Voltage Regulators

A voltage regulator is a circuit that holds its output voltage steady while its input voltage and its load change. This article explains why circuits need one, the numbers that describe how well a regulator works, and the two big families: linear regulators such as the 7805, LM317 and AMS1117, which burn off the extra voltage as heat, and switching regulators such as buck and boost modules, which convert it with an inductor.

1. A short recap

• Power is voltage times current, P = V × I, measured in watts (W) (recap). Real sources have internal resistance, so their voltage sags as the load draws more current (recap).
• A linear power supply turns mains into DC with a transformer, a bridge rectifier, a filter capacitor and a regulator; the capacitor leaves a ripple (recap).
• A Zener diode fed through a resistor holds a fixed voltage, but only for small loads (recap).
• A transistor lets a small current or voltage control a large current (recap), and can work as a fast switch (recap).
• An inductor stores energy in its magnetic field; buck and boost converters use that to change a DC voltage (recap).
• A small ceramic capacitor beside each chip, a decoupling capacitor, supplies fast bursts of current (recap).
• LED constant-current drivers regulate current instead of voltage (recap).

2. Why we regulate

Raw DC is rarely steady. A battery's voltage falls as it discharges (recap). A rectified supply carries ripple, a repeating dip each time the filter capacitor discharges between peaks; article 8's 9 V supply swings by about 1.9 V, 120 times a second.

Most electronics wants one exact voltage. A USB port supplies a nominal 5 V,^[1] and the Raspberry Pi Pico microcontroller board runs its chip and input/output pins from 3.3 V.^[2] Feed chips too little and they misbehave; too much and they can be damaged. A regulator sits between the messy source and the load and delivers the fixed voltage (Figure 1).

3. The numbers that describe a regulator

Datasheets use the same handful of terms. Texas Instruments (TI) defines them this way:^[3]

• Line regulation: how much the output changes when the input changes. The 7805's is typically 3 mV for an input anywhere from 7.5 to 20 V.^[4]
• Load regulation: how much the output changes when the load current changes. The 7805's is typically 10 mV from 5 mA to 1.5 A.^[4]
• Dropout voltage: the smallest input-to-output difference at which the regulator still regulates. Below it, the output follows the input down.^[3]
• Quiescent current (or ground current): the current the regulator uses for itself, which flows in at the input but not out to the load.^[3] The 7805's is up to 8 mA.^[4]
• Power supply rejection ratio (PSRR), or ripple rejection: how strongly the regulator blocks ripple on its input from reaching its output, at a given frequency.^[3] It is quoted in decibels (dB); every 20 dB is a factor of ten.

The 7805's ripple rejection is typically 80 dB at 120 Hz, and at least 62 dB.^[4] 80 dB is a factor of 10,000, so 1.9 V of ripple would shrink to about 0.19 mV, or to about 1.5 mV at the 62 dB minimum (both calculated).

4. Linear regulators: how they work

A Zener regulator wastes current: its resistor must pass enough for the load and the Zener at all times. A common fix lets the Zener drive the base of a transistor, which carries the load current, so the Zener handles only the small base current. The output is then about 0.7 V below the Zener voltage, the transistor's base-emitter drop.^[5]

An integrated series pass regulator refines that idea with four blocks (Figure 2):^[6]

• a voltage reference, a small, very stable internal voltage;
• a pass element, a transistor in series with the load that drops the extra voltage;
• a pair of sampling resistors, a divider that feeds back a fixed fraction of the output;
• an error amplifier that compares that fraction with the reference and drives the pass element.

If the output sags, the sampled voltage falls below the reference, and the amplifier turns the pass transistor further on until they match again. If the output rises, it does the reverse. This continuous correction is called negative feedback.

5. Fixed regulators: the 78xx and 79xx

The 78xx family are three-pin positive regulators. The last digits give the voltage: TI's µA7800 series offers 5, 8, 10, 12, 15 and 24 V.^[7] The 79xx family regulates negative voltages, such as −5, −12 and −15 V, for circuits that need a supply below ground.^[8]

The 7805 gives 4.8 to 5.2 V at room temperature, up to 1.5 A, from inputs up to 35 V. Its dropout is 2 V at 1 A, so the input must stay at 7 V or more.^[4] It has current limiting (about 2.1 A when shorted) and shuts down if its chip reaches 150 °C; TI calls it "essentially indestructible".^[4] Why so much? Classic regulators such as the LM317 use an NPN Darlington pass element (two transistors in tandem),^[9] and a Darlington needs at least about 1.6 V across it to regulate.^[6]

In the TO-220 package (a plastic body with a metal mounting tab), pin 1 is the input, pin 2 is ground and pin 3 is the output, and the tab is ground (Figure 3).^[7] Other regulators differ: on the LM317 the tab is the output,^[9] and on the LM79xx it is the input.^[8] Check the datasheet before wiring.

TI measures the 7805 with 0.22 µF on the input and 0.1 µF on the output; the µA7800 sheet uses 0.33 µF in. The input capacitor is needed if the regulator is far from the supply's filter capacitor. The output capacitor is not needed for stability but improves the response to sudden load changes.^[4]​^[7] The 79xx is stricter: it requires an output capacitor of 1 µF solid tantalum or 25 µF aluminum for stability, plus 2.2 µF on the input if it is more than 3 inches from the filter capacitor.^[8] Place them close to the pins.

6. The adjustable LM317

The LM317 sets its output with two resistors, from 1.25 to 37 V at up to 1.5 A. It holds 1.25 V, its reference V_REF, between its output and adjust pins. R1 sits across that gap, and R2 runs from the adjust pin to ground (Figure 4).^[9] TI gives:

V_OUT = 1.25 V × (1 + R2 ÷ R1) + I_ADJ × R2

I_ADJ, the small current out of the adjust pin, is typically 50 µA and usually negligible.^[9] TI's examples use R1 = 240 Ω.^[9]

Worked example: 3.3 V from 9 V. All results are calculated.

• Choose R1 = 240 Ω. We need 1 + R2/R1 = 3.3 ÷ 1.25 = 2.64, so R2 = 1.64 × 240 = 394 Ω. The nearest standard value is 390 Ω.
• Check: 1.25 × (1 + 390/240) = 3.28 V, plus 50 µA × 390 Ω = 0.02 V, gives about 3.30 V.
• Headroom: 9 − 3.3 = 5.7 V, above the 3 V TI recommends for full current.^[9]
• The LM317 needs a minimum load current to regulate: 3.5 mA typical, 10 mA at most, specified with 40 V across it.^[9] R1 alone draws 1.25 ÷ 240 = 5.2 mA, which covers the typical figure even with nothing connected.

TI suggests 0.1 to 1 µF at the input if the regulator is more than about 6 inches from the filter capacitors, and an optional 1 µF at the output for better transient response.^[9] A 10 µF capacitor from the adjust pin to ground raises its ripple rejection to about 80 dB at 120 Hz.^[9]

7. Heat and efficiency

A linear regulator passes the full load current while dropping the difference between input and output, so it turns this much power into heat:^[3]

P = (V_IN − V_OUT) × I_OUT

Worked example: 12 V to 5 V at 1 A with a 7805. All results are calculated.

• Heat: (12 − 5) × 1 = 7 W. The load gets 5 W, so the efficiency is 5 ÷ 12 ≈ 42%. Quiescent current adds up to about 0.1 W more (8 mA × 12 V).
• Without a heat sink, a TO-220 7805 rises 54 °C above the surrounding air for every watt.^[4] 7 W would mean a 378 °C rise. Long before that it reaches 150 °C and shuts down.
• TI gives the limit as P_MAX = (T_J(MAX) − T_A) ÷ θ_JA.^[4] At 25 °C that is (150 − 25) ÷ 54 ≈ 2.3 W.

The symbol θ (theta) is thermal resistance: degrees of temperature rise per watt flowing through a part. T_J is the junction (chip) temperature and T_A the ambient (surrounding air) temperature. Heat passes from the chip to the case (θ_JC, 4 °C/W for the TO-220 7805),^[4] through a thin thermal pad or grease to a heat sink, and from the sink to the air. The resistances add like resistors in series (Figure 5).^[10]

To keep the chip at or below 125 °C in 25 °C air, the total must be at most (125 − 25) ÷ 7 ≈ 14 °C/W. Take away the case's 4 °C/W, and the pad and sink together need about 10 °C/W or less (calculated). Heat-sink makers state that figure for their products.^[4] In a warm box, the budget shrinks. TI warns that published junction-to-air figures depend on the test board, so treat them as estimates.^[10]

Article 8's 9 V supply made 2.9 W and needed a sink for the same reason (recap). When the drop times the current runs to watts, a switching regulator (section 9) is usually the better tool.

8. Low-dropout regulators (LDOs)

A low-dropout regulator, or LDO, keeps regulating with a much smaller input-to-output gap. The trick is the pass element. A Darlington needs at least about 1.6 V. An LDO uses a single PNP transistor or a P-channel MOSFET as its pass element, and typically works with under 0.5 V.^[6] A MOSFET in dropout acts like a small resistor, so the dropout is roughly the load current times its on-resistance; one TI LDO drops typically 350 mV at 1 A.^[3] MOSFET LDOs also draw very little quiescent current.^[6]

The popular AMS1117 sits in between. Its maker calls it low dropout and guarantees at most 1.3 V (1.1 V typical) at 0.8 A, higher above that. It delivers up to 1 A, comes in fixed 1.5, 1.8, 2.5, 2.85, 3.3 and 5.0 V versions and an adjustable one, and accepts at most 15 V in.^[11] Its quiescent current is 5 mA typical, and the AMS1117-3.3's ripple rejection is 72 dB typical.^[11] Unlike the 7805, it needs an output capacitor: the datasheet says 22 µF of solid tantalum ensures stability in all conditions.^[11]

Use an LDO when the input is only a little above the output, as when making 3.3 V from 5 V, or when a battery should be used down to nearly the output voltage. A smaller gap also means less heat and better efficiency for the same current.^[3]

9. Switching regulators

A switching regulator does not burn off the extra voltage. It switches the input on and off thousands to millions of times a second and uses an inductor and capacitor to smooth the pulses (recap). An ideal switch wastes no power: when it is on, almost no voltage is across it, and when it is off, no current flows. The output is set by the duty cycle D, the fraction of each cycle the switch is on, which a control chip adjusts by feedback much like a linear regulator's error amplifier.

Buck (step-down)

A buck converter lowers the voltage. With the switch on, current flows from the input through the inductor to the load and the inductor stores energy. With the switch off, the inductor keeps the current flowing through a diode, giving up that energy (Figure 7).^[12] TI estimates D = V_OUT ÷ (V_IN × η), where η (eta) is the efficiency, using 90% as an example.^[12] For 12 V to 5 V that gives D ≈ 0.46 (calculated).

Some converters replace the diode with a second transistor switch, a synchronous design.^[12] A diode always drops its forward voltage, but a MOSFET switched on at the right moment drops much less, so synchronous converters are more efficient. ROHM notes that older diode designs barely top 80%.^[13]

Boost (step-up) and buck-boost

A boost converter raises the voltage. The switch connects the inductor across the input, storing energy; when it opens, the inductor's voltage adds to the input and pushes current through the diode into the output capacitor (Figure 8; recap).^[14] TI estimates D = 1 − V_IN × η ÷ V_OUT.^[14] Power is conserved, minus losses, so a boost draws more current from its input than it delivers.

A buck-boost converter can go either way, which suits a battery whose voltage crosses the output. The Pico board uses one, an RT6150, to make 3.3 V from about 1.8 to 5.5 V.^[2] The simplest inverting buck-boost gives a negative output, larger or smaller than the input depending on D.^[15]

Efficiency, noise and layout

Switchers typically reach about 80 to 95% efficiency. ROHM puts the maximum near 95%, falling significantly at light load,^[16] and TI uses 90% as a working estimate.^[12] TI's LM2596 buck datasheet gives 73% (3.3 V out) and 80% (5 V out) from 12 V at 3 A.^[17] A bench test of a 12 V to 5 V module at 1 A measured 82%.^[18] Compare that with the 7805's 42%: for 5 W out, about 1.1 W of heat instead of 7 W (calculated; Figure 9).

The price is noise. The switching leaves output ripple at the switching frequency, plus sharp spikes at each edge, and gives off electromagnetic interference (EMI), radio-frequency noise that products must keep within legal limits. ROHM advises avoiding switchers in noise-averse applications.^[16] Layout matters: TI says the high-current paths must be short, wide traces, with the parts close to the chip over a ground plane, and the feedback wiring kept away from the inductor.^[17]​^[19] A small input capacitor belongs right at the chip's input pin.^[19]

10. Ready-made modules

Small boards with the chip, inductor, diode and capacitors already fitted are cheap and popular:

• LM2596 buck. TI's part takes up to 40 V in, delivers 3 A, and switches at a fixed 150 kHz. The adjustable version sets its output with V_OUT = 1.23 V × (1 + R2 ÷ R1), from 1.2 to 37 V.^[17]
• MT3608 boost. It runs from 2 to 24 V in, gives up to 28 V out, switches at 1.2 MHz and quotes up to 93% efficiency. Its feedback reference is 0.6 V.^[20]

On adjustable modules, a small screw-turned trim potentiometer (a variable resistor) takes the place of a feedback resistor. Turn it to set the output, and measure with a meter before connecting anything. One tester points out that a failed pot usually goes open circuit, and the output then jumps to its maximum, which can destroy whatever is connected.^[18]

Cautions. Treat headline ratings as upper limits, not working values.

• The MT3608's datasheet title says 2 A, but its limit is a 4 A switch current. TI shows a boost's maximum output current is roughly that switch limit times (1 − D), minus a ripple term.^[20]​^[14] At 5 V to 12 V with 85% efficiency, D ≈ 0.65, so the ceiling is about 1.4 A at most (calculated), and heat may limit it further.
• Many cheap "LM2596" boards are counterfeit. Testers have found them switching at about 50 to 53 kHz instead of 150 kHz, with roughly double the ripple of a genuine chip.^[18] Another batch ran at 18 kHz with 400 mV of ripple and scorching heat below the rated 3 A, although it still held its output voltage.^[21]
• TI warns that parts bought outside authorized sources may be counterfeit or unreliable.^[22] For anything important, buy from a reputable seller and test before trusting the rating.

11. Linear or switching?

12. References and shunt regulators

Every regulator contains a voltage reference, and references are also sold as parts on their own. The TL431 is the classic. It is a shunt regulator: like a Zener, it sits across the load and sinks whatever current keeps the voltage fixed, from 1 to 100 mA. Its reference is 2.495 V typical, and two resistors set any voltage from there up to 36 V, with V_O = (1 + R1 ÷ R2) × V_REF. TI calls it an excellent replacement for Zener diodes in many applications.^[23] A common job is in isolated mains supplies, where it compares the output with its reference and signals the mains side through an optocoupler.^[24]

13. Choosing a regulator

Start with four numbers: the lowest and highest input voltage, the output voltage, and the maximum load current. Then check the dropout, the heat and any noise needs.

USB 5 V to 3.3 V for a microcontroller. A USB port should hold its supply between 4.75 and 5.25 V,^[1] though a bus-powered hub port may give as little as 4.4 V and newer rules allow up to 5.5 V.^[25] An AMS1117-3.3 at 4.75 V has 1.45 V of headroom, enough for its 1.3 V maximum dropout at 0.8 A;^[11] at 4.4 V it has only 1.1 V and may drop out (calculated). At 300 mA from 5 V it makes (5 − 3.3) × 0.3 ≈ 0.5 W of heat (calculated), which its maker says a well-copper-clad board can spread at as little as 45 °C/W, about a 23 °C rise (calculated).^[11] A modern LDO with a few hundred millivolts of dropout gives more margin. Many boards already include a regulator, like the Pico's buck-boost,^[2] so read the board's documentation first. For more about the chips themselves, see Microcontrollers.

A 12 V car to 5 V. A car's supply is hostile. TI's automotive guide shows a normal range up to 16 V, 24 V during jump starts, up to 42 V in a load dump (the battery disconnecting while the alternator is charging), dips as low as 4 V while the engine cranks, and −14 V if the battery is connected backward.^[26] A 7805 at 1 A would burn 7 to 11 W from 12 to 16 V (calculated), and 42 V exceeds both its 35 V and the LM2596's 40 V ratings.^[4]​^[17] So use a buck converter rated for the full input range, with reverse-polarity and transient protection in front of it, or a ready-made car USB adapter. Protection parts are covered in article 17, Fuses and Circuit Protection.

14. Testing a regulator

• Output under load. Measure the output with no load, then with a known load at the full current. A power resistor works: 5 Ω draws 1 A at 5 V and dissipates 5 W, so it gets hot.^[18] The output should stay within the datasheet's tolerance (recap).
• Low input. Lower the input until the output starts to fall. That shows the real dropout, and whether your supply's worst case is safe.
• Ripple with a scope. Use an oscilloscope (recap) with AC coupling, which blocks the steady DC and shows only the ripple.^[27] Measure right at the output capacitor with the shortest possible ground, such as a ground spring or "tip and barrel" contact; a long ground lead picks up noise and adds false spikes.^[28]​^[27] Richtek says the scope's 20 MHz bandwidth limit is fine for most uses, but noise-sensitive circuits need full bandwidth.^[28] For switchers, the ripple frequency also reveals the true switching frequency; that is how the fake LM2596 chips were caught.^[18]
• Thermal check. Run at full load for several minutes and measure the case temperature with a thermometer probe rather than a finger.

Common failures. A wrong pinout (sections 5 and 6) can feed the input straight to the load. If a 7805's ground pin comes loose, its output rises toward the input voltage and can damage the circuit.^[4] A missing or wrong output capacitor can make an LM79xx or AMS1117 oscillate.^[8]​^[11] A large output capacitor can discharge backward through the regulator if the input is shorted; TI suggests a protection diode across the regulator, pointing from output back to input, above 10 µF on the 7805 and above 100 µF on the LM79xx.^[4]​^[8] An overloaded LM317 shuts down and resets itself, so an output that cuts in and out points to overload or heat.^[9]

15. Safety

• Hot parts. A regulator can run near its 150 °C shutdown point.^[4] Don't touch it under load, and allow for heat when mounting it. On the 78xx the tab is ground, but on the LM317 and LM79xx it is live, and so is any heat sink bolted to it.^[7]​^[9]​^[8]
• Reversed input. Input transients more than 0.8 V below ground can damage a 7805.^[4] A series diode protects against a reversed supply but costs about 0.6 V; a MOSFET wired for reverse protection wastes far less (recap).^[29]
• Input capacitors. Electrolytic capacitors must go in the right way round and be rated above the highest input voltage (recap).
• Mains supplies. The regulator end of a supply is low voltage, but the transformer and rectifier side of a mains supply is not. Work on mains only with the right training and precautions (recap; recap).

The whole story in brief

• A regulator holds its output steady despite changes in input, load and ripple.
• Datasheets describe it by line and load regulation, dropout, quiescent current and PSRR.
• Linear regulators use a reference, error amplifier and pass transistor. The 7805 needs 2 V of headroom; the LM317 is set by two resistors.
• Linear heat is (V_IN − V_OUT) × I; check it against the thermal resistance and add a heat sink.
• LDOs work with a smaller gap; the AMS1117 needs about 1.1 to 1.3 V and a 22 µF output capacitor.
• Buck, boost and buck-boost switchers reach roughly 80 to 95% efficiency but add ripple and noise; layout matters.
• Treat cheap module ratings with caution, and test output, ripple and temperature.

Coming up in this section: 16 Integrated Circuits.

Sources

All links were checked in October 2026. External links open in a new tab.

1. USB Implementers Forum, "Power Delivery Design Issues for Hi-Speed USB on Motherboards" (PDF): https://www.usb.org/sites/default/files/power_delivery_motherboards.pdf
2. Raspberry Pi Ltd, Raspberry Pi Pico Datasheet (PDF), power chain and RT6150 buck-boost regulator: https://datasheets.raspberrypi.com/pico/pico-datasheet.pdf
3. B. S. Lee, Texas Instruments, "Understanding the Terms and Definitions of LDO Voltage Regulators" (SLVA079, PDF): https://www.ti.com/lit/an/slva079a/slva079a.pdf
4. Texas Instruments, LM340, LM340A and LM7805 Family Wide V_IN 1.5-A Fixed Voltage Regulators data sheet (SNOSBT0L, PDF): https://www.ti.com/lit/ds/symlink/lm340.pdf
5. Tony R. Kuphaldt, Lessons In Electric Circuits, Volume III (Semiconductors), chapter 4, "Bipolar Junction Transistors" (the common-collector amplifier as a voltage regulator): https://www.ibiblio.org/kuphaldt/electricCircuits/Semi/SEMI_4.html
6. Texas Instruments, "Technical Review of Low Dropout Voltage Regulator Operation and Performance" (SLVA072, PDF): https://www.ti.com/lit/an/slva072/slva072.pdf
7. Texas Instruments, µA7800 series (uA7805 to uA7824) positive voltage regulators data sheet (SLVS056P, PDF): https://www.ti.com/lit/ds/symlink/ua78.pdf
8. Texas Instruments, LM79XX Series 3-Terminal Negative Regulators data sheet (SNOSBQ7C, PDF): https://www.ti.com/lit/ds/symlink/lm79.pdf
9. Texas Instruments, LM317 3-Pin Adjustable Regulator data sheet (SLVS044Z, PDF): https://www.ti.com/lit/ds/symlink/lm317.pdf
10. Texas Instruments, "Semiconductor and IC Package Thermal Metrics" (SPRA953D, PDF): https://www.ti.com/lit/an/spra953d/spra953d.pdf
11. Advanced Monolithic Systems, AMS1117 1A Low Dropout Voltage Regulator data sheet (PDF, copy hosted by Handsontec): https://www.handsontec.com/dataspecs/IC/AMS1117.pdf
12. Texas Instruments, SLVA477C, "Basic Calculation of a Buck Converter's Power Stage" (PDF): https://www.ti.com/lit/an/slva477c/slva477c.pdf
13. ROHM TechWeb, "Differences between Synchronous and Nonsynchronous Rectifying DC-DC Conversion": https://techweb.rohm.com/product/power-ic/dcdc/92/
14. Texas Instruments, SLVA372D, "Basic Calculation of a Boost Converter's Power Stage" (PDF): https://www.ti.com/lit/an/slva372d/slva372d.pdf
15. Texas Instruments, "Understanding Inverting Buck-Boost Power Stages in Switch Mode Power Supplies" (SLVA059B, PDF): https://www.ti.com/lit/an/slva059a/slva059a.pdf
16. ROHM TechWeb, "Advantages vs Disadvantages in Comparison with Linear Regulator" (switching regulator basics): https://techweb.rohm.com/product/power-ic/dcdc/88/
17. Texas Instruments, LM2596 SIMPLE SWITCHER Power Converter 150-kHz 3-A Step-Down Voltage Regulator data sheet (SNVS124G, PDF): https://www.ti.com/lit/ds/symlink/lm2596.pdf
18. K6JCA (amateur radio blog), "Counterfeit LM2596 Regulator Boards" (2018), bench measurements of low-cost modules: https://k6jca.blogspot.com/2018/02/counterfeit-lm2596-regulator-boards.html
19. Texas Instruments, AN-1149, "Layout Guidelines for Switching Power Supplies" (SNVA021C, PDF): https://www.ti.com/lit/an/snva021c/snva021c.pdf
20. Aerosemi, MT3608 2A High Efficiency 1.2MHz Current Mode Step-Up Converter data sheet, Rev 1.1 (PDF, hosted by Datasheets.com): https://static.datasheets.com/doc/ec69a9ea-fa22-4383-b0f6-12a9a4ae558c-aerosemi-mt3608-ds.pdf
21. Hackaday, "Cheap Power Supplies With Fake Chips Might Not Be That Bad" (November 2023): https://hackaday.com/2023/11/10/cheap-power-supplies-with-fake-chips-might-not-be-that-bad/
22. Texas Instruments, "Anti-counterfeit": https://www.ti.com/quality-reliability/quality/anti-counterfeit.html
23. Texas Instruments, TL431, TL432 Precision Programmable Reference data sheet (SLVS543S, PDF): https://www.ti.com/lit/ds/symlink/tl431.pdf
24. Texas Instruments, "Shunt Regulator Design Procedures for Secondary Feedback Loop in Isolated Converter" (SLVAF37, PDF): https://www.ti.com/lit/an/slvaf37/slvaf37.pdf
25. USB Implementers Forum, "USB2.0/3.2/BC1.2 Drop Droop Test for non-USB Type-C Products", version 1.4.1 (PDF): https://www.usb.org/sites/default/files/USB20_32_BC12_Drop_Droop_1_4_1.pdf
26. Texas Instruments, Analog Applications Journal, "Designing the front-end DC/DC conversion stage to withstand automotive transients" (SLYT707, PDF): https://www.ti.com/lit/an/slyt707/slyt707.pdf
27. Tektronix, "Probing Techniques for Accurate Voltage Measurements on Power Converters with Oscilloscopes": https://www.tek.com/en/documents/application-note/probing-techniques-accurate-voltage-measurements-power-converters-oscillos
28. Richtek, "Output Ripple Measurement Methods for DC-DC Converters" (design tips, PDF): https://www.richtek.com/Design%20Support/~/media/DT_PDF/Ripple_measurement_tips.pdf
29. J. Falin, Texas Instruments, "Reverse Current/Battery Protection Circuits" (SLVA139, PDF): https://www.ti.com/lit/an/slva139/slva139.pdf