• Please review our updated Terms and Rules here
  • From now on we will require that a prefix is set for any items in the sales area. We have created regions and locations for this. We also require that you select a delivery option before posting your listing. This will hopefully help us streamline the things that get listed for sales here and help local people better advertise their items, especially for local only sales. New sales rules are also coming, so stay tuned.

Zilog System 8000

For those interested in fiddling with the Z8000, the Olivetti M20 can be had relatively cheaply, but a little hard to find in the US. It isn't a mini like setup but still has reasonable abilities. Without too much effort some have expanded to 512K of memory. The memory management is not all that good because there as not a working MMU for it when the M20 was designed. That is why it can't do a full 1Meg as the processor design was capable. To have both Data and separate Instruction areas, allowing one to execute a full 64K while having a full 64K of data at the same time, meant that execution areas needed to have two memory maps. This was because you couldn't write directly to Instruction space without mapping it as Data space first. Of course you could pay the penalty of a full address access but that would slow it down.
Dwight
 
July 2021, and I just saw a Z8000 system again on eBay for $599 and $300 shipping but, wouldn't you know it
Only three boards in it? Does it have a CPU board? There is no board in slot 1. And there are no memory board neither the memory controller.
Yep. Same problems. Likely the same exact machine, I think. Model 20 #05-0066-00

On a side note, the first Tandy1000 came out with an upgrade socket for the option of a Z380, which was "vaporware", but, after scrapping the project, the team later went on to make the Z8000. Tandy stopped putting in the upgrade socket soon, maybe even before the first run finished.
I don't know enough of the technical details, but I'd love to mess around with one of these early 1000s and see if any later Zilog chips like the Z8000 could work, maybe with an adapter of some kind? The speed might ultimately be very disappointing, and of course software next to non-existent, (maybe adapt linux? run or recomplie Z80 CP/M versions and software?) I can dream! :)
 
It is the same (door stop) machine.

Why not run CP/M-8000 - there is a version out there... You just need to tweak the BIOS for your non-existent Tandy...

Dave
 
Won't the System 8000 run some flavor of Unix? I recall using an Onyx box doing development back in the day.

yes, the problem has been finding a copy of the software
the other problem is the seller sold off the important boards
I have the CPU board
 
I have three tapes with ZEUS OS and SADIE. But the problem is how to read them. The format is not GCR, but MFM. And the drive has a head with four fixed tracks. A couple of years back I got help reading a tape into a logic analyzer dump. Then I ran it throug a small utility I wrote to convert into a file which could untar.

I have a spare drive which I intend to fiddle with but the rubber-roller is very hard and for some reason the motor axis is not mounted in the centre of it. Some day I will get to it.
 
Getting a kennedy 6455 tape drive with a fixed 4-track head working with four read preamps has been on my todo list for a while to deal with Onyx/Zilog/DEI pre-QIC tapes
These early drives don't serpentine, so you can read the whole tape with one forward pass if you have four preamps.

I have a manual for them http://bitsavers.org/pdf/kennedy/64xx
 
Last edited:
The Olivetti M20 was a Z8000 machine. I and mostly Chris Groessler ported the CP/M8000 to the Olivetti M20. I've had it running on one of my M20s but it has been several years since I had it up and running. I added a TRS80 controller and ST251 hard drive to it as well. Before connecting any random Z8000 to it one must also have the ability on the machine to split the data and instruction memory into two 64K ram banks, as it expects this. The Z8000 was really intended to run with a memory management chip but it had too many design issues and never made it to production. The M20 did the split with a memory ROM. This worked relatively well except one had to give up 1/2 of the Z8000's potential memory addressing. This was because to load data from the disk, the memory needed to be mapped to data memory and then remapped to instruction memory to execute it. This was done by the address decoder ROM with the various segment addresses. If it had been done with a memory management chip as originally intended, it could have been switched with a control register in that chips. To add to the complexity of the ROM, the video memory was dual ported into the main memory for the color. On top of this, there were mother boards with two banks of 16K DRAMs or two banks of 64K DRAMs. Because the expansion RAM slots were decoded on the memory board for easier control of the black/white or color options. This was all put into the memory control ROM. It was all quite complicated since the address one used by the CPU was only the address use by the RAM in the low 14 bits. Bits 14 and 15 were combined with the segment to determine, through the ROM, which physical memory would be used.
Anyway, the CP/M8000 that is out there expected the split instruction/data memory system. Not all Z8000 systems were setup that way.
Sorry for all the long winded description.
Dwight
 
the first Tandy1000 came out with an upgrade socket for the option of a Z380, which was "vaporware", but, after scrapping the project, the team later went on to make the Z8000. Tandy stopped putting in the upgrade socket soon, maybe even before the first run finished.
I think you're referring to the TRS-80 Model 4, the first version of which had a socket for the never-released Z800 CPU:

http://www.trs-80.org/model-4/

I had an original Tandy 1000 and there was no unused CPU socket. It didn't even have a socket for a 8087 math co-processor.
 

Then, someone that had the low level code should be able to adapt it to that platform. It is just a mapping issue. CP/M is relatively simple at what low level functions it needs. I suspect if one used the B/W mapping that the M20 used, it wouldn't take that much to adapt it. I believe the interface used the SYS CALL for all the I/O and disk interface.
It is more project than I'd want to tackle but then I'm not a heavy software type. I'm more at home in the hardware.
Dwight
 
Not sure if this will help but the Commodore 900 uses a Z8001 CPU and 8010 MMU as well.

I was fortunate enough to have the source code for Coherent (a port for the C900) on my hard drive. Thanks to Poul-Henning Kamp at DataMuseum.dk, he extracted and published an image of my hard drive here: http://phk.freebsd.dk/misc/cbm900/ Maybe some of that code could help?

There was a MOS Technology internal memo containing a "Custom MMU proposal" for the C900 here that contains some information that may be useful: http://vintagecomputer.ca/files/Comm...20Proposal.pdf

Hope this might help someone.
Santo
 
yes, the problem has been finding a copy of the software
the other problem is the seller sold off the important boards
I have the CPU board
Oh Boy,

Just stumbled on this forum after I purchased a System 8000 on Ebay, and man now I feel like the punchline.

I just bought the same system that is the subject of the last few posts.

@Al Kossow,
Don't suppose you would have a price in mind for that CPU board?

*Don't mind the egg on my face, it's just a side-effect of my addiction to buying vintage systems.*

-Casey
 
I figured that might be the case, my second approach to this problem would be perhaps replicating the CPU and memory PCB and creating a modern-day equivalent of those components while I wait for originals to crop up.

Al, would you be willing to take high resolution photos of your cpu and memory boards? I may be able to trace the schematics in kicad, and make an inventory of the IC and passive components.

-Casey
 
Back
Top