Choosing a Board and Going Further

Basic Electronics

Course 1 · Chapter 11 · Choosing a Board and Going Further

Basic Electronics 10 used four boards, each for one circuit, and named the reason each time. This last chapter turns that habit into a method. You will compare the boards on the same axes (logic level, speed, memory, I/O, power, wireless, price, ease of use), put numbers on the one resource every mobile robot runs out of (battery energy), match a board to each kind of robot job, learn to read a datasheet and to derate a part for heat, set up a small bench of test equipment, and learn to solder a joint that will last. The chapter ends with a recap of the whole course and the path on to the robotics courses.

All prices, clock speeds, memory sizes and currents below are typical or approximate figures from memory. They change with the board revision, the maker, the shop and the year, so treat them as a guide to the order of magnitude and check the maker’s page or the datasheet before you buy or design around a number. The logic levels, the ratios and the formulas do not change.

1. The boards side by side

The table of Basic Electronics 10 compared pin current, ADC, PWM, interrupts and pull-ups. This table adds the other axes that decide a purchase, and three more boards that you will meet. Read it as a map, not a league table: no board wins every row.

Board	Chip and speed (typical)	Logic	Memory (typical)	Wireless	Price (approx.)	Ease of use
Arduino Uno (R3)	ATmega328P, 8-bit, 16 MHz	5 V	2 KB RAM, 32 KB flash	None	Official about 20 to 25 pounds; clones about 6 to 10	Easiest: huge tutorial base, forgiving, simulated in Tinkercad Circuits
Arduino Nano (same chip)	ATmega328P, 16 MHz	5 V	2 KB RAM, 32 KB flash	None	About 4 to 20 pounds (clone to official)	As the Uno, in a breadboard-sized package
Raspberry Pi Pico	RP2040, dual-core Cortex-M0+, up to 133 MHz	3.3 V	264 KB RAM, 2 MB flash (board)	None (the Pico W adds Wi-Fi and Bluetooth)	About 4 pounds (Pico W about 6)	Easy: MicroPython or C, drag-and-drop programming over USB
ESP32 dev board	Dual-core Xtensa, up to 240 MHz (other ESP32 variants differ)	3.3 V	About 520 KB RAM, 4 MB flash (board)	Wi-Fi and Bluetooth built in	About 5 to 10 pounds	Moderate: very capable, but pin quirks (input-only pins, boot pins, ADC2 and Wi-Fi) catch beginners
Teensy 4.x	Cortex-M7, about 600 MHz	3.3 V (pins are not 5 V tolerant)	1 MB RAM, 2 to 8 MB flash	None on the board	About 25 to 35 pounds	Moderate: Arduino-style software, very fast, for jobs like a balancing robot or audio
Raspberry Pi 4 or 5	Quad-core Cortex-A72 (Pi 4, about 1.5 GHz); faster on the Pi 5	3.3 V, not 5 V tolerant	1 to 8 GB RAM; memory card for storage	Wi-Fi, Bluetooth, Ethernet	About 35 to 75 pounds, more for larger memory	Moderate: a full Linux computer, a keyboard and screen help, loose timing
Raspberry Pi Zero 2 W	Quad-core Cortex-A53, about 1 GHz	3.3 V	512 MB RAM	Wi-Fi, Bluetooth	About 15 pounds	As the Pi, smaller and cheaper, less memory

Reading the table: three ideas that save money and time

Clock speed is not the point. A 16 MHz Uno reads a pin and turns on a motor driver with microsecond-level timing that a 1.5 GHz Raspberry Pi running an operating system cannot promise. A microcontroller runs one program with no operating system, so its timing is predictable; a Pi runs many programs and its timing is not. Speed matters for vision, maths and networking; predictability matters for motors and encoders (Basic Electronics 10).
Logic level decides what you can connect directly. A 5 V board matches 5 V sensors and 5 V logic ICs; a 3.3 V board matches the TB6612 and DRV8833 motor drivers and most modern sensors, and needs a divider or a level shifter for 5 V signals (Basic Electronics 8 and Basic Electronics 10).
The cheapest board is rarely the cheapest project. A 5-pound board that needs three extra modules, a level shifter and an hour of debugging costs more than a 10-pound board that has the radio built in. Count the whole bill of parts and your own time.

2. Power: a budget per board and a battery runtime

A robot dies when its battery does, and the battery that matters is usually the one that powers the logic for hours while the motors only run in bursts. The method is the same as the current budget of Basic Electronics 9: list every load, add the currents, check the peak against what the supply can deliver, then convert to energy and time.

 Energy in a battery:      E (Wh) = V(nominal) x C (Ah)          Usable energy:    E(usable) = E x f         f = 0.8 is a sensible planning figure (the pack is not run to empty, and it sags under load)
 Power drawn from the battery by a load behind a converter:      P(batt) = P(load) / efficiency        P(load) = V(rail) x I(rail)
        a linear regulator (Basic Electronics 9) has efficiency = V(out) / V(in)  (and its input current is simply equal to its output current);  a buck converter is typically 85 to 95 %
 Runtime:      t = E(usable) / P(total)

 Typical board-only power (approximate, idle to light work, at the 5 V supply, the board alone with nothing attached):
        Board                      current at 5 V       power       Notes
        Arduino Uno                about 45 mA          0.225 W     the power LED, the regulator and the USB chip use some of it
        Raspberry Pi Pico          about 25 mA          0.125 W     sleep modes can lower it a great deal
        ESP32 dev board            about 100 mA         0.50 W      Wi-Fi active, average; radio transmit bursts reach 250 mA or more; deep sleep is far lower
        Raspberry Pi 4             about 600 mA         3.0 W       idle; heavy load, a camera and USB devices push it towards 1.2 A or more (the supply must be 3 A class)

 Example 1, how long does each board alone run from a 10 000 mAh power bank (3.7 V cell, 85 % usable after its own conversion losses)?
        E = 3.7 V x 10 Ah = 37 Wh        E(usable) = 37 x 0.85 = 31.45 Wh
        Uno:          31.45 / 0.225 = 139.8 h   (about 5.8 days)
        Pico:         31.45 / 0.125 = 251.6 h   (about 10.5 days)
        ESP32 (Wi-Fi): 31.45 / 0.50  =  62.9 h   (about 2.6 days)
        Raspberry Pi 4: 31.45 / 3.0  =  10.5 h
        The factor of 300 between a Pico and a Pi 4 is why a camera robot carries a big battery and a sensor logger does not.

A complete power budget for one robot

 A phone-driven robot: an ESP32, two ultrasonic sensors, three IR line sensors and one servo, on a 5 V logic rail from a buck converter (90 % efficient), and two motors straight from a 2S LiPo (7.4 V nominal, 2200 mAh)

   Load on the 5 V rail           average       peak
   ESP32 dev board (Wi-Fi on)     120 mA        250 mA
   2 x HC-SR04 (15 mA each)        30 mA         30 mA
   3 x IR line sensor (20 mA)      60 mA         60 mA
   1 x servo (SG90 class)         100 mA        360 mA      (the peak is the stall-ish moment of a move; the servo's own wiring and capacitor follow Basic Electronics 9)
   Total                          310 mA        700 mA

   The buck converter must supply at least the peak, with margin:  700 mA is 70 % of a 1 A part (use a 1.5 or 2 A part for comfort)

   Battery energy:   E = 7.4 x 2.2 = 16.28 Wh        E(usable) = 16.28 x 0.8 = 13.02 Wh
   Logic power:      P(load) = 5 V x 0.310 A = 1.55 W        P(batt) = 1.55 / 0.9 = 1.722 W
   Motors:           two motors averaging 0.3 A each on the 7.4 V pack:    P = 0.6 A x 7.4 V = 4.44 W    (driver losses ignored)
   Total:            P = 1.722 + 4.44 = 6.162 W

   Runtime:          t = 13.02 / 6.162 = 2.11 h   (about 127 minutes)

   What the budget tells you:  the motors are 72 % of the drain (4.44 / 6.162), so a better board saves little; the lever is the motors' duty cycle and the gearing.
   The same robot with a linear regulator instead of the buck converter:  the battery supplies the rail current itself, 0.31 A x 7.4 V = 2.294 W instead of 1.722 W, and the regulator dissipates (7.4 - 5) x 0.31 = 0.74 W as heat.
   Total 6.734 W gives 13.02 / 6.734 = 1.93 h, 18 minutes less, plus a warm regulator.

Budget the peak and the average separately

The average sets the battery life. The peak sets the size of the regulator, the wire, the fuse and the capacitor, and decides whether the board brown-outs when a servo starts (Basic Electronics 9). A design that fits on the average but not on the peak is the usual cause of a robot that resets whenever it moves.

3. The best board for each robot job

Basic Electronics 10 gave a short table of four jobs. This one widens it. Each row names a board, the reason and the one thing to watch; “beside” means that a second board handles the part the first is bad at, which is how many robots are built.

Robot job	Best board	Why, and the catch
A first robot: line follower or sumo with a few 5 V analogue sensors, no wireless	Arduino Uno or Nano	5 V logic and ADC match 5 V sensors; the most tutorials; forgiving of mistakes. Catch: 2 KB of RAM is small, and 8-bit maths is slow for anything fancy.
A precise driving robot: two motors with encoders, a PID loop, 20 kHz PWM, a battery monitor	Raspberry Pi Pico	PWM on any pin at 20 kHz, interrupts on every pin, 3.3 V drivers, a 12-bit ADC, low cost. Catch: no wireless (use a Pico W if you need it), few ADC pins.
A robot controlled from a phone or a web page; a Wi-Fi sensor node	ESP32	Radio and two cores on the chip, many pins. Catch: input-only pins, ADC2 off while Wi-Fi runs, noisy ADC, higher power than a Pico.
A camera robot: following a face, a colour or a line from images; speech; a web dashboard	Raspberry Pi 4 or 5, with a Pico or an Arduino beside it for the motors	Needs an operating system and memory for image libraries; the Pi cannot give hard timing or an ADC, so the small board drives the motors and reads the encoders and sensors. Catch: 3 to 6 W, a 3 A supply, a long boot time, and shutting it down properly.
A small, light robot that needs vision but little else	Raspberry Pi Zero 2 W	The same software at a fraction of the weight and power (about 1 W). Catch: 512 MB of RAM limits the vision work.
A self-balancing robot (a tilt sensor read and acted on 200 times a second)	Pico, ESP32 or Teensy	A fast, predictable loop and floating-point maths. Catch: an 8-bit Uno can do a simple version but runs out of speed and memory first.
An arm with six servos	Pico, or an Arduino Mega or Uno with a servo driver board	Many timed pulses at once: a Pico has 16 PWM channels; a PCA9685 servo-driver module (16 channels over I2C) adds servos to any board. Catch: servo power is its own supply, never the board (Basic Electronics 9).
A sensor logger or beacon that must run for weeks on batteries	Pico (or an ESP32 in deep sleep)	Low power, sleep modes. Catch: the board’s own regulator and LEDs, not the chip, often dominate the sleep current, so a bare chip or a low-power board beats a general dev board.
A robot with a spinning range sensor, mapping or ROS software	Raspberry Pi 4 or 5, with a microcontroller for the motors	The robotics software and the network drivers run on Linux. Catch: the loose timing again, so the real-time part goes to the small board.

A quick decision procedure

What must it do that is hard? If the answer is vision, speech or a screen, a Raspberry Pi goes in; if it is precise timing or analogue input, a microcontroller goes in; if it is both, use both.
What logic level are the parts? 5 V parts favour an Uno; 3.3 V parts favour a Pico or ESP32; a mixture means dividers or level shifters (Basic Electronics 8).
Does it need a radio? Yes, and nothing else special: ESP32 or Pico W. No: avoid paying for one.
Do the pin and current budgets fit? Count analogue inputs, PWM outputs, interrupt pins and the total current (Basic Electronics 10), and the power budget of section 2.
How experienced are you? A board with plenty of examples gets you running sooner; first projects should not need a pin-quirks chart.

4. Reading a datasheet

Every part you will ever use has a datasheet, and every datasheet has the same few parts. The skill is knowing which table to trust for which question. Download the datasheet from the maker (the part number is printed on the part); a distributor’s copy is usually the same document.

Features and description (page 1). What the part is for and its headline figures. A summary only; never design from it.
Absolute maximum ratings. The limits that destroy the part: supply voltage, pin voltage, current, temperature. These are not operating conditions. You design to stay well inside them (Basic Electronics 10 used “supply plus 0.3 V” for a pin).
Recommended operating conditions. The ranges within which the part is guaranteed to meet its other specifications: this is the table to design to.
Electrical characteristics. Each parameter has minimum, typical and maximum columns and, importantly, test conditions. The typical value is what a part usually does at 25 degrees C; the minimum and maximum are the guarantees. Design so the circuit works at the worst case in the direction that hurts: the largest value if too much is bad, the smallest if too little is bad.
Graphs. Curves such as forward voltage against current, or resistance against temperature, show what the table does not (typically measured at one temperature).
Package and pinout. Which leg is which, and the size and spacing. Check this before wiring, especially for the TMP36 of Basic Electronics 10, whose reversed pins make it hot in seconds.
Application circuit. The maker’s recommended wiring, including the capacitors and resistors the part needs.

Worked example: sizing a logic-level MOSFET from its datasheet

 An illustrative logic-level N-channel MOSFET (typical of the kind in Basic Electronics 6; every figure is an example, check the real datasheet of the part you buy):
        Gate threshold V(GS)(th):                    1.0 min,  1.7 typ,  2.5 max V     (at I(D) = 250 uA: the gate voltage where it only just starts to conduct, NOT a "fully on" voltage)
        On-resistance R(DS)(on), V(GS) = 4.5 V:       0.035 typ, 0.044 max ohm
        On-resistance R(DS)(on), V(GS) = 2.5 V:       0.048 typ, 0.060 max ohm
        Continuous drain current I(D), 25 C case:     4 A            Absolute maximum V(DS): 30 V        Absolute maximum V(GS): +/- 12 V

   Question: can a 3.3 V Pico pin drive it to switch the 6 V motor M1 (running 0.25 A, stall 1.09 A)?
        The pin gives 3.3 V, above the 2.5 V test point, so use the 2.5 V row and its WORST CASE (maximum) R(DS)(on) = 0.060 ohm.   (the threshold of 2.5 V maximum is far below 3.3 V: the gate is well past threshold)
        Voltage lost in the switch at stall:    V = I R = 1.09 x 0.060 = 65 mV      (1.1 % of the 6 V supply: negligible)
        Power in the switch at stall:           P = I^2 R = 1.09^2 x 0.060 = 0.0713 W        running:  0.25^2 x 0.060 = 3.75 mW        no heatsink needed
        The R(DS)(on) rises when hot (roughly 1.5 times at 100 C, a typical order of magnitude):  0.060 x 1.5 = 0.090 ohm,  P = 1.09^2 x 0.090 = 0.107 W   still tiny.
        The drain-source voltage rating of 30 V is 5 times the 6 V supply, and the motor's switch-off spike is clamped by the flyback diode of Basic Electronics 5: safe.
        Verdict: suitable. Had the only R(DS)(on) given been "at V(GS) = 10 V", the datasheet would be telling you that this part is NOT a logic-level choice for a 3.3 V pin.

Derating: the datasheet number is not the number to use

Ratings are given at a stated temperature, usually 25 degrees C, and fall as the part gets hotter. Derating is the reduction. Datasheets give a derating curve or a rule such as “derate linearly from 70 degrees C to zero at 155 degrees C”; as a formula, with the power rating P_rated given at T_rated and falling to zero at T_max:

 Linear derating:      P(allowed at T) = P(rated) x (T(max) - T) / (T(max) - T(rated))          for T between T(rated) and T(max)
 Thermal resistance:    T(junction) = T(ambient) + R(theta,JA) x P          so      P(allowed) = (T(junction max) - T(ambient)) / R(theta,JA)

   Example 1, a 0.25 W resistor rated at 70 C, zero at 155 C (a common curve), running in a warm enclosure at 100 C:
        P(allowed) = 0.25 x (155 - 100) / (155 - 70) = 0.25 x 55 / 85 = 0.162 W      (65 % of the rating)
        A resistor dissipating 0.2 W in that box would be overloaded even though 0.2 W is under "0.25 W".

   Example 2, a small plastic-cased transistor, R(theta,JA) = 200 C/W (typical of the small TO-92 package; read the real figure), T(junction max) = 150 C, ambient 40 C:
        P(allowed) = (150 - 40) / 200 = 0.55 W

   Example 3, a 7805-style linear regulator (Basic Electronics 9) dropping 12 V to 5 V at 0.3 A, in a TO-220 package with no heatsink, R(theta,JA) = 65 C/W (typical order), T(junction max) = 125 C, ambient 40 C:
        Dissipation:  P = (12 - 5) x 0.3 = 2.1 W          Temperature rise = 65 x 2.1 = 136.5 C          T(junction) = 40 + 136.5 = 176.5 C   ->  over the limit (the regulator would shut itself down)
        The most it can dissipate:  P(max) = (125 - 40) / 65 = 1.31 W          so the largest current at this drop:  I = 1.31 / 7 = 0.187 A
        Cures:  add a heatsink (lowers R(theta)), lower the input voltage, or use a buck converter, whose loss is a few per cent instead of 58 %  ((12 - 5) / 12).
   A good design habit:  run a part at no more than half its derated rating.

5. Test equipment: what to buy first

You can build a long way with three instruments. A fourth item, the breadboard and its wires, is cheap and was covered in Basic Electronics 2.

The digital multimeter (DMM). The first purchase (Basic Electronics 2 showed how to use it). A cheap one is adequate for continuity, voltage and resistance, but check its accuracy specification, which is written as ±(a % of reading + b digits). The percentage scales with what you measure; the digits are a fixed amount at the meter’s resolution on that range, and matter most for small readings.
A bench power supply with a current limit. Adjustable voltage, and a current limit you set before connecting the circuit. A first power-up with the limit at, say, 100 mA turns a wiring mistake into a lit LED on the supply instead of a burnt part. It is the second purchase; until then, batteries and a fuse (Basic Electronics 9) do the job.
An oscilloscope or a logic analyser. A scope draws voltage against time and is the only instrument that shows a PWM waveform, ripple, ringing or a bounce. Two numbers matter: bandwidth and sample rate. A logic analyser shows only high and low, many channels at once, and is the tool for serial, I2C and SPI traffic; cheap USB ones with 8 channels cost about 10 pounds, and a small USB scope with 20 MHz bandwidth about 50 to 70 pounds (approximate figures).

 Multimeter accuracy:      error = (a / 100) x reading + b x (resolution of the range)
   A 3 1/2-digit meter on its 20 V range displays 0.01 V steps. Spec: +/-(0.5 % of reading + 2 digits). A reading of 5.00 V:
        error = 0.005 x 5.00 + 2 x 0.01 = 0.025 + 0.020 = 0.045 V       the true value is between 4.955 V and 5.045 V   (0.9 % of the reading)
   A small reading is hit harder by the digits:  0.50 V on the same range:  0.005 x 0.50 + 0.02 = 0.0225 V = 4.5 %.   Use a lower range (the 2 V range has 0.001 V steps) when the signal is small.
   Practical consequences:  a 5 % resistor measures to within a few per cent on a good range, so a meter cannot tell a 4.7 k from a 4.8 k at the 5 % tolerance (Basic Electronics 3), but it can spot a wrong band at once.

 Oscilloscope bandwidth (the frequency at which the displayed amplitude has fallen to 70.7 % of the true one):
   rise time of the scope itself:    t(r, scope) = 0.35 / BW             20 MHz scope: 0.35 / 20e6 = 17.5 ns
   a signal edge appears slower:     t(displayed) = sqrt( t(signal)^2 + t(scope)^2 )         a 10 ns logic edge:  sqrt(10^2 + 17.5^2) = 20.2 ns   (so the scope is the limit, not the circuit)
   For a square wave, you need a bandwidth of at least 5 times the fundamental to see a reasonable square corner:  a 20 MHz scope suits a fundamental up to 20 / 5 = 4 MHz.  A 20 kHz motor PWM or a 1 MHz I2C clock is well within it.
   Use the x10 probe setting for most work: it presents about 10 Mohm and a few pF to the circuit, loading it far less than the x1 setting; compensate the probe on the scope's square-wave test terminal.
   Logic analyser:  sample at 4 times the signal frequency at the very least, 10 times if you want to see edge timing: a 24 MHz analyser is fine for a 1 MHz bus, not for a 20 MHz one.

6. Soldering: making a joint that lasts

A breadboard is for testing; a robot that moves needs soldered connections, because a push-fit lead shakes loose. Soldering melts an alloy (solder) so that it wets both the part’s lead and the board’s pad and, on cooling, forms a metal joint. The flux inside the solder wire cleans the metal so that the solder can wet it.

Equipment: a temperature-controlled iron or station (about 25 to 60 W, with a fine conical or small chisel tip), solder wire with a flux core (0.6 to 0.8 mm for electronics), a damp sponge or brass wool to clean the tip, desoldering braid, a stand, and ventilation (a small fan or fume extractor to draw the flux smoke away from your face).
Temperature: the usual tin-lead solder (Sn63/Pb37) melts at about 183 degrees C; lead-free solder (for example the SAC305 alloy) melts at about 217 to 220 degrees C. A tip set to roughly 320 to 350 degrees C works for both (typical figures). Too cool, and the joint is dull and lumpy; too hot, and pads lift and parts are damaged.
Technique: tin the tip with a little solder; touch the tip so that it heats both the lead and the pad for one to two seconds; feed the solder wire into the junction of lead, pad and tip, not onto the tip; let a small amount flow; remove the solder first, then the iron; hold the part still until the joint is solid (a second or two). The whole joint takes about two to four seconds; a longer time risks damaging the part or the board.
A good joint is shiny (a lead-free joint is a little duller), shaped like a small smooth cone or volcano that follows the lead up from the pad, with the lead visible inside it. A bad joint is a dull, grainy ball that sits on top (a cold joint: poor contact, intermittent), or a blob that bridges to the next pad (a short). Reheat it with a touch of fresh solder; remove surplus with braid.
Heat-sensitive parts: the first parts to fit are the low-profile ones (resistors, diodes), and the last the tall ones (headers, capacitors). Use sockets for ICs, and respect the polarity of diodes, LEDs and electrolytic capacitors (Basic Electronics 5).

Soldering safety

The tip is at 300 to 350 degrees C: always return the iron to its stand, never lay it on the bench, and never grab a falling iron. Wear safety glasses (solder spits and wire clippings fly), solder in a ventilated place, and wash your hands after handling tin-lead solder. Unplug the iron when you finish. If you burn yourself, cool the burn under running cold water for at least ten minutes.

7. Going further: the next two courses

Basic Electronics ends where a robot’s electronics begins. Two further courses on this site are planned (neither has been written yet) and are written to follow this one:

Robotics Fundamentals (planned): the mechanical and control ideas of a robot, which use the parts of this course as building blocks: sensors and what they measure, actuators and drive systems, kinematics, feedback and control, and how the electronics of a robot is laid out and tested.
Programming for Robotics (planned): the software side, which builds on the boards of Basic Electronics 10 and this chapter: reading sensors and driving motors in code, loops and timing, state machines and simple control programs.

Until they appear, the best next step is the practical one: pick one of the board jobs in section 3, build it on a breadboard, then make it permanent with solder. Each project teaches the same lesson as this whole course: a number you calculated, a reading you measured and a limit you read from a datasheet should agree, and when they do not, the disagreement is where the learning is.

8. Course complete: a recap of Chapters 1 to 10

Chapter	Title	The one thing to keep
1	Electricity Basics	Voltage pushes, current flows, resistance opposes: V = I R and P = V I are in every calculation after this.
2	Circuits and Measurement	Read a schematic, use Kirchhoff’s laws, wire a breadboard, measure with a multimeter (voltage across, current through).
3	Resistors and Dividers	Resistor values and power ratings; the divider V_out = V_in R2 / (R1 + R2); pull-ups and pull-downs.
4	Capacitors and Inductors	RC time constants, filters, decoupling, and the back-EMF of an inductor that makes a switch-off spike.
5	Diodes and LEDs	One-way current, an LED always needs a resistor, flyback and polarity protection.
6	Transistors as Switches	A small current or voltage at a base or gate switches a larger load; choose a logic-level MOSFET for a 3.3 V pin.
7	Analogue Signals and Op-Amps	Analogue versus digital, ADC resolution, the op-amp as buffer, amplifier and comparator, and noise.
8	Digital Logic Basics	Logic levels and noise margins, gates, Schmitt triggers, counters, the 555 timer and level shifting.
9	Motors, Relays and Power	Motor model, PWM, H-bridges, servos, relays, batteries, regulators, and the current budget that keeps a robot from resetting.
10	Microcontrollers in Circuits	A board is a part with limits: pin current, ADC and PWM resolution, pull-ups, level shifting, and safe motor driving.
11	Choosing a Board and Going Further	Choose by logic level, timing, radio, power and price; budget the energy; read datasheets at the worst case and derate; test and solder.

9. Challenges

Challenge 1: Choose a board for three robots

For each robot, choose a board (Uno, Pico, ESP32 or Raspberry Pi, with a second board if needed), give two reasons from this chapter, and estimate the logic runtime. (A) A sumo robot: four 5 V IR sensors on analogue pins (20 mA each), two motors, no wireless, built by a beginner; the board takes 45 mA, the pack is a 2S LiPo (7.4 V, 1000 mAh) feeding the Uno’s own linear regulator. Find the regulator’s dissipation and the runtime of the logic alone (80 % usable). (B) A rover driven from a phone: the ESP32 averages 120 mA, a servo 80 mA and two sensors 30 mA on a 5 V rail from a buck converter (90 %) on a 7.4 V, 2200 mAh pack. Find the rail current, the battery power for the logic and the logic runtime (80 % usable); and say why GPIO34 is a sensible pin for a battery-monitor divider. (C) A camera rover that follows a coloured ball: a Raspberry Pi 4 (3.0 W), a camera (0.25 W) and a Pico driving the motors (0.15 W), on a 10 000 mAh, 3.7 V power bank (85 % usable). Find the total power and the runtime; then repeat with a Raspberry Pi Zero 2 W (assume 0.8 W for the board under load) and say what you would lose.

View solution

Challenge 2: Read a datasheet table

A sample red LED datasheet (illustrative figures) gives: absolute maximum continuous forward current 30 mA; peak current 100 mA at 1 kHz and 10 % duty; power dissipation 75 mW at 25 degrees C, derated by 1.0 mW per degree C above 25 degrees C; forward voltage at 20 mA of 1.7 V minimum, 1.9 V typical and 2.2 V maximum. (a) The LED is driven from a 3.3 V Pico pin and the current must never exceed 10 mA, whatever the LED’s forward voltage: find the minimum resistor and pick the next E12 value (100, 120, 150, 180, 220 ohm). (b) With that resistor, find the current for the minimum, typical and maximum forward voltages, and the power in the LED in the worst case. (c) At 70 degrees C ambient, find the allowed power, and the largest current if V_f is 1.9 V; at what temperature does the allowed power reach zero? (d) A pulsed 100 mA, 10 % duty at 1 kHz: find the average current and the pulse width, and say whether it is within the ratings. (e) Three such LEDs are driven from three Pico pins: is the Pico’s total pin budget (about 50 mA) respected?

View solution

Challenge 3: Plan a bench toolkit with a budget

You have 120 pounds. Approximate prices: digital multimeter 20; temperature-controlled soldering station 28; solder wire 8; desoldering braid and flux pen 6; side cutters and wire strippers 10; safety glasses 3; two breadboards and a jumper kit 12; component assortment 12; tweezers 5; helping hands 10; fume extractor fan 14; USB logic analyser 10; silicone mat 8; bench power supply (30 V, 5 A) 55; 20 MHz USB oscilloscope 60. (a) Mark the first eight items as essentials; find their total and what is left; choose the extras that give the best value for health and safety and show the new total. (b) Plan the next two purchases (the supply and the scope) and, saving 15 pounds a month, say how many months it takes. (c) A multimeter is specified ±(0.8 % of reading + 3 digits) on a 2 V range (0.001 V steps): find the error for a reading of 1.000 V and of 0.100 V, as percentages of the reading. (d) The same meter has ±(1.0 % + 2 digits) on its 20 kΩ range (0.01 kΩ steps): find the range for a displayed 4.70 kΩ and check whether a 5 % resistor labelled 4.7 kΩ fits within it. (e) For the 20 MHz scope, find the scope’s rise time, the displayed rise time of a 10 ns logic edge, and the highest square-wave fundamental that meets the 5-times rule.

View solution

Quick reference

Choosing: hard thing first (vision: Pi; timing or analogue: microcontroller; both: both); then logic level, radio, pin and current budgets, and experience; Uno 5 V and easy; Pico 3.3 V and precise; ESP32 wireless; Pi software-heavy
Energy: E = V × C; usable = 0.8 E; P(batt) = P(load)/efficiency; t = E(usable)/P(total); linear regulator efficiency = V_out/V_in; budget the peak and the average separately
Datasheets: design to Recommended Operating Conditions and the worst-case min/max column; Absolute Maximum is the limit that destroys; check the test conditions, the pinout and the application circuit
Derating: P(allowed) = P_rated (T_max − T)/(T_max − T_rated); T_j = T_a + R_θJA P; run a part at no more than half its derated rating
Test gear: meter error = (a/100) × reading + b × resolution; scope t_r = 0.35/BW and bandwidth at least 5 × the fundamental; logic analyser at least 4 × the signal; set the supply’s current limit first
Soldering: tip about 320 to 350 °C; heat lead and pad together 1 to 2 s; feed the solder to the joint, then remove solder, then the iron; shiny cone is good, dull ball is cold; ventilation, glasses, stand
Next: Robotics Fundamentals and Programming for Robotics (both planned)

That completes Basic Electronics. You can now read a schematic, calculate the part values of a circuit, check them with a meter or a simulator, choose a board for a robot and say why, and read a datasheet critically. Robotics Fundamentals and Programming for Robotics, both planned, will use all of it.