Ken Shirriff Powers an IBM 604 Thyratron Module

A thyratron-powered pluggable module from the 1948 IBM 604 Electronic Calculator has been brought back to life by Ken Shirriff, offering a vivid look at how modular vacuum-tube computing once worked—and why that design philosophy still shows up decades later.

A pluggable module sitting in the wrong decade is usually just museum clutter. This one, however, has been turned on.

In an early slice of computing history—the transition period between 1940s electromechanical accounting systems and the arrival of fully digital computers—the IBM 604 Electronic Calculator arrived with a clear promise: accounting. calculating. and engineering tasks. It wasn’t built for the kind of complex instruction chains people now expect from computers. Loops and advanced features weren’t its thing. But it did rely on a modular architecture that made the machine easier to maintain, expand, and repair.

Recently, Ken Shirriff took a close look at one of those swappable pieces and powered it up.

What makes the IBM 604’s setup feel surprisingly modern is its emphasis on plug-and-play functionality. The design used modular units that could be swapped in and out. each carrying a vacuum tube and its associated components. The point wasn’t elegance—it was practicality. Modules or cards with specific functions could be combined in different ways to increase flexibility. keep costs down. and make repairs straightforward.

According to the account of the system, the IBM 604 used a total of about 1,250 vacuum tubes, and all of them were found on these pluggable modules.

The module Shirriff worked with contains a thyratron tube. The thyratron. described here as a high-current switch and rectifier. is the kind of component you can almost feel in your hands: it’s meant to change states under power. In a short demonstration video, the thyratron is shown switching a lamp on and off. The accompanying explanation details what’s happening inside the module as it does its job.

The IBM 604 itself may have had limits. Its functionality was “rather limited,” and it was constrained by punch card input and output speed. Even so, it was a commercial hit, with over 5,600 units produced.

The story doesn’t stop at vacuum tubes, either. A transistorized prototype version followed in 1954, built with 2,200 transistors and using 95% less power. That prototype became the foundation for the IBM 608, described as the world’s first commercial all-transistorized calculator.

But the semiconductor era moved so fast that the IBM 608 didn’t stay in place for long. By the time a newly released product reached the market, the pace of semiconductor technology meant it was already on the way to obsolescence.

Still, the modular ideas behind the IBM 604 didn’t disappear. The same fundamental approach—swapping modules to extend what a system can do—kept reappearing across decades. From the 1950s Bendix G-15 to modern PCs with pluggable RAM and expansion cards. and onward to mainframes where hot swapping even entire CPU modules is treated as a feature rather than a novelty. the through-line remains unmistakable.

And in the simplest sense, that’s what makes Shirriff’s powered-up test resonate: it’s not just electronics running again. It’s a glimpse of a design philosophy—built to be changed—that still defines how people expect machines to behave.

IBM 604 Ken Shirriff thyratron vacuum tubes modular computing IBM 608 transistorized calculator pluggable modules Bendix G-15 hot swapping expansion cards