Why
that part?

A project photograph can look like a drawer of unrelated objects. The trick is to stop asking “What are all these names?” and ask “What needs to happen here?” A door needs to detect a person, decide to open, move, and stop at the right position. Those jobs lead you toward a sensor, a controller, a motor driver, a motor and a mechanism.

There are three trails to follow: energy from a source to a load, information from an input through a decision to an output, and force through a mechanism. A MOSFET belongs to the power-control trail. A bearing belongs to the mechanical trail. A sensor starts an information trail. A PCB and its connectors hold the electrical trails together.

Recognition sketches, not identification guarantees. A package shape does not establish its internal function or leg order.

A component, module and system are different levels. A MOSFET is a component. A motor-driver module contains components and connectors. A car combines modules, wiring and mechanics. Learning the layers prevents two common mistakes: assuming a bare chip includes everything its module includes, and treating the module as a magic box whose connections need no explanation.

SOURCE TRAIL · Source foundation: 1N.1, PDF pp. 31–33; 26N project survey. The component atlas and mechanical selection material are added for your goal.

A battery has two terminals. Voltage is the potential difference between them. Connect a complete conducting path and current can flow. A break anywhere in the series path stops the steady current everywhere in that path. Current does not arrive at a lamp, get used up, and disappear: charge continues around the circuit while energy is transferred to light and heat.

A resistor obeys V = I × R approximately over its normal operating range. For 3.3 V across 1 kΩ: I = 3.3 / 1000 = 0.0033 A = 3.3 mA. An LED is not a resistor. In the workbench we use a simplified 2.0 V red-LED drop, leaving 1.3 V across the series resistor. That makes about 3.9 mA with 330 Ω.

Power is P = V × I. A resistor dissipates I²R. Double its current and the heat becomes four times as large. A “¼ W” marking describes a dissipation rating under stated conditions, not a current it always consumes.

SOURCE TRAIL · 1N.2, printed pp. 5–10 / PDF pp. 33–38. Workbench uses an original, explicitly simplified LED example.

A node is a set of points electrically joined together. Wires can wander all over a drawing and still belong to one node. Components separate nodes. On a breadboard, metal strips hidden under the holes create nodes for you. Putting both legs of a resistor into the same connected strip bypasses its intended job.

The drawing is a top view of a simplified mini board with continuous side rails. Real boards can have split rails, different orientations or no rails. With power disconnected, use continuity mode to prove which holes connect. A four-leg pushbutton usually has two internally joined pairs: rotating it can turn your intended switch into a permanent connection.

A schematic can omit long return wires by placing identical ground symbols at several points. They still mean an electrical connection. The ground in a battery project is usually the chosen local zero-volt reference, not a literal wire to the soil.

SOURCE TRAIL · 1L.1–1L.2, printed pp. 25–27 / PDF pp. 53–55; 1N.4.5, printed pp. 20–21 / PDF pp. 48–49.

A pull-up resistor connects an input to a positive logic supply so it has a defined state when a switch is open. A button to ground overrides it when pressed. At 3.3 V with 10 kΩ, pressing the button draws only 0.33 mA through the resistor. Without the resistor, connecting the input directly to 3.3 V would make pressing that button short the supply.

A voltage divider uses two series resistors. The middle voltage is Vin × Rbottom / (Rtop + Rbottom) when the attached load draws negligible current. “Negligible” is doing important work in that sentence.

The load sits in parallel with the lower resistor. A 10 kΩ load makes their combined resistance 5 kΩ, so the output becomes 3.3 × 5 / (10 + 5) = 1.10 V. This is why a divider can scale a sensor signal but usually cannot power a motor or a changing electronic load. Source resistance and load resistance are part of the design.

Read common markings: 4k7 means 4.7 kΩ; 1R0 means 1 Ω; a common three-digit SMD code 103 means 10 followed by three zeros = 10 kΩ. Other coding systems exist. Confirm a loose resistor with a meter; do not trust colour under poor lighting.

SOURCE TRAIL · 1N.3–1N.4 and 1L.2, PDF pp. 39–55. Values and workbench are adapted to a 3.3 V learning circuit.

When a capacitor charges through a resistor, its voltage rises quickly at first and more slowly as it approaches the supply. The current falls as the voltage difference across the resistor shrinks. It does not keep charging forever at the starting current.

The time constant is τ = R × C. After one τ, voltage has reached about 63% of its final change; after five τ, about 99%. For 10 kΩ and 100 µF, τ is one second. A larger resistor or capacitor makes the response slower. An RC low-pass filter uses the same behavior to smooth rapid changes, at the cost of delay.

A common ceramic marking 104 means 100,000 pF = 100 nF = 0.1 µF. A can marked “100 µF 16 V” specifies capacitance and a voltage limit. The stripe on many aluminium electrolytics marks the negative terminal; other capacitor families can mark polarity differently. Read the part data.

SOURCE TRAIL · 2N.1–2N.2 and 2N.4.2, printed pp. 51–55 and 75 / PDF pp. 79–83 and 103–104.

A forward-biased silicon diode often drops roughly 0.6–0.8 V at modest current; that is a rough operating observation, not a universal fixed number. Schottky diodes can have a lower forward drop but different leakage and voltage limits. LEDs vary with colour, current and temperature. That is why a series resistor or current driver belongs with an LED.

A motor winding or relay coil is inductive: its current cannot change instantaneously. Open its switch and it can raise the voltage sharply in an attempt to keep current flowing. A flyback diode across a DC coil gives that current a controlled loop. Its cathode K goes to the positive supply; anode A goes to the switched low side. During normal on-time, it is reverse biased.

The MOSFET’s built-in body diode is not automatically a replacement for the diode across a low-side-switched coil: its location and direction differ. A diode also changes coil release behavior; ordinary flyback suppression can slow relay release. Faster release designs use a different clamp chosen within the driver and coil limits.

SOURCE TRAIL · 3N.5, PDF pp. 146–147; 4L.5.2, printed pp. 175–176 / PDF pp. 203–204. The protection circuit is redrawn and explained for a small DC load.

An NPN bipolar transistor has base, collector and emitter. A base resistor limits drive current. A useful switching design deliberately provides enough base current to saturate the transistor, within the GPIO and transistor limits. Do not size a switch using only a typical small-signal gain number.

An N-channel enhancement MOSFET has gate, drain and source. Positive gate-to-source voltage enables conduction. The insulated gate needs little steady current, but each transition requires charge to move. Large gate charge and rapid switching may call for a separate gate-driver IC.

*The shown 100 Ω gate resistor and 100 kΩ pull-down are illustrative starting values for a small, slowly switched load. The series resistor affects edge rate and peak gate current; the pull-down prevents a floating gate during reset. Pick the final values from drive capability, gate charge and measured switching behavior.

Suppose a MOSFET has 0.05 Ω on-resistance under the actual drive and temperature conditions. At 1 A, conduction loss is I²R = 0.05 W. At 3 A, it becomes 0.45 W. A headline “30 A” rating does not tell you that a tiny uncooled assembly can sustain 30 A. Thermal conditions and switching losses still matter.

Look at the complete package drawing before wiring. A metal tab can be electrically connected to the drain. A TO-220 silhouette does not guarantee a gate–drain–source order, and an SMD adapter can reorder labels. The diagram shows terminal roles, not left-to-right physical legs.

SOURCE TRAIL · 4N.5 and 4L.5, PDF pp. 194–204; 12N.1–12N.2 and 12L.1, PDF pp. 493–498 and 514–516. MOSFET package and threshold footnotes were inspected visually.

NO means normally open; NC means normally closed; COM is the moving contact’s common terminal. In a simple normally-off lamp circuit: load supply positive → COM; NO → lamp; lamp return → load supply negative. Applying the correct coil drive closes the circuit. The coil does not generate the voltage at NO.

A bare relay normally has no “logic input.” Its coil may need tens of milliamps or more, plus a flyback path. A relay module may add a transistor, diode, indicator and optocoupler. Its input may be active-low or incompatible with 3.3 V. Check the module schematic; an optocoupler on a board does not guarantee isolation if grounds or jumpers reconnect the sides.

A printed AC contact rating does not establish the same DC switching rating. An inductive load or motor startup can be more demanding than a resistive lamp. Read the load category and endurance conditions. Keep all learning experiments on low-voltage DC; mains switching calls for a different level of enclosure and protection design.

SOURCE TRAIL · 12L introduction, PDF p. 514, motivates comparison with a relay. Coil/contact selection and practical relay wiring are added material; see OMRON references in the source map.

A linear regulator behaves like a controlled series element, losing the excess voltage as heat. Its approximate dissipation is (Vin − Vout) × Iout. A buck switches to step voltage down; a boost steps it up; a buck-boost can regulate across input voltages above and below its output. Switching converts energy—it cannot create extra power.

A 9 V to 5 V linear regulator at 0.2 A dissipates 0.8 W. Whether that is tolerable depends on package, copper, airflow and temperature. Also check dropout: a regulator needs sufficient headroom to maintain its output. Capacitor type and value can affect stability; use the manufacturer’s application circuit, not a random “bigger is better” capacitor.

Budget a motor supply from loaded current, startup or stall current, and simultaneous loads. A microcontroller reset when a motor starts often indicates supply sag or a poor return path. Place local capacitors near the driver as specified, and avoid routing motor return current through the logic board or a breadboard rail.

Battery Ah describes stored charge. Wh ≈ nominal volts × Ah is a more useful energy comparison across voltages. Runtime estimates must allow for varying load, conversion loss and cutoff. Design for the pack’s full-charge voltage as well as nominal voltage; use a matching commercial charger and properly assembled pack.

SOURCE TRAIL · 11N.1–11N.3 and 11N.6, PDF pp. 461–478. Heat, dropout and conversion concepts are condensed; battery and wiring decisions are added.

A brushed DC motor usually has two power leads. Reverse the voltage polarity and rotation usually reverses. A gearmotor adds a gearbox. A positional hobby servo adds gears, sensing and a controller; a continuous-rotation servo interprets its command differently. A stepper needs sequenced winding currents. A brushless motor needs electronic commutation, often provided by an ESC. These drivers are not interchangeable.

When a brushed motor spins it generates back EMF opposing the applied voltage. At startup or stall there is little back EMF, so current can be much higher. No-load current describes a lightly spinning motor, not the hardest case. A driver sized only for that figure may fail or shut down.

PWM rapidly alternates drive states. Frequency is how often the period repeats; duty is the active fraction. It changes the average electrical drive, but duty is not RPM. A stalled motor at low duty may still heat. Reverse only after stopping for your first bench experiments; rapid reversal can create large current and return energy to the supply.

Choose a driver with documented sustained current and startup handling at your supply and cooling conditions. “2 A peak” is not “2 A indefinitely.” Check its truth table, sleep/enable state, fault behavior and current-limit implementation. A gate driver moves MOSFET gate charge; a motor driver may include the power bridge; a software driver is code. The shared word does not make these the same thing.

SOURCE TRAIL · 8L.4 PWM motor drive, PDF pp. 368–369; 10N/10L motor feedback, PDF pp. 435–457; 26N mechanical devices, PDF pp. 1072 onward. Driver selection is an added practical synthesis.

For a simple external gear pair, reduction = driven teeth / driving teeth. The output turns in the opposite direction. A 12-tooth input driving 24 teeth gives 2:1 reduction: half speed and about twice torque before losses. Several stages multiply their ratios. If a gearmotor specification already gives output RPM, do not divide it by the internal ratio again.

Choose meshing gears with compatible module and pressure angle, plus appropriate tooth form, material and load capacity. Correct ratio alone does not make two gears compatible. A D-shaft and matching clamping hub can keep a wheel from slipping; a plain hole of approximately the right size cannot guarantee torque transmission.

Wheel ground speed is approximately π × diameter × RPM / 60. Wheel force is approximately axle torque / wheel radius. Bigger wheels travel farther per turn but need more torque for the same push. Tires can slip, so available motor torque is not the only limit.

A bearing’s bore fits the shaft; outside diameter fits the housing; width affects the assembly spacing. Radial load pushes across the shaft; axial load pushes along it. Select for the actual load direction, speed and environment. A deep-groove ball bearing often carries radial and some axial load, but its exact ratings govern. A bushing may be sufficient for a simple slow mechanism.

SOURCE TRAIL · Mechanical supplement. Source book: 10L mechanism example, PDF p. 453, and 26N survey. Gear and bearing selection references: KHK and SKF in the source map.

A thermistor changes resistance with temperature; it often needs a divider. A potentiometer provides a position-related divider voltage. A Hall sensor responds to a magnetic field. An encoder produces rotation-related pulses. An IMU sends measured motion over a digital bus. “Digital” means there is a defined electrical and protocol interface, not that every three-pin sensor can be plugged into the same socket.

An op-amp can buffer or amplify a small signal; a comparator reports which of two input voltages is larger. Some sensor modules provide both AO (analog) and DO (thresholded digital) outputs. Their potentiometer may set the comparison threshold rather than calibrating the sensor’s true physical measurement.

Hysteresis gives a threshold two boundaries. For a light-controlled output, you might switch on above 2.0 V and switch off below 1.5 V. Between them, keep the previous state. That avoids rapid chattering when noise nudges a signal around one boundary. Those voltages are an example, not universal sensor thresholds.

SOURCE TRAIL · 6N feedback and op-amps, PDF pp. 273–289; 8N.2 comparators, PDF pp. 348–354; 18N.1–18N.2, PDF pp. 717–724. Sensor interface survey is added material.

A schematic symbol might label a transistor’s pins G, D and S. Its PCB footprint gives pads numbers 1, 2 and 3. The library mapping must connect those identities correctly for the exact package. A beautiful board with a reversed footprint is still wrong. Check top-versus-bottom views, pin-1 markers and mirrored connector drawings.

A via connects copper layers. A ground plane provides a broad reference/return conductor. A test point lets you measure a useful node. Decoupling parts need short connections to the pins they support. Fast current loops should be compact; long paths add resistance and inductance.

“JST connector” is not a complete specification: many distinct connector families exist. Generic friction-fit jumpers are convenient for bench signals but poor choices for high motor current and a vibrating car. A screw terminal also needs correct wire preparation and tightening; a wire merely touching the metal is not a reliable connection.

Move from breadboard to soldered perfboard or a suitable carrier when the circuit works. Inspect joints under magnification, check continuity while gently moving cables, and label both ends of every power cable. Fit standoffs so conductive hardware cannot touch exposed copper. Avoid loose strands and unsupported heavy wires.

SOURCE TRAIL · Practical construction extends 1L.1, PDF pp. 53–54, 4L.6 noise, PDF pp. 204–205, and 9N/9S wiring and stability discussions. Connector and footprint selection are additional material.

You do not have to read every page of every datasheet. Begin with your job: supply range, expected current, control voltage, switching rate, environment and physical constraints. Then inspect the pinout, recommended operating conditions, electrical characteristics and typical application. Treat absolute maximum ratings as boundaries against damage, not desirable operating points.

Here is a fictional comparison for a small 5 V, 1 A load switched by a 3.3 V GPIO. Assume voltage ratings are otherwise sufficient; the incomplete data are deliberate.

Now choose a hypothetical 5 V relay whose coil power is 0.36 W. Coil current is approximately 0.36 / 5 = 72 mA. That points you toward a driver stage, not direct GPIO drive. The contact rating is a separate question determined by the load. The relay’s “5 V” label alone never answers both questions.

For mechanics, make the same kind of decision sheet: shaft diameter and shape, bearing dimensions, rated load, hub attachment and mounting clearance. An exact dimension with a tolerance is more useful than “small bearing” or “fits an RC car.”

A typical curve illustrates behavior; a minimum/maximum specification states a guarantee under its stated test conditions. A module seller’s description cannot replace its schematic or a thermal evaluation. Use curves to understand trends, then design with margins and verify on the bench.

SOURCE TRAIL · 12L.1, PDF pp. 514–516, emphasises drive conditions, misleading current headlines and dissipation. The candidate comparison and selection-card method are original exercises.

Imagine a small knob-controlled fan. It is a better first decoding exercise than a full car: one motor, one direction, no radio. You want the knob to change speed. Before writing code, draw three paths: knob voltage to controller; controller command to switch; motor-supply energy through the motor and switch back to the source.

This is a system view. The complete MOSFET circuit in chapter 7 adds the gate pull-down and identifies the gate resistor; the actual driver also needs its specified supply decoupling. Do not treat this overview as a bare-part construction recipe. Use either a fully specified discrete design or a documented low-side driver module compatible with your load and 3.3 V logic.

There are two supplies in this bench arrangement: the controller’s normal USB-derived logic supply, and a separate current-limited motor supply. Their grounds share a reference. Their positive rails are not casually joined. Motor current should return by an adequate conductor to its supply, rather than wandering through the controller board.

Now pretend something fails. If the motor never moves, first measure motor supply at the driver, then the enable/control state, then output behavior and the mechanism. If the controller resets only on startup, look at supply droop and return paths. If the motor turns but the wheel does not, inspect the hub. Each symptom points to a trail, which tells you where to measure.

SOURCE TRAIL · Synthesis of source topics 1N, 4L.5, 8L.4, 11N and 12N. The fan system and diagnostic exercise are original, created for your component-understanding goal.

The STM32 replaces manual signal generation and observation. GPIO can read a switch or command a driver. An ADC measures an allowed analog voltage. A timer creates PWM or counts encoder edges. I²C, SPI and USART exchange structured data. None of these peripherals removes the need to choose a suitable power circuit.

For your STM32F411RE CMSIS learning, retain the path you already use: physical signal → MCU pin → GPIO clock → pin mode or alternate function → peripheral clock → peripheral configuration → observable signal. For example, PA6 can carry TIM3_CH1 through alternate function 2. Its Nucleo D12 header label is another name for that board connection—not a second independent signal.

A timer generates a repeated count. In a common edge-aligned PWM arrangement, frequency = timer_clock / ((PSC + 1) × (ARR + 1)). The compare value chooses duty. Before calculating, establish the actual timer clock, including the applicable APB timer-clock behavior. A memorized register value is not a substitute for the clock configuration.

Start with a button that lights the onboard LED. Then read a potentiometer without a motor connected. Next observe PWM on a logic analyzer or oscilloscope. Finally connect a proven driver stage with its load secured. That ordering makes a wrong pin, bad sensor reading and underpowered motor three separate problems.

The useful change in your understanding is this: a line of C changes a hardware register; a register selects or controls an electrical signal; that signal commands a power device; the power device changes current; current produces force. When you can explain every link, the project has stopped being a collection of mysterious parts.

SOURCE TRAIL · The source’s microcontroller portion uses other devices and interfaces. This short STM32 translation is added; pin mapping was checked against ST’s F411 datasheet and Nucleo manual in the earlier edition. It is conceptual guidance, not new tested firmware.