When Tandy ended production of the Color Computer, development of the CoCo should have ended with it.

There would be no Color Computer 4. No next-generation GIME from Tandy. No official multi-megabyte CoCo 3. And eventually, no new supply of the proprietary components upon which the machines depended. But something unusual happened.

The people using the Color Computer simply kept developing it themselves.

Third-party development certainly didn't begin when Tandy left the market. During the CoCo's commercial life, an extensive aftermarket had already developed around it. Companies such as Disto, Ken-Ton, Burke & Burke, PBJ, Speech Systems and many others produced hardware that went well beyond Radio Shack's own catalog.

After Tandy left, however, those developers were no longer merely supplementing an actively developed commercial computer. Increasingly, they were the platform's hardware and software developers.

For more than three decades, individuals, small companies and user groups continued creating hardware and software for the CoCo.

Hard drives gave way to CompactFlash and SD cards. The 6809 gave way to the more capable Hitachi 6309. Memory expanded from kilobytes into megabytes. Modern video systems replaced aging RF output. Development environments on modern computers made writing new CoCo software considerably easier. FPGA implementations began replacing—and extending—the custom hardware itself.

And as original components became increasingly difficult to obtain, the emphasis expanded from upgrading the Color Computer to preserving it.

Today there are replacements for proprietary circuitry, newly manufactured versions of discontinued peripherals, recreated motherboards and open-hardware designs that can help keep Color Computers operating long after the remaining supply of original parts is exhausted.

Tandy stopped developing the Color Computer.

The CoCo community never did.

One institution became particularly important to the post-Tandy community: the annual "Last" Chicago CoCoFEST!

At the 1991 RainbowFEST, Lonnie Falk announced what was being called "The Last Chicago Fest."

Members of the Glenside Color Computer Club decided the community deserved at least one more gathering. Dave and Nancy Myers of CoCoPRO! organized the First Annual "Last" Chicago CoCoFEST! in 1992.

Of course, calling something the First Annual Last CoCoFEST! contained a joke from the beginning.

The event was successful enough to require another "Last" CoCoFEST! And another. And another.

The joke became part of the identity of the Color Computer community. The supposedly final gathering became a place where developers could demonstrate prototypes, exchange technical information, introduce hardware, distribute software and put new products directly into the hands of people who were still using their Color Computers.

It also became something of a living contradiction.

Every year the community gathered for the "Last" CoCoFEST!—and somebody invariably showed up with something new for the CoCo.

The Radio Shack floppy system transformed the usefulness of the Color Computer, but by the late 1980s its capacity was becoming increasingly restrictive.

Third-party developers responded with hard-drive interfaces.

Ken-Ton, Disto, Burke & Burke and the Glenside Color Computer Club were among those developing mass-storage solutions using technologies ranging from SCSI to IDE and PC-style hard-disk controllers.

Disto produced SCSI hardware. Ken-Ton also offered SCSI solutions. Burke & Burke approached the problem differently with the CoCo XT, allowing a Color Computer to use PC XT-style hard-disk controller hardware. Glenside eventually produced an IDE interface.

For a computer originally designed around cassette tape and floppy disks, even a modest hard drive represented an enormous leap.

More importantly, these systems helped create the software and conventions that later generations of CoCo storage would inherit.

An important part of that history is RGB-DOS.

RGB-DOS emerged during the early hard-drive era as a way of making large storage devices accessible from an environment familiar to Disk BASIC users.

Years later, Boisy Pitre continued that line of development as HDB-DOS.

HDB-DOS eventually became much more than software associated with one particular hard drive. It was adapted to several generations of CoCo storage hardware and later to DriveWire.

That established a pattern that appears repeatedly throughout the post-Tandy story:

New hardware frequently tried to preserve compatibility with existing CoCo software rather than requiring users to abandon everything that came before it.

Few organizations have had as much influence on post-Tandy Color Computer development as Cloud-9.

Founded in the 1990s, Cloud-9 brought together Mark Marlette, concentrating primarily on hardware, and Boisy Pitre, concentrating largely on software.

Over the years its work encompassed storage, memory, operating-system support and communications.

Rather than simply keeping old peripherals working, Cloud-9 continued treating the Color Computer as a platform that could still be developed.

Cloud-9's TC³ SCSI interface represented a newer generation of SCSI storage for the CoCo.

SCSI had already demonstrated how dramatically large-capacity storage could change a Color Computer. Cloud-9 carried that idea forward with newer hardware and software support.

For a machine whose original mass-storage medium had been cassette tape, the ability to work comfortably with hard disks was a remarkable transformation.

But another storage technology was becoming inexpensive, compact and readily available. IDE—and particularly CompactFlash—was almost perfectly suited to vintage computers.

Cloud-9's SuperIDE became one of the better-known mass-storage interfaces developed for the Color Computer.

It supported IDE devices as well as CompactFlash, providing solid-state storage without abandoning the hard-drive-oriented software ecosystem that had already developed.

Programmable Flash on the interface could hold DOS or ROM images, and a real-time-clock option was available.

With HDB-DOS and OS-9/NitrOS-9 support, a CompactFlash card could essentially become a Color Computer hard drive.

For the CoCo, CompactFlash was nearly ideal. It was silent. It used little power. It had no moving parts. And even a relatively small card provided an amount of storage that would have seemed absurd during the Color Computer's commercial lifetime.

The CoCo had moved from cassette tape to floppy disk, then hard disk, and finally to solid-state mass storage.

But storage didn't necessarily have to reside inside—or even next to—the computer.

In 2003, Boisy Pitre created DriveWire as a solution for the Color Computer at a time when floppy drives were disappearing and convenient modern storage solutions for vintage machines were still uncommon.

The idea was remarkably elegant.

Instead of putting a hard disk in the Color Computer, put the virtual disks on another computer and make them accessible through the CoCo's serial connection.

No SCSI controller. No IDE controller. No hard disk.

The host computer became the storage system.

DriveWire developed into far more than a simple virtual floppy arrangement. Later versions expanded the protocol and host software to provide additional communications and networking services.

DriveWire demonstrated that a Color Computer could access PC-based virtual storage hosted on another machine—a concept later developers would take in new directions.

Roger Taylor is difficult to place into a single category because his contribution to continued CoCo development crossed several of them.

He developed programming tools. He developed games. He developed storage and communications hardware. Eventually, his work extended into FPGA implementations of the Color Computer itself.

That breadth makes his work a particularly good example of how post-Tandy development stopped being neatly divided between "hardware people" and "software people."

One problem facing programmers for an obsolete computer is the development environment itself.

Original assemblers and editors running directly on the CoCo still work, but they aren't necessarily the most efficient way to develop substantial new software decades later.

Taylor developed a succession of modern development environments:

Portal-9 >> Rainbow IDE >> Phoenix IDE

These environments allowed programmers to write Color Computer software on modern PCs using cross-development tools rather than requiring the entire development process to take place on the target machine.

CCASM, Taylor's cross-assembler, became another part of that environment.

Source code could be edited on a modern computer, assembled, converted into appropriate CoCo formats and then tested using an emulator or real hardware.

The development computer had changed. The target was still the CoCo.

Taylor didn't merely create tools that allowed other people to develop software. He continued developing CoCo software himself.

One of his best-known CoCo 3 games was Jeweled, a Bejeweled-style puzzle game designed for a 512K CoCo 3.

Other games followed as well.

That matters because hardware development alone doesn't keep a computer platform alive. Someone has to continue writing software for it.

Taylor participated in both sides of that equation.

Taylor later developed another approach to storage and communications through CoCoNet and the MicroSD Drive Pak.

The Drive Pak brought MicroSD storage directly to the Color Computer while CoCoNet expanded the concept into a broader local and remote storage environment.

Taylor also experimented with serial and wireless communications, including Bluetooth-based connectivity.

His work therefore spans an unusually broad portion of the modern CoCo story:

Tools to develop software. Software to run on the machine. Hardware to expand it. Communications to connect it. And eventually programmable logic capable of recreating it.

Another approach to preservation is represented by Australian TRS-80 enthusiast Ian Mavric and The Right Stuff.

Mavric's work frequently addresses a very practical question:

What happens when we still want to use the original hardware, but the peripheral required to use it is becoming scarce?

One answer was the RE-FD502.

Rather than replacing floppy disks with virtual storage, Mavric developed a newly manufactured floppy-disk controller so Color Computer users could continue operating real floppy drives and real disks without depending upon an increasingly scarce original Radio Shack controller.

That represents a very different philosophy from DriveWire.

DriveWire asks: Can we eliminate the physical disk?

The RE-FD502 asks: What if we still want to use it?

Mavric has also reproduced the Deluxe RS-232 Program Pak and worked with 6309 processor upgrades, CoCo 3 memory upgrades, repair and refurbishment.

His Experimenter's CoCo Cartridge has also provided a basis for custom Program Pak and diagnostic use.

This kind of work is easy to overlook because it doesn't necessarily transform the Color Computer into something radically new.

Instead, it performs another equally important job:

It keeps the original machine usable.

Darren Atkinson approached modern storage from yet another direction.

Instead of simply presenting a large storage device to the CoCo, CoCoSDC was designed to behave sufficiently like the floppy-controller hardware that existing Color Computer software expected.

That distinction is critical.

Many CoCo programs—particularly games and software using custom disk routines—didn't politely ask Disk BASIC to read a file. They communicated with the floppy-controller hardware directly.

Simply providing a filesystem on an SD card wasn't enough.

CoCoSDC therefore combined modern solid-state storage with floppy-controller compatibility, allowing an SD card to contain disk images while retaining compatibility with a remarkable amount of existing software.

Ed Snider participated in early production of assembled hardware, while Tim Lindner designed an enclosure.

A Color Computer once surrounded by boxes of floppy disks could now carry an enormous software collection on a tiny SD card.

Once again: Compatibility with yesterday's software became part of tomorrow's hardware.

The CoCo 3 shipped with 128K. Tandy supported expansion to 512K.

Third-party developers decided that wasn't enough.

Tony DiStefano of Disto developed some of the most ambitious early CoCo 3 memory systems, eventually reaching 1MB and 2MB.

This wasn't simply a larger version of Tandy's memory board.

The GIME's normal memory-management architecture was designed around the CoCo 3's standard addressable memory. Making substantially more RAM available required additional circuitry and a different method of selecting it.

Disto demonstrated that an architectural limit wasn't necessarily a practical one.

If the existing hardware couldn't address what developers wanted, they changed the hardware around it.

A computer officially supported to 512K could now operate with megabytes of memory.

Another extraordinary upgrade came from an unexpected source.

Hitachi manufactured the 6309 as a highly compatible alternative to Motorola's 6809.

For years it was generally regarded as exactly that—a replacement processor.

Then developers discovered that the processor contained additional registers, instructions and a native operating mode beyond the documented 6809 architecture.

Suddenly the 6309 wasn't merely a replacement. It was an upgrade.

The next challenge was putting those hidden capabilities to practical use on the CoCo.

Chris Burke of Burke & Burke became one of the important early developers bringing those capabilities to Color Computer users through PowerBoost and associated OS-9 work.

The 6309 eventually became deeply associated with NitrOS-9, where software could take advantage of its additional capabilities.

Decades later, CMOS 63C09 processors remain desirable CoCo upgrades.

The CPU Tandy never officially put into a Color Computer became one of the platform's longest-lived performance enhancements.

Cloud-9 also continued pushing CoCo 3 memory development.

Its TRIAD SRAM systems replaced conventional DRAM-based expansion with modern static RAM.

And Cloud-9 did not stop at 512K.

A 2MB TRIAD SRAM configuration carried megabyte-class CoCo 3 expansion into a newer generation of hardware.

The progression had become remarkable:

128K stock CoCo 3 >> Tandy 512K >> third-party 1MB/2MB systems >> modern SRAM-based multi-megabyte systems.

More recently, Richard Lorbieski and BoysonTech continued that development with the Boomerang family.

Modern SRAM designs provide CoCo 3 memory expansion without depending upon banks of aging DRAM.

Newer designs have pushed the idea farther, combining expanded memory with other desirable modern upgrades, including the 63C09.

Hardware that once required several boards and early-1990s components can increasingly be recreated with smaller, cooler and more efficient modern circuitry.

Jim Brain of RETRO Innovations approached the memory problem differently.

Rather than simply constructing a larger RAM board, CocoMEM explores extensions to the Color Computer's memory-management architecture itself.

By operating at the GIME interface, the hardware can participate directly in memory-management operations and extend what the original architecture can do.

At this point, developers are no longer merely asking: "How much RAM can we attach to Tandy's computer?"

They're asking: "What if we redesign the way the Color Computer manages memory?"

All of this hardware development raises an obvious question:

What do you do with it?

One answer is exactly what computer enthusiasts have always done.

Write better software.

Commercial Color Computer development faded after Tandy left the market, but game development never completely disappeared.

More importantly, modern programmers eventually began targeting the expanded Color Computer rather than limiting themselves to configurations Tandy originally sold.

That represents a major change. The aftermarket stopped being merely a collection of optional upgrades. It began defining new target platforms for CoCo software.

The CoCo had already demonstrated how far an expanded machine could be pushed near the end of its commercial era.

In 1989, Dave Dies, working as Phantom Software, developed Sinistaar, published by Sundog Systems for the 512K CoCo 3.

It was inspired by Williams' Sinistar arcade game and provided CoCo 3 owners with an impressively ambitious interpretation of the concept, including digitized sound and speech.

It also established an interesting historical bookend. Decades later, programmers would return to Sinistar again.

One of the landmark post-commercial projects appeared in the 2000s.

Known throughout the community as Sock Master, the developer brought the original arcade Donkey Kong program to the 512K CoCo 3 by translating the arcade game's Z80 program for the 6809-based Color Computer.

That distinction matters.

This wasn't simply another Donkey Kong-inspired game. The project worked from the actual arcade program.

The result became one of the most memorable demonstrations of what a 512K CoCo 3 could accomplish long after commercial development had supposedly ended.

Glen Hewlett took the concept of translating arcade software considerably farther.

His work has involved studying original arcade programs and adapting their code and behavior to the Color Computer 3.

The challenge goes well beyond translating processor instructions. An arcade game may use a different CPU, completely different graphics hardware, different sound hardware and an entirely different memory architecture.

The original game logic can sometimes be preserved. Everything that communicates with the original hardware has to be understood and adapted.

Projects such as Pac-Man, Space Invaders, Defender and Robotron demonstrate just how successful that approach can be.

This isn't emulation. The Color Computer itself is executing software adapted to run on its own hardware.

Hewlett's Rick Dangerous pushes the concept in a different direction.

The game began as an effort to reverse-engineer the Amiga version. Hewlett disassembled portions of the original 68000 program and decoded graphics and game structures.

He subsequently discovered that XRick, an open-source C implementation of Rick Dangerous, provided a much more practical basis for the game logic.

The finished CoCo version therefore combines his Amiga reverse-engineering work with a conversion of XRick's C implementation into optimized CoCo 3 assembly rather than being a literal Amiga 68000 transcode.

What makes the project especially important to this story are its hardware requirements:

6309 processor.

2MB of RAM.

The completed version, released in August 2026, uses a 256×200 16-color display, two simultaneous audio channels during gameplay and three for music on the title and splash screens.

Think about what that means.

Tandy never sold a CoCo 3 / 6309 / 2MB computer.

Third-party developers effectively created that machine. And decades later, software developers began writing programs that require it.

The upgrade had become a platform.

Not all modern CoCo development is about converting existing games.

Nick Marentes has continued treating the Color Computer as a machine worth writing new games for.

His work stretches across decades and includes numerous titles taking advantage of expanded CoCo 3 configurations, including machines equipped with 512K and the 6309.

That represents an equally important form of continued development.

Preservation keeps old software running.

Development gives the computer something new to run.

And the story continues today.

The CoCo has already had one Sinistar-inspired game in Sundog's 1989 Sinistaar.

Now Vincent Tran is developing a new Sinistar project for the Color Computer.

That gives the machine a remarkable historical progression:

Williams Sinistar >> Sundog/Phantom Software Sinistaar >> Vincent Tran's modern Sinistar project.

The current project also reinforces something that has become increasingly obvious throughout this article.

The Color Computer isn't merely being displayed. It isn't merely being emulated. And it isn't merely being repaired.

People are still developing software for it.

Modern digital video didn't begin the CoCo's history of third-party video expansion.

Long before FPGA hardware and HDMI displays, developers were already finding ways around the limitations of the Color Computer's built-in display system.

One of the most interesting examples was the PBJ WordPak family.

The original PBJ WordPak and WordPak II provided an independent display system for text-intensive applications.

Instead of relying upon the CoCo's built-in VDG, the cartridge contained its own CRT-controller hardware and video memory.

Software support allowed the CoCo to use an 80-column display, making the machine considerably more practical for word processing, programming and other text-intensive applications.

The arrival of the CoCo 3 created a problem, however.

The original WordPak's $FF98 address conflicted with the CoCo 3's GIME registers.

A later WordPak-RS, sold through Radio Shack, moved the interface to $FF76, allowing it to coexist with the CoCo 3.

Decades later, the WordPak story resumed.

An original WordPak-RS obtained from Gene Heskett was reverse-engineered specifically so the rare hardware could be reproduced.

The board was traced and recreated in KiCad. Prototypes were manufactured and debugged. A replacement character-generator ROM was created, and new NitrOS-9 and Disk Extended Color BASIC 80-column drivers were written.

The resulting project deliberately recreated the original hardware and published information necessary for others to manufacture it.

That is preservation in a particularly pure form:

Find rare hardware. Understand it. Reproduce it. Document it. Make it possible for somebody else to reproduce it again.

But the project didn't end with copying the past.

The later WordPak 2+ explores what a new Color Computer video cartridge could become.

Built around the Yamaha V9958 with 128K of dedicated video RAM, it supports the CoCo 1, 2 and 3 and provides 32-, 40- and 80-column text modes, hardware sprites, horizontal and vertical scrolling, RGB output and substantially enhanced graphics modes.

Disk BASIC and OS-9/NitrOS-9 software support accompany the hardware.

The two projects together almost perfectly summarize the post-Tandy development philosophy:

Preserve what existed. Reproduce what is disappearing. Then continue developing it.

For the CoCo 1 and CoCo 2, video itself has become a practical preservation problem.

Those machines were designed around Motorola's MC6847 Video Display Generator and television technology of the late 1970s.

RF output made perfect sense when nearly every household had a television and dedicated computer monitors were expensive.

More than forty years later, RF is a liability.

Developers initially produced composite, RGB and other modifications. Eventually they began attacking the problem digitally.

CoCoVGA, developed through the work of Brendan Donahe and Ed Snider, represented a major change in approach.

Instead of merely converting an analog signal after the CoCo generated it, programmable hardware could observe the machine's digital video activity and generate a clean modern display.

That created an interesting preservation problem.

Many CoCo games intentionally exploited NTSC artifact colors to create colors that technically weren't present in the underlying bitmap.

A modern digital video system therefore has to reproduce not merely what the computer generated—but what an analog television did to it.

That is preservation at the level of behavior rather than simply electronics.

AC Lovell of AC's 8 Bit Zone took the concept considerably farther with CoCoDV.

Its immediate advantage for the CoCo 1 and CoCo 2 is modern digital video.

But calling CoCoDV merely a video adapter seriously understates what it does.

CoCoDV introduces graphics capabilities that the original VDG never possessed, including tile-based graphics, hardware sprites and smooth horizontal and vertical scrolling.

It also addresses artifact-color compatibility and expands the machine's audio possibilities.

The question is no longer merely: "How do we display a CoCo on a modern television?"

It becomes: "What could a CoCo 1 or 2 have done if development of its graphics and sound hardware had continued?"

The 6809 is still running the computer. The graphics system around it has entered another century.

The CoCo 3 presents a different problem.

Tandy replaced much of the CoCo 1 and 2's video and memory-support hardware with the custom GIME—Graphics Interrupt Memory Enhancer.

GIME provided the CoCo 3's enhanced graphics, 80-column text, interrupts and memory management.

It was a tremendous improvement.

It also created a future preservation problem.

GIME is proprietary silicon.

No semiconductor manufacturer is going to restart production of original Tandy GIME chips.

Traditionally, when a GIME failed, the replacement came from another CoCo 3. Eventually that becomes unsustainable.

So developers decided to build another GIME.

The GIME-X project was a collaboration between Gary Becker and Ed Snider.

Becker handled the programmable-logic/Verilog implementation. Snider designed and built the physical hardware.

The immediate preservation value is obvious. A dead GIME should not have to mean a dead CoCo 3.

But FPGA technology creates another possibility.

Once the GIME exists as programmable logic, why stop at duplicating it?

Additional video modes, expanded memory capabilities and faster operation can be explored while retaining compatibility with the original architecture.

The replacement can eventually become: A better GIME than the GIME.

Snider has since largely stepped away from active hardware development, but the substantial body of CoCo hardware he designed, produced or collaborated upon remains an important part of the platform's modern history.

GIME-Z, developed through AC's 8 Bit Zone, represents another modern programmable-logic approach to replacing the increasingly scarce CoCo 3 GIME.

Again, the objective isn't merely to convert the CoCo's video output. The custom logic itself is being recreated.

This completes an interesting progression:

Use Tandy's custom silicon. Expand Tandy's custom silicon. Recreate Tandy's custom silicon. And once the last step becomes possible: Improve it.

Video chips aren't the only proprietary components capable of killing a Color Computer when they fail.

This is where the work of Rocky Hill, known online as qbancoffee, becomes particularly important.

Rocky and qbancoffee are the same person—not two separate developers.

His work asks a fundamental preservation question:

How do we continue repairing these computers when proprietary parts no longer exist?

Rocky's Pepper replaces the Color Computer's custom SALT—Supply and Level Translator circuitry.

SALT handles several important functions, including supply regulation, RS-232 interfacing, level translation, cassette input and cassette-relay control.

Pepper recreates those functions with obtainable components.

That is not merely an upgrade. It's a way of preventing a failed proprietary component from turning an otherwise repairable CoCo into a parts machine.

Rocky's Daiquiri tackles another proprietary component used in later Color Computers, recreating functions associated with the machine's DAC, sound multiplexing and joystick circuitry.

Again, the objective is reproducibility.

A proprietary chip may disappear.

The knowledge required to reproduce its function doesn't have to disappear with it.

Rocky's preservation work goes considerably farther.

Complete Color Computer motherboards have been recreated in KiCad.

That means we've progressed from:

Repairing the CoCo >> Replacing obsolete CoCo circuitry >> Manufacturing another CoCo motherboard.

The limiting factor increasingly isn't the fiberglass circuit board. It's the remaining supply of proprietary components.

And other projects are attacking those components as well.

A heavily expanded Color Computer traditionally depended upon another piece of Radio Shack hardware: the Multi-Pak Interface.

The MPI turned the CoCo's single cartridge connector into four slots, allowing combinations of disk controllers, serial interfaces, sound hardware, ROM Paks and other devices to remain connected simultaneously.

Unfortunately, original Multi-Paks are now vintage hardware themselves.

Modern developers approached the problem in different ways.

Ed Snider's mini-MPI and MEGA mini MPI reinterpreted the concept using modern hardware and additional capabilities.

Other developers chose another route: Reproduce the original Multi-Pak architecture.

Open-hardware reproductions mean the basic design can survive even when the person currently manufacturing a board decides to stop.

Someone else can obtain the design. Someone else can manufacture it. Someone else can repair it. Someone else can improve it.

That's fundamentally different from the way third-party hardware usually disappeared during the 1980s.

The Color Computer's built-in sound hardware was simple, but that didn't prevent third-party developers from turning it into a surprisingly capable music computer.

Speech Systems became particularly important.

Among its more ambitious products was the Symphony-12, a twelve-voice synthesizer.

Software turned the combination into something far more sophisticated than a computer that beeped through a television speaker.

But like so much specialized CoCo hardware, original Symphony-12 units eventually became difficult to obtain.

Decades later, Jim Brain of RETRO Innovations revisited the concept with the Philharmonic-12.

Philharmonic-12 recreates the essential architecture of the original hardware while maintaining compatibility with existing software and surviving music files.

That's important. The software and music already exist.

A modern reproduction gives them hardware on which to live again.

I've assembled and used a Philharmonic-12 myself, using original Symphony software and archived MUS music files.

That experience demonstrates something easily lost when talking only about circuit boards:

Hardware preservation can also preserve creative work.

Speech Systems and Rulaford Research also produced MIDI hardware and software that allowed the Color Computer to communicate with external synthesizers.

Lester Hands wrote much of the software associated with that ecosystem.

Programs such as Lyra and other CoCo MIDI applications turned the computer into a practical composition and sequencing machine.

As the original interfaces became scarce, compatible replacement hardware appeared.

Jim Brain's MIDI Maestro provides a modern MIDI interface while deliberately retaining compatibility with existing CoCo MIDI software.

I've used MIDI Maestro with Lyra and an external Casio synthesizer as part of my own SepTandy project.

The CoCo involved in that project contained a 6309 processor, CoCoDV, CoCoSDC and newly manufactured MIDI hardware while running software rooted in the original Speech Systems/Rulaford era.

That one machine almost summarizes this entire article.

Technology from several generations was working together instead of requiring everything that came before it to be discarded.

The logical endpoint of much of this development is implementation of the Color Computer itself in programmable logic.

Gary Becker's CoCo3FPGA demonstrated that the architecture could be recreated inside an FPGA and then extended.

Roger Taylor also moved into FPGA development with his CoCo-on-a-chip work.

This creates an interesting philosophical question:

How much can you change before a CoCo stops being a CoCo?

There probably isn't one correct answer.

From a preservation standpoint, however, something extremely important has happened.

The Color Computer no longer exists solely inside pieces of silicon manufactured forty years ago.

Its architecture can be described in programmable logic. It can be studied. It can be recreated. It can be modified. And it can be carried forward.

Perhaps the most important change in modern CoCo development isn't any particular processor, memory board or video system.

It's how the hardware itself is being preserved.

During the commercial era, a product could effectively disappear when its manufacturer closed.

Schematics might be unavailable. PAL equations could be lost. PCB artwork might exist only as film stored in a filing cabinet that eventually went into a dumpster.

Modern development can work very differently.

Schematics can be stored online. KiCad projects can generate new PCBs. Gerber files can be downloaded. Firmware can be archived. Programmable-logic designs can be preserved alongside software.

Rocky Hill's recreated motherboards can be manufactured again.

The WordPak-RS recreation provides manufacturing information so builders don't have to depend upon one person continuing to sell the board.

Multi-Pak designs can survive their current manufacturer.

Modern MIDI hardware can be documented and reproduced.

This gives the Color Computer something the original machine never really had:

A path toward hardware preservation through reproducibility.

No article could realistically identify every person who contributed hardware, software, documentation, manufacturing, testing and preservation work to the Color Computer.

That would be another article by itself.

But some names appear repeatedly throughout this story: Tony DiStefano, Chris Burke, Mark Marlette, Boisy Pitre, Roger Taylor, Ian Mavric, Darren Atkinson, Tim Lindner, Brendan Donahe, Ed Snider, Gary Becker, AC Lovell, Rocky Hill, Jim Brain, Richard Lorbieski, Lester Hands, Gene Heskett, Sock Master, Glen Hewlett, Nick Marentes, Dave Dies and Vincent Tran, along with many others whose contributions deserve recognition.

Some designed hardware. Some wrote operating systems and drivers. Some wrote games. Some reverse-engineered obsolete hardware. Some manufactured boards. Some documented systems. Some archived software. Some organized gatherings. And many did several of those things.

The Color Computer survived because those contributions overlapped.

A hardware designer could build an interface because documentation survived. A programmer could support that interface because hardware was available. An archivist could preserve the old software. Another developer could build replacement hardware compatible with it.

And somebody could take all of that work and create something none of the previous developers had imagined.

Currently, if you go to the CoCo fan pages on Discord or other social media, you'll find numerous others discussing hardware and game creation today. Grant, Paul and a few others on Discord are reglars discussing games. Every Thursday night, the CoCo Nation does the Game On Challenge - both new and old games are selected. I am truly sorry for leaving names out - those I have, I will try to add back in later. My point is, game development hasn't slowed much. What has changed, is game quality. Nick's recent pinball game, The 8-bit guy release Quarx for ALL of the CoC models (I think), Grantb (Discord) has released several new games recently that are great playing games.

Perhaps the secret wasn't simply that people kept developing new Color Computer hardware. It was that so many of them respected what had already been created. A modern storage controller tries to run yesterday's disk software. A replacement floppy controller lets yesterday's physical disks keep spinning. A replacement sound board plays yesterday's music files. A newly manufactured MIDI interface runs software written for hardware that vanished decades ago. A recreated WordPak runs software descended from the original WordPak environment. Modern Multi-Paks accept old cartridges. A replacement GIME first tries to behave like the original GIME before extending it.

And modern development environments still produce software for the same processor architecture.

Even the modern games demonstrate the same continuity. A 6309 and two megabytes of modern SRAM don't make the machine cease to be a CoCo. They give programmers a bigger CoCo to write for. That's an important distinction. While modern upgrades were once considered almost heretical by some, today the community rarely treats modernization as an excuse to discard the past. Instead, each generation of hardware gave the previous generation another chance to keep working—and the next generation of programmers something new to explore. The only real concern among CoCoNutz today is whether some of this new hardware will ever see software to fully utilize the new features.

There is something remarkable about the Color Computer's longevity.

The original CoCo wasn't designed to be an exotic engineering workstation. It was an affordable home computer that connected to a television, stored programs on cassette tape, ran at less than one megahertz, and had graphics hardware whose limitations programmers spent years learning to exploit.

Perhaps that relative simplicity is one reason the Color Computer survived. People could understand it—and once they understood it, they changed it.

When Radio Shack didn't offer the hard drive they wanted, somebody built one. When 512K wasn't enough, somebody figured out how to install megabytes. When programmers discovered the greater capabilities of Hitachi's 6309 processor, they rewrote software to take advantage of it. When developing directly on the original machines became cumbersome, modern cross-assemblers and development environments took their place.

As original hardware aged, the community simply kept solving problems. SD-card interfaces replaced unreliable mechanical disks, while new floppy controllers were created for people who still wanted to use the real thing. Rare 80-column hardware was reverse-engineered and recreated, then followed by more capable successors. Scarce music and MIDI cartridges gained modern replacements.

When modern televisions could no longer display the CoCo's original video, developers created new digital video solutions—and then went further, adding graphics capabilities the original machines never had. As custom Tandy components began to fail, replacements were designed. Eventually, even entire motherboards were recreated.

And when programmers got their hands on all this expanded hardware, they began writing software that actually needed it.

A 512K CoCo became a target machine. Then the 6309 became a target processor. Today, a game such as Glen Hewlett's Rick Dangerous can specify a 2MB, 6309-equipped CoCo 3 as its minimum system—a Color Computer configuration Tandy never manufactured.

Meanwhile, programmers continue producing new software for the platform. Vincent Tran is developing Sinistar while this article is being written.

FPGA technology has pushed things even further. Once it became practical to recreate the Color Computer's custom logic, the question changed from "Can we make another one?" to "Can we make a better one?"

Radio Shack and Tandy created the Color Computer, but its development didn't end when they stopped manufacturing it. For more than three decades, engineers, programmers, small businesses, user groups and individual enthusiasts have continued the work.

That is why a computer introduced in 1980 can sit on a desk today with a 6309 processor, megabytes of RAM, SD-card storage, digital video, hardware sprites, MIDI hardware and newly manufactured replacement circuitry—and still run software written when the machine was new.

Perhaps the annual "Last" Chicago CoCoFEST! is the perfect symbol for the whole story. For more than three decades, the CoCo community has been holding its "Last" CoCoFEST! while somebody, somewhere, is designing the next piece of hardware or writing the next piece of software.

Tandy stopped building the CoCo.

The community never stopped developing it.