By the middle of the 1970s, linear regulators were already everywhere. Fixed-output parts like the 7805 had a simpler power design than discrete-transistor regulators, but they still locked designers into predefined voltages and encouraged one regulator per rail. 

National Semiconductor took a different approach. Instead of selling another fixed-output device, it released the LM317, an adjustable positive regulator that could deliver more than 1.5 A and set any output voltage from roughly 1.25 V up to 37 V using two resistors.

An LM317 in operation with a heat sink. Image used courtesy of Gelpgim22 via Wikimedia Commons (CC BY-SA 4.0)

Backed by the idea that microprocessors, memory, and analog ICs were proliferating, and systems increasingly needed odd voltages that did not map neatly to standard regulator SKUs, the LM317 arrived as a general-purpose solution at a moment when engineers wanted fewer part numbers, not more.

A Three-Terminal Regulator That Floats

What set the LM317 apart was not its pass transistor or its current rating, but how it defined regulation. Instead of regulating its output relative to ground, the LM317 regulated the voltage between its output pin and the adjust pin. Internally, the device maintained a reference of about 1.25 V across those two terminals. Everything else followed from that choice.

With a resistor from output to adjust and another from adjust to ground, the regulator forced a constant current through the divider. As a result, the output voltage became a function of the resistor ratio rather than a fixed internal setpoint. National’s application literature quickly standardized on a 240-ohm resistor from the output to adjust, ensuring a few milliamps of programming current and keeping the device in regulation even at light loads.

Schematic of the LM317. Image used courtesy of Peter S. via Wikimedia Commons (CC BY-SA 3.0)

This floating architecture had two important consequences. First, it allowed a single device to cover a wide voltage range without internal trimming for each value. Second, it meant the LM317 did not care where ground was, only that the input-to-output differential stayed within its limits. Designers could regulate high-voltage supplies just as easily as low-voltage ones, provided the device itself never saw more than its rated differential.

Internally, the LM317 used an op amp driving a Darlington pass transistor, along with current limiting, thermal shutdown, and safe operating area protection. These safeguards were not an afterthought. National explicitly designed the part so that protection remained active even if the adjust pin was accidentally disconnected, a critical detail for field service and lab environments.

Predictable Performance, Useful Quirks

In use, the LM317 behaved like a well-mannered linear regulator. Typical line regulation was on the order of 0.01% per volt, and load regulation around 0.1%. Ripple rejection was respectable on its own and could be pushed much higher by bypassing the adjust pin with a capacitor. National’s own examples showed ripple rejection approaching 80 dB with a modest electrolytic.

The LM317’s pin configurations. Image used courtesy of Texas Instruments

The device also had many quirks. Because its internal bias current was sourced from the output rather than ground, the LM317 required a minimum load current to stay in regulation. Early documentation often cited 10 mA as a worst-case requirement. In practice, this meant that ultra-light loads could cause the output to drift unless a bleeder resistor was added.

Another practical quirk was protection diodes. Large output capacitors, especially when paired with high input voltages, could dump charge back through internal structures during power-down, so National’s application notes recommended simple diodes to steer that current safely. This advice has been repeated in datasheets ever since. 

None of this slowed adoption by engineers; once understood, the device was predictable, forgiving, and hard to kill.

From Bench Supplies to System Rails

The LM317’s legacy stems from its widespread use. In lab power supplies, it enabled adjustable outputs with minimal circuitry, sometimes paired with a small negative rail to allow adjustment down to 0 V. In battery chargers, it doubled as a constant-current source by placing a resistor between output and adjust and taking the load from the adjust pin. In audio and instrumentation gear, it provided clean rails without the switching noise that early DC-DC converters could introduce.

Just as importantly, it changed how engineers thought about regulators. Voltage selection stopped being a catalog decision and became a schematic one, an approach that was carried forward into later adjustable regulators and even into modern switching controllers, many of which still use a 1.2-V or 1.25-V internal reference and an external divider to set output voltage. 

Today, despite its legacy, the LM317 feels almost invisible. It is not fast, not efficient by modern standards, and not especially compact compared with integrated switching regulators. Yet it remains in production decades after its introduction, fabricated by multiple manufacturers long after National Semiconductor itself disappeared.

Its endurance comes from the same place as other long-lived analog parts. The LM317 solved a real problem cleanly, established a pattern engineers could rely on, and stayed close enough to ideal behavior that its limitations were easy to manage. It did not try to be clever; it succeeded in being useful.