• Please review our updated Terms and Rules here

TRS80 Model 5

danielbooneamerica

danielbooneamerica

Experienced Member
Joined
Mar 3, 2020
Messages
408
Location
Wayne National Forest
All over internet I see a wide range of discussion of what a Model 5 should have if it were to be built, what Radio Shack thought next model should have and what users would like to have.

Now here we are all these years later and we have no replacement TRS80 computer nor do we even have an agreed upon design.

All that being said i thought i would start a wish list.

* Note this is a hardware list only. We can assume drivers to support hardware would be developed also.

eZ80 CPU running @50Mhz executing code with 0 wait states.
Serial port.
SPI port.
VGA.
keyboard.
lots of RAM.
FLASH BOOT & DISK.
*Very important* TRS80 style bank memory & lots of it, think big people.

....this is a start.
 
Lots of possibilities, the model 4 was intended to eventually use the Z800 so a new design could use the Z280 although a FPGA design might be faster, then there was the Z380 which is a Z80 compatible 32 bit cpu but at the time the successor to the model 4 would have been a development of the color computer.
Or even an x86 design.
 
mnbvcxz said:
Lots of possibilities, the model 4 was intended to eventually use the Z800 so a new design could use the Z280 although a FPGA design might be faster, then there was the Z380 which is a Z80 compatible 32 bit cpu but at the time the successor to the model 4 would have been a development of the color computer.
Or even an x86 design.
Click to expand...

I am not sure about color computer line replacing trs80 if that is what you meant. RS had an image problem using coco for small businesses.

That being said for sure color would have been added to trs80.

Z380/280 is pretty well dead end streets. I am looking more for a wish list of what we can and need to get designed and built now.

There is mass duplication of effort all over web. I wrote a new bios/sysres and have most that tree shaken out so let's gets this party rolling.

I liked this LISP board but they dont plan to support banked memory and seem to feel 1mb is all anyone should need. At this point I can see it's a good starting point but more computer architecture thought must be given to our design. The LISP board would not have all the features we need nor expansion built in design.

Trs80 bus is pretty tried and proven making a good starting point on i/o. I am amused how easily most modern peripherals have been attached for one reason or another. I can verify a trs80 has been connected to a motion system of a full flight level D flight simulator for testing.

While some of the old trs80 group is still out of old folks home, is it possible we could design the next step?
 
danielbooneamerica said:
All over internet I see a wide range of discussion of what a Model 5 should have if it were to be built, what Radio Shack thought next model should have and what users would like to have.

Now here we are all these years later and we have no replacement TRS80 computer nor do we even have an agreed upon design.

All that being said i thought i would start a wish list.

* Note this is a hardware list only. We can assume drivers to support hardware would be developed also.

eZ80 CPU running @50Mhz executing code with 0 wait states.
Serial port.
SPI port.
VGA.
keyboard.
lots of RAM.
FLASH BOOT & DISK.
*Very important* TRS80 style bank memory & lots of it, think big people.

....this is a start.
Click to expand...


It was not this sim but one like it we connected model 4 and using basic did what we call "exercise motion". How flexible is that!

https://m.youtube.com/watch?v=h8KbtmMmhGg
 
Given that the Z800 continued to never materialize, I think an official TRS-80 model 5 (in an alternate reality where the Tandy 1000 never came to be and MS-DOS wasn't a thing) would have had a new main CPU (perhaps a 68000, if the bus timings aren't too far off from the Z80 to make it work?) and a Z80 co-processor onboard or on an add in card for backwards compatibility. Kinda like a Commodore 128 on crack.

If I were to dream up a Model 5 in Current Year, I think I'd pretty much just want a copy of the Model 4, except with a spot where you could slot in a Raspberry Pi that could tri-state the CPU and take over the bus and all the peripherals and run whatever you wanted in emulation but with access to the TRS-80s display/input/disks, or could just act as a Linux box. It would have to have that Model 4 "aesthetic" though, and that olds-kool 70s keyboard. And in native Model 4 mode the Pi would act as a floppy emulator when needed. It would also have to have a built-in CRT, same as the Model 4, but that could probably never happen nowadays. :p

Dumb Question: Where does the "80" in "TRS-80" come from? The Model 4 and the DT-1 were the only ones that were 80 column, right?
 
bladamson said:
Dumb Question: Where does the "80" in "TRS-80" come from? The Model 4 and the DT-1 were the only ones that were 80 column, right?
Click to expand...

:) Maybe the original plan was to use the Z64 and have 80 columns but then the intern got the specs swapped and used the Z80 with 64 columns! :):p
 
danielbooneamerica said:
I liked this LISP board but they dont plan to support banked memory and seem to feel 1mb is all anyone should need. At this point I can see it's a good starting point but more computer architecture thought must be given to our design. The LISP board would not have all the features we need nor expansion built in design.
Click to expand...

There was LISP Board for the TRS80's? The Business Models? You mean like a MacIvory Lisp Machine for the Mac? Details please!
 
You'd want to look at what the ZX Spectrum Next did, for sure (another Z80-based system).
 
wmcbrine said:
You'd want to look at what the ZX Spectrum Next did, for sure (another Z80-based system).
Click to expand...

FPGA's are cool and can do almost everything, but also a bit boring by now... hardware development basically becomes software development.

But the Spectrum Next looks indeed amazing!
 
Last edited:
I've personally looked into several possibilities for something like this, and have thought through designs using Z280, Z380, and eZ80. I have chips for all of these; eZ80 and Z380 are current production, too. Yes, the Z8038018FSG is current and active at both Mouser and Digikey; I got two today, in fact, from Digikey and already had two from Mouser from a couple of years ago.... If you're wanting to think big, it doesn't get much bigger than Z380's 4GB address space!

Having said that, eZ80 has a lot going for it: speed, software support (thanks to the TI-84CE), current and active, Ethernet capability, and a clean bus design. It's faster than Z380 thanks to shorter addressing prefix instructions, and requires fewer chips for a useful design. I have several modules, all the eZ80F91 variant.

BUT, having said that, Z280 is the one I really want. I have some prototype boards from plasmo as well as an operating CPU280 board and several additional new boards that need building, along with 30-40 Z280 chips ready to go on boards. Of these three, the Z280 would be the easiest to fit as an in-Z80-socket upgrade for the TRS-80 thanks to compatible DRAM refresh (Z380 implements CAS before RAS, eZ80 doesn't do any DRAM refresh).

I'm personally planning to prototype all three, it'll give me an excuse to learn kicad and roll some PCBs. I'm also working on the Z8S180 XLR8r workalike, but I'll probably get the Z280 one done first. I'm looking at a simple RC2014 form-factor for a Z380; it would be one of the very few Z380 hobbyist designs if I'm successful. The hardest part is the 4GB of RAM: the biggest EDO 5V DRAM chips are 64Mx8, and that would be 32 chips if I've done my math correctly. It might seem easy to try to use modern PC RAM, but the voltage difference plus blocking/unblocking the 64bit RAM interface to the 16bit bus of the Z380 might be tricky. FB-DIMM would be interesting, and single 4GB modules are available, but the AMB lanes on FB-DIMMS run at 4Gb/s....a bit fast....

Chris Brock has a fully-ported MP/M for eZ80, and many of those techniques would be useful.
 
A system based on the eZ80 would probably be more true to the spirit of the original "Model 5" concept than a fully FPGA-realized machine like the ZX Spectrum Next. That said... I'll admit I don't know much about the eZ80, but skimming through the datasheet for it I see several places where it's said that I/O ports 0080-00FFh are dedicated to the chip's onboard peripherals. This would seem to conflict with where the Model III/4 mapped all the important port-mapped hardware? This looks like it might be a dealbreaker, but I'm sure someone who's actually used the CPU knows the answer to this off the top of their head. (It looks like the Z280 had an I/O paging register to solve this problem, but the eZ80 doesn't seem to?)

Beyond that, I simply don't know how good the eZ80 would be at running the existing base of TRS-80 software without patches. Ideally you'd be able to do that seamlessly while modifying the OS/BASIC/etc. so parts of it can run in the enhanced 24 bit mode but, again, I don't know how easy it is to shift gears like that with the eZ80. I guess this PDF does say it's possible.

As for the rest of the hardware you'd obviously want an enhanced video chip that in addition to faithfully emulating as many existing TRS-80 video standards as possible (IE, everything from the 1K 64x16 text-and-pseudographics mode up to the TRS-80 Model III/4 high-res mods) would implement some higher-res graphics, although I'd probably suggest drawing the line at late-80's-early-90's pixel counts because you're still dealing with a relatively weak CPU by modern standards. These days the most obvious way of implementing a video chip like this would of course to be inside of an FPGA; it might be possible to use a relatively small one, though, if the ambitions are modest enough.

(An idea that might be worth looking at would be to implement "sprites" in the form of programmable color tiles that can be mapped on top of the semigraphics blocks; done right it might be possible to convert the "game engines" of old Big 5, etc, games to display more impressive visuals with relatively little effort.)
 
At some point, as we talk about video, we would probably end up reinventing the MSX TurboR, just with eZ80 instead of R800. The MSX had an uncanny resemblance, in my opinion, to what the supposed Model 5 would have been.
 
280 380 etc are nice but my first thoughts is that if a project was to have wide spread support then folks in other communities must want to use same platform (CP/M, LISP & other Z80 based systems).

eZ80 has wide acceptance and supported by Zilog online still. So eZ80 already finds wide acceptance and would find wide support I think.

That being said, to my knowledge you might be offering the only 280 or 380 SBC in market place. Thus far I have found no general purpose 280 or 380 SBC for sale in production, that is a shame.

I see many projects on different websites but most go unfinished and I am sure that is for an assortment of reasons.

My project is to stage of being solder ready to run a new version of the OS.
 
lowen said:
At some point, as we talk about video, we would probably end up reinventing the MSX TurboR, just with eZ80 instead of R800. The MSX had an uncanny resemblance, in my opinion, to what the supposed Model 5 would have been.
Click to expand...

Yeah, that is kind of the problem, really. While technically high-res graphics add-ons existed for the TRS-80 line practically speaking there's very little software that makes use of it, none of it really mainstream. Creating enhanced color capabilities that really meaningfully relate to the "TRS-80" part of the machine's ancestry is almost as much a philosophical challenge as a technical one.

It is funny that you mention the MSX line because there were several add-ons for the TRS-80 that interfaced a TI 9918A graphics chip to it, and I think some of them even included a sound chip. Again, I don't think there was any commercial software created for these things but I suppose technically you could use the fact they existed as an excuse to implement enhanced graphics in Yamaha V9958-ish sort of way.

Realistically, though, given what a character-oriented machine the TRS-80 was I think an emphasis on features like programmable graphics character sets might be the most immediately useful approach to enhancing the video system.
 
It is correct eZ80 I/O is direct conflict with TRS80. Correct this I/O cannot be relocated.

For model 4 mode under an advanced OS such as TRS/LSDOS this is no matter, apps do not talk directly to hardware but thru OS.

OS only communicates with hardware via BIOS. BIOS required a complete rewrite.

Legacy software is still 16 bit program counter software but with advanced instructions we can access 24b addresses for such things as overlays, Spooling, RAM DISKS, FLASH DISK space etc.
 
And as I recall the TRS80 hires has original IBM PC beat. All we need to do is use same standard at least as a starter. One reason being much legacy software including basic exists for original TRS80 hires mode.

Original TRS80 video chip had some bugs such as scrolling. This is why scrolling was done in software in TRS80. Some things like that could now be eliminated thus freeing precious bytes of low core.
 
danielbooneamerica said:
For model 4 mode under an advanced OS such as TRS/LSDOS this is no matter, apps do not talk directly to hardware but thru OS.
Click to expand...

Incompatibility with Model 3 software would be a hard deal-breaker for me when it comes to interest in a "Model 5", but maybe I'm not the target audience.

I'm not particularly familiar with the Model 4's "BIOS", but it seems to me that having to patch any and all software that does port I/O (which is what using the eZ80 would amount to) would be a non-trivial problem; it would surprise me greatly if *every* piece of Model 4 software was completely well behaved and only used operating system calls, and literally all its I/O is between 80 and FFh. Serial, floppy, cassette/sound, video chip, memory banking controls...

As for the TRS-80 Hi-Res having the IBM PC "beat", well, I guess, technically? It's 640x240 instead of 640x200, but that's not a *lot* to crow about. (It's kind of interesting they chose a resolution that requires an oddball 20k of RAM; IBM clearly simply went with the most that would fit in an even 16k.)
 
You must log in or register to reply here.
Back
Top