The Feedback Loop: How the Driver Knows Its Own Output Voltage

A constant voltage driver must hold its output at 24V regardless of input variation, load change, or temperature drift. It does this automatically, without user adjustment, because it contains a closed feedback loop: a circuit that measures the output, compares it to a reference, and adjusts the switching duty cycle to correct any error. The feedback loop is the heart of the driver's regulation, and its design quality shows up in the specs buyers actually read — line regulation, load regulation, output voltage tolerance, and transient response.

The loop has three stages. First, a precision reference sets the target. In small flyback drivers this is usually the TL431, an integrated shunt reference that produces an accurate 2.5V reference and an amplifier, configured with two resistors to divide the 24V output down to 2.5V at the reference input. If the output rises to 24.2V, the divided voltage exceeds 2.5V, and the TL431's output pulls the feedback signal toward the off state. If the output falls to 23.8V, the opposite happens. The divider ratio is what sets the output voltage: change the resistor values, and the output changes. This is how a 24V platform is reconfigured for 12V or 36V variants — not by redesigning the transformer, but by changing the feedback divider.

Second, the error must cross the isolation barrier. The primary side (mains) and secondary side (SELV) must remain galvanically isolated — no direct wire can carry the feedback signal across. An optocoupler bridges the barrier: an LED on the secondary side shines on a phototransistor on the primary side. The TL431 modulates the LED current; the phototransistor reads it on the primary side and adjusts the duty cycle of the switch. The optocoupler carries the error signal across the isolation while maintaining the safety barrier. Its current transfer ratio — how much secondary current translates to primary current — is itself a source of drift over temperature and age, which is why the loop is designed to tolerate a range of optocoupler characteristics.

Third, the primary controller chip reads the feedback signal and adjusts the switch on-time. The chip runs the switching converter at a fixed frequency (or in quasi-resonant mode), and the feedback voltage sets the duty cycle. A stronger feedback signal means the chip reduces the duty cycle, delivering less energy to the output; a weaker signal means it increases the duty cycle. The whole loop runs at the switching frequency — every 15 microseconds at 65kHz — continuously correcting the output.

The design challenge is loop stability. A feedback loop that corrects too slowly produces sloppy regulation and a drooping output under load changes. A loop that corrects too aggressively overshoots and oscillates — the output hunts up and down, and the loop rings. The compensation network (a few passive components around the TL431 and the optocoupler) is tuned so that the loop's gain and phase produce stable regulation across the load range. A well-compensated loop gives tight line and load regulation (typically ±2 percent), clean transient response when the load changes, and no audible or oscillating output. A poorly compensated loop shows up in practice as output voltage drift, output ripple that changes with load, or audible noise at certain operating points.

For a buyer, the output of the loop is visible on the datasheet. Line regulation says how much the output changes as the input sweeps across its range (200V to 264V); load regulation says how much it changes as the load goes from no load to full load. Tight numbers — ±1 or ±2 percent — indicate a properly compensated loop. A driver with ±5 percent regulation, or one whose output voltage varies visibly as the load changes, has a loop designed to a looser standard. The feedback loop is invisible inside the plastic case, but its performance is printed on the nameplate — and that is the spec the buyer can check.