Two trackballs, and the Atari Trak-Ball that turned out to be a mouse

A trackball — like the Atari Trak-Ball and the Wico Command Control on my bench — is really just an upside-down mouse with a big heavy ball, and the idea predates home gaming by decades. The first one was a British radar device — Ralph Benjamin's "roller ball," built in 1946 for plotting targets — and a 1952 Canadian system actually used a five-pin bowling ball as the roller. By 1978 the idea reached the arcade in Atari's Football, and a few years later it was the soul of Centipede. Naturally, Atari wanted one in the living room.

Two of them ended up on my bench: an Atari Trak-Ball and a Wico Command Control Trackball. They plug into the same nine-pin Atari port and roll the same way, but they're different beasts.

The Atari Trak-Ball

Atari's was designed by Dan Kramer, who started by retooling real arcade trackballs as a hobby and went on to build Atari's whole Trak-Ball line — including the Atari 5200 Pro-Line version, still regarded as one of the best-built home controllers ever made. The hardware was genuinely good.

The catch — what Scott Stilphen fairly called the trackball's "failed promise," in his Atari Compendium write-up — was software. A true trackball needs a game written to read it, and only a handful did: Centipede, Missile Command, Crystal Castles, a few others. In those, it was magic. In everything else, the Trak-Ball fell back to a joystick-emulation mode and rolled like a mushy stick. The earliest CX22 didn't even have the mode switch — the case has the hole and the markings, but the board doesn't support it; Atari only made about 15,000 of those single-mode units before the later CX22 and the CX80 added a working joystick/trackball switch on the back.

The Wico

The Wico was the third-party rival — an arcade-grade mechanism with a bigger, smoother ball and the heft to match, cross-compatible across Atari and Commodore. Plenty of people preferred its feel to Atari's. Even Kramer had a (begrudging) opinion of it: he called the Wico trackball "a current vampire" — his way of saying it pulls a lot of power.

And the Wico earned that arcade-grade feel honestly. Wico Corporation was a Niles, Illinois outfit — practically up the road from this workshop — that spent decades as the coin-op industry's go-to for replacement parts, and by the late 1970s was making the arcade joysticks that ended up in cabinets everywhere. Its Command Control home line, this trackball included, carried that arcade DNA straight into the living room. Which is probably why, of the two on my bench, I reach for the Wico every time. It just feels right.

The two trackballs on the bench: Atari's wedge-shaped Trak-Ball (left) and the Niles-built Wico Command Control (right).

Opening it up

The one that wouldn't work, though, was the Atari — my CX22. Dead on both the 2600 and the 8-bit: no movement, no buttons, nothing. So I took it apart expecting a snapped wire or a tired encoder. Instead I found a tidy little mystery.

Inside the genuine Atari board — the silkscreen still reads CX22 ASSY · Atari © 1983 — someone had run a small loom of bodge wires straight onto the µPC339C, the quad comparator that cleans the optical encoder's pulses into usable quadrature. They'd tapped the trackball's X and Y signals and the buttons right off that chip and rerouted them out to the cable in the Atari ST's mouse pinout. And they weren't shy about it: flip the case over and the original Atari label is hand-marked in ballpoint — "ST MOUSE BALL — 5-11-87." Somebody turned this trackball into an ST mouse in the spring of 1987 and wrote the date on it.

There's a nice irony in that: Atari's own trackball, quietly repurposed into a pointer for Atari's own later computer — and clearly a daily driver for whoever did the deed.

Inside the CX22 — the steel roller and the genuine Atari board, with a previous owner's bodge-wire loom running off to the side.

The smoking gun: wires soldered straight onto the µPC339C comparator to tap the raw quadrature.

Flip it over and the date's right there — "ST MOUSE BALL, 5-11-87."

The Atari ST mouse pinout — and what got lifted

The Atari ST reads its mouse on the keyboard's joystick port as raw quadrature — the encoders' two-phase pulse trains, straight, with no processing. Here's the full nine-pin map, and where each line came from inside the rewired CX22:

Here's why those four "tapped at the comparator" lines are the whole trick. A trackball and an ST mouse begin life identically: two optical encoders, one per axis, each throwing two pulse trains ninety degrees out of phase — quadrature. In the CX22 those raw pulses get squared up by the µPC339C comparator, and then the cart's own downstream logic converts them into the "direction-and-motion" format Atari's trackball-aware games expect on pins 1–4 (or, flip the switch, into plain joystick up/down/left/right).

The ST wants none of that processing. It wants the raw quadrature. So the previous owner reached in before the conversion logic and soldered onto the four comparator outputs directly — X0, X1, Y0, Y1 — and ran them straight to pins 1–4 in the ST's order, with the buttons on 6 and 9 and power and ground on 7 and 8. That little loom of wires on the chip is the entire hack.

If you want to trace it down to the silicon: the µPC339C is a garden-variety quad comparator — NEC's member of the LM339 family — in a 14-pin DIP. Each of its four comparators squares up one encoder phase, so its four outputs — pins 1, 2, 13, and 14 — are the clean quadrature lines (X0, X1, Y0, Y1). That's exactly the set of pins the bodge wires land on.

And that's exactly why it was dead on my 2600 and my 8-bit: pins 1–4 were now handing those machines raw ST-style quadrature instead of the direction-and-motion (or joystick) signals they were looking for. Same connector, same ball, a completely different language.

From ball to plug: the rewire taps the comparator's raw quadrature and routes it to the Atari ST mouse pinout.

The µPC339C is a 14-pin quad comparator; its four outputs — pins 1, 2, 13, 14 — are the quadrature the hack lifts.

What now

It's a clean little hack, and honestly a fun one — a CX22 makes a surprisingly good ST pointer. But I want my trackball back, so I'll wire it back to Atari Trak-Ball spec and put it in front of Missile Command where it belongs. The ST-mouse wiring goes in my notes, in case I ever want to send it back the other way.

Lesson for the bench: when a vintage controller "doesn't work," check what it's actually wired to do before you start chasing faults. Sometimes the last owner already finished a project on it — just not the one you were expecting.

Built and bench-tested at the CRT workshop on real Atari hardware.

Frequently asked

What's the difference between the Atari Trak-Ball and the Wico Command Control trackball?
Both plug into the nine-pin Atari port and roll the same way, but the Atari Trak-Ball (designed by Dan Kramer) and the Wico Command Control differ in feel—the Wico is an arcade-grade mechanism with a bigger, smoother ball, cross-compatible across Atari and Commodore. Wico Corporation was a Niles, Illinois outfit that had long supplied the coin-op industry.

Why did the Atari Trak-Ball feel like a mushy joystick in most games?
A true trackball needs a game written to read it, and only a handful were—Centipede, Missile Command, Crystal Castles, and a few others. In everything else the Trak-Ball fell back to a joystick-emulation mode. The earliest CX22 (about 15,000 units) didn't even have a working mode switch; the later CX22 and the CX80 added one.

How was a CX22 Trak-Ball rewired into an Atari ST mouse?
Someone soldered a small loom of wires onto the CX22's µPC339C quad comparator to tap the raw X0, X1, Y0, and Y1 quadrature before the cart's conversion logic, then routed those to pins 1–4 in the Atari ST mouse order, with the buttons on pins 6 and 9 and power and ground on 7 and 8. The ST wants raw quadrature, which is why the modified unit worked as an ST mouse but was dead on the 2600 and 8-bit.

What does the µPC339C do in the CX22 Trak-Ball?
The µPC339C is a 14-pin quad comparator (NEC's member of the LM339 family) that squares up the optical encoders' pulses into clean quadrature; its four outputs, on pins 1, 2, 13, and 14, are the X0, X1, Y0, and Y1 lines the ST-mouse rewire taps.

Further reading

• The Atari Compendium — "The Failed Promise of the VCS/2600 Trak-Ball Controller" (Scott Stilphen)
• ANTIC Interview 403 — Dan Kramer, Atari Trak-Ball Controllers
• Atari joystick port — Wikipedia (the DE-9 standard, including ST mouse mode)
• Wico Corporation: A Historical Primer (Museum of the Game)
• NEC µPC339C datasheet — quad comparator, 14-pin DIP