Skills at the bench

Top 10 ways to increase your practical engineering knowledge

By Bulan Sarkar, founder of ElectronicsInfoline (since 2000) · About a 11-minute read

Reading builds theory; practical knowledge comes from measuring things, and from being wrong about them first. Here are ten ways to get it, each with a short drill you can do on a kitchen-table bench, and the answer the drill should give, so you can check your own reading against it.

You need a multimeter, a breadboard and a handful of parts; one drill needs an oscilloscope. The meter figures are from one popular meter, the Fluke 17B+. Yours will have its own table, and the first drill is finding it.

Measure more than you read. The ten ways: know your meter's accuracy, see how it loads a circuit, check parts against their data sheets, predict a result before you measure it, learn the oscilloscope and its probe, work out the watts, repair dead gear, keep a notebook, order a real PCB, and write up what you built.

Each one has a drill. Example: put a meter across the lower half of a 1 MΩ + 1 MΩ divider on 10 V. You expect 5.00 V; a 10 MΩ meter shows 4.76 V, 4.8 % low, because the meter becomes part of the circuit it measures.

The drills in numbers

5 V reading, Fluke 17B+: ±55.0 mV
1 MΩ divider, meter across it: 4.76 V
10 kΩ with 100 µF, ±tolerances: 0.76 s–1.26 s
5.0 ns edge seen on a 1X probe: 12.2 ns

The ten ways at a glance

The first six are bench drills with a number to hit. The last four are habits, and they make the first six stick.

Figure 1. The ten ways and the drill that goes with each. Blue: bench drills with a computed answer. Green: habits.

1. Know your meter, including what it can't tell you

A multimeter reading is a range, not a number. The specification table gives accuracy as ±(a percentage of the reading + a number of counts), where one count is the last digit on the current range. The Fluke 17B+ reads DC volts to ±(0.5 % + 3 counts) (Fluke).

Drill: find your meter's table (manual or maker's page) and work out the band on two readings. On the Fluke 17B+, 3.3 V sits on the 4 V range with 1 mV resolution: ±19.5 mV, or 0.59 %. A 5 V rail needs the 40 V range, where a count is 10 mV, and the band grows to ±55.0 mV, 1.1 %. The higher voltage is read less accurately.

Figure 2. What a Fluke 17B+ reading of 3.3 V and of 5 V really means: the true voltage is anywhere in the blue band. Each bar is scaled to its own reading.

Our take: once you know the band, you stop chasing the last digit. A 5 V rail that reads 4.96 V is in spec on this meter; a change of 10 mV between two readings tells you nothing.

2. See what the meter does to the circuit

A voltmeter draws a little current, so it behaves like a resistor across whatever you measure. The Fluke 17B+ has a DC input impedance of more than 10 MΩ (Fluke). On a low-impedance node that disappears; on a high-impedance one it does not.

Drill: build three dividers from a 10 V supply, each two equal resistors: 10 kΩ, 100 kΩ and 1 MΩ. Each should give 5.0 V. Put the meter across the lower resistor. It sits in parallel with it, so the lower half becomes smaller and the reading drops.

Figure 3. The 1 MΩ divider with the meter connected. The meter's 10 MΩ sits across the lower resistor and pulls the midpoint from 5.00 V down to 4.76 V.

Figure 4. How far the meter pulls each divider low. At 100 kΩ the error (0.50 %) already matches the meter's own 0.5 % accuracy.

The 1 MΩ divider reads 4.76 V: 4.8 % low, ten times the meter's own error. That is how people end up hunting a fault that their meter put there. The meter loading lesson works through the general case.

3. Check real parts against their data sheets

A 4.7 kΩ resistor with a gold band is a 5 % part: anything from 4.465 kΩ to 4.935 kΩ is in spec. Measured on the Fluke 17B+'s 40 kΩ range (±0.5 % + 2 counts) (Fluke), the meter adds ±43.5 Ω of its own.

Drill: measure ten resistors from one bag, write each reading down, and check every one sits inside the band. Then do the same with the data sheet of something you use every week, an LED or a regulator, and find the one figure that limits your circuit.

Figure 5. A 4.7 kΩ 5 % resistor: the band the maker allows, and the much narrower band the meter adds to any one reading. Both bars share one scale.

The meter is good enough to sort resistors: its band is about a tenth of the part's. Our resistor colour code calculator decodes the bands if they are hard to read.

4. Predict the result before you measure it

The habit that teaches most is writing the expected value down first. When the measurement disagrees, you have found either a mistake in your model or a fault in the circuit, and both are worth knowing. A simulator is good for the same thing; our circuit simulator guide lists free ones.

Drill: charge a 100 µF capacitor through 10 kΩ from 5 V. The time constant is τ = RC = 1.00 s; after one τ the capacitor reaches 63.2 % of the supply, 3.16 V. That is the prediction. Now the tolerances: the resistor is ±5 % and an ordinary electrolytic is ±20 % (Nichicon), so τ can be anywhere from 0.76 s to 1.26 s. Time the climb to 3.16 V with a phone stopwatch and the meter.

Figure 6. The capacitor voltage for the fastest (τ = 0.76 s), nominal (1.00 s) and slowest (1.26 s) parts. A stopwatch time to 3.16 V anywhere in that span is a correct circuit.

Discharge the capacitor fully between runs, through a resistor, and wait at least 6.3 s (five τ of the slowest case) when you charge it to full. If your time lands outside the span, check the capacitor first. The RC time constant calculator redoes the sum for other parts.

5. Get time on an oscilloscope, and learn its probe

A scope shows what a meter averages away: edges, ringing, ripple, noise. The first thing it teaches is that the probe is part of the measurement. The RIGOL DHO804 is a 70 MHz scope; its standard PVP3150 probe is rated 35 MHz on the 1X setting and 150 MHz on 10X (RIGOL).

Drill: look at a fast logic edge, about 5.0 ns, with the probe on 1X and then on 10X. Each part of the chain adds a rise time of 0.35 / bandwidth: the scope 5.0 ns, the probe 10.0 ns on 1X and 2.3 ns on 10X. They add as the root of the sum of squares, so the screen shows 12.2 ns on 1X and 7.4 ns on 10X.

Figure 7. One 5.0 ns edge on a 70 MHz scope: the 1X probe shows it as 12.2 ns, the 10X probe as 7.4 ns.

Leave the probe on 10X unless you are looking at millivolts, and compensate it on the scope's square-wave output first. If you are choosing a scope, our oscilloscope guide works out bandwidth, sample rate and memory for a home bench.

6. Work out the watts before you power up

Every resistor turns P = V² / R into heat. A common through-hole resistor is rated 0.25 W, and that rating assumes cool air; we work to 50 % of it.

Drill: before you power anything, write the power next to every resistor that drops more than a volt or two. Then touch each one after a minute (low voltage only). The ones near 100 % of their rating are the ones that feel hot.

Figure 8. Power in a 0.25 W resistor as a share of its rating. Red: above our 50 % margin. 100 Ω across 5 V sits exactly at the rating and runs hot.

220 Ω on 5 V is 114 mW: fine. 100 Ω on 5 V is 250 mW, the full rating, and 470 Ω on 12 V is 306 mW, over it. At 12 V a 0.25 W part needs at least 1152 Ω to stay inside the margin, so the next standard value up, 1.2 kΩ, or a bigger resistor. The power rating lesson covers derating with temperature.

7. Fix dead gear

A broken gadget is a circuit with one known fault, which makes it a better teacher than a kit that works first time. Start with battery and USB-powered things: a toy, a torch, a power bank, a cheap speaker. Leave anything that plugs into the mains until you have been trained for it, because mains gear holds lethal voltages, and some of it stays charged after you unplug it.

Drill: before you open it, write down what should be there. Then measure in this order: the battery or input voltage on load, the fuse or polyfuse, the regulator output, then the parts that get hot. A flat cell, a broken wire at the connector and a blown fuse are common faults, and quick to rule out. Continuity and in-circuit tests cover the meter side.

8. Keep a lab notebook

The drills above only work if the predictions and readings are written down. A notebook entry needs five things: date, the circuit (a sketch is enough), what you expected, what you measured with which instrument and range, and what you changed next.

Drill: redo drill 2 and write it up as a notebook page. Next month you will not remember which divider read 4.76 V; the page will. If you want ready pages, our lab notebook templates are printable A4 sheets with these columns, and a filled-in example.

9. Design a small board and have it made

A breadboard teaches circuits; a PCB teaches layout, footprints and fab tolerances. Pick something you have already built on a breadboard, such as a 555 flasher, and lay it out on two layers.

Drill: before you order, print the layout at 1:1 and push the real parts through the paper. Wrong footprints show up immediately. Our PCB design guide gives the track and via rules, and how PCBs are made explains the hole and track tolerances that come back from the factory.

Tips and checks

Replace the meter's battery as soon as the low-battery symbol appears.
Short the probes together on the lowest ohms range before you measure a small resistance, and subtract the lead reading.
Use a current-limited supply for anything new. A bench supply with its limit set low, say 100 mA, protects a new circuit from a wiring mistake.
When a reading surprises you, measure it a second way: a different range, a different meter, or across a known resistor.
Buy parts in tens. Ten resistors of one value are the sample for drill 3; our guide to buying components in India lists where.

Questions people ask

How can an electronics engineering student get practical knowledge?

Do small bench drills with a known answer: meter accuracy, meter loading, RC timing, resistor power. Predict each result, measure it, and write both down. Repairing dead gear and ordering a small PCB of your own come next.

Which instrument should I buy first?

A multimeter, and learn its specification table. A current-limited bench supply comes next. An oscilloscope is worth it once you work with signals that change faster than a meter can follow.

Why does my meter read low on a high-value divider?

The meter's own input resistance, about 10 MΩ, sits in parallel with the resistor you are measuring across. On a 1 MΩ divider from 10 V that pulls the reading from 5.00 V to 4.76 V.

Is simulation enough?

No. A simulator uses ideal or modelled parts at nominal values. Real parts have tolerances, instruments load circuits, and wires have resistance. Use a simulator to make the prediction, and the bench to test it.

Lessons behind the drills

Sources and assumptions

Fluke, 17B+ Digital Multimeter, product page specifications (fluke.com, en-in) (DC volts 4.000 / 40.00 / 400.0 / 1000 V ranges, 0.001 / 0.01 / 0.1 / 1 V resolution, ±(0.5 % + 3); resistance 40.00 kΩ range 0.01 kΩ resolution ±(0.5 % + 2); DC volts input impedance >10 MΩ (nominal); 4000-count display)
Nichicon, UVR series aluminium electrolytic capacitors, specifications table (capacitance tolerance ±20 % at 120 Hz, 20 °C; −40 to +85 °C)
RIGOL, DHO800 Series Digital Oscilloscope data sheet (EN), pp. 6–8 (DHO804 70 MHz, 5 ns rise; input 1 MΩ with 15 pF; PVP3150 probe 35 MHz at 1X and 150 MHz at 10X)

Read on 13 October 2026. Our own choices, not taken from any source: the drill values (3.3 V and 5 V readings, 10 V dividers, a 4.7 kΩ 5 % resistor, 10 kΩ with 100 µF, a 5.0 ns edge), the meter's input taken as exactly 10 MΩ (the table says more than that, so real loading is slightly less), the 0.25 W rating and the 50 % margin, and the 0.35 / bandwidth rise-time rule with root-sum-of-squares, which assumes single-pole responses. This page rebuilds a 2019 ElectronicsInfoline article of the same title; none of its text is reused.