive itself.
 
Chuck(G) said:
I figured out how to get a keyboard LED to blink under keyboard control, so I taped a phototransistor over the LED and bit-banged the data out of the machine. It took more than a day, but the machine stayed up.
Click to expand...

This is an Epic tale, love to here stories like this!
 
The issue seems to be confined only to the video board itself, since swapping the terminal ROM into the main CPU board results in the same underscores. I have piggybacked 4116 RAM chips on top of the video RAM, but it does not change output.

Just following traces, it appears most data lines from the terminal board to the video board are going into a N8T245N Transceiver, which is also fed into another identical chip on the other end of the board, and a 74S240N Buffer/Line Driver. There is a couple more traces headed to a 74LS139N Decoder, a 7438N Logic gate, and a few other traces to various other logic.

Would you find it logical to try replacing those chips first?
 
You must log in or register to reply here.
Back
Top