The X16 (and 6502) a historical perspective

[voidstar]

When I was preparing the "domesticating the computer" story, I tried to find an early photo of the PET computers being produced (c. 1977). I was never able to find one (but did come across a story about the PET case coming from the metal cabinet factory Commodore had in Canada).

The earliest "confirmed" photos of "appliance computer" production I came across was from 1978. Shown below on the left -- Tandy had purchased a JC Penny store in the Fort Worth, TX area, and began the TRS-80 production (which I say was the most successful pre-1980s microcomputer, being the first to 100,000 units sold; things obviously accelerated differently after 1980).

trs_x16_history.jpg (592.95 KiB) Viewed 3543 times

From there, you see the process got a little more streamlined. The next photo in the series is from the Tandy Color Computer (c. 1980). That started in the US (Fort Worth) but did move to South Korea eventually (and then got moved back awhile later).

The image below is the "Liquid Solder" system that Tandy used (which was operated by a Vietnamese immigrant, but I'd have to re-read all the Intercom issues from around that year to dig up his name again -- there is at least one photo of him operating it).

trs80history.jpg (147.72 KiB) Viewed 3543 times

The X16 DevBoard being made also in the DFW area is very neat. That's why I tend to view the X16 as a hypothetical CoCo4, that combines the expansion bus similar to Apple, and SID audio of the C64 - making the X16 a much better CoCo3. [ the CoCo3 also has "weird" video resolutions, like 80x24, 64x24, 32x24, 32x16 and also its interesting GIME graphics -- but Tandy then put the "good audio chip" into the Tandy 1000!! ]. Obviously, the X16 heritage is closer to the C64 (with PETSCII + Commodore BASIC and 6502 instructions), just geographically there are ties to the old Tandy systems. Either way, the X16 is a "better late 1980s system" that would have been a dream system to get.

[mortarm]

Actually, 80x24 (and 80x25) was a pretty common text mode as a number of VDTs used it.

I'm glad that geography is the only thing shared with the Coco. Never liked those micros.

[kelli217]

The Coco 1 and 2 weren't so great, but the 3 was quite a nice machine. Much better graphics, 80 columns, lowercase, lots more RAM, and a keyboard with a Ctrl key and even a couple of function keys... way better for running OS-9, which took advantage of all the bells and whistles of the 6809.

Speaking of bells and whistles... nobody ever seemed to make use of the 6-bit DAC, which was quite sophisticated for the time, and was there from the start with the Coco 1. You only ever seemed to hear the 1-bit square wave tone generator from any of the in-house software. But there are a few demos. You did NOT need the Orchestra-90 cartridge.

[voidstar]

I'm glad that geography is the only thing shared with the Coco. Never liked those micros.

Was it the "uranium green" standard background color? Many people were thrown off by that.

Or the 32 columns was awkward also - not even 40! But it didn't dawn on me until years later that the prior TRS-80's (Model 1, 3) were 64 column (as was the old IBM 5100) - in a weird way, having the number columns be a power of 2 makes sense (in the IBM5100 it has a switch to flip between LEFT32, RIGHT32, or the the full64 - its registers were 16-bit, which aligned nicely when using the LEFT32/RIGHT32 modes {and while in REGISTER mode}).

The CoCo's also lacked a memorable PETSCII-equivalent. Instead it had "semigraphics" which were a pretty boring set of banner/border styles. They wasted a lot of character codes just having different colors of the same symbols.

I had the Orch-90 cartridge- wasn't till years later I appreciated the 6-bit DAC more. But this reminds me: One thing I learned recently from 8BitGuy's Apple video is how in the very early days, there was apparently another existing Apple-named company in the music industry, and they made some kind of agreement that Apple Computer "wouldn't enter the music industry" (hence one reason why earlier on that system had relatively weak audio; until of course "non-AppleComputer" expansion options came about)


There was also a talk earlier this year about some new microcoded/pipelined 6809 (CPU that the CoCo's used), relevant since you can stay lower power and low cost (pin wise) and gain even more performance.

[8_bit_memories]

Speaking of bells and whistles... nobody ever seemed to make use of the 6-bit DAC, which was quite sophisticated for the time, and was there from the start with the Coco 1. You only ever seemed to hear the 1-bit square wave tone generator from any of the in-house software. But there are a few demos. You did NOT need the Orchestra-90 cartridge.

While I never owned a CoCo, I did have several Z-80-based TRS-80s, and both the Orchestra-85 and Orchestra-90 products.

Jon Bokelman (principal author of the original product) ported a player to Windows, which runs fine under Wine as long as 32-bit support is enabled.

I wonder what it would take to port/recreate a player for the X16 without Bokelman's help, and if it's worth the effort.

[Ser Olmy]

Speaking of bells and whistles... nobody ever seemed to make use of the 6-bit DAC, which was quite sophisticated for the time, and was there from the start with the Coco 1. You only ever seemed to hear the 1-bit square wave tone generator from any of the in-house software. But there are a few demos. You did NOT need the Orchestra-90 cartridge.

Oh but you did need it, if you wanted the CoCo to do literally anything else while also playing back half-decent audio.

There's nothing besides the CPU feeding the 6-bit DAC, meaning if you want to play a simple 1000 Hz square wave you have to send data to the DAC 2000 times/second, and time it exactly right too. Want something more advanced than a square wave? Then you'll need a lookup table, and the number of write operations required suddenly increase by about an order of magnitude.

One way to get this semi-right is to put the playback routine in the HSYNC interrupt handler, which leaves precious few CPU cycles left for some actual gameplay. Oh, and you'll have to put the joystick handler in there too, otherwise you'd have to disable interrupts in order to read the joystick axes (as the same DAC is used to generate values that are fed to a comparator), which would throw off the timing.

And that's why games on the CoCo tended to have rubbish round.

[voidstar]

The "double-speed" of the CoCo3 (and extra RAM) maybe allowed that 6-bit DAC to be more useful than the original 16KB CoCo1. The CoCoTracker (on a CoCo3) as example:
https://www.youtube.com/watch?v=49uB068i3ZU

It's true, many overlook that things like keyboard input, serial bit-to-byte conversions, or showing a font on the screen are all really "hardware-accelerated" features outside the CPU. But, a "general processor" could do these things if it had to.