Switches and Relays

Every circuit needs a way to be turned on and off. A switch does it by hand or by machine. A relay does it with an electromagnet, so that one circuit controls another. This article covers how both are built, how to read their names and ratings, why contacts bounce and spark, how to drive a relay safely, solid-state relays, and the safety rules for switching mains.

A relay is the coil from article 5 put to work. Current through the coil makes a magnetic field, and the field pulls a piece of iron that moves a set of contacts. Everything article 5 said about coils, including the voltage spike when their current is cut, applies to every relay.

About the analogies. Everyday comparisons appear only in shaded "Analogy" boxes beside the real physics, and each box says where the comparison stops working.

Mains switching can kill. Switches and relays are often used on 120 V or 240 V AC. Experiment with low-voltage supplies, and read section 13 before connecting anything to mains.

1. A short recap

2. What a switch is, physically

An electrical switch is any device that interrupts the flow of current. It has two states: closed (on, conducting) or open (off).[1] Every mechanical switch has the same few parts (Figure 1):

Inside a simple pushbutton switch Cross-section of a normally open pushbutton. A button on top pushes a plunger down onto a metal bridge, the moving contact, which a return spring holds up. Below each end of the bridge is a fixed contact on a terminal that leaves the bottom of the insulating housing. While released there is a gap between the bridge and the fixed contacts, so the circuit is open; pressing the button pushes the bridge down onto both fixed contacts and closes the circuit. Actuator (button) Moving contact (bridge) Return spring Terminals Housing (insulating) Gap: open circuit Fixed contacts Terminals Press: the bridge moves down and joins both fixed contacts
Figure 1. Inside a simple normally open pushbutton, shown released. Pressing the button pushes the bridge down onto both fixed contacts and closes the circuit. Simplified drawing; the part names follow Kuphaldt.

Contact materials. Contacts are usually silver or a silver alloy, whose conduction is barely affected by surface corrosion. Gold resists corrosion best but carries less current and can "cold weld" if pressed hard.[1] Relay makers use silver-tin oxide where welding is a risk and gold cladding for very small currents.[2]

Contact resistance. Even a closed contact has a little resistance, called its contact resistance.[3] More contact force lowers it.[4] Omron guarantees no more than 0.1 Ω (100 milliohms) for one small relay.[5]

Snap action. In a snap-action switch, a spring flips the contacts over suddenly at a set point, however slowly you move the actuator. That keeps arcs short and reduces wear.[4]

3. Poles and throws

Switches are named by two numbers.[1]​[3] Poles are the separate circuits the switch controls (the moving contacts). Throws are the positions each pole can connect to (the fixed contacts per moving contact). "Single" means one and "double" two, giving four common types (Figure 2):[6]​[3]

Poles and throws: SPST, SPDT, DPST and DPDT switch symbols Four schematic symbols. SPST: one blade between two terminals, a plain on-off switch with two terminals. SPDT: a common terminal whose blade connects to either of two terminals, three terminals in all. DPST: two on-off blades joined by a dashed line so they move together, four terminals. DPDT: two changeover switches joined by a dashed line, six terminals. SPST Single pole, single throw 2 terminals: on/off SPDT Single pole, double throw COM 3 terminals: one or the other DPST Double pole, single throw 4 terminals: two on/offs DPDT Double pole, double throw COM COM 6 terminals: two changeovers Dashed line: both poles move together, worked by one actuator
Figure 2. The four common pole-and-throw combinations and their terminal counts. Sources: Kuphaldt, SparkFun and E-Switch.

The common terminal (COM) connects to the moving contact. Rotary switches can have many throws: "single-pole, five-throw" is five positions on one pole.[1] A multi-position switch either opens the old connection before making the new one (break-before-make, the usual kind) or briefly bridges both (make-before-break), used where the circuit must never be open even for an instant.[1]

4. Momentary or maintained; normally open or closed

A maintained (latching) switch stays where you put it, like a wall light switch. A momentary switch is active only while held, like a doorbell button, and a spring returns it.[6]​[1] Some pushbuttons latch on and off with alternate presses.[1]

A switch's normal state is its state with nothing pressing on it, as it sits on the shelf. A normally open (NO) contact closes when actuated. A normally closed (NC) contact opens when actuated (Figure 3).[1] "Normal" does not mean "usual": a limit switch may sit pressed all day, but its labels still describe it unpressed. For relays, NO and NC describe the contacts with the coil unpowered.[7]

Normally open and normally closed pushbuttons, released and pressed Four pushbutton symbols. Top row, normally open: released, the contact bar sits above the two contacts and the circuit is open; pressed, the bar is pushed down onto them and current flows. Bottom row, normally closed: released, the bar touches both contacts and current flows; pressed, the bar is pushed away and the circuit opens. Released (normal state) Pressed Normally open (NO) Normally closed (NC) Open: no current Closed: current flows Closed: current flows Open: no current
Figure 3. Normally open and normally closed pushbuttons. "Normal" means released, with nothing pressing on the switch. Source: Kuphaldt.

5. Kinds of switch

Switching one light from two places

A staircase light uses two 3-way switches, the electrician's name for SPDT switches. Two wires called travelers join their throws. The lamp is on when both switches select the same traveler and off when they differ, so flipping either one changes the state (Figure 4). For more locations, 4-way switches go between them; each connects the travelers straight through or crossed.[10]

Two 3-way switches controlling one lamp Two copies of a staircase lighting circuit. The supply feeds the common terminal of SPDT switch A. Two traveler wires join A's two throws to the two throws of SPDT switch B, whose common terminal feeds the lamp, which returns to the supply. Top: both switches select traveler 1, so there is a complete path and the lamp is on. Bottom: switch B has been flipped to traveler 2 while A still selects traveler 1, so the path is broken and the lamp is off. Both switches up: one complete path, lamp on traveler 1 traveler 2 Switch A Switch B Supply ON Switch B flipped: the path is broken, lamp off traveler 1 traveler 2 Switch A Switch B Supply OFF Colored line: the path current follows. Flipping either switch always changes the lamp.
Figure 4. Two 3-way (SPDT) switches joined by two travelers. The lamp lights when both switches select the same traveler. Source: Matthews, Wake Forest University.

6. Ratings: voltage, current, AC and DC

Every switch and relay has a maximum voltage and current. Three things limit a contact: heat from current through it when closed, sparking as it opens or closes, and the voltage across the open gap.[1] A heavy-duty toggle switch might be rated 2 A at 250 V or 4 A at 125 V.[6]

Why DC is harder to break than AC

As contacts part, current can keep flowing through hot, ionized air in the gap: an arc, which burns the contacts. With AC the arc tends to go out by itself, because 60 Hz current stops and reverses 120 times a second. DC never passes through zero, so a DC arc can keep burning (Figure 5). Switches are therefore rated higher for AC than DC.[1]​[4]

Why an AC arc goes out and a DC arc may not Two graphs of current against time, with a dashed line where the contacts open. AC, 60 hertz: after the contacts open the arc carries the current only until the sine wave next reaches zero, then the arc goes out and the current stays at zero. DC: the current never passes through zero, so the arc can keep carrying it after the contacts part. At 60 hertz a zero crossing comes at most 8.3 milliseconds after the contacts open. AC (60 Hz) time 0 current contacts open current reaches zero: arc goes out arc DC time 0 current contacts open arc can keep burning no zero crossing to help Solid line: current through closed contacts. Dashed: current carried by the arc after the contacts part. At 60 Hz a zero comes at most 1/120 s (8.3 ms, calculated) after the contacts open. Illustrative waveforms.
Figure 5. When contacts open, an arc can carry the current on. An AC arc tends to go out at the next current zero; a DC arc has none. Illustrative waveforms; the principle is from Kuphaldt and Omron.

The Omron G5LE power relay can switch 250 V AC but only 125 V DC. Its resistive rating is 10 A at 120 V AC but 8 A at just 30 V DC.[11] Never apply an AC rating to a DC load. (Recap: AC mains.)

Inductive loads such as motors and solenoids make arcing worse: as the contacts open, the coil drives its current on and the voltage across the gap jumps.[1] That is article 5's kickback again. The G5LE's AC rating halves to 5 A for an inductive load.[11]

Inrush current. Some loads draw much more current for a moment at switch-on: 10 to 15 times the running current for an incandescent lamp, 5 to 10 times for a motor, 20 to 40 times for a capacitive load.[2] A 60 W, 120 V bulb draws 0.5 A once lit but could draw 5 to 7.5 A at switch-on (calculated). Lamps, motors and solenoids can weld contacts, melting them together so they stay closed.[2]

Snubbers and wetting current

A snubber, a capacitor and resistor in series across the contacts, reduces arcing. The capacitor absorbs the voltage rise as the contacts open, and the resistor limits its discharge current when they close. Old car ignition points used one, called a condenser.[1] Panasonic suggests starting at 0.5 to 1 µF per amp and 0.5 to 1 Ω per volt.[2] (Recap: RC circuits.)

Too little current is also a problem. A small spark at each operation cleans the film that forms on contacts. The minimum current that keeps a contact healthy is its wetting current; for tiny signals, choose gold contacts.[1] The gold-alloy G5V-1 relay is specified down to 1 mA at 5 V,[5] the silver-alloy G5LE only to 100 mA.[11]

7. Contact bounce and debouncing

Closing contacts hit, spring apart and hit again several times before settling. This contact bounce lasts milliseconds, too short to see on a lamp.[1] A microcontroller is fast enough to see every bounce and count one press as many (Figure 6).

Jack Ganssle tested 18 switches. They bounced for 1.6 ms on average and up to 6.2 ms, excluding two poor ones, one of which bounced for 157 ms on release. A relay bounced for 0.4 to 2.9 ms.[12] Removing the bounce is debouncing:

Contact bounce and an RC plus Schmitt trigger debouncer Three traces against time from 0 to 30 milliseconds. Top, the switch pin, pulled up to 5 volts: when the button is pressed at 3 milliseconds it bounces between 5 volts and 0 several times before settling at 0 at about 7 milliseconds. Middle, the voltage after an RC filter with a 10 millisecond time constant: it falls smoothly, pausing slightly during the bounces, and crosses the lower threshold at about 17 milliseconds. Bottom, the Schmitt trigger output: a single clean step from 5 volts to 0 at that moment. The bounce pattern and thresholds are illustrative. Switch pin raw: bounces 5 V 0 After RC filter slow, smooth fall 5 V 0 lower threshold upper threshold Schmitt trigger output one clean step 5 V 0 0 5 10 15 20 25 30 Time (ms) after the start; button pressed at 3 ms
Figure 6. A bouncing switch (top), the same signal through a 10 kΩ, 1 µF RC filter (middle) and a Schmitt trigger's clean output (bottom). The bounce pattern and thresholds are illustrative; the RC curve is calculated from them. Method: Ganssle.

A button on a microcontroller input also needs a pull-up resistor from the pin to the supply, with the switch from the pin to ground. Open, the pin reads high; pressed, low. The resistor stops the closed switch shorting the supply.[6] More in the Microcontrollers pages.

8. Relays: a switch moved by a coil

A coil that pulls a movable piece of iron is a solenoid, and the moving piece is the armature. A relay is a solenoid whose armature moves switch contacts.[7] It builds directly on article 5's coil and magnetic field, which listed relays among a coil's uses.

How a relay is built

A typical relay has a coil of fine copper wire on an iron core; a hinged iron armature held off the core by a spring; and contacts: a moving common contact on the armature, a fixed NC contact it rests on, and a fixed NO contact it moves to (Figure 7).[7]​[14] Power the coil and the core pulls the armature in, so COM leaves NC and touches NO. Remove the power and the spring pulls it back.[7] You can usually hear the click.

Inside an electromagnetic relay Cutaway of a relay with its coil switched off. A coil of copper wire is wound around an iron core that stands on an L-shaped iron frame. A flat iron armature is hinged at the top of the frame and lies across the top of the core with a gap; a spring from the frame holds it up. On the free end of the armature is the moving contact, resting against the fixed normally closed contact above it, with the fixed normally open contact below. A flexible lead joins the armature to the common pin. An arrow shows that the armature end is pulled down onto the normally open contact when the coil is energized. Separate pins come out for the coil and for the NO and NC contacts. Frame (iron) Spring Hinge Flexible lead to COM Iron core COM pin Coil pins Armature (iron, pivots on the hinge) NC contact (fixed) Moving contact: pulled down onto NO when the coil is on NO contact (fixed) Coil of copper wire NO and NC pins
Figure 7. A relay with its coil off. Energizing the coil pulls the armature down, moving the common contact from NC to NO. Simplified drawing based on Kuphaldt and Omron.

Two things make relays useful. A small coil current controls a much larger contact current: Kuphaldt's example has a 12 V DC coil switching 480 V AC, with coil currents typically well under 1 A and industrial contacts rated at least 10 A. And the coil and contacts are electrically isolated, sharing no wires.[7] The G5V-1 withstands 1,000 V AC between coil and contacts, the G5LE 2,000 V AC.[5]​[11]

Analogy: a doorbell for a floodgate. A light press on a small button gets a heavy gate opened on a completely separate channel, and the person pressing never touches the water.

Where it breaks down: a doorbell button does not fight back when released. A relay coil stores energy in its magnetic field and kicks back a voltage spike when its current is cut, so the control side needs protection. And the "gate" is not perfect: contacts arc, wear and can weld.

Coil voltage and current

A relay is chosen by its rated coil voltage. The 5 V G5V-1 has a 167 Ω coil drawing 30 mA, about 150 mW;[5] by Ohm's law, 5 V ÷ 167 Ω ≈ 30 mA (calculated). It also comes with 3 to 24 V coils; the 12 V version is 960 Ω and 12.5 mA.[5] Coil resistance is specified at 23 °C, ±10 percent.[5] Copper's resistance rises about 0.4 percent per °C, so a hot coil needs more voltage to operate.[2]

Pick-up and drop-out

The armature needs a minimum coil current to pull in, the pick-up (pull-in) value. Once in, close to the core, it needs much less to stay, so the current must fall well below pick-up before it drops out. The gap between the two is hysteresis.[7] The G5V-1 is guaranteed to operate by 80 percent of rated voltage (4.0 V) and to release below 10 percent (0.5 V) (Figure 8).[5] In between, its state depends on the individual relay and on its history, so drive the coil at its rated voltage and switch it fully off.

Relay pick-up and drop-out: hysteresis in a 5 volt Omron G5V-1 Graph of relay state, released or operated, against coil voltage from 0 to 6 volts. Shaded bands mark what the datasheet guarantees: at or below 0.5 volts (10 percent of rated) the relay is released, and at or above 4.0 volts (80 percent) it is operated; 5 volts is the rated voltage. An illustrative relay is drawn as a loop: as the voltage rises it stays released until it picks up at 3.0 volts, and as the voltage falls it stays operated until it drops out at 1.5 volts. 0 1 2 3 4 5 6 Coil voltage (V), 5 V relay Operated (COM to NO) Released (COM to NC) 0.5 V (10%): must release 4.0 V (80%): must operate 5 V rated voltage rising: picks up falling: drops out Shaded: guaranteed by the datasheet. Lines: one possible relay (illustrative pick-up 3.0 V, drop-out 1.5 V).
Figure 8. Pick-up and drop-out for a 5 V Omron G5V-1. The shaded limits are from the datasheet (operates by 80 percent, releases below 10 percent of rated voltage); the loop is one illustrative relay.

Contact ratings, life and speed

Relay contacts follow all of section 6. The G5V-1 signal relay is rated 0.5 A at 125 V AC or 1 A at 24 V DC; the G5LE power relay, 10 A at 120 V AC.[5]​[11] Datasheets give a mechanical life with no load and a shorter electrical life at full load: 5,000,000 and 100,000 operations for the G5V-1. It takes up to 5 ms to operate and 5 ms to release.[5]

Driving a relay from a microcontroller

A microcontroller pin cannot safely supply a coil's 30 mA, so a transistor does it (Figure 9). Article 7 worked this example: a 2N3904 with a 1.2 kΩ base resistor switches a 5 V G5V-1 from an Arduino pin, as a low-side switch between coil and ground. A logic-level MOSFET can do the same (recap).

When the transistor turns off, the coil's kickback can destroy it (recap). A flyback diode across the coil, band (cathode) toward the positive supply, gives the current a loop to die away in (recap). Panasonic calls the diode absolutely necessary in a transistor-driven relay circuit. It slows the release to 2 to 5 times the catalog time; a Zener diode in series with it shortens that.[2]

Driving a relay from a microcontroller pin with a transistor and flyback diode Schematic in two halves. Control side: a microcontroller output pin feeds the base of a 2N3904 NPN transistor through a 1.2 kilohm resistor. The transistor's emitter goes to ground and its collector to the bottom of a 5 volt, 167 ohm, 30 milliamp relay coil whose top connects to +5 volts. A 1N4148 flyback diode sits across the coil with its cathode toward +5 volts. A dashed line shows the coil mechanically moving the relay contact. Load side, separate: a 12 volt battery, a lamp and the relay's normally open contact form their own loop, sharing no wire with the control side. Control side (5 V) +5 V Flyback diode 1N4148 Relay coil 5 V, 167 Ω, 30 mA 2N3904 1.2 kΩ Microcontroller output pin coil moves the contact Load side (separate) NO contact + 12 V Lamp No wire is shared with the control side
Figure 9. A 2N3904 transistor driving a G5V-1 relay coil from a microcontroller pin, with a 1N4148 flyback diode, as worked out in article 7. The contacts switch a separate 12 V lamp circuit. Coil values: Omron G5V-1 datasheet.

The contacts form a separate circuit. In Figure 9 they switch a 12 V lamp on its own supply, sharing no wire with the microcontroller.

Latching relays

An ordinary relay is on only while its coil is powered. A latching relay stays in its last position with the coil unpowered; a pulse sets it and another resets it. One-coil types use pulses of opposite polarity; two-coil types have separate set and reset coils.[2] They save power, but the circuit must set a known state at power-up.

Reed relays

A reed relay is a reed switch inside a coil, sometimes with a built-in flyback diode; the coil's field replaces the magnet.[15] Its sealed contacts have a very long mechanical life, especially when switching small signals.[15]​[9]

Contactors

A relay built to switch large power is a contactor. Its contacts are usually normally open, so the load is off whenever the coil is. Its commonest job is starting motors, with three main contacts for a three-phase supply, often 480 V or more, plus a small auxiliary contact for the control circuit.[7]

9. Solid-state relays

A solid-state relay (SSR) has no moving parts (Figure 10). Its input drives an LED whose light crosses an insulating gap to a light-sensitive device; together they form an optocoupler, which provides the isolation. That device switches the output semiconductor:[7]​[16]​[14]

Inside a solid-state relay for AC loads Simplified schematic of a triac solid-state relay. Input side: the input terminals drive an LED through a resistor. Its light crosses an isolation gap, the only link between input and output, to a light-triggered triac driver, which fires the gate of the main triac. The triac connects the two output terminals, with a snubber, a resistor and capacitor in series, across them. Outside, the output terminals are in series with a load and an AC supply. A dashed arrow through the snubber shows that a small leakage current flows even when the SSR is off. Inside the SSR package Input + Input − from a controller LED isolation gap: only light crosses light-triggered driver triac snubber R C Load AC Dashed arrow: even when the SSR is off, a small leakage current flows through the snubber and the triac, enough to light a small lamp faintly or give a shock. Simplified; real SSRs add more parts.
Figure 10. A simplified triac solid-state relay. The optocoupler (LED plus light-triggered driver) gives isolation; the snubber across the output is one reason an "off" SSR still leaks. Sources: Kuphaldt and Omron.

Advantages. No contacts to wear, no sparking, no noise, and faster switching.[7]​[16] Omron quotes about 0.2 ms for a MOSFET relay against 3 to 5 ms for a mechanical one.[14] Zero-cross SSRs turn on only near the point where the AC voltage passes through zero, and a triac turns off when the current does, avoiding spikes and electrical noise.[7]​[16]

Disadvantages.

10. Transistor, relay or SSR?

Choosing an electronic switch
Transistor or MOSFETElectromechanical relaySolid-state relay
Isolation between control and loadNone; grounds are joined (recap)Yes, coil to contacts[7]Yes, through the optocoupler[16]
SpeedMicroseconds (recap)Milliseconds[5]About 0.2 ms (MOSFET relay)[14]
WearNo moving partsAbout 100,000 operations at full load[5]​[11]No contacts to wear[16]
Off-state leakageSmallNone[16]Up to milliamps[17]

11. Testing switches and relays

Test with the power off and the part isolated from the rest of the circuit.[18]

12. Everyday uses

13. Safety

Use parts rated for the job. Anything that switches mains must be rated for that voltage and current, with AC ratings for AC loads. (Recap: household AC and mains safety.)

Contacts can weld. Inrush from lamps, motors and capacitive loads can weld contacts shut, so "off" does nothing.[2] Choose relays with margin, and never rely on one relay alone to stop something dangerous.

An "off" SSR is not isolation. Current leaks through an off SSR and its snubber, so Omron says to turn off the input and load power and check that it is safe before wiring or replacing one.[16] Some leak up to 10 mA.[17] For comparison, 1 mA can be felt, many safety rules take 5 mA as the limit, and 10 to 20 mA can make muscles clamp so a person cannot let go.[19] Because SSRs usually fail shorted, Omron advises a contactor or circuit breaker on the load supply that can cut power if the SSR fails, never the SSR alone.[16]

Never trust a switch alone. Contacts weld and SSRs fail shorted, so a device's on-off switch is not a safety disconnect. Kuphaldt says the disconnect should be separate from the regular switch and used only to make the system safe.[20] Before working on wiring:

Relay coils kick back. Fit every transistor-driven DC relay coil with a flyback diode.[2]

The whole story in brief

Coming up in this section: 13 Batteries and Cells.

Sources

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

  1. Tony R. Kuphaldt, Lessons In Electric Circuits, Volume IV (Digital), chapter 4, "Switches": https://www.ibiblio.org/kuphaldt/electricCircuits/Digital/DIGI_4.html
  2. Panasonic Industry, "Relay Technical Information" (PDF): https://industry.panasonic.eu/storage/custom-upload/Components/Relays/technical_information_relay_en.pdf
  3. E-Switch, Switches: A Comprehensive Guide (switch whitepaper, version 5.0, PDF): https://www.e-switch.com/wp-content/uploads/2025/07/Switch-Whitepaper-v5.0.pdf
  4. Omron Electronic Components, "Switches: Technology" (basic knowledge): https://components.omron.com/us-en/products/basic-knowledge/switches/technology
  5. Omron, G5V-1 low-signal relay datasheet (PDF): https://omronfs.omron.com/en_US/ecb/products/pdf/en-g5v_1.pdf
  6. SparkFun Electronics, "Button and Switch Basics": https://learn.sparkfun.com/tutorials/button-and-switch-basics/all
  7. Tony R. Kuphaldt, Lessons In Electric Circuits, Volume IV (Digital), chapter 5, "Electromechanical Relays": https://www.ibiblio.org/kuphaldt/electricCircuits/Digital/DIGI_5.html
  8. Adafruit Learning System, "Make It Switch" (all pages): https://learn.adafruit.com/make-it-switch?view=all
  9. Standex Detect, "Reed Switch Operational Characteristics" (Reed Technology Academy): https://standexdetect.com/resources/reed-technology-academy/reed-switch-characteristics/reed-switch-operational-characteristics/
  10. Brian Matthews, Wake Forest University, "Tutorial: 3-Way Switches and 4-Way Switches": https://www.matthews.sites.wfu.edu/courses/p230/switches/SwitchesTut.html
  11. Omron, G5LE PCB power relay datasheet (PDF): https://omronfs.omron.com/en_US/ecb/products/pdf/en-g5le.pdf
  12. Jack G. Ganssle, "A Guide to Debouncing" (part 1, switch bounce measurements): http://www.ganssle.com/debouncing.htm
  13. Jack G. Ganssle, "A Guide to Debouncing, Part 2: Solutions": http://www.ganssle.com/debouncing-pt2.htm
  14. Omron Electronic Components, "Relays: Basics" (basic knowledge): https://components.omron.com/us-en/products/basic-knowledge/relays/basics
  15. Pickering Electronics, "Introduction to Reed Relay Basics (Part 1)": https://www.pickeringrelay.com/introduction-reed-relay-basics-part-1/
  16. Omron, Technical Guide for Solid-state Relays (PDF): https://files.omron.eu/downloads/latest/manual/en/csn_ssr_tg_technical_manual_en.pdf
  17. Omron, Solid State Relays Guide Book (PDF): https://edata.omron.com.au/eData/Relays/Y108-E1-1.pdf
  18. Fluke, "A Guide to Continuity Testing with a Multimeter": https://www.fluke.com/en-us/learn/blog/digital-multimeters/how-to-test-for-continuity
  19. OpenStax, College Physics 2e, section 20.6, "Electric Hazards and the Human Body": https://openstax.org/books/college-physics-2e/pages/20-6-electric-hazards-and-the-human-body
  20. Tony R. Kuphaldt, Lessons In Electric Circuits, Volume I (DC), chapter 3, "Electrical Safety": https://www.ibiblio.org/kuphaldt/electricCircuits/DC/DC_3.html
  21. U.S. Occupational Safety and Health Administration, Controlling Electrical Hazards (OSHA 3075, PDF): https://www.osha.gov/sites/default/files/publications/OSHA3075.pdf