Electronics Engineer interview questions and answers

An ideal op-amp has infinite open-loop gain, infinite input impedance, zero output impedance and infinite bandwidth. With negative feedback this gives two working rules: no current flows into the inputs, and the voltages at the two inputs are equal.

Inverting amplifier: the non-inverting input is grounded, so the inverting input sits at 0 V (a virtual ground). The input current Vin/Rin must all flow through Rf, so Vout = −(Rf/Rin) × Vin. With Rf = 100 kΩ and Rin = 10 kΩ, the gain is −10. Non-inverting: the input goes to the + pin, and the gain is 1 + Rf/Rg, always at least 1, with no phase inversion. A voltage follower is the case Rf = 0, gain 1.

Biasing sets a steady DC operating point, the Q-point, so the transistor stays in its active region for the whole input signal. Without it, the negative half of an AC signal would switch the transistor off and the output would be clipped. Voltage-divider bias with an emitter resistor is the most common, because the emitter resistor keeps the Q-point stable when β or temperature changes.

In a common-emitter amplifier the input goes to the base and the output is taken at the collector. It gives both voltage and current gain, and the output is 180° out of phase with the input. With the emitter resistor bypassed by a capacitor, the voltage gain is roughly −RC / re, where re ≈ 25 mV / IE at room temperature.

The cutoff frequency is fc = 1 / (2πRC). For R = 1 kΩ and C = 100 nF, RC = 1000 × 100 × 10⁻⁹ = 10⁻⁴ s, so fc = 1 / (2π × 10⁻⁴) ≈ 1.59 kHz. At fc the output is 1/√2 of the input, about 70.7%, which is −3 dB, and the output lags the input by 45°.

Above fc a first-order filter rolls off at −20 dB per decade, so at 15.9 kHz the signal is down about 20 dB. For a steeper roll-off you cascade stages or use an active filter, such as a Sallen-Key with an op-amp, which also avoids the loading problems of passive RC stages. Swapping R and C gives a high-pass filter with the same fc.

The Barkhausen criterion: around the feedback loop, the gain magnitude must be at least 1 (exactly 1 for steady amplitude), and the total phase shift must be 0° or 360°. In practice the loop gain starts slightly above 1 so oscillation builds up from noise, and the amplifier's non-linearity, or an automatic gain control, limits the amplitude.

In an RC phase-shift oscillator, an inverting amplifier gives 180°, and three RC sections give the other 180° at one frequency. For three equal sections of series capacitors and shunt resistors, that frequency is f = 1 / (2πRC√6), and the network attenuates the signal by 29 there, so the amplifier needs a gain of at least 29. Wien bridge and crystal oscillators are the other common types to know.

A linear regulator drops the extra voltage as heat. Its dissipation is (Vin − Vout) × I = (12 − 5) × 1 = 7 W, and its efficiency is about Vout / Vin = 5/12, roughly 42%. Seven watts needs a large heatsink, so a linear regulator is a poor choice here. A buck converter switches the input and filters it with an inductor and capacitor; it typically reaches 85 to 95% efficiency, so it loses well under 1 W.

The trade-off is that a buck produces switching ripple and electromagnetic noise, and needs careful layout. Linear regulators and LDOs are quiet and simple, best when the voltage drop or current is small. For sensitive circuits like an ADC reference, a common design is a buck followed by an LDO.

Any Boolean function can be built using only NAND gates, or only NOR gates, which is why they're called universal. That's useful in practice: one gate type is cheaper to manufacture, and NAND is the natural gate in CMOS.

With NAND: NOT A is a NAND gate with both inputs tied to A, since (A·A)' = A'. AND is a NAND followed by a NAND used as an inverter: ((A·B)')' = A·B, two gates. OR uses De Morgan's theorem: invert each input with a NAND, then NAND the results, (A'·B')' = A + B, three gates. Interviewers often follow up by asking for XOR, which takes four NAND gates.

A combinational circuit's output depends only on its present inputs, with no memory: adders, multiplexers, decoders and encoders. A sequential circuit's output depends on its present inputs and its past, because it stores state: flip-flops, registers, counters and state machines.

A latch is level-sensitive: a D latch passes its input to the output the whole time its enable is high, and holds it when enable goes low. A flip-flop is edge-triggered: it samples its input only at a clock edge, rising or falling. That's why synchronous designs use flip-flops: every stage changes at the same instant, which makes timing predictable, while latches can let changes ripple through while enabled.

When J and K are both 1, a JK flip-flop toggles. In a level-triggered JK flip-flop, if the clock stays high longer than the flip-flop's propagation delay, the output toggles, feeds back to the inputs, and toggles again, repeatedly, for as long as the clock is high. The final state is unpredictable. That's the race-around condition.

It's avoided by making sure the output can change only once per clock: use a master-slave JK flip-flop, where the master follows the inputs while the clock is high and the slave updates only when it goes low, or use an edge-triggered flip-flop. Keeping the clock pulse shorter than the propagation delay works in theory but is impractical.

In an asynchronous, or ripple, counter only the first flip-flop gets the clock, and each next flip-flop is clocked by the previous one's output. It's simple, but the delays add up along the chain, so the outputs settle at different times and decoding them can give brief glitches, which limits its speed. In a synchronous counter all flip-flops share one clock and logic decides which ones toggle, so all outputs change together, and it's faster.

A mod-N counter needs n flip-flops where 2ⁿ ≥ N. For mod-10, 2³ = 8 is too few and 2⁴ = 16 is enough, so 4 flip-flops, counting 0 to 9 and resetting at 10. The decade counter IC 7490 is the classic example.

Setup time is how long the input must be stable before the clock edge; hold time is how long it must stay stable after it. If either is violated, the flip-flop can go metastable: its output hovers between 0 and 1 for an unpredictable time before settling, and circuits reading it may see different values.

Setup limits the clock speed: the clock period must be at least clock-to-Q delay plus the longest logic delay plus setup time. Hold violations don't depend on frequency, so they're fixed by adding delay to short paths. A signal from another clock domain, or a push button, has no fixed timing, so pass it through a synchroniser of two flip-flops in series, which gives metastability time to resolve. For multi-bit data, use a handshake or an asynchronous FIFO.

A microprocessor is mainly a CPU on a chip: it needs external RAM, ROM, I/O ports and timers to make a working system. It's built for general-purpose computing with high processing power, as in a PC. A microcontroller puts the CPU, RAM, flash or ROM, I/O ports, timers, ADC and serial interfaces on one chip, so it can run a dedicated task with very few extra parts.

So microcontrollers are cheaper, smaller and use less power, and they're found in washing machines, remote controls, cars and meters. The 8085 is a classic microprocessor and the 8051 a classic microcontroller; today ARM Cortex-M chips, such as STM32, are the common microcontrollers in industry.

The 8051 is an 8-bit microcontroller with 4 KB of on-chip ROM, 128 bytes of on-chip RAM, four 8-bit I/O ports (P0 to P3, 32 pins in all), two 16-bit timers/counters, one full-duplex serial port and five interrupt sources: two external, two timers and the serial port. It uses a Harvard architecture, with separate program and data memory, each addressable up to 64 KB externally.

Its internal RAM holds four register banks of R0 to R7 in the lowest 32 bytes, a bit-addressable area from 20H to 2FH, and general-purpose RAM; the special function registers sit from 80H to FFH. Mention that Port 0 needs external pull-up resistors, and that the 8052 adds a third timer and more memory.

The 8085 has a 16-bit address bus, but to save pins the lower 8 address bits share pins AD0 to AD7 with the data bus. At the start of a machine cycle the pins carry the low address byte, and ALE (Address Latch Enable) goes high; an external latch such as the 74LS373 captures that byte, and the pins then carry data.

The 8085 has five interrupts. TRAP is non-maskable, with the highest priority, and is both edge and level triggered. RST 7.5 is edge triggered, and RST 6.5 and 5.5 are level triggered; these three are maskable and vectored. INTR has the lowest priority and is non-vectored: the interrupting device supplies the instruction. The priority order is TRAP, RST 7.5, 6.5, 5.5, INTR.

In memory-mapped I/O, devices get addresses in the same address space as memory, and the processor reaches them with ordinary memory instructions. Any instruction that works on memory works on a device register, but the devices use up part of the memory address range. Most modern microcontrollers, including ARM Cortex-M, work this way: peripheral registers are just addresses.

In I/O-mapped, or isolated, I/O, devices have a separate address space reached with special instructions. The 8085 uses IN and OUT with an 8-bit port address, so it can address 256 input and 256 output ports, and the IO/M signal tells external hardware whether an address is for memory or I/O. It keeps memory space free but offers fewer instructions for devices.

With polling the processor checks the device's status repeatedly in a loop. It's simple and predictable, but it wastes CPU time, and the response is only as fast as the loop. With interrupts the device signals the processor when it needs attention, so the CPU can do other work or sleep, and responds quickly, at the cost of more complex code and care with shared data.

Use interrupts for events that are rare, unpredictable or urgent: a received byte, a button press, an emergency stop, or waking a battery device from sleep. Polling suits devices that are always ready or checked at a fixed rate anyway, or very fast events where interrupt overhead would be too high. For bulk data transfer, DMA beats both.

Current is measured in series: switch off, break the circuit, move the red lead to the A or mA jack, and connect the meter so the current flows through it. Start on the highest range if you don't know the current. Voltage is measured in parallel, across the component, with the red lead in the V jack.

The common mistake is leaving the lead in the current jack and then measuring a voltage. In current mode the meter has very low resistance, so connecting it across a supply is nearly a short circuit, and it blows the fuse or damages the meter. Also measure resistance and continuity only with the power off, and discharge large capacitors first.

A common rule is a scope bandwidth of three to five times the highest frequency you need to see, because a square wave contains harmonics well above its fundamental. For fast edges, use bandwidth ≈ 0.35 / rise time: a 1 ns edge needs about 350 MHz, and more to measure it accurately.

A 10x probe divides the signal by 10, but presents a much higher input impedance, about 10 MΩ instead of 1 MΩ, and much less capacitance, so it disturbs the circuit less and passes higher frequencies. Compensate it on the scope's calibration square wave first. Keep the ground lead short for fast signals, and remember the scope's ground is connected to mains earth, so never clip it to a non-earthed high-voltage point.

Most function generators have a 50 Ω output impedance, and their displayed amplitude assumes a 50 Ω load. With a 50 Ω load, the internal 50 Ω and the load form a voltage divider, so the generator produces 2 Vpp internally to deliver 1 Vpp. A scope input is high impedance, 1 MΩ, so almost no voltage drops inside the generator, and you see the full 2 Vpp.

The fix is to set the generator's output to High-Z (load) mode, so the display matches what you measure, or terminate the line with 50 Ω. Interviewers ask this to see whether you understand source and load impedance, which also matters for cables and RF work.

An oscilloscope shows a signal against time; a spectrum analyser shows its power against frequency, with a much wider dynamic range, so it can see a weak signal next to a strong one. Use it to check a transmitter's frequency, power and harmonics, find interference and spurious tones, measure the bandwidth of a modulated signal, and do EMI pre-compliance checks on a product before sending it for certification.

The resolution bandwidth setting controls how closely spaced signals you can separate, and lowering it also lowers the noise floor, at the cost of slower sweeps. Many oscilloscopes have an FFT mode, which is fine for low frequencies and quick checks, but it has far less dynamic range.

In amplitude modulation the carrier's amplitude follows the message; in frequency modulation the carrier's frequency varies with it, while the amplitude stays constant. AM needs a bandwidth of twice the highest message frequency, 2fm. FM needs more: by Carson's rule, about 2(Δf + fm), where Δf is the peak frequency deviation. For FM broadcast, with Δf = 75 kHz and fm = 15 kHz, that's about 180 kHz.

In return, FM is much less affected by noise, because most noise changes amplitude, which an FM receiver's limiter removes. That's why FM radio sounds cleaner. AM is simpler and travels farther at medium-wave frequencies, which is why it's still used for long-distance broadcasting.

A signal with no frequency components above fmax can be reconstructed exactly from its samples if the sampling rate is more than twice fmax. 2fmax is called the Nyquist rate. Telephone speech, limited to about 3.4 kHz, is sampled at 8 kHz; audio up to 20 kHz is sampled at 44.1 kHz for CDs.

If you sample too slowly, frequencies above half the sampling rate appear as false lower frequencies in the samples. That's aliasing, and it can't be removed afterwards. To prevent it, put an analog low-pass anti-aliasing filter before the ADC to remove everything above half the sampling rate, and sample with some margin above the Nyquist rate, since real filters don't cut off sharply.

Shannon's theorem gives the maximum error-free data rate of a channel with bandwidth B and signal-to-noise ratio S/N: C = B log₂(1 + S/N). It shows you can trade bandwidth for SNR, and that no coding scheme can beat this limit.

First convert the SNR from decibels: 30 dB = 10^(30/10) = 1000. Then C = 3000 × log₂(1001) ≈ 3000 × 9.97 ≈ 29.9 kbps. So a 3 kHz telephone channel with 30 dB SNR can't carry much more than about 30 kbps. A common mistake is putting 30 straight into the formula instead of 1000. Mention that doubling the bandwidth roughly doubles capacity, while doubling the SNR adds only about one bit per second per hertz.

They differ in bits per symbol. BPSK has two phases and carries 1 bit per symbol. QPSK has four phases, 2 bits per symbol, so it doubles the data rate in the same bandwidth with the same bit error rate per unit of energy per bit. 16-QAM varies both amplitude and phase over 16 points, carrying 4 bits per symbol.

Higher orders pack the points closer together, so less noise is needed to push a symbol into its neighbour's region: they need a higher SNR for the same error rate. That's why Wi-Fi and mobile networks use adaptive modulation, switching to high-order QAM when the signal is strong and dropping to QPSK or BPSK at the cell edge.

Tell it in a clear order: the problem, the block diagram, your part, the result. For example: "Our project was [project], which [what it does]. It has [sensor], a [microcontroller], a [power supply or driver stage] and [display or wireless module]. I designed the [part you designed, for example the signal-conditioning circuit] and soldered and tested the board. Our first version had [problem, for example a noisy reading], which I traced with the oscilloscope and fixed by [solution]. It finally [measured result]."

Expect deep follow-up questions on any component you name, so mention only what you can explain.

Show commitment with evidence, since interviewers worry you'll leave for an IT offer. For example: "I've chosen core because I enjoy hardware: designing and debugging circuits is what I spend my own time on. I built [project or hobby circuits], took [lab, internship or certification], and I'm comfortable with tools like [simulation software, PCB design tool, oscilloscope]. I want to grow into a [design / testing / product] engineer, and this role gives me hands-on work from the start."

Don't speak negatively about IT jobs; just make clear that core electronics is a deliberate choice, not a fallback.

Answer honestly, and if you're willing, show you understand why it matters. For example: "Yes. I know production and testing run in shifts, and working on the line is the fastest way to learn how products are really built and why they fail. During my [internship or industrial visit] at [company or plant], I saw [what you saw]. I'm also fine with travelling to customer sites for installation or support."

If you have a real constraint, say so politely now rather than after joining; interviewers value that more than a yes you can't keep.

Use STAR and show a methodical, calm approach. "A week before a customer shipment, [number]% of boards failed the functional test (situation). I was responsible for the test analysis (task). I grouped the failures, found they all came from one reel of [component], measured it against the datasheet, and confirmed a supplier batch problem; we reworked the affected boards and added an incoming inspection check (action). We shipped on time, and the failure rate dropped to [number] (result)."

End with what you changed in the process, which shows you think beyond the one fix.

Show that you act quickly and check the replacement properly. First confirm the stock with the distributors and the manufacturer's notice, and tell the project lead with the expected impact on schedule. Then find alternatives: a pin-compatible part from another manufacturer, or a newer part from the same family. Compare the datasheets parameter by parameter, not just the headline values, build and test a few boards with it, and update the bill of materials and documentation.

For example: "When our [regulator] became unavailable, I qualified a pin-compatible part, tested it at temperature extremes, and we avoided a board redesign." Mention designing with a second source in future.