Buffer Gate Explained: Signal Restoration and the Relay That Spanned a Continent

In the previous post in this series, we built the NOT gate — a relay wired to invert its input. Energize the coil, and the output goes off; release it, and the output comes on. NOT is powerful precisely because it opposes: it turns truth into falsehood, presence into absence, 1 into 0.

Now consider the opposite idea. What if the relay just agreed with its input? Energize the coil — output on. Release — output off. Output faithfully mirrors input, all the way down the line. The gate that does this is called the buffer, and in pure logic it is the identity function: $Out = A$, always, without exception. A gate that changes nothing.

So why build it? Why name it? Why include it in a series about building a computer?

Because “changes nothing logically” and “does nothing physically” are very different statements — and understanding that gap is how real machines stay reliable across distances, across loads, and across time.

A normally-open contact, wired straight through

To build a buffer from a relay, you need one wire, one relay, and the right choice of contact.

In earlier posts, we’ve used both of the relay’s contact types. Normally closed (NC) was the NOT gate: the contact bridges the output circuit at rest, and the energized coil pulls it open. Normally open (NO) is the other personality: the contact sits open at rest, and the energized armature closes it.

Wire the NO contact into your output circuit and the relay becomes a buffer:

• Coil off (input = 0): no magnetism, the armature rests in its spring-held position, the NO contact remains open, the output circuit is broken — output is 0.
• Coil on (input = 1): the electromagnet overcomes the spring, the armature swings down, the NO contact bridges the output circuit — output is 1.

Output equals input. The full truth table occupies exactly two rows:

Compare this with the NOT gate’s two-row table and notice the difference: NOT has 0→1 and 1→0; the buffer has 0→0 and 1→1. They use the same relay, the same coil, the same armature. The only choice is which contact pin you connect your output wire to. One terminal inverts. The other restores. That single wiring decision is the entire difference between a NOT gate and a buffer.

The fact that changes everything

Here is the physical detail that turns the buffer from a curiosity into an essential building block: the coil circuit and the contact circuit carry completely separate currents.

A tiny, feeble signal through the coil — barely enough to tug the armature — throws the switch in the contact circuit. And the contact circuit is connected to its own local supply, quite independent of where the input came from. The output lamp does not draw from the input source. It draws from the local battery. The input signal’s only job is to move the armature. Once the armature moves, the local supply takes over and drives the output at full strength.

This means the buffer is not merely copying a signal. It is rebuilding it. The output is not an echo of the input; it is a brand-new full-strength pulse, regenerated at the point of reception. A weak, degraded, half-exhausted input energizes the coil, the armature snaps down, and a fresh robust output is born on the other side of the switch.

That is the trick. And it is, almost exactly, the trick that made the telegraph work — decades before anyone used the word “logic gate.”

A continent wired by repeaters: 1861

On October 24, 1861, the first transcontinental telegraph line sent its inaugural message from Sacramento to Washington, D.C. The line spanned roughly 3,700 kilometers — through the Rocky Mountains, across the Nevada desert, over the Sierra Nevada — and it worked because of relay repeater stations placed at intervals along the way.

A telegraph signal sent from St. Louis would fade into inaudibility after a few hundred miles. The copper wire leaked. The resistance accumulated. By the time the pulse reached a relay station, it was a whisper — too weak to actuate anything downstream. But that whisper was just strong enough to energize a relay coil. The armature pulled down, closed a normally-open contact, and a fresh battery at the relay station launched a new full-strength pulse toward the next leg of the journey. That new pulse carried on until it too faded, and the next station rebuilt it again.

The message traveled coast to coast through twenty or so relay hops, each one performing the same invisible act: weak input energizes coil, coil closes contact, contact releases fresh local power as output. The identity function, implemented in copper and iron, scaled to continental distances. Within days of the line’s completion, the Pony Express — which had covered the same route in ten days on horseback — was out of business.

If you have been with this series since the beginning, you will recognize that moment. In Post 1, we met Joseph Henry and his relay repeater, the foundational idea behind the whole device: use the faint arriving signal not to do the work, but to throw a switch, and let a fresh local battery carry the signal onward. That was the invention of the relay. The buffer gate is that same idea, reduced to a truth table. We have come full circle.

Three reasons to build a gate that copies

Armed with that history, the three engineering reasons for the buffer become concrete rather than abstract.

1. Signal restoration. A real signal is never perfectly strong at every point in a circuit. Wire has resistance. Connections degrade. Long chains of gates accumulate loss. The buffer inserts a point of regeneration: no matter how weak the signal arriving at the coil, the signal leaving the contact is sourced from a local supply and is full strength. Every repeater station on the transcontinental line was doing this. Modern digital chips do it with driver circuits and repeaters at the same scale, for the same reason.

2. Fan-out. An output driving multiple downstream coils must supply current to all of them simultaneously. As more loads attach to a single source, the voltage droops, timing deteriorates, and gates near the end of the chain may fail to switch cleanly. A buffer interposes: the upstream signal drives one buffer coil, and the buffer’s contact circuit — backed by its own supply — drives all the downstream loads at full strength. The signal source sees only one load. The downstream stage sees only full-strength output. One becomes many without degradation.

3. Isolation and timing. Because the coil circuit and the contact circuit are electrically separate, a buffer is also a circuit isolator. Noise, transients, and fault conditions on the output side cannot propagate backward into the input. The two circuits share only the mechanical link of the armature — and that mechanical link introduces a small, predictable transit time. The armature physically swings through its arc, and that tiny delay is knowable and stable. In a relay machine built from many cascaded stages, that deterministic per-stage delay becomes a timing resource: engineers who knew the relay’s actuation time could predict, to the millisecond, how long a chain of twenty relay buffers would take to propagate a signal from one end to the other.

Try it yourself

Below is a live relay wired as a buffer: the input switch energizes the coil through the coil circuit, and the normally-open contact closes to light the output lamp through a separate circuit. Before you touch anything, state your prediction: does the lamp come on when you close the switch, or does it go off?

Three things to notice as you play:

1. Hold the buffer and NOT gate side by side. The NOT gate’s lamp is on when the input is off; this lamp is off when the input is off. They are mirror images — same relay, same coil, different contact terminal. The entire difference between the gate that inverts and the gate that restores is one wire’s worth of choice.

2. Imagine it as a repeater station. Pretend the input switch represents a telegraph signal that has traveled five hundred miles and arrived exhausted. The coil barely needs that whisper — the armature doesn’t care how tired the signal is, only whether there is enough current to pull it. Once it snaps down, the local contact circuit drives the lamp (the outgoing signal) at full strength. You are running a simplified version of the 1861 transcontinental telegraph, in miniature, in your browser.

3. Turn the sound on and count the clicks. One click when the coil energizes, one when it releases. In a relay machine with a dozen buffer stages in series, you could hear the signal propagating forward as a rolling wave of clicks — each one a point of regeneration along the path.

The unsung hero of large machines

In a logic textbook, the buffer gets perhaps half a page. It is the simplest possible gate: $Out = A$. What could there be to say?

Quite a lot, it turns out — but only once you stop thinking about gates as abstract symbols and start thinking about them as physical devices that carry real current and live inside real machines. Charles Petzold makes this point throughout CODE: the power of abstraction is that it lets you think clearly at higher levels, but the cost is that the physical reality underneath disappears from view. The buffer is where that hiding is most instructive to undo.

In the relay computers of the 1940s — machines like the Harvard Mark I, which contained some 3,304 relays — buffers were present at every stage boundary, at every point where a signal crossed from one power domain to another, and at every juncture where a single output had to drive multiple downstream coils. Without them, the signal integrity that makes cascaded logic reliable would collapse. A 1 would sometimes arrive as a borderline “maybe 1” at the fifth gate in a chain; a “maybe 1” would produce the wrong output; the wrong output would propagate forward and corrupt every subsequent computation.

The NOT gate makes the smart decision — invert. The AND gate enforces a condition — both. The buffer does neither of those glamorous things. It simply ensures that the decision made two gates back arrives correctly at the gate two stages ahead, at full strength, on time, uncontaminated. That unglamorous job is what keeps a computer working through thousands of cascaded operations.

What you just learned — and what’s next

From one relay, wired through its normally-open contact:

• The buffer is the identity gate. $Out = A$. True copies true; false copies false.
• The output is not an echo but a regeneration. Fresh current from the local contact circuit makes the buffer a signal restorer as much as a signal copier.
• Three practical functions: (1) signal restoration — a weak input produces a full-strength output; (2) fan-out — one input drives many loads via a buffered local supply; (3) isolation and timing — electrical separation of input and output, plus a deterministic actuation delay.
• Historical grounding: the 1861 transcontinental telegraph worked by chaining exactly this circuit across a continent — buffering and regenerating at every station along the way. The buffer gate is the logical distillation of that engineering insight.

In Post 8 of Building a Computer from Relays, we will meet the gate that surprises everyone who encounters it for the first time: the NAND gate. NAND is sometimes called the “universal” gate — from NAND alone, in sufficient combinations, you can build AND, OR, NOT, and even the buffer itself. Every logic function that exists is derivable from this one configuration of one relay. Understanding why that is true, and what it means for building a computer, is one of the most satisfying moments in all of digital logic.

Ready to keep building? Every circuit in this series lives in the DigiSim Relay Lab — open the buffer next to the NOT gate, wire them in series, and listen to the relay tell you what it’s doing.