Voltage, Current, Resistance, and Power
Ohm's law is easy. Power is where the smoke comes from.
🎯 By the end you'll understand
Voltage as a difference between two points (never a property of one), current as a flow that must return, and resistance as their ratio. The three forms of the power equation and which one to reach for at the bench. The LED series-resistor calculation worked end to end, and what happens to the same resistor when the supply becomes 12 V. What a resistor's power rating actually promises, how it derates with temperature, and why a part that is correct by Ohm's law can still be the wrong part. And how these quantities map onto Beacon's board — including a 1.85 µA leak the canon says Rev A ships with.
KEY INSIGHTOhm's law tells you the current. Power tells you whether the part survives.
A resistor does not know what voltage it is rated for. It knows how many watts it is turning into heat, and how fast that heat can leave. Everything else is bookkeeping.
Here is the bench scene. An engineer builds a status LED on a 3.3 V rail, calculates a 130 Ω series resistor from Ohm's law, picks a 0805 chip resistor because that is what is in the drawer, and it works perfectly. Months later the same subcircuit is copied onto a board with a 12 V input for an industrial customer. The calculation is repeated correctly — the current goes up, the engineer notes the LED is "a bit bright" — and the resistor discolours, the solder joint cracks, and the board comes back from the field with a brown halo on the silkscreen. Ohm's law was applied correctly both times. Nobody computed the power. This lesson is about the difference between those two calculations, and why the second one is the one that matters.
Four quantities, one law, one equation you must never skip. Voltage (V) is a difference in electrical potential between two points. Current (I) is the rate of charge flow through a path, and it always returns. Resistance (R) is the ratio V/I for a component. Power (P = V·I = I²R = V²/R) is the rate at which electrical energy becomes heat, and it is the only one of the four that destroys parts. Dissipation is calculated, not felt: a component rated far above its voltage can still cook on power.
CORE CONCEPTVoltage is between two points. Current returns. Resistance is the ratio.
These three words are used so casually that people forget what they actually mean, and the forgetting produces real bugs. Let's define each one in the way that prevents those bugs.
Voltage: a difference, never a property of one point
Voltage is electrical potential energy per unit charge, and like any potential energy it only means something as a difference. "The rail is at 3.3 V" is shorthand for "the rail is 3.3 V above the point we agreed to call 0 V — ground." Change the reference and every number changes. This is not pedantry. When a multimeter says a node is at 3.3 V, it is telling you about the node and about where you clipped the black lead. When two boards with separate supplies are connected by a signal wire but no ground wire, "3.3 V on board A" and "3.3 V on board B" have no defined relationship, and the receiver sees garbage — a floating-ground bug you will meet properly in Lesson 1.7.
Current: a flow that must return
Current is the rate at which charge passes a point, measured in amperes (1 A = 1 coulomb per second). The word that matters is through. Current is a property of a path, not a point, and every path is a loop: whatever leaves the positive terminal of a battery must come back to the negative one. There is no such thing as "sending 20 mA into an LED" without 20 mA coming out the other side and finding its way home. When a return path is missing — a ground pin not connected, a sensor with no common ground to the microcontroller — the current does not flow "a little"; it does not flow at all, or it finds a return path you did not design, through an input-protection diode, for instance, and destroys it. Every current you budget in this course is also a return current you must route.
Resistance: the ratio, and Ohm's law
For an ideal resistor, the current through it is proportional to the voltage across it, and the constant is the resistance:
V = I · R ⇔ I = V / R ⇔ R = V / I
Notice the prepositions: voltage across the resistor, current through it. A 130 Ω resistor with 1.3 V across it passes 10 mA. That is all Ohm's law says, and for a resistor at a fixed temperature it is exact enough to bet a product on. The law does not apply to LEDs, diodes, or transistors, whose current-voltage relationship is exponential rather than linear — which is why an LED needs a resistor in the first place, and why you cannot "just connect an LED to 3.3 V" and expect a sensible current.
EXAMPLEThe water analogy — and exactly where it breaks
Voltage is pressure, current is flow rate, resistance is a narrow pipe. The analogy is genuinely useful: it explains why a bigger pressure difference drives more flow, why a narrower pipe passes less, and why pressure is measured between two points (a pressure gauge reads relative to something). Use it. But know its three failure points, because each one corresponds to a real bug.
1. No return path, no current. Water can leave a tank and pool on the floor. Charge cannot: a wire connected to +3.3 V and to nothing else carries zero current, however "high the pressure." Beginners who imagine current being pushed into a pin forget that it must be drawn out of another. 2. No pressure without flow, in a wire. A closed tap holds pressure with no flow; a wire with no current has (nearly) no voltage across it, because V = I·R and R is tiny. When you measure 0 V across a trace, it does not mean the trace is dead — it means nothing is flowing through it, or it is a very good conductor. 3. Water has mass and momentum; charge in a wire effectively does not. Turn off a valve and pipes hammer. Open a switch and current stops in nanoseconds — unless there is an inductor in the loop, in which case the analogy suddenly works again and the flyback spike (Lesson 1.5) is exactly water hammer. The analogy is a tool for intuition, not a substitute for the equations.
One more place the analogy misleads: it suggests that a "high voltage" is dangerous and a "low voltage" is safe. For parts, what is dangerous is power. A 1.5 V AA cell shorted through a thin wire delivers amperes and melts insulation; a 3.3 V rail through a 1 MΩ resistor delivers 3.3 µA and nothing happens. The next section makes that precise.
HOW IT WORKSPower: the quantity that destroys parts
Power is the rate at which energy is converted, in watts (1 W = 1 joule per second). In a resistor, all of that energy becomes heat, and the heat has to leave through the package into the board and the air. If it leaves slower than it arrives, the temperature rises until either the two balance or the part fails. The electrical power delivered to any two-terminal element is:
P = V · I
with V across the element and I through it. For a resistor, substituting Ohm's law gives two more forms that are algebraically identical and practically different:
P = I² · R P = V² / R
Which form you use depends on what you know. If you know the current through a series element (a sense resistor, a series resistor), use I²R — and notice the square: doubling the current quadruples the heat. If you know the voltage across a shunt element (a bleed resistor across a rail, a pull-up held low), use V²/R. If you know both, P = V·I is quickest and works for non-linear parts too — it is how you compute the dissipation in an LED, a diode, a regulator, or a MOSFET, none of which obey Ohm's law.
Worked example: the LED series resistor, end to end
Decide the current, then find the LED's forward voltage
A standard red indicator LED is comfortable at 10 mA (bright enough for a status light; well under a typical 20 mA absolute maximum for a small indicator LED). At that current its forward voltage Vf is about 2.0 V — read this from the LED's datasheet I-V curve, because it varies by colour: red and infrared sit near 1.8–2.0 V, green and blue nearer 3.0 V or more. The forward voltage is not a design choice; it is a property of the semiconductor.
The resistor takes what is left
The supply is 3.3 V and the LED takes 2.0 V, so the resistor must drop 3.3 − 2.0 = 1.3 V while passing 10 mA. Ohm's law: R = 1.3 V / 0.010 A = 130 Ω. That is an E24 value; 120 Ω or 150 Ω from the E12 series would give 10.8 mA or 8.7 mA, both fine. Pick 130 Ω for the rest of the example.
Now — and this is the step people skip — compute the power
PR = I²R = (0.010)² × 130 = 13 mW. Or P = V·I = 1.3 × 0.010 = 13 mW, same thing. The LED itself dissipates 2.0 V × 10 mA = 20 mW. An 0402 chip resistor is typically rated 1/16 W = 62.5 mW, so the resistor is at about 21 % of its rating. Any package works. The circuit is fine, and it is fine because you checked, not because 3.3 V "sounds low."
The same resistor at 12 V
Someone reuses the subcircuit on a 12 V board. The LED still drops about 2.0 V (slightly more at higher current), so the resistor now drops 10 V: I = 10 / 130 = 77 mA. Two things are wrong at once. The LED is at almost four times its 20 mA maximum and will fail — quickly and brightly. And the resistor dissipates I²R = (0.077)² × 130 = 0.77 W. A 0805 is rated 1/8 W = 0.125 W; this is 6× over. Even a through-hole 1/4 W part is 3× over. If the LED fails short, the resistor sees the full 12 V: 92 mA and 1.1 W. It discolours, its resistance drifts upward as the film degrades, and eventually it opens — a slow, ugly failure that looks like "the LED stopped working" to whoever finds it.
Every time you write down a resistor value, write the power next to it, in the schematic or in your notes. It takes five seconds. "130 Ω, 13 mW" is a design; "130 Ω" is a guess about the supply voltage that will be wrong the day someone changes the supply. Experienced engineers argue about many things, but not about this one.
Power in things that are not resistors
P = V·I is general. A linear regulator dropping 4.2 V to 3.3 V at 40 mA dissipates (4.2 − 3.3) × 0.040 = 36 mW — in a small package, harmless. The same regulator dropping 12 V to 3.3 V at 200 mA dissipates 8.7 × 0.2 = 1.74 W, which will destroy a SOT-23 part in seconds. A charger dropping 5 V to 3.7 V at 500 mA dissipates 0.65 W. A MOSFET carrying 1 A with 50 mΩ on-resistance dissipates I²R = 50 mW; the same MOSFET half-on at 2 Ω dissipates 2 W and is, as Lesson 1.5 puts it, a heater. In every case the question is identical: how many watts, into what package, at what ambient temperature?
SEE ITLive circuit bench — the correct resistor is still the wrong part
Below is a bench with a supply, a resistor, an optional LED in series, and a choice of package. The model is simple and honest: current from Ohm's law, dissipation from I²R, and a film temperature from a thermal model that matches the standard thick-film derating curve (100 % rated power at 70 °C ambient, derating linearly to zero at 155 °C film temperature). Set up the 3.3 V LED circuit first. Then change only the supply.
What to notice. With the LED at 3.3 V, all four packages are comfortable — Ohm's law gave you a number and the power check confirmed it. Push the supply to 5 V and the current rises to 23 mA: the LED is already past its rating while the resistor is still fine, which is the reason "the LED got brighter" is a warning, not a feature. At 12 V the 0805 is at six times its rating and the film temperature is off the chart; and even a through-hole 1/4 W part is three times over its rating — at this power nothing in the list survives. Then switch to "resistor only" and lower the value: a 10 Ω resistor across 3.3 V passes 330 mA and dissipates 1.1 W — proof that a low voltage is not a safe voltage. Finally, set 100 °C ambient and watch the allowed power fall to about 65 % of the printed rating — a part sitting at 70 % of its rating is suddenly over its limit with no change to the circuit: this is what a derating curve means.
DEFINITIONWhat "rated 1/8 W" actually promises
A resistor's power rating is a promise with conditions attached, and the conditions are the whole story. The standard thick-film chip resistor rating means: this part may dissipate its rated power continuously at an ambient temperature of 70 °C or below, mounted on a specified test board with specified copper, and under those conditions its film will not exceed about 155 °C. Above 70 °C ambient the allowed power falls linearly to zero at 155 °C. That is the derating curve, and manufacturers (Vishay, Yageo, Panasonic, KOA) publish essentially the same one for standard thick-film parts.
| Package | Typical thick-film rating | Rated at | Power at 100 °C ambient (derated) | Bench reality |
|---|---|---|---|---|
| 0402 | 1/16 W (62.5 mW) | ≤ 70 °C | ≈ 40 mW | Fine for signal paths, pull-ups, LED resistors at 3.3 V |
| 0603 | 1/10 W (100 mW) | ≤ 70 °C | ≈ 65 mW | The default on Beacon; comfortable for everything below ~50 mW |
| 0805 | 1/8 W (125 mW) | ≤ 70 °C | ≈ 81 mW | Sense resistors, higher-current LED drive |
| 1206 | 1/4 W (250 mW) | ≤ 70 °C | ≈ 160 mW | Where a 0805 runs warm; note some 1206 are only 1/8 W |
| Axial through-hole | 1/4 W (250 mW) | ≤ 70 °C | ≈ 160 mW | Breadboards; leads carry heat better than a chip's end caps |
Three refinements experienced engineers add. First, the test board matters: a 0603 on a big ground pour runs cooler than the same part on a thin trace, because the pads and copper are the heat path. Datasheet ratings assume a specific test pattern, and a dense 35 × 25 mm board with parts crowded together is not that pattern. Second, a rule of thumb many teams adopt is to keep continuous dissipation under 50 % of the rating — not because the datasheet is lying, but because the derating curve, board density, and pulses conspire. Third, pulse ratings are separate: a chip resistor can survive a short pulse far above its continuous rating (the film heats faster than it can fail), which is why a 0603 across a 100 nF capacitor discharge is usually fine. The pulse curves are in the datasheet too, and Lesson 1.6 teaches you to find them.
BEACONBeacon's power map — where the watts and the microamps live
Beacon is a low-power device, so the interesting question is not "what gets hot" but "where does each quantity matter." The table applies this lesson's four quantities to the board as fixed in the project canon.
| Element | V across | I through | P = V·I | Which quantity bites |
|---|---|---|---|---|
| MCP73831 charger, mid-charge | 5.0 − 3.7 = 1.3 V | 500 mA | 0.65 W | Power — the hottest part on the board; thermal foldback |
| XC6220 LDO, active burst (breadboard stages) | 4.2 − 3.3 = 0.9 V | ≈ 40 mA | 36 mW | Harmless — but its ≈ 8 µA quiescent current is not (Lesson 4.2) |
| Battery divider, 2 × 1 MΩ, if left connected | 3.7 V | 1.85 µA | 6.8 µW | Current — 1.85 µA is 23 % of the entire 8 µA sleep budget |
| USB D± series resistors, 22 Ω | mV | mA, briefly | negligible | Neither — they exist for signal integrity, not power |
| WS2812B-2020 LED, full white | 3.3 V | tens of mA | ≈ 0.1 W | Current — must be cut in sleep by the PMOS (Lesson 1.5), or it dominates the budget |
| ESP32-C3 module, BLE burst | 3.3 V | ≈ 40 mA avg (peaks higher) | ≈ 0.13 W for 45 ms | Energy per burst, not heat — 45 ms is too short to warm anything |
The divider row deserves a closer look because it is the canon's first confessed defect. Two 1 MΩ resistors in series across a 3.7 V cell pass 3.7 / 2,000,000 = 1.85 µA, permanently, whenever they are connected. At 6.8 µW that is thermally nothing; as current it is nearly a quarter of the board's 8 µA sleep budget, which is why the design switches the divider with a small PMOS so it only draws current during the few milliseconds of a battery measurement. Rev A's post-mortem (Lesson 6.9) records that the divider leaks about 1.8 µA more than planned — which is exactly this number, and tells you before you read the schematic what went wrong: the switch is not actually disconnecting it. The habit from this lesson — compute I through everything, not just P — is what lets you predict that.
Beacon's breadboard in Lesson 1.8 uses a plain red LED and the XC6220 regulator before the WS2812B and the buck arrive. Design the LED resistor now: 3.3 V rail, red LED at 5 mA (dim is fine for a status light, and cheaper in energy) → R = (3.3 − 2.0) / 0.005 = 260 Ω → 270 Ω, P = 0.005² × 270 = 6.75 mW. Any package. Write "270 Ω, 6.8 mW" on the schematic. When you measure the current in Lesson 1.8 and get 4.6 mA instead of 5, the reason is the LED's real Vf at that current (probably 2.05–2.1 V) — a datasheet-conditions lesson for 1.6.
WEAKNESSWhere this bites
Chip resistors carry a voltage rating (often 50 V for 0603, 150 V for 0805) and beginners treat it as the safety limit: "12 V is below 150 V, fine." The voltage rating is about breakdown across the part, not about heat; a 130 Ω 0805 at 12 V is far below its voltage rating and six times over its power rating. Capacitors are the same the other way round: a 16 V ceramic on a 3.3 V rail can be perfectly safe on voltage and still lose most of its capacitance to DC bias (Lesson 1.4). Read every rating as "the limit of one specific failure mode" and check the others separately.
A 0.1 Ω current-sense resistor at 1 A dissipates 0.1 W — fine in a 0805. At 3 A, intuition says "three times more"; the square says nine times: 0.9 W, and the 0805 is destroyed. The same square is why a 2× current overload is a 4× thermal overload, why sense resistors are chosen by power before precision, and why the "slightly higher" current of the 5 V LED example is more dangerous than it looks. Whenever a current changes, square the ratio before deciding the change is small.
"The pin is at 3.3 V." Measured against what? The most expensive version of this mistake is connecting two separately-powered boards by a data line without a ground: each board's "3.3 V" is defined only relative to its own ground, and the two grounds can sit volts apart. The receiver sees a signal swinging around an undefined reference, sometimes works, and one day a transient pushes it past an absolute maximum. The related bench version: measuring a "0 V" across a trace and concluding the trace is dead, when it just carries no current. Voltage is between two points. Always name both.
KEY INSIGHTHow this connects
- What an Embedded System Is (1.1) — the electrical and thermal failure classes from the taxonomy are the two halves of this lesson: wrong current is electrical, and P = V·I is the bridge to thermal.
- Kirchhoff's Laws and Circuit Analysis by Hand (1.3) — "the drops sum to the source" becomes a law, and the 2 × 1 MΩ divider from the power map is analysed properly: why it reads low into an ADC and why the fix is not a smaller divider.
- Passive Components for Real (1.4) — the power rating and derating curve here are one of several fine-print clauses on a resistor; tolerance, temperature coefficient, and the capacitor's DC-bias derating are the rest.
- Diodes, Transistors, and MOSFETs as Switches (1.5) — P = V·I applied to parts that do not obey Ohm's law, including the half-on MOSFET that becomes a heater and the PMOS that cuts Beacon's LED in sleep.
Every calculation in the rest of the course reduces to these four quantities. A pull-up value is a current at a voltage; a trace width is a resistance and a power; a regulator's efficiency is a ratio of powers; a battery life is a current integrated over time. When a later lesson's number looks arbitrary, come back here and ask: what is V, what is I, what is R, and what is the P nobody wrote down?
GLOSSARYKey terms introduced
- Voltage (V)
- Electrical potential difference between two points, in volts. Always a difference; "3.3 V" is shorthand for "3.3 V above the agreed reference (ground)."
- Current (I)
- Rate of charge flow through a path, in amperes. A property of a loop, not a point; every current has a return path, designed or otherwise.
- Resistance (R)
- The ratio V/I for a linear element, in ohms. Ohm's law V = I·R holds for resistors and wires, not for diodes, LEDs, or transistors.
- Power (P)
- Rate of energy conversion, in watts. P = V·I for any element; P = I²R = V²/R for resistors. The quantity that becomes heat and destroys parts.
- Forward voltage (Vf)
- The voltage across a diode or LED when conducting, set by the semiconductor, not the circuit (≈ 2.0 V for red, ≈ 3.0 V+ for blue/white). Read from the datasheet I-V curve at the intended current.
- Power rating
- The continuous dissipation a part can tolerate under specified conditions (typically ≤ 70 °C ambient on a test board). Above that ambient the allowed power derates linearly to zero at the maximum film temperature (≈ 155 °C for thick film).
- Derating curve
- The published relationship between ambient temperature and allowed power (or voltage, or current). Turns a single rated number into the function it actually is.
ACTIVE RECALLCheck your understanding
Answer in your own words first, then reveal.
1A green LED (Vf ≈ 3.0 V) is to run at 8 mA from a 5 V rail. Find the resistor and its dissipation, and pick a package. Then repeat with the same resistor on a 3.3 V rail — what happens?›
Resistor drop = 5.0 − 3.0 = 2.0 V; R = 2.0 / 0.008 = 250 Ω → 240 Ω or 270 Ω. Dissipation P = I²R = 0.008² × 250 = 16 mW (or 2.0 V × 8 mA = 16 mW). A 0402 (62.5 mW) is at 26 %; any package works. On a 3.3 V rail the resistor has only 0.3 V to drop: I = 0.3 / 250 = 1.2 mA. The LED is barely lit and — this is the useful part — the circuit is not damaged, just non-functional. Lowering the supply is safe; raising it is what destroys parts. Also notice that Vf on the green LED's datasheet curve will be somewhat lower at 1.2 mA than at 8 mA, so the real current is a little higher than 1.2 mA; Ohm's law only applies to the resistor.
2Why is a 1.5 V AA cell more dangerous to a thin wire than a 3.3 V rail through a 1 MΩ resistor? Which of the four quantities decides?›
Power decides, and power depends on current as much as voltage. The AA cell has a low internal resistance (a fraction of an ohm) so a shorting wire draws amperes: P = I²R in the wire and the cell's own resistance can be several watts, enough to melt insulation. The 3.3 V rail through 1 MΩ delivers I = 3.3 µA and P = V²/R = 10.9 µW — nothing. Voltage alone tells you neither; the water analogy's "high pressure is dangerous" instinct is exactly what misleads here. Ask how much current will flow and multiply.
3Beacon's 2 × 1 MΩ battery divider dissipates only 6.8 µW. Why does the canon bother switching it off with a PMOS, and what does the Rev A post-mortem's "1.8 µA extra leak" tell you before you look at the schematic?›
Because the constraint on Beacon's sleep state is current, not power. The divider's 3.7 V / 2 MΩ = 1.85 µA is trivial as heat but is 23 % of the entire 8 µA sleep budget, which is why it is switched so it only conducts during a battery measurement. A measured extra leak of about 1.8 µA is the divider's full-time current almost exactly — so before opening the schematic you can say the PMOS switch is not actually isolating the divider in sleep: either the switch is on the wrong side, its gate is not being pulled to the right rail in deep sleep, or the ADC pin's own path provides a return. The lesson: compute I through every element, not only P, and compare each I against the budget it lives in.
4A 0805 resistor is dissipating 90 mW on the bench at 25 °C, well under its 125 mW rating. The product must operate inside a sealed enclosure that reaches 85 °C. Is the part still adequately rated? Show the derating.›
No. The standard thick-film curve gives 100 % up to 70 °C, then linear to zero at 155 °C. At 85 °C ambient: allowed fraction = (155 − 85) / (155 − 70) = 70 / 85 ≈ 0.82, so the 0805 is now a 0.82 × 125 = 103 mW part. 90 mW is 87 % of that — technically inside, with no margin for board density, pulse loading, or a hot neighbour. Most teams would move to a 1206 (250 mW → 206 mW at 85 °C, 44 % loaded) or halve the dissipation. The general lesson: a rating is a curve, and the bench at 25 °C is the most flattering point on it.