Voltage Comparator: Deep Guide

November 12 2025
Ersa

Guide to voltage comparator ICs: how it works, inverting vs non-inverting, hysteresis sizing, pull-up and window detection, selection tips.
 

A voltage comparator compares two input voltages and drives a binary output for threshold/window detection, edge timing, and front-end protection. This hub is intent-first: overdrive vs propagation-delay, outputs & pull-ups, hysteresis, CMVR/clamps, window planning, interfacing, validation, application playbooks, selection matrix, BOM and FAQs—so readers can progress from “understand” to “select & procure”.

Quick Browse

Browse sections

What & When

A voltage comparator is an analog decision device that compares two input voltages and drives a binary output. Used in threshold and window detection, timing edges, and front-end protection, a voltage comparator IC asserts when the non-inverting input exceeds the inverting input and de-asserts otherwise.

How it works

  1. Two inputs (V+, V−) and one digital-style output (Vo).
  2. Huge open-loop gain turns tiny ΔVIN into rail switching.
  3. Propagation delay shrinks with larger overdrive and faster devices.
  4. Add positive feedback (hysteresis) to stop chatter near threshold.

Key features & applications

  • Open-loop switching (not linear op-amp feedback).
  • Threshold detection using fixed or divided Vref.
  • Building block for window checks and ADC timing edges.
  • Hysteresis to suppress ripple-induced toggling on slow ramps.
Voltage comparator IC with V+, V−, Vref and binary output Vo on a deep blue background
Binary output asserts when V+ exceeds V−.
Trip Logic

If V+ > V− → Vo = HIGH; otherwise LOW. Inverting topology flips the sense.

Speed vs Overdrive

Propagation delay TPD decreases as input overdrive ΔVIN increases; at tiny overdrive expect about 1.2–1.5× nominal TPD.

Noise Hardening

Vhys ≥ 3× Vin,pp (≥ 4× for automotive) to suppress ripple-induced chatter; large source impedance can dilute hysteresis—buffer or reduce source R.

CMVR Reality

Near the rails the valid input common-mode range can shrink and TPD can worsen. Verify RRIO/CMVR and re-measure timing at corner conditions.

Jump to interface choices · see delay vs overdrive


 

Overdrive vs Propagation Delay

Goal: Given shortest pulse width and ΔVIN, decide if pulses will be missed.

  • ΔVIN < 5 mVTPD ≈ 1.2–1.5× nominal (near-threshold slows switching).
  • ΔVIN ≥ 100 mV → approaches device minimum TPD.
  • Sweep ΔVIN × Temperature × VDD; record minimum and worst-case TPD.
  • If 2 × TPD ≥ τ_min (shortest pulse), use a faster device or increase ΔVIN.
Practical Bound

For single-shot capture, require pulse_min ≥ 3 × TPD_max to absorb asymmetry and output filter.

Cold-Corner Risk

Cold often lengthens TPD 10–30%. Validate with small ΔVIN at −40 °C (or device minimum).

Jitter Budget

Comparator adds edge uncertainty ≈ σ_T ≈ σ_noise / (dV/dt) at threshold. Increase slew or hysteresis.

Propagation delay vs input overdrive ΔVIN with tiny, typical and large regions; narrow-pulse loss shaded
Delay falls as overdrive grows; tiny-ΔVIN risks missing narrow pulses.

Detect tiny-ΔVIN pulses →


Outputs & Pull-Up Choices (OD/PP)

Goal: Choose open-drain vs push-pull; size pull-up so rise time does not clip the shortest pulse.

  • Open-drain: easy level mixing and wired-AND; rise time set by RPU × CL.
  • Push-pull: symmetric edges for higher speeds; ensure level-domain compatibility.
  • Rule of thumb: t_r ≈ 2.2 · RPU · CL; require t_r ≤ 0.1 · τ_min.
  • Back-solve RPU,max and current I ≈ VPU/RPU; for multi-sources ensure ΣI_sink ≤ IOL(max).
Sizing Shortcut

Given τ_min and CL, set RPU ≤ 0.045 · τ_min / CL (uses 0.1× rule with 2.2 factor).

Cross-Domain Pull-up

When level-shifting, pull-up to the receiving domain; confirm absolute max on the comparator’s output.

Wired-AND Budget

For N sinks, ensure N · Ileak + Ipullup(HIGH) ≤ IOH of the receiving gate; check low-level VOL at IOL.

Open-drain comparator with pull-up resistor and load capacitance; rise time set by R × CL
Pull-up and load set rise time; keep it well below the shortest pulse.

See pull-up sizing →


Hysteresis for Noise Immunity

Goal: Choose “enough” hysteresis for ripple/slow-ramp rails and avoid chatter.

  • Baseline: Vhys ≥ 3 × Vin,pp; automotive: ≥ 4 ×.
  • Large source impedance reduces effective hysteresis; buffer or lower the source resistance.
  • Sampling window must cover typical ripple, not rare spikes.
  • Asymmetric thresholds can dodge ripple peaks/valleys.
Design Formula

For inverting Schmitt: Vhys ≈ VOUT,pp · R1/(R1+R2). Tune by the feedback ratio; keep bias currents in budget.

Slew-Rate Link

Edge stability needs Vhys > σ_noise · 6 and dV/dt high enough that σ_T ≪ τ_min/10.

Sensor Front-Ends

For slow sensors (NTC, shunts) place RC pre-filter before comparator; set f_c below ripple, above event bandwidth.

Comparator with hysteresis (Schmitt trigger) showing dual thresholds rejecting ripple noise
Dual thresholds (VTH+ / VTH−) reject ripple around the trip point.

Hysteresis formulas & 10 cases →


CMVR, Input Protection & Clamp

Goal: Keep inputs safe and valid near rails and during over-voltage events.

  • Near rails, confirm input common-mode range (CMVR) or RRIO tolerance.
  • When ESD/clamp diodes conduct, limit Iclamp ≤ 1 mA via series resistance.
  • For ±12 V, +48 V, or ISO 7637-2 transients, combine series R with TVS/RC.
  • Small ΔVIN near CMVR edges can lengthen TPD — re-measure at corners.
±12 V Input

Series R ≈ 12–22 kΩ; add small RC (e.g., 1–4.7 nF) to tame spikes.

+48 V Lines

Series R ≈ 47–100 kΩ plus SMBJ class TVS; check input leakage/headroom.

Automotive Surge

Series R + 600–1000 W TVS; validate under ISO 7637-2/ISO 16750 pulses.

High-voltage input protection for low-voltage comparators; clamp current kept below 1 mA
Keep clamp current small and verify CMVR near rail limits.

High-voltage inputs — clamp & CMVR →


Window Comparator (Two Comparators)

Goal: Detect “in-range / out-of-range” and budget threshold errors.

  • Two comparators plus a divider define VTH_L and VTH_H.
  • Error stack (approx.): σ ≈ √(Vref%² + Rdiv%² + VOS%²); add temperature drift linearly.
  • Open-drain OR outputs share one pull-up; expect asymmetric delays.
  • Supervision parts with internal reference/hysteresis reduce drift and PCB spread.
Tolerance Budget

Treat divider tolerance as RSS with reference and offset; for tight windows use 0.1% resistors and low-drift ref.

Delay Asymmetry

With open-drain, LH and HL paths differ (passive pull-up). Check both against minimum pulse width.

Noise Margin

Set VTH_H − VTH_L > worst-case ripple + offsets + temp drift to avoid window flapping.

Window comparator with upper and lower thresholds and stacked error contributors
Upper/lower thresholds define the pass band; track reference and divider errors.

Window error budget details →

Interfacing

Connect comparators to downstream logic with clear rules for current budgeting, rise/fall shaping, and level-domain matching. Focus on wired-AND behavior, pull-up placement, and module board caveats (LED + large pull-ups).

Wired-AND comparators sharing one pull-up; pull-up in receiving domain; series damping at driver
Pull up to the receiving domain; check ΣIsink and use a 33–100 Ω series resistor near the driver.
Wired-AND Budget

For N sources sharing one pull-up: ΣI_sink ≤ IOL(max) of the active comparator. Estimate I_sink ≈ Σ[(VPU−VOL)/RPU,i] + ΣIleak,i.

Rise-Time Bound

Use tr ≈ 2.2·RPU·CL and require tr ≤ 0.1·τmin to avoid pulse clipping.

Level Domain Choice

Pull up to the receiving domain (MCU/FPGA). Check output absolute max and VIH/VIL windows.

Series Damping

Place a 33–100 Ω series resistor near the driver to tame ringing/EMI on cables/backplanes.

Worked Example — 3.3 V bus, N=3 sources

Given RPU=4.7 kΩ, CL=50 pF, target τmin=20 ns. Rise time tr≈2.2·4.7k·50p≈517 nsfails. Need RPU ≤ 0.045·τmin/CL ≈ 18 Ω (use 100–330 Ω practical + driver sink check), or reduce CL/buffer the node.

Module wiring note
  • Common “comparator modules” ship with large pull-ups (10–100 kΩ) and an LED to VPU → slow edges and level mismatch.
  • Fixes: (1) replace pull-up to meet tr bound; (2) move pull-up to the receiving domain; (3) remove/retime the LED or buffer it.
  • Check: power-up order, reverse-current paths, leakage vs. divider error, MCU VIH/VIL at temperature.
Comparator module pitfalls: LED to VPU and large pull-up causing slow edges and level mismatch; recommended fixes shown
Typical module traps and remedies: pull-up value/domain and LED buffering.

Debounce & slow inputs →


Validation & Production Notes

Comparator timing depends strongly on ΔVIN, temperature, and supply. Validate with a 3-D sweep, then lock production with statistics (CPK) and sound measurement practice.

Validation flow: ΔVIN × Temperature × VDD sweep to measure TPD, thresholds, hysteresis and jitter; production stats with CPK
Sweep ΔVIN × Temp × VDD → log TPD/thresholds/hysteresis/jitter → qualify with CPK and control charts.
Timing Criterion

Require max(TPD_LH,TPD_HL) ≤ spec and pulse_min ≥ 3×TPD_max.

Hysteresis Criterion

Vhys ≥ 3×Vin,pp (automotive ≥ 4×) and |ΔVhys|/Vhys ≤ 10% across PVT.

Jitter Budget

Edge uncertainty ≈ σT ≈ σnoise / (dV/dt); target σT ≤ τmin/10.

Measurement Notes

Scope BW ≥ 5×(1/tr), 10× low-C probe or 50 Ω input; de-embed cable delay; tight local decoupling (0.1 µF + 1 µF).

ΔVIN point (mV) Temp (°C) VDD (V) TPD_LH / TPD_HL (ns) VTH+ / VTH− (V) Vhys (mV) Jitter σT (ns) Pass/Fail criterion Samples (n) Lot / Unit / Notes
2 −40 VDD_min         TPD_max & jitter within spec 30 Lot#, Unit#, Fixture
5 25 VDD_typ         Vhys ≥ 3×Vin_pp 30 Lot#, Unit#, Fixture
100 125 VDD_max         σT ≤ τ_min/10 30 Lot#, Unit#, Fixture
Sampling & CPK

Per lot/line/shift start with n=30. Compute CPK on TPD, VTH, Vhys; target CPK ≥ 1.33 (≥1.67 preferred).

MSA & Control

Run GR&R for fixtures; track with X-bar/R or I-MR charts; define stop/containment when rules trip.

Common Pitfalls

Tiny ΔVIN inflates TPD; module LEDs slow edges; CMVR edge + temp causes chatter; cross-domain pull-up over-volts MCU; probe loading shifts thresholds.

Download artifacts: CSV validation log and PDF checklist
Provide both raw data (CSV) and human checklist (PDF) alongside scripts.

Patterns & Deep-Dive Previews

Hub cards preview the key comparator patterns. Each card gives a one-paragraph summary, one golden rule, one risk, and one CTA into a spoke page. We keep this section concise to preserve the main page’s vertical intent.

Hysteresis Comparator

Positive feedback creates two thresholds (VTH+ / VTH−) that suppress chatter around the trip point, stabilizing edges on ripple-rich or slow-ramp rails.

Rule: Vhys ≥ 3×Vin,pp (auto ≥4×) Risk: Source impedance reduces effective Vhys
Hysteresis design →

Pull-Up Sizing (OD/PP)

Choose open-drain for level mixing or push-pull for symmetry. Size the pull-up to keep rise time far below the minimum pulse width budget.

Rule: tr ≈ 2.2·R·CL; tr ≤ 0.1·τmin Risk: Missed narrow pulses
Pull-up sizing →

Window Error Budget

Stack reference, divider tolerance, input offset and temperature drift to bound upper/lower thresholds and their pass-band timing skew.

Rule: σ ≈ √(Vref%² + Rdiv%² + Vos%²) Risk: OD OR → asymmetric delays
Window budget →

Debounce & Slow Inputs

For relays/switches/optos, combine RC timing with Schmitt thresholds to reject bounce and slow edges without distorting functional timing.

Rule: τ ≈ 5–10 ms (mechanical) Risk: Over-/under-debounce
Debounce patterns →

Tiny ΔVIN & Narrow Pulses

Overdrive-dependent propagation delay dominates near threshold. Validate minimum detectable width across ΔVIN, temperature and VDD.

Rule: ΔVIN < 5 mV → TPD ↑ Risk: Corner-case misses
Detect narrow pulses →

High-Voltage Inputs & CMVR/Clamp

Protect low-voltage comparators with series-R, TVS/RC, and proper biasing; verify common-mode range near rails and re-measure delay vs. overdrive.

Rule: I_clamp ≤ 1 mA Risk: Clamp skew & delay
HV inputs & CMVR →

Applications (10 domains)

Comparator use cases grouped by topology, key specs, executable rules, risks, and validation notes. Each card previews the essentials and links to an application-specific IC shortlist for procurement.

Power-Good / UV-OV Supervision

Window comparator to assert PG within bounds and flag faults.

  • Topology: Window + open-drain to PG tree
  • Key specs: Vhys 1–3%, CMVR near rails, low Iq
  • Rules: Vhys ≥ 3×Vin,pp; include divider/ref in σ
  • Risks: Power-up order; OD OR delay asymmetry
  • Validation: cold-crank/surge/ripple sweeps
IC shortlist →

Zero-Cross Detection

Edge detection for AC sensing and synchronous control.

  • Topology: Non-inverting + mid-rail bias
  • Key specs: TPD, input protection to ±HV
  • Rules: Add Schmitt; filter HF noise
  • Risks: CMVR edge, high dv/dt injection
  • Validation: fMAX vs TPD; noise/THD injection
IC shortlist →

Battery Window / Cell Guard

In-range detection for cell voltage protection and alarms.

  • Topology: Window + OD OR to fault line
  • Key specs: offset & drift; built-in refs
  • Rules: σ ≈ √(Vref%² + Rdiv%² + Vos%²)
  • Risks: High-R divider leakage
  • Validation: temp sweep + long-term drift
IC shortlist →

Over-Current / Shunt Trip

Trip on shunt voltage or with a front-end current sense amp.

  • Topology: Inverting threshold
  • Key specs: ΔVIN sensitivity, TPD
  • Rules: Ensure ΔVIN ≥ 5–10 mV
  • Risks: Cold/low-VDD misses
  • Validation: narrow-pulse train sweep
IC shortlist →

Motor/Encoder Edge Recovery

Square up sensor edges on long/noisy cables for MCU capture.

  • Topology: Schmitt + OD/PP
  • Key specs: Vhys, ESD, EMI
  • Rules: Vhys ≥ 2× noise-pp
  • Risks: Ringing → add 33–100 Ω
  • Validation: EMI & cable length sweep
IC shortlist →

Sensor Window Guard (Temp/Pressure)

In-range/out-of-range guard for industrial sensors.

  • Topology: Window + bias
  • Key specs: offset, ref accuracy, Iq
  • Rules: Prefer built-in precision refs
  • Risks: Drift → false windows
  • Validation: temp cycling + ADC cross-check
IC shortlist →

Clock Restoration / Squaring

Restore degraded clocks for digital domains.

  • Topology: High-speed PP comparator
  • Key specs: tPD, tr/tf symmetry, fMAX
  • Rules: fMAX ≈ 1/(tr+tf+tPD_LH+tPD_HL)
  • Risks: Low input slew → jitter
  • Validation: eye/jitter vs slew sweep
IC shortlist →

ADC SOC / Timing Qualifier

Clean, deterministic edge for start-of-conversion or gating.

  • Topology: Non-inv + Schmitt; PP
  • Key specs: deterministic delay, CMVR
  • Rules: pulse_min ≥ 3×TPD_max
  • Risks: ΔVIN ≈ 0 → meta-like flutters
  • Validation: ΔVIN vs SOC timing plot
IC shortlist →

Windowed RPM / Speed Guard

Speed in-range detection with noise-immune thresholds.

  • Topology: Zero-cross + window + debounce
  • Key specs: noise immunity, Vhys
  • Rules: Set τ per sensor; Vhys per ripple
  • Risks: Mechanical bounce false alarms
  • Validation: vibration & spectrum tests
IC shortlist →

Mains Presence / Brownout

Detect line presence and brownout with safe isolation margin.

  • Topology: Rectified sense + divider + Schmitt
  • Key specs: input protection, CMVR, isolation
  • Rules: I_clamp ≤ 1 mA; creepage/clearance
  • Risks: Surge events → TVS + series-R
  • Validation: surge/ESD/EFT compliance
IC shortlist →

Selection Matrix — 7-Brand Shortlist

① Lock by speed (TPD) & ΔVIN
Match shortest pulse width & overdrive: High-speed ≤10 ns / Mid 10–100 ns / General 0.1–1 µs / Ultra-low-power ≥1 µs.
② Interface & power
OD vs PP, pull-up domain & CL, rise time vs τmin; RRIO/CMVR near rails; Iq cap for battery.
③ Qual/pack/stock
AEC-Q100, package height, pin semantics, and “in-stock” band (High/Medium/Low) for procurement.
Card fields: TPD bin (ns) · Iq (µA) · CMVR (RRIO/rail headroom) · Output (OD/PP/IOUT) · VOS grade · AEC-Q100 · Package height · Stock (H/M/L)

Texas Instruments

TLV3501 / TLV3502
Role: High-speed PP

~4.5 ns PP outputs, good tr/tf symmetry; ideal for clock restore/edge recovery at high fMAX.

  • TPD: ~4.5 ns (bin ≤10 ns) · Iq: mid
  • CMVR: RRIO; Output: PP, IOUT symmetric
  • VOS: low; AEC-Q100: —; Height: low
  • Stock: Medium
Why: Deterministic, fast, square edges → narrow-pulse integrity.
TLV3201 / TLV3202
Role: Fast general PP

~40 ns PP, RRIO; balanced speed/power for SOC triggers and timing qualifiers.

  • TPD: ~40 ns (bin 10–100 ns) · Iq: low-mid
  • CMVR: RRIO; Output: PP
  • VOS: low; AEC-Q100: —; Height: low
  • Stock: High
Why: “Default fast PP” with wide domain fit and easy routing.
LM2903-Q1 / LM393-Q1
Role: Automotive OD

Dual OD classics for PG/window trees; massive ecosystem, easy wired-AND and second-source.

  • TPD: ~130–300 ns · Iq: low
  • CMVR: wide; Output: OD, strong sink
  • AEC-Q100: Y; Height: varied (SOIC/TSSOP/SOT)
  • Stock: High
Why: Proven, cheap, everywhere; great for supervision networks.

STMicroelectronics

TS3011
Role: High-speed PP

~8 ns PP with RRIO; robust for clock squaring and fast edge restoration.

  • TPD: ~8 ns · Iq: mid
  • CMVR: RRIO; Output: PP
  • AEC-Q100: —; Height: low
  • Stock: Medium
Why: Fast & RRIO for compact high-speed domains.
TS3021 / TS3022
Role: High-speed OD

Open-drain high-speed; easy wired-AND and level mixing with pull-up domain control.

  • TPD: ~20 ns class · Iq: mid
  • Output: OD; CMVR: wide
  • AEC-Q100: —; Height: low
  • Stock: Medium
Why: OD speed with flexible domain interfacing.
TS881
Role: Ultra-low-power

nA-µA class Iq for battery/wearables; perfect for slow sensors & duty-cycled sampling.

  • TPD: µs bin · Iq: ultra-low
  • CMVR: wide; Output: OD/PP variants
  • Stock: Medium
Why: Battery life first; speed is secondary.

onsemi

NCS2200 / NCV2200
Role: Ultra-low-power OD

µA Iq with OD output for watchdog/threshold; automotive NCV variant available.

  • TPD: µs bin · Iq: ultra-low
  • Output: OD; AEC-Q100: Y (NCV)
  • Stock: Medium
Why: Battery projects needing OD & car variants.
LM2903 / LM393 / LM311
Role: Industrial/auto staples

Core OD families with broad packages & second-source; LM311 as classic single with faster edges.

  • TPD: ~80–300 ns (LM311 faster) · Iq: low
  • Output: OD; AEC-Q100: options
  • Stock: High
Why: Easy availability; good for window/PG nets.

Microchip

MCP6561/2/6
Role: Fast PP, RRIO

~20–50 ns PP, RRIO; great for ADC SOC/timing gates with deterministic edges.

  • TPD: ~20–50 ns · Iq: low
  • Output: PP; CMVR: RRIO
  • Stock: High
Why: Solid “fast PP” with good availability.
MCP6541/2
Role: Ultra-low-power

Very low Iq for battery sensors; OD/PP variants; tolerant to slow edges with Schmitt configs.

  • TPD: µs bin · Iq: ultra-low
  • Output: OD/PP; CMVR: wide
  • Stock: Medium
Why: Great for low-power windows/debounce.

Renesas (Intersil)

ISL55110 / ISL55123
Role: Ultra-high-speed

Super-fast comparators for sampling/clock restore where jitter budget is tight.

  • TPD: sub-ns–few ns · Iq: higher
  • Output: PP; CMVR: spec-dependent
  • Stock: Medium
Why: When fMAX or jitter dominates.
ICL7665
Role: Dual thresholds

Classic dual-window supervisor; simple cell/battery guards without MCU.

  • TPD: ~100 ns–µs · Iq: low
  • Output: OD/PP options; AEC: project-dependent
  • Stock: Medium
Why: Minimalist window protection block.

NXP

LM393 / LM2903 (NXP)
Role: Industrial OD

Workhorse OD comparators with broad packaging and second-source paths.

  • TPD: ~130–300 ns · Iq: low
  • Output: OD; CMVR: wide
  • Stock: High
Why: Easy multi-vendor alignment.
LM311 (NXP)
Role: Faster single OD

Classic single OD with faster response; suits legacy sockets & simple level detection.

  • TPD: tens of ns · Iq: low-mid
  • Output: OD; CMVR: wide
  • Stock: Medium
Why: Drop-in OD with speed headroom.

Melexis

MLX90xxx Sensor ICs (with internal comparator)
Role: Automotive sensing

For projects bound to Melexis brand, use sensor ICs with built-in comparators to realize thresholds/edges without a discrete comparator.

  • TPD: internal path; Iq: per device
  • AEC-Q100: Y (sensor lines); Height: per package
  • Stock: Low–Medium
Note: For general discrete comparators prefer TI/ST/onsemi/Microchip; keep Melexis as domain-specific option.

BOM & Procurement

Required fields

  • Vrail / Level domains (cross-domain pull-up?)
  • Vth / Vhys targets (incl. tolerances)
  • TPD requirement (shortest pulse, ΔVIN scenario)
  • Output type / Pull-up domain / CL estimate
  • Iq limit (battery) · Temp drift limit
  • AEC-Q100 (Y/N) · Package height · Second-source plan

Risks & mitigations

  • Pin/semantic mismatch (OD/PP, VO, EN/REF) — verify before layout
  • EOL/NRND lifecycle status — avoid fragile SKUs early
  • Lead time / MOQ for small builds — plan buffer SKUs
  • Lot variation in TPD vs ΔVIN, VOS, Vhys — include corner sampling
  • Second-source alignment of Vth/Vhys/CMVR/IOUT — lab cross-check

Submit your BOM

Send your rails, thresholds, speed & interface constraints. We’ll return a 7-brand aligned shortlist within 48 hours.

/submit-bom →
Example SKUs (why they fit)
  • TI: TLV3501 — fast PP for edge/clock; LM2903-Q1 — auto OD for PG/window; TLV7011 — low-power RRIO for battery.
  • ST: TS3011 — ~8 ns PP RRIO; TS3021 — OD high-speed, wired-AND; TS881 — ultra-low-power.
  • onsemi: NCS2200/NCV2200 — µA Iq OD; LM2903 — industrial/auto staple.
  • Microchip: MCP6561 — ~50 ns PP for SOC/ADC; MCP6541 — ultra-low-power for slow sensors.
  • Renesas: ISL55110 — ultra-high-speed, tight jitter; ICL7665 — dual thresholds for simple guards.
  • NXP: LM393/LM2903 — OD second-source paths; LM311 — faster single OD legacy fit.
  • Melexis: MLX90xxx sensor ICs with internal comparator — when brand lock requires integrated solution.

FAQs — Voltage Comparator

How large should hysteresis be for 20–50 mV ripple rails?
Use a practical floor of Vhys ≥ 3×Vin,pp for general rails and ≥ 4× for automotive noise. Source resistance reduces effective hysteresis, so buffer or lower the source impedance if trip points still chatter. Validate at worst ripple and temperature corners with the scope’s time-trend to confirm a clean single transition band.
Further reading: Hysteresis design
What’s the relation between tiny ΔVIN and propagation delay—how do I avoid missing narrow pulses?
Propagation delay stretches when ΔVIN is tiny: below ~5 mV, expect TPD ≈ 1.2–1.5× nominal and stronger temperature dependence. Keep the shortest pulse ≥ 2× worst-case TPD, increase ΔVIN with gain or bias, or select a faster comparator. Verify at cold corners and minimum supply where delay typically peaks.
Further reading: Overdrive vs Delay
How do I size the pull-up resistor so rise time doesn’t clip the shortest pulse?
For open-drain outputs use tr ≈ 2.2·RPU·CL and require tr ≤ 0.1·τmin. This yields RPU,max = 0.1·τmin/(2.2·CL). Check ΣI_sink ≤ IOL(max) when multiple lines share the node, and confirm level-domain compatibility if the pull-up is not on the comparator’s supply rail.
Further reading: Outputs & Pull-Up
What’s a safe input clamp for high-voltage comparator ICs (±48 V into low-voltage parts)?
When ESD or clamp diodes conduct, limit current aggressively: use Iclamp ≤ 1 mA as a simple design rule. Typical series resistors are ~12–22 kΩ for ±12 V and ~47–100 kΩ for +48 V domains, adding TVS and small RC where surge standards apply. Re-check CMVR margins near the rails.
Further reading: CMVR & Clamp
What’s inside a “hysteresis comparator” and how does it differ from adding RC?
True hysteresis uses positive feedback to create fixed dual thresholds, VTH+ and VTH−, guaranteeing one clean transition band. A front-end RC only filters; on slow ramps or large ripple it can still dither around the single threshold. Prefer built-in or feedback hysteresis, then add RC for bandwidth shaping.
Further reading: Hysteresis design
When should I choose open-drain vs push-pull outputs?
Choose open-drain for level-mixing, wired-AND and multi-source lines, accepting R·C-limited rise times. Choose push-pull for symmetric edges and higher speeds. As a simple rule, if your τmin is near 100 ns or faster, push-pull is often safer; for cross-domain or multi-device buses, open-drain remains ideal.
Further reading: Outputs & Pull-Up
How do I plan window thresholds and include Vref/Rdiv/Vos/temperature drift?
Start from the functional limits, then budget tolerances with σ ≈ √(Vref%² + Rdiv%² + VOS%²). Add temperature contributions linearly if specified that way, and keep the pass window ≥ 6σ to avoid edge escapes. Remember open-drain ORing creates asymmetric delays that can skew effective window timing.
Further reading: Window thresholds
How do I wire comparator “modules” to 3.3 V MCUs when the supply is 5 V?
Many modules ship with 5 V pull-ups and an LED path that slows edges. Move the pull-up to 3.3 V or replace the on-board resistor, remove or bypass the LED path, and recompute tr with tr ≈ 2.2·RPU·CL. Add a small 33–100 Ω series resistor near the MCU pin to tame reflections on long runs.
Further reading: Interfacing notes
How do I verify thresholds/hysteresis across −40~+125 °C with CPK≥1.33?
Sweep ΔVIN × temperature × VDD and log VTH+, VTH−, Vhys and TPD using properly bandwidth-limited probes and local decoupling. Use at least 30 units per lot and compute process capability; only release if CPK ≥ 1.33 at the worst corner. Archive raw CSV and scope setups for traceability.
Further reading: Validation plan
What’s a safe acceptance criterion for tiny ΔVIN detection at low supply?
Define the shortest pulse ≥ 2× TPD measured at the minimum VDD and cold temperature with the smallest ΔVIN. Target false-miss probability below your system rate (e.g., < 10⁻⁶ per hour) and allocate ≥ 20% margin for 3σ drift. Add hysteresis to stabilize edges without erasing true narrow events.
Why not reuse an op-amp comparator circuit diagram for high-speed edges? (brief)
Op-amps are optimized for linear operation; saturating them yields long recovery and inconsistent delays. Dedicated comparators have output stages, internal hysteresis and propagation paths tuned for rapid rail-to-rail switching. For jitter budgets and sharp timing edges, a comparator IC is the reliable and repeatable choice.
Further reading: Overdrive vs Delay
What are common pitfalls in second-sourcing across brands (polarity, pinout, hysteresis method)?
Watch for inverted logic, OD vs PP substitutions, CMVR headroom near rails, and different hysteresis implementations or magnitudes. Identical packages can hide pin-semantic changes. Run A/B curves for VTH+/VTH−, Vhys and TPD vs ΔVIN, then freeze a cross-brand test sheet before production release.
Ersa

Anastasia is a dedicated writer who finds immense joy in crafting technical articles that aim to disseminate knowledge about integrated circuits (ICs). Her passion lies in unraveling intricate concepts and presenting them in a simplified manner, making them easily understandable for a diverse range of readers.