Electronics Guide

The Schmitt Trigger

The Schmitt trigger is the quintessential hysteresis circuit, converting a slowly changing or noisy input into a clean digital output. It is named after Otto Schmitt, who invented it in 1934 as a graduate student studying the propagation of nerve impulses in squid, and who published it in 1938 as "A Thermionic Trigger." The circuit uses positive feedback to create two distinct threshold voltages.

In an op-amp Schmitt trigger, the positive feedback network connects from the output to the non-inverting input. When the output is high, the positive feedback raises the threshold that the input must exceed to cause a transition; when the output is low, the threshold is lowered. The difference between these thresholds is the hysteresis width, set by the ratio of resistors in the feedback network. In the common inverting configuration, with the divider returned to ground and an output that saturates symmetrically, the two trip points sit at plus and minus the saturated output voltage times the feedback divider ratio, so the hysteresis width is twice that product. Returning the divider to a reference voltage instead of ground shifts both trip points by the reference voltage times one minus the divider ratio, without changing the width. With outputs saturating at ±12 V, 10 kΩ from the non-inverting input to ground, and 100 kΩ from the output, the divider ratio is 10/110, about 0.091, so the trip points sit near +1.09 V and −1.09 V and the hysteresis is about 2.2 V. A designer trades hysteresis for sensitivity simply by adjusting the ratio.

The hysteresis width should be chosen based on the expected noise amplitude. Setting the hysteresis slightly larger than the peak-to-peak noise ensures clean switching without false transitions. Too much hysteresis, however, reduces the circuit's sensitivity to legitimate signal changes.

Transistor-Based Hysteresis Circuits

Discrete transistor Schmitt triggers use the regenerative action of a coupled transistor pair. In the classic emitter-coupled configuration, the collector of the input transistor drives the base of the second transistor through a resistive divider, and the two emitters share a common resistor. As the input rises past the upper threshold, the input transistor begins to conduct and its collector voltage falls, turning the second transistor toward cutoff. The current through the shared emitter resistor drops, the emitter voltage falls with it, and the input transistor's base-emitter voltage rises further, so the transition reinforces itself until both transistors have reached their opposite states.

The threshold voltages depend on the transistor bias conditions and the shared emitter resistor value. Unlike op-amp versions where hysteresis can be set precisely with resistor ratios, discrete designs require careful analysis of transistor operating points. Temperature variations affect the thresholds, making discrete Schmitt triggers less precise but often faster than their op-amp counterparts.

Applications of Hysteresis

Hysteresis circuits find applications wherever noise immunity at switching thresholds is important:

• Signal Conditioning: Converting sine waves or noisy signals to clean square waves for digital processing
• Threshold Detection: Preventing oscillation in level-crossing detectors and comparators
• Switch Debouncing: Eliminating multiple transitions from mechanical switch contacts
• Thermostat Control: Providing temperature dead bands that prevent rapid cycling of heating or cooling systems
• Voltage Monitoring: Creating reliable power-on reset circuits with definite switching points

The SR Latch

The Set-Reset latch is the simplest memory element, constructed from two cross-coupled gates or transistors. Each gate's output connects to the other's input, creating a positive feedback loop that maintains whichever state was last established. The Set input forces the latch to its "1" state, while the Reset input forces it to "0". In the NOR form, asserting both inputs together is forbidden: it forces both outputs low, and if the two inputs are then released at the same moment, the state the latch falls into is unpredictable.

The cross-coupled configuration ensures that once set, the latch remains set even after the Set input is removed. The positive feedback regeneratively maintains the state. This memory behavior persists as long as power is applied, making SR latches useful for storing status information, implementing control flip-flops, and building more complex sequential circuits.

Analog Latches and Sample-Hold Circuits

Positive feedback also enables analog memory functions. Regenerative comparators can capture and hold an analog decision, latching their output to indicate which input was larger at a specific moment. These circuits are essential in analog-to-digital converters where successive approximation requires holding comparison results.

Sample-and-hold and track-and-hold circuits, by contrast, store an analog level as charge on a capacitor rather than by regeneration, so their hold accuracy depends on low-leakage switches and a high-impedance buffer rather than on feedback. Positive feedback enters the data-converter signal chain one stage later, in the regenerative comparator that follows the hold. Once the capacitor freezes the sampled input, cross-coupled devices lock the comparison to a definite logic level, and it is that latched decision, not the stored analog voltage itself, that positive feedback maintains.

Metastability Considerations

When input signals change near the switching threshold, bistable circuits can enter a metastable state where they hover between their two stable conditions. This metastability is an inherent consequence of positive feedback systems having an unstable equilibrium point between stable states.

Metastability resolves itself over time as noise or small asymmetries push the circuit toward one stable state. Near the balance point the circuit behaves like the regenerating latch described in the next section, so the probability that it is still unresolved after a time t falls as e^−t/τ, where τ is the regeneration time constant. There is always some probability of remaining metastable for any given duration, and a synchronizer's mean time between failures grows exponentially with the resolution time it is allowed:

MTBF = e^t_r/τ / (T₀ f_clk f_data)

Here t_r is the time allowed for resolution, T₀ is the width of the metastability window, and f_clk and f_data are the clock and data rates. Every additional τ of settling time multiplies the MTBF by e, about 2.7. System designers therefore allow sufficient settling time or add a second synchronization stage, which buys nearly a full clock period of resolution time.

Latching Comparators

A latching comparator samples its input difference at a clock edge and uses positive feedback to rapidly amplify this difference to a full logic swing. The regenerative action continues until the output saturates at one supply rail or the other. Once latched, the output holds its state regardless of subsequent input changes until the next clock edge.

The StrongARM latch, named after the DEC StrongARM microprocessor that used it as a sense amplifier, is a popular regenerative comparator topology in analog-to-digital converters. During the reset phase, its cross-coupled nodes are precharged. When the clock activates the comparator, the differential input pair creates a small imbalance, and cross-coupled inverters then amplify it regeneratively to full swing. Once the decision is made, no static current flows.

Near its balance point a cross-coupled pair acts as a linear amplifier with positive feedback, so a small initial imbalance ΔV₀ grows exponentially:

ΔV(t) = ΔV₀ e^t/τ

The regeneration time constant τ is roughly the capacitance at each regenerating node divided by the transconductance of the cross-coupled devices, C/g_m. Reaching a logic swing V_L therefore takes τ ln(V_L/ΔV₀), and each tenfold reduction in the starting imbalance adds a further τ ln 10, about 2.3τ. Taking τ = 10 ps for illustration, a 1 mV imbalance reaches 1 V in about 69 ps and a 1 µV imbalance in about 138 ps. The decision time grows only logarithmically as the input shrinks, which is why a latch resolves ordinary inputs quickly, yet an input balanced almost exactly can hold it undecided for arbitrarily long.

Dynamic Comparators

Dynamic comparators operate in two phases: reset and evaluation. During reset, internal nodes are initialized to known states. During evaluation, positive feedback amplifies the input difference. This clocked operation eliminates static power consumption, and the clock edge itself defines the sampling instant, so no separate sample-and-hold is needed.

The double-tail comparator splits the circuit into an input stage and a latch stage, each with its own clocked tail transistor. The input pair can then be sized for gain and low offset while the latch draws its own current for fast regeneration, and with fewer transistors stacked between the supply rails, the circuit works from lower supply voltages than a single-tail design such as the StrongARM latch. The separation also shields the input from the latch's large, fast swings. This topology achieves excellent speed-power trade-offs in modern CMOS processes and is widely used in low-voltage, high-speed data converters.

Offset Compensation Techniques

Regenerative comparators amplify any input offset along with the signal, making offset cancellation essential for precision applications. Auto-zeroing techniques store the offset during a calibration phase and subtract it during normal operation. Chopper stabilization modulates the signal to a frequency where offset and low-frequency noise can be filtered out.

In a successive-approximation converter, a single comparator makes every decision, so its offset shifts all codes equally and appears only as an offset of the whole converter. In a flash ADC, where many comparators work in parallel, each comparator's offset moves its own threshold and so distorts the transfer characteristic, and trimming every comparator individually becomes impractical. Instead, designers rely on redundancy and digital calibration, using extra comparator levels and post-processing to compensate for offset errors.

Avalanche Pulse Generators

Avalanche transistors can generate extremely fast pulses with rise times well under one nanosecond. The transistor's collector is held above its collector-emitter breakdown voltage but below its collector-base breakdown voltage, a region in which the device presents negative resistance. A capacitor charges through a resistor toward the supply. When a trigger pulse reaches the base, or when the rising voltage itself starts avalanche multiplication, the transistor switches on regeneratively and dumps the capacitor into the load as a sharp pulse. After the discharge, the transistor recovers, the capacitor recharges, and the cycle can repeat.

The avalanche process is inherently noisy, and a free-running avalanche circuit shows timing jitter set by the statistical nature of carrier multiplication. This jitter, while undesirable for precision timing applications, makes avalanche circuits useful sources of randomness, and avalanche noise from reverse-biased junctions is a common entropy source in hardware random number generators.

Selecting Avalanche Transistors

Not every transistor is suitable for avalanche operation, and few are specified for it. The circuit deliberately operates the transistor beyond its rated collector-emitter breakdown voltage, BV_CEO, while staying below its collector-base breakdown voltage, so the part must survive repeated avalanche discharges within that window and keep enough current gain to sustain regeneration. The ZTX415 is specifically designed for avalanche mode: Diodes Incorporated rates it at a BV_CEO above 100 V, a BV_CES above 260 V, and 60 A of peak avalanche current in 20 ns pulses, although its July 2026 data sheet marks it as not recommended for new designs. The 2N2369, a 15 V high-speed switching transistor, is widely used in avalanche pulsers even though its data sheets do not specify the mode, so builders select individual devices by test.

Avalanche breakdown carries a positive temperature coefficient, unlike the negative coefficient of low-voltage Zener breakdown, so a rising junction temperature raises the avalanche voltage. The positive coefficient discourages current from crowding into a hot spot during the avalanche itself, but reliability is still bounded by average power dissipation and peak current. Excessive repetition rate or current density can trigger localized heating and second breakdown that destroys the junction. Limiting average dissipation, controlling peak current, and adequate heat sinking are therefore essential for reliable operation.

Applications of Avalanche Circuits

Avalanche circuits serve specialized needs where their unique characteristics provide advantages:

• Time-Domain Reflectometry: Generating sub-nanosecond steps for measuring cable lengths and locating faults
• Laser Diode Drivers: Producing fast current pulses for pulsed lasers, including time-of-flight laser rangefinders
• Pockels Cell and Streak Camera Drivers: Switching the fast high-voltage pulses that Q-switch lasers and sweep streak-camera deflection plates
• Nuclear Instrumentation: Generating calibration pulses that mimic radiation detector signals
• Random Number Generation: Exploiting avalanche noise for true random bit sequences
• EMC Testing: Creating fast transients for immunity testing

Principles of Super-Regeneration

In a super-regenerative receiver, positive feedback is adjusted to just beyond the oscillation threshold. A quench oscillator periodically reduces the gain below the oscillation point, resetting the circuit. When the quench allows gain to increase, oscillations build up from noise. If an external signal is present at the resonant frequency, oscillations build up faster because they start from a larger initial amplitude.

Because the oscillation grows exponentially from a starting amplitude set by the signal, the time it takes to reach full amplitude depends on the logarithm of the ratio between that amplitude and the signal: a signal ten times stronger arrives sooner by a fixed interval, not in a tenth of the time. In the logarithmic mode, where each burst reaches full amplitude before the quench ends it, the detected output therefore varies roughly with the logarithm of the input, which gives the receiver a wide dynamic range much as automatic gain control would. In the linear mode, the quench cuts each burst off before it reaches full amplitude, so the output stays proportional to the input, but the operating point must then be held carefully. The quench frequency sets a basic trade: a lower rate allows more build-up time and so more gain, but it limits the modulation bandwidth the receiver can pass.

Design Considerations

The quench waveform shape significantly affects receiver performance. Sinusoidal, sawtooth, and square quench waveforms are all used, and the rate at which the quench carries the circuit through the oscillation threshold influences both selectivity and sensitivity. Because the quench samples the received signal once per cycle, its frequency must be at least twice the highest modulation frequency to satisfy the Nyquist criterion. In practice it is set well above that, commonly in the tens of kilohertz and above the audible range so that the quench tone can be filtered out, while remaining far below the radio frequency so that each burst can decay before the next begins.

Super-regenerative receivers radiate energy at the tuned frequency due to the oscillation during build-up. This re-radiation can interfere with nearby receivers and may violate regulatory requirements. Shielding, antenna isolation, and low radiation designs help mitigate this issue.

Modern Applications

More than a century after its invention, the super-regenerative receiver has been displaced by integrated superheterodyne and low-IF receivers in most products, but it survives where cost, simplicity, and power matter more than selectivity:

• Garage Door Openers and Doorbells: Cost-effective receivers for short-range remote control
• Hobby Radio Control and Toys: Simple receivers built around a single active stage
• ISM Band Receiver Modules: Inexpensive receivers for 315 MHz and 433 MHz links
• Wake-Up Receivers: Always-on receivers, a research focus for wireless sensor networks, that wake a more capable radio when a signal is detected

Regenerative Amplification

A regenerative amplifier couples energy back into a tuned circuit in phase with the oscillation. As feedback increases, the circuit's effective resistance decreases, raising the Q factor. At the regeneration threshold, the net resistance becomes zero and the circuit oscillates. Operating just below this threshold achieves maximum Q enhancement while maintaining stability.

If the feedback restores a fraction T of the energy the tuned circuit loses each cycle, only the fraction 1 − T of the loss remains, and the Q rises in inverse proportion:

Q_eff = Q / (1 − T)

Here Q is the tuned circuit's unassisted quality factor and T is the loop gain at resonance, which approaches but must stay below unity. Practical Q multiplications of 10 to 100 times are achievable, requiring T = 0.9 and T = 0.99 respectively, so a modest coil with a Q of 50 becomes an effective Q of 500 to 5,000. At a 455 kHz intermediate frequency, where the bandwidth is the center frequency divided by Q, that narrows a 9.1 kHz response to between 910 Hz and 91 Hz. The hundredfold end requires holding the loop gain within 1 percent of unity, which is the root of the stability problems described next.

Stability Considerations

Q-multiplied circuits are inherently sensitive to component variations and environmental changes. Any drift that increases loop gain above unity causes oscillation, while drift that decreases gain reduces the enhancement. Automatic gain control systems help maintain operation near the optimum point.

Temperature compensation is particularly important because both active device gain and resonator components drift with temperature. Supply variation moves the device gain as well, so Q-multipliers are usually given a well-regulated supply and a regeneration control that can be trimmed in use.

Applications

Q-multiplication finds use wherever high selectivity is needed without expensive high-Q components:

• Radio Receivers: Sharpening the selectivity of simple tuned circuits; the regenerative detector is itself a Q-multiplier, and add-on Q-multipliers sharpened the intermediate-frequency response of communications receivers
• Active Filters: The equal-component Sallen-Key section uses positive feedback from its output to raise its pole Q, which equals 1/(3 − K) and rises without limit as the amplifier gain K approaches 3
• Integrated RF Filters: Q-enhanced LC filters add a cross-coupled negative-resistance pair to restore the low Q of on-chip inductors, at a cost in noise and dynamic range
• Oscillators: Carried past the threshold, the same loss cancellation becomes the start-up mechanism of an LC oscillator

Creating Negative Resistance

Active circuits can synthesize negative resistance through positive feedback. The common op-amp version uses the non-inverting input as its terminal. A divider of R₂ from the output to the inverting input and R₁ from the inverting input to ground sets a non-inverting gain of 1 + R₂/R₁, and a third resistor R connects the output back to the terminal. A positive voltage applied to the terminal drives the output to a larger positive voltage, so current flows back through R and out of the terminal instead of into it. The terminal therefore delivers more current as its voltage rises, and its input resistance is negative:

R_in = −R R₁ / R₂

With R₁ = R₂, the terminal presents −R. The circuit is stable only while the total resistance connected across the terminal is smaller than the magnitude of R_in. Connect a larger one and the net resistance turns negative, the positive feedback wins, and the output runs to a supply rail.

The magnitude of the synthesized negative resistance is set by resistor ratios, so it can be specified precisely, something not possible with devices such as the tunnel diode, whose negative resistance varies with the bias point. The op-amp's bandwidth, in turn, limits the frequency range over which the synthesis holds.

Negative Impedance Converters

The same circuit, in its general form, is the negative impedance converter (NIC). Replacing R with an impedance Z gives Z_in = −Z R₁/R₂, so an impedance connected at one port appears negated, and scaled by the resistor ratio, at the other. A capacitor becomes a negative capacitor and an inductor a negative inductor. These are non-Foster elements, so called because their reactance falls with rising frequency, whereas Foster's reactance theorem requires the reactance of every lossless passive network to rise. NICs are also used to cancel parasitic impedances. Simulating an inductor with a capacitor is the work of related circuits: the gyrator, which inverts an impedance rather than negating it, and the generalized impedance converter.

In antenna work, an NIC can cancel the reactance of an electrically small antenna over a bandwidth that the Bode-Fano limit denies any passive matching network. The technique has been demonstrated in research, but stability is hard to guarantee, because the circuit is stable only for certain terminations, and the converter's own noise adds directly to a receiver's input.

Loss Cancellation

Negative resistance can cancel the positive resistance representing losses in a resonant circuit. Placing a negative resistance in parallel with a tank whose losses are modeled as a parallel resistance R_p raises the effective parallel resistance, and with it the Q, which is Q-multiplication seen from the other side. Carried past full cancellation, the same arrangement becomes an oscillator. A cross-coupled transistor pair presents about −2/g_m across its tank, and start-up requires g_mR_p > 2, the negative-resistance form of the oscillation condition that governs most integrated LC oscillators.

Complete loss cancellation sits exactly at the edge of oscillation, so an amplifier or filter keeps a small net positive resistance for stability, while an oscillator is given a start-up margin and relies on amplitude limiting to bring the net resistance back to zero. The ability to tune the negative resistance provides a convenient way to optimize circuit performance or to compensate for component aging.

Tunnel Diode Characteristics

The tunnel diode's current-voltage characteristic shows a region where current decreases as voltage increases, creating negative differential resistance. The region is bounded by a peak current at the lower voltage and a valley current at the higher voltage, and it lies at low forward voltages, typically below 0.5 V. Practical devices operate at a few milliamperes and a few tenths of a volt, so the available signal swing, and with it the output power, is small.

Tunnel diode switching times can be in the picosecond range because current flow involves quantum tunneling rather than minority carrier diffusion; the junction capacitance, charged by the diode's small current, sets the practical limit. This makes tunnel diodes among the fastest semiconductor devices available, though their small current capacity limits applications to low-power circuits.

Amplifier Design

A tunnel diode amplifier uses the negative resistance region to provide gain. The diode is biased in the negative resistance region, where small signal variations cause amplified responses. In the usual microwave form, a circulator separates the incident and reflected waves, and a diode presenting a negative resistance of magnitude R_n to a line of characteristic impedance Z₀ reflects more power than it receives, with a power gain of

G = [(R_n + Z₀) / (R_n − Z₀)]²

With R_n = 60 Ω and Z₀ = 50 Ω, G = (110/10)² = 121, about 20.8 dB. The gain grows without limit as R_n approaches Z₀, and the circuit is stable only while R_n exceeds Z₀, so gain and stability trade directly.

Stability requires careful attention to bias and termination. If the total resistance seen by the diode becomes negative at any frequency, oscillation results. Resistive loading, frequency-selective networks, and careful layout minimize parasitic resonances that could cause spurious oscillations.

Applications and Limitations

Tunnel diode amplifiers serve niche applications where their unique properties provide advantages:

• Low-Noise Preamplifiers: Tunnel diode reflection amplifiers served as low-noise microwave front ends, including in satellite receivers, until gallium arsenide field-effect transistors displaced them
• High-Speed Triggers: Picosecond switching enables ultra-fast pulse circuits
• Frequency Multipliers: The nonlinear characteristic generates harmonics efficiently
• Microwave Oscillators: Simple oscillators reaching tens of gigahertz

Manufacturing challenges and limited availability have restricted tunnel diode adoption. Modern alternatives including resonant tunneling diodes (RTDs) offer similar characteristics with better manufacturability, extending the frequency range into the terahertz region.

Stability Management

• Define Operating Boundaries: Clearly establish the conditions under which the circuit should oscillate, latch, or amplify
• Account for Tolerances: Component variations can push marginally stable circuits into oscillation
• Consider Temperature Effects: Gain changes with temperature can alter positive feedback levels
• Analyze All Frequencies: Ensure stability at frequencies outside the intended operating range
• Include Start-Up Provisions: Circuits that should oscillate must reliably start under all conditions

Noise Considerations

• Noise Amplification: Positive feedback amplifies noise along with signal, potentially limiting dynamic range
• Jitter in Timing Circuits: Regenerative switching times depend on noise, causing timing uncertainty
• Bandwidth Selection: Limiting bandwidth reduces integrated noise without sacrificing necessary performance

Testing and Verification

• Measure Hysteresis: Verify threshold voltages match design intent
• Check Over Voltage and Temperature: Ensure proper operation across the full operating range
• Probe for Oscillation: Look for unwanted oscillation at frequencies above the intended bandwidth
• Verify Margins: Confirm adequate margin exists between operating point and instability