TRS-80 Fragmentation

Peter Cetinski pete at pski.net
Fri Apr 27 08:55:26 CDT 2018


Excuse my long post, but I get excited whenever I can talk about the Model 16. :)  I will expound a bit on what was already mentioned.

The Model 16 was an engineering marvel.  It was released in 1982 in the same form factor as the venerable Model II.  It was essentially an upgraded Model II.  In fact, the Model II could be upgraded to mostly Model 16 specs via an upgrade kit.  The 16 was a dual processor machine with 2 independent computer systems running in parallel, one a Z80 and the other a MC68000.  The Z80 side of the machine ran just as it did on the Model II.  This was a big advantage in that the machine could run the entire Model II library of programs.  The 68K subsystem consisted of a CPU card and one or more memory cards which shared their own independent bus (via ribbon cables) from the Z80 bus.  

When the 16 was running with a 68K OS, the Z80 subsystem controlled all I/O via the main computer bus.  The Z80 and 68K communicate via shared memory in the 68K memory space with both sides essentially rapidly polling certain locations for requests and responses placed in memory.  There was a facility available in the 16 for the 68K to interrupt the Z80, but since the Model II with the Model 16 upgrade did not have this capability, no operating systems took advantage of this feature AFAIK since the percentage of machines using the upgrade was significant.  One cool feature is that the Z80 could bank switch in 16K of any location on the 68K memory so Z80 programs run on the Model 16 could use the 68K memory without a 68K OS. 

When it was released in 1982, the Model 16 came with TRSDOS-16.  This was essentially a MC68000 runtime that ran on the 68K boards.  With TRSDOS-16, the Z80 side of the machine ran the Z80 TRSDOS-II OS.  A huge issue was that the Model 16 was released with almost no software ready to take advantage of the 68K.  This was a classic case of the hardware way ahead of the software.  This resulted in Tandy actually including a copy of the Assembler 16 with every Model 16 sold so that customers could write their own software.  TRSDOS-16 had many limitations, a few of which was that it was only single user and that the assembler used non-standard 68K mnemonics.

Tandy knew they needed a multi-user system for the system to succeed, and was considering Unos as the solution when for whatever reason they could not make that happen.  They then pivoted to XENIX which was finally released maybe a year or so after the machine.  The initial releases were rushed and as a result very buggy.  Tandy lost a lot of market time advantage due to this fumble.  However, XENIX ultimately did well on the platform.  So much so that a year later the Model 16B, which was a 68K version of the Model II’s successor, the Model 12, was the most popular Unix based workstation on the market.  An upgrade of the machine to the Tandy 6000 a year or so later saw an increase in processing power.  But by then, we all know the story of the IBM PCs architecture dominance of the market and Tandy’s attempts to succeed in that area.  This caused internal business conflicts which eventually doomed the MC68000 architecture at the company.




More information about the cctalk mailing list