Concepts
Terms you meet on every schematic but rarely see explained: bootstrap, snubber, dead time... Each with a small circuit you can play with.
What is bootstrap? How is the upper MOSFET driven?
Bootstrap makes a 'floating' supply that rides up with the midpoint to drive the upper MOSFET. The capacitor charges while the lower MOSFET is on, and that energy drives the gate while the upper MOSFET is on.
FreeWhat is a snubber? How does an RCD snubber work?
The snubber catches the spike energy from the transformer's leakage inductance the moment the MOSFET turns off, stores it in the capacitor and turns it into heat in the resistor. That keeps the drain voltage within what the MOSFET can withstand.
🔒 repair packWhat is dead time?
Dead time is the short interval when both the upper and lower MOSFETs are off (usually a few hundred nanoseconds). It prevents the short circuit and also lets the inductor current swing the midpoint by itself, so the MOSFETs turn on with no voltage across them.
🔒 repair packStart-up resistor and auxiliary winding: how the IC gets its power
When plugged in, the start-up resistor slowly charges the VCC capacitor. When VCC reaches the start threshold the IC runs; shortly after, the transformer's auxiliary winding takes over powering it.
🔒 repair packWhat is ESR? Why can a capacitor look 'fine' but be faulty?
ESR is the small series resistance inside every capacitor. In switch-mode supplies the capacitor charges and discharges tens of thousands of times a second, so as ESR grows, output ripple and capacitor heating rise quickly.
🔒 repair packHow do the current-sense resistor (Rs) and the CS pin work?
The current-sense resistor turns MOSFET current into a small voltage. The IC watches it on every pulse and ends the pulse when it reaches the threshold. It's both overcurrent protection and, on current-mode ICs, the output regulation itself.
🔒 repair packOpto-coupled feedback: how is the output held steady?
The voltage divider scales the output for the TL431; the TL431 compares it with its 2.495 V reference and turns the difference into opto LED current. The light crosses the isolation barrier, and the mains-side IC shortens or lengthens the pulse accordingly.
🔒 repair packInput protection: fuse, NTC, varistor, X and Y capacitors
The fuse prevents fire on a short, the NTC limits the big switch-on current, the varistor clamps lightning and sudden surges, and the X/Y capacitors and common-mode choke form the noise filter.
🔒 repair packSeries bulb: a safe first power-up after a repair
An incandescent bulb in series with the device limits the current to the bulb's rating on a short and takes the voltage itself. Its brightness also shows the device's state: dim means healthy, steadily bright means there's still a short.
🔒 repair packWhat is AVR? How do transformer taps correct the voltage?
AVR (Automatic Voltage Regulator) keeps the output near 230 V when the grid is low or high. In UPSs it's done with a tapped autotransformer: one winding with several taps. Whichever tap the relays connect the grid to, the output rises or falls by that ratio. Since no battery is used, the battery is spared during short and long sags. Wall-mounted 'voltage stabilisers' use the same idea; the big ones move a sliding contact with a servo motor instead of relays (stepless).
🔒 repair packSoft-start: why does a supply power up slowly?
Soft-start begins the duty cycle at zero at switch-on and slowly raises it to its working value. The IC charges the capacitor on its SS pin with a small constant current; the SS voltage sets the widest pulse allowed. So there's no surge in the input current and the output rises without overshoot. Once SS reaches the top, the feedback loop takes over.
🔒 repair packPWM and duty cycle: how switching on and off sets the power
In PWM the switch is either fully on or fully off; the frequency stays fixed and only the pulse width changes. The fraction of each period spent on is the duty cycle (D). The average voltage the load sees is D × the supply voltage: at 12 V, 25 % duty gives 3 V and 75 % gives 9 V. Because the switch never sits in between, it barely heats up; it's far more efficient than dropping power with a resistor or linear regulator.
🔒 repair packTransformer and inductor saturation: why do MOSFETs blow for no reason?
A ferrite core can only carry energy up to a certain magnetic flux (B). Below that limit the winding resists fast changes in current and the current draws a clean ramp. At the limit (saturation) the core can't take more flux, the inductance collapses and almost nothing limits the current: it multiplies within a few hundred nanoseconds. If current limiting can't keep up, the MOSFET, Rs and sometimes the IC all go. Hot ferrite saturates earlier.
🔒 repair packHiccup protection: why does the board go 'tick tick'?
Hiccup means the supply runs briefly on an overload or short, stops for a long time, then tries again. In most flybacks this happens by itself: the start-up resistor charges the VCC capacitor and the IC starts; but with the output shorted the auxiliary winding can't feed VCC, VCC drops and the IC stops. The cycle repeats. The average power stays small, so the MOSFET and diode don't burn. Some ICs do the same with a dedicated timer.
🔒 repair packSine PWM (SPWM): how is a clean sine made from DC?
In SPWM a slow sine (the wanted output) is compared with a fast triangle (the carrier, 4–20 kHz). While the sine is above the triangle, the bridge connects the load one way; below it, the other way. So the pulses are wide near the sine's peaks and narrow near its troughs. The LC filter (or the motor winding) averages the pulses, leaving a sine. Modulation index m = sine amplitude ÷ triangle amplitude; m sets the output amplitude and the reference sine sets its frequency. Above m = 1 the peaks are clipped.
🔒 repair packWhat is an IGBT? How is it different from a MOSFET?
An IGBT (Insulated Gate Bipolar Transistor) is a switch whose input is driven by voltage through an insulated gate like a MOSFET, and whose output conducts like a bipolar transistor. When on, a VCE(sat) of ≈1.5–2.5 V remains, changing little with current; a MOSFET's loss grows with the square of the current, I² × RDS(on). At 600–1700 V and tens to hundreds of amps the IGBT runs cooler. The price is speed: at turn-off a 'tail current' flows until the stored charge clears, so IGBTs usually run at 2–20 kHz. It has no inherent body diode; modules add a fast anti-parallel diode to every IGBT. The gate turns on with +15 V and is held off with −5…−15 V so noise can't turn it on.
🔒 repair packWhat is an isolation transformer? How do you scope the mains side?
In the grid, neutral is bonded to earth in the panel; a bench scope's probe clip is earthed through its plug too. The 'hot GND' on an SMPS primary (the bridge's negative) connects to L through a diode during one half of the mains cycle. Clipping onto it shorts L straight to earth. A 1:1 isolation transformer feeds the device from a winding that has no link to earth: even if the clip ties one point to earth, there's no path for current to return. A differential probe measures the difference between two points without connecting to earth at all. Remember: an isolation transformer doesn't protect you from touching two points at once, and it makes any RCD behind it useless.
🔒 repair packHow do a contactor and an overload relay work? The seal-in circuit
A contactor is a relay built for big currents: when coil A1–A2 is energised, three main contacts (1/L1–2/T1, 3/L2–4/T2, 5/L3–6/T3) and the auxiliaries move together. An auxiliary contact's last digit gives its type: 3-4 is NO (13-14), 1-2 is NC (21-22). A thermal overload has one bimetal per phase; if the motor current stays above the setting for long enough, the bimetals heat, bend, open the 95-96 NC contact and close the 97-98 NO contact. In a seal-in circuit, contact 13-14 wired in parallel with the start button keeps feeding the coil after the button is released; stop (NC) or the overload stops the motor by breaking that path. When power fails and returns, the motor doesn't restart by itself.
🔒 repair packWhat is 4–20 mA? How does a current loop work and how do you measure it?
In a 4–20 mA loop the 24 V supply, the transmitter, the field cable and the PLC input are all in series. A 2-wire transmitter takes its power from the loop too, and sets the current it draws according to what it measures: the bottom of the range is 4 mA, the top 20 mA. Since the current is the same everywhere in a series circuit, cable resistance and length don't change the reading, and it shrugs off noise. A 250 Ω resistor at the PLC input turns the current into 1–5 V. Making zero 4 mA (a 'live zero') does two jobs: the transmitter powers itself from that current, and 0 mA (or < 3.6 mA) is recognised as 'wire broken / sensor fault'. With 0–10 V signals a broken wire looks just like 'value zero'.
🔒 repair packHow do earthing and a residual current device (RCD) work? Why does it trip?
A residual current device (RCD, also called an RCCB or GFCI) passes the L and N conductors together through a toroid. In a healthy circuit the outgoing and returning currents are equal and their fields cancel. If some current leaks to earth through the frame and PE, or through a person's body, a difference (IΔ) appears between L and N; once it exceeds 30 mA the RCD opens L and N together within 300 ms at most, within 40 ms at 150 mA. An RCD doesn't limit the current, it limits the time; nor does it protect against overcurrent. Earthing (PE) holds the frame at earth potential in a fault and gives the leak a far easier path than a body. Together they protect you.