Precision 50/60 Hz Notch Filters (Twin-T & Active)
← Back to:Active Filters & Signal Conditioning
A precision 50/60 Hz notch is a repeatable, deep and stable hum killer: it suppresses mains interference without damaging the useful band. The real win comes from controlling mismatch, loading, and temperature drift with a verification-and-production strategy that keeps depth and center frequency consistent.
H2-1. What a Precision 50/60 Hz Notch Is
A precision 50/60 Hz notch is a narrow, deep rejection placed exactly at mains frequency to suppress hum without damaging the wanted passband. “Precision” does not mean “a visible dip in simulation” — it means the notch remains deep, centered, and repeatable across tolerance, temperature, and loading.
What 50/60 Hz hum looks like (in practice)
- Dominant single tone at 50 Hz or 60 Hz, often with a few harmonics (100/120, 150/180…)
- Usually appears as a narrowband interference that steals headroom and pollutes FFT bins
- If front-end is already clipping/rectifying, harmonics grow and a notch alone rarely “finishes the job” (system-level coupling must be fixed first)
When a notch is the right tool (fast decision rule)
- Use a notch when the interference is mainly a linear, narrowband 50/60 Hz tone and the signal chain must preserve passband gain.
- Fix coupling first when measurements show obvious overload artifacts (flat-topping, asymmetry, or strong harmonic comb) — a notch will not restore headroom already lost.
Depth
How much the 50/60 Hz component is reduced at f0. Depth is the “headroom recovery” knob.
Verify: inject 50/60 Hz → measure residual (Vrms or dBFS).
Center accuracy
How well the notch center stays on 50/60 Hz under tolerance and tuning. A deep notch off-center is effectively “missing.”
Verify: f0 error (Hz) at room + corners.
Stability
Depth and f0 must hold across temperature and loading (probe, ADC input network, cable changes).
Verify: sweep temperature + swap load → depth/f0 drift.
Passband impact
A notch that “works” but damages passband gain/phase can break measurement fidelity and calibration.
Verify: insertion loss + phase/group-delay ripple.
Scope (this page)
- Twin-T / active notch structures, depth limits from mismatch, temperature/trim strategy, and verification hooks.
- How to write measurable targets: residual hum, f0 accuracy, drift, and passband damage.
Non-goals (intentionally not expanded here)
- Full EMI/ground-loop mitigation playbooks (system page).
- General biquad synthesis and tunable SVF notch theory (architecture pages).
Field hooks (fast alignment)
Quick check: inject a clean 50/60 Hz tone and measure residual at the ADC input (Vrms or dBFS).
Fix: treat the notch as a “precision block” — isolate loading, use matched ratios, and plan trimming if depth must hold across corners.
Pass criteria: residual hum at f0 stays below the system budget at worst-case gain and temperature.
Fix: treat the notch as a “precision block” — isolate loading, use matched ratios, and plan trimming if depth must hold across corners.
Pass criteria: residual hum at f0 stays below the system budget at worst-case gain and temperature.
H2-2. Specs That Actually Matter (Depth, BW, Passband Damage)
Good notch design starts with measurable targets. The notch depth is set by the residual-hum budget, the notch width must cover mains drift and tuning error, and the passband must remain predictable for calibration and time-domain fidelity.
Spec priority rule (design backwards)
- Set residual-hum budget (Vrms or dBFS) at worst-case gain → this determines required depth.
- Cover mains drift + tuning error → this determines required notch width and acceptable f0 error.
- Constrain passband damage (gain/phase) so calibration and time-domain fidelity remain valid.
Practical spec sheet (what to write and how to verify)
| Metric | Why it matters | Typical target (set by budget) | How to verify |
|---|---|---|---|
| Depth at f0 | Recovers headroom; reduces hum bin energy. | Commonly 40–80 dB, or residual < X dBFS. | Inject 50/60 Hz; measure residual Vrms or FFT bin. |
| f0 error (Δf0) | A deep notch off-frequency behaves like no notch. | Typically within a small fraction of notch width. | Measure center frequency at room + corners. |
| Notch width | Must cover mains drift + tuning error without passband damage. | “Wide enough to not miss,” but “not so wide it hurts.” | Sweep around f0; confirm attenuation stays < spec. |
| Passband flatness / insertion loss | Avoids gain error and preserves calibration scaling. | Set max ripple / loss in the usable band. | Measure gain vs frequency away from notch edges. |
| Phase / group-delay ripple | Protects waveform shape and timing (measurement/audio). | Set max phase / group-delay deviation in passband. | Sweep phase or measure impulse/step response consistency. |
| Drift vs temp / load | Defines “precision” and prevents field surprises. | Specify worst-case depth/f0 over temperature and load set. | Temp sweep + load swap; check depth/f0 stability. |
Pass criteria template (drop-in wording):
Residual hum at 50/60 Hz < X (Vrms or dBFS) at worst-case gain and temperature; notch center error < Y Hz; passband gain error < Z dB.
Residual hum at 50/60 Hz < X (Vrms or dBFS) at worst-case gain and temperature; notch center error < Y Hz; passband gain error < Z dB.
Field hooks (spec sanity)
Quick check: measure residual 50/60 Hz (dBFS or Vrms) at worst-case gain; confirm notch center does not miss the interference.
Fix: widen notch only enough to cover drift; improve ratio matching and isolation before chasing extreme “simulated depth.”
Pass criteria: residual hum + passband gain/phase limits hold across temperature and the expected load set.
Fix: widen notch only enough to cover drift; improve ratio matching and isolation before chasing extreme “simulated depth.”
Pass criteria: residual hum + passband gain/phase limits hold across temperature and the expected load set.
H2-3. Where 50/60 Hz Enters (So You Don’t Over-Filter)
A 50/60 Hz notch is most effective when the interference behaves like a linear, narrowband tone added to the signal. Before adding more filter order, identify where the hum enters and decide whether a notch is a primary fix or only a cleanup tool.
Fast decision rule (3-minute triage)
- Classify the entry as CM, DM, or REF injection using one quick swap/test.
- Check linearity: if harmonics form a comb (100/120/180…), the chain is likely overloading/rectifying.
- Pick the role: notch as primary (DM tone) vs cleanup (CM/REF dominated).
CM
Common-Mode coupling
Symptom: hum changes noticeably with long cables, body proximity, or measurement setup; both inputs move “together.”
Quick check: compare balanced vs unbalanced hookup (or temporarily improve symmetry) and observe hum change; compare differential measurement vs single-ended reading.
If notch helps? Maybe — notch can remove residual tone, but CM problems often need front-end symmetry/CM rejection and layout/ground fixes.
Quick check: compare balanced vs unbalanced hookup (or temporarily improve symmetry) and observe hum change; compare differential measurement vs single-ended reading.
If notch helps? Maybe — notch can remove residual tone, but CM problems often need front-end symmetry/CM rejection and layout/ground fixes.
DM
Differential-Mode injection
Symptom: FFT shows a clean 50/60 Hz bin that scales with front-end gain like a real input component.
Quick check: replace the sensor with a known dummy impedance; if hum drops sharply, DM entry through sensor/cable is likely. Reduce input amplitude/gain to confirm linear scaling.
If notch helps? Yes — DM single-tone behavior is the best match for a precision notch, provided the chain is not overloading.
Quick check: replace the sensor with a known dummy impedance; if hum drops sharply, DM entry through sensor/cable is likely. Reduce input amplitude/gain to confirm linear scaling.
If notch helps? Yes — DM single-tone behavior is the best match for a precision notch, provided the chain is not overloading.
REF
Reference / ground-path injection
Symptom: hum varies drastically with instrument grounding, probe ground clip, or outlet/earth reference; “same circuit, different setup” changes results.
Quick check: change the measurement reference point (single-point vs another node) and compare hum; compare an isolated measurement method vs a grounded one.
If notch helps? Maybe/No — a notch can hide the tone but does not fix an unstable reference path; the underlying injection may still break repeatability.
Quick check: change the measurement reference point (single-point vs another node) and compare hum; compare an isolated measurement method vs a grounded one.
If notch helps? Maybe/No — a notch can hide the tone but does not fix an unstable reference path; the underlying injection may still break repeatability.
When a notch becomes “over-filtering” (treating symptoms only)
Red flag: a harmonic comb (100/120, 150/180…) grows strongly or time-domain waveforms show clipping/rectification.
In this case, removing only the 50/60 Hz bin rarely restores headroom — the chain must regain linearity first.
Non-goals (kept out on purpose)
- Full grounding/shielding playbooks and EMI compliance methods.
- RLD and bio-safety reference-drive details (application pages).
Field hooks (quick action)
Quick check: swap to a dummy sensor impedance and compare the 50/60 Hz bin; change measurement reference and compare again.
Fix: treat CM/REF injection as a repeatability risk; reserve notch depth for DM single-tone cleanup when the chain is linear.
Pass criteria: hum entry class is identified and the notch role (primary vs cleanup) is justified by measurements.
Fix: treat CM/REF injection as a repeatability risk; reserve notch depth for DM single-tone cleanup when the chain is linear.
Pass criteria: hum entry class is identified and the notch role (primary vs cleanup) is justified by measurements.
H2-4. Passive Twin-T Notch: The Core Mechanism
A Twin-T notch uses two complementary T networks in parallel. Around the notch center, the two paths produce conditions that cancel at the output, creating a deep rejection. In real hardware, notch depth is usually limited by ratio mismatch and source/load disturbance rather than by the topology itself.
What sets f0 (without long derivations)
The notch center is set by the RC time constant plus ratio matching between the two T sections.
A common normalized form is f0 ≈ 1 / (2πRC) when the complementary sections use matched ratios.
Precision is dominated by how well these ratios hold across tolerance and temperature.
Pros
- Simple, low-cost notch for fixed hum frequencies (50/60 Hz).
- Works well as a dedicated “cleanup block” when the chain remains linear.
- Easy to combine with buffering or a gain stage when insertion loss must be recovered.
Cons (physics, not opinions)
- Insertion loss is common in passive implementations and must be budgeted.
- Depth is mismatch-limited: cancellation requires accurate ratios, not just “close values.”
- Source/load sensitivity: a different load can shift depth and center unless isolation is designed in.
Best fits
- Fixed 50/60 Hz suppression where notch stability is more important than extreme narrowness.
- Chains that can provide buffer isolation at least on one side of the Twin-T network.
- Systems that value predictable passband behavior and measurable production verification.
Field hooks (first-order sanity)
Quick check: measure notch depth change when swapping one “ratio-critical” R or C to a tighter-matched part.
Fix: prioritize ratio matching and isolation before chasing extreme simulated depth; budget insertion loss early.
Pass criteria: notch center is on target and depth remains within spec under the expected load condition.
Fix: prioritize ratio matching and isolation before chasing extreme simulated depth; budget insertion loss early.
Pass criteria: notch center is on target and depth remains within spec under the expected load condition.
H2-5. Active Notch Variants (Buffered, Bridged, Regenerated Twin-T)
“Making the notch active” is not about adding complexity — it is about solving specific hardware limits: load sensitivity, insertion loss, insufficient depth, or a need to shape effective Q. Each variant improves a different pain point and introduces a different verification risk.
Selection anchor (choose the role first)
- Buffered: keep depth stable when the load or probe changes.
- Bridged: reduce passband loss and improve “fit” in a signal chain.
- Regenerated: push deeper notches with feedback — must prove stability margin.
Variant
Buffered Twin-T
hi-Z in
low-Z out
Gains
- Depth and f0 become repeatable against probe/ADC input changes.
- Reduces “board surprises” caused by source/load interaction.
Pitfalls
- Op-amp noise/distortion can dominate if signal swing is large.
- Output swing and common-mode limits can reduce usable headroom.
When to use
Use when depth shifts with loading or measurement setup, and a notch must remain stable across the expected load set.
Variant
Bridged / Active-assisted Twin-T
lower loss
better fit
Gains
- Improves passband gain behavior versus purely passive insertion.
- Balances notch performance with chain-level gain planning.
Pitfalls
- Op-amp GBW/phase margin affects both notch and passband integrity.
- Depth is still cancellation-limited; ratio matching remains critical.
When to use
Use when passband insertion loss is unacceptable and a modest active assist can restore predictable gain while keeping the notch practical.
Variant
Regenerated Twin-T (feedback-deepened)
feedback
stability
Gains
- Achieves deeper notches for the same matching quality.
- Can increase effective rejection without widening the notch excessively.
Pitfalls
- Positive feedback can oscillate when phase margin is insufficient.
- Temperature and component spread can change loop gain and move behavior.
- Verification must include stability checks, not only depth.
When to use
Use when very deep rejection is mandatory and stability margin can be proven across temperature and load corners.
SVF-notch note (1-line redirect)
If notch frequency/Q must be tunable or multi-response outputs are needed, use an SVF-notch architecture (see the State-Variable/Notch page).
Field hooks (verification focus)
Quick check: sweep load/probe conditions and record depth/f0 shift; for regenerated variants, also check for peaking/ringing near the notch region.
Fix: buffer first to stabilize behavior; use regeneration only when stability can be proven across temperature and spread.
Pass criteria: depth target is met and the design remains stable (no oscillation/peaking) under worst-case conditions.
Fix: buffer first to stabilize behavior; use regeneration only when stability can be proven across temperature and spread.
Pass criteria: depth target is met and the design remains stable (no oscillation/peaking) under worst-case conditions.
H2-6. Depth Is Limited by Mismatch (Tolerance → Notch Depth Budget)
Notch depth is a cancellation problem. Any ratio mismatch leaves a finite residual at the notch center. For first-pass budgeting, residual often scales with mismatch ε, and an order-of-magnitude rule is: Depth(dB) ≈ 20·log10(1/ε) (useful for selecting matching grade, not as a strict derivation).
Depth budgeting (practical 4-step loop)
- Define target residual at 50/60 Hz: < X mVrms (or < X dBFS) at worst-case gain.
- Convert to required depth using expected hum amplitude at the notch input.
- Allocate mismatch budget ε that can achieve the depth target (matching grade selection).
- Map ε to parts/actions: ratio-critical R/C, layout symmetry, and load isolation plan.
Error-source checklist (why “sim deep” becomes “board shallow”)
- Ratio mismatch (R): tolerance + temperature tracking differences; prefer resistor networks/arrays for tracking.
- Ratio mismatch (C): dielectric/tempco spread; prefer C0G/NP0 and matched pairs where cancellation is critical.
- Parasitics: pad/trace asymmetry, leakage paths (humidity/flux), and unintended coupling that disturbs the effective ratios.
- Source/load disturbance: Rs/RL changes alter the cancellation condition; buffering can convert “sensitive” to “repeatable.”
- Temperature & aging: depth and f0 drift when parts do not track over the expected operating profile.
What to improve first (decision rules)
If depth shifts with load
Improve isolation (buffering, controlled interface impedances) before buying tighter capacitors.
If depth is poor at room
Improve ratio matching (networks/arrays, matched pairs, symmetry) — matching usually dominates depth.
If depth is fine at room but fails over temperature
Improve tracking: same series/geometry, thermally similar placement, and avoid asymmetric self-heating.
If only a small correction is needed
Add trim points to center f0; treat trim as a centering tool, while depth remains primarily matching-limited.
Production strategy (one-knob vs two-knob trim)
- One-knob trim: adjust a single element to center f0; fastest and most repeatable for production.
- Two-knob trim: adjust center and depth; higher labor/fixture complexity, reserved for extreme rejection targets.
Field hooks (fast debugging)
Quick check: measure depth sensitivity to swapping one ratio-critical R or C; a large change indicates mismatch-limited depth.
Fix: prioritize ratio matching and tracking (networks/arrays, C0G pairs) and isolate source/load before adding regeneration.
Pass criteria: residual at f0 meets the threshold across tolerance/temperature/load corners defined by the product.
Fix: prioritize ratio matching and tracking (networks/arrays, C0G pairs) and isolate source/load before adding regeneration.
Pass criteria: residual at f0 meets the threshold across tolerance/temperature/load corners defined by the product.
H2-7. Temperature Drift & Compensation (Hold f0 and Depth)
A precision 50/60 Hz notch is only “precision” when both center frequency and depth hold across temperature. In practice, depth stability is usually dominated by ratio tracking (how parts drift together), not by the absolute tempco of a single part.
Where drift comes from (notch-relevant only)
- R tracking: ratio drift from tempco mismatch and non-uniform self-heating.
- C tracking: tempco spread and dielectric behavior; C0G/NP0 is preferred for predictable tracking.
- Layout thermal gradient: identical values placed in different temperature zones break cancellation symmetry.
- Aging / absorption: long-term change can shift the effective ratios and raise the residual tone floor.
Compensation ladder (from simplest to strongest)
Level 1
Component tracking (most reliable)
- Use resistor networks/arrays for ratio tracking and thermal proximity.
- Use C0G/NP0 (or stable film) and matched pairs for ratio-critical capacitors.
- Place ratio-critical parts in the same thermal zone with symmetric routing.
Level 2
Analog trim shaping (NTC/PTC concept)
- Introduce a controlled temp-dependent element to counter the drift direction.
- Use when a fixed BOM cannot meet the f0 window across the operating range.
- Keep the scheme simple; avoid temperature-dependent behavior that cannot be production-verified.
Level 3
Digital calibration (trim + store + re-verify)
- Use digipot or a switched RC bank to adjust the effective ratio.
- Store coefficients in EEPROM and validate with a re-measure step.
- Best when notch behavior must hold across wide temperature and unit-to-unit spread.
See also
Calibration hooks (digipot / RC bank / EEPROM) belong to the dedicated Calibration Hooks page.
Temperature acceptance template (write as measurable rules)
Center stability
Δf0/ΔT ≤ X (ppm/°C or Hz/°C) over Tmin → Tmax.
Depth at worst temperature
Depth ≥ Y dB (or residual < Z mVrms) at the worst-case temperature corner.
Passband damage
Insertion loss change ≤ A dB, ripple ≤ B dB (or group-delay ripple ≤ C).
Re-test rule
Re-measure at the reference temperature after a full sweep to detect irreversible drift and hysteresis.
Field hooks (temperature-proofing)
Quick check: sweep temperature and log f0 shift and residual; if depth collapses while f0 stays near target, tracking mismatch is likely.
Fix: prioritize tracking (networks/arrays + C0G pairs + thermal symmetry); add trim only when a fixed BOM cannot meet the window.
Pass criteria: Δf0/ΔT and depth/residual limits are met at the worst-case temperature corners.
Fix: prioritize tracking (networks/arrays + C0G pairs + thermal symmetry); add trim only when a fixed BOM cannot meet the window.
Pass criteria: Δf0/ΔT and depth/residual limits are met at the worst-case temperature corners.
H2-8. Loading, Source Impedance, and Op-Amp Non-Idealities
Notch behavior can change dramatically when the source, load, or amplifier changes. Most “unexpected” shallow notches are caused by impedance disturbance, while most ringing/side-lobes are caused by active-stage limits (phase margin or headroom).
Source (Rs)
Source impedance disturbs cancellation
Symptom: f0 shifts and depth changes when the signal source or upstream stage changes.
Quick check: swap to a low-impedance source (or add a known series resistor) and compare depth/f0 sensitivity.
Fix: buffer the network input or define a controlled interface impedance so cancellation conditions remain stable.
Quick check: swap to a low-impedance source (or add a known series resistor) and compare depth/f0 sensitivity.
Fix: buffer the network input or define a controlled interface impedance so cancellation conditions remain stable.
Load (RL)
Load collapses depth (probe/ADC effect)
Symptom: depth looks good unloaded but becomes shallow when a probe, cable, or ADC input network is connected.
Quick check: emulate several RL values and record depth change; large sensitivity indicates load-driven cancellation loss.
Fix: buffer the output, or add a controlled isolation interface (Riso/RC) so the load becomes predictable.
Quick check: emulate several RL values and record depth change; large sensitivity indicates load-driven cancellation loss.
Fix: buffer the output, or add a controlled isolation interface (Riso/RC) so the load becomes predictable.
GBW / PM
Active variants can peak or ring
Symptom: side-lobes, peaking near the notch, or time-domain ringing; regenerated designs may approach oscillation.
Quick check: reduce feedback strength (or gain) and see if peaking diminishes; sensitivity suggests phase margin limits.
Fix: select an amplifier with adequate phase margin in this configuration, reduce regeneration, and verify stability across corners.
Quick check: reduce feedback strength (or gain) and see if peaking diminishes; sensitivity suggests phase margin limits.
Fix: select an amplifier with adequate phase margin in this configuration, reduce regeneration, and verify stability across corners.
Headroom
Overload creates harmonics the notch cannot remove
Symptom: 100/120 Hz and higher harmonics rise with amplitude, indicating clipping/rectification-like behavior.
Quick check: lower input amplitude or gain and observe whether harmonics drop nonlinearly.
Fix: increase headroom, reduce passband gain, and place the notch where the chain stays linear.
Quick check: lower input amplitude or gain and observe whether harmonics drop nonlinearly.
Fix: increase headroom, reduce passband gain, and place the notch where the chain stays linear.
Ib / leakage
High R makes bias and leakage visible
Symptom: low-frequency noise/offset changes with humidity/temperature; depth drifts over time despite “same values.”
Quick check: reduce resistor magnitude scale and compare; clean/rehydrate differences often expose leakage sensitivity.
Fix: lower resistor values, use networks/arrays, keep symmetry, and apply leakage control practices for ratio-critical nodes.
Quick check: reduce resistor magnitude scale and compare; clean/rehydrate differences often expose leakage sensitivity.
Fix: lower resistor values, use networks/arrays, keep symmetry, and apply leakage control practices for ratio-critical nodes.
Failure-mode map (fast priority)
Shallow notch → Rs/RL + matching
Peaking/ringing → phase margin
LF noise rises → high R / leakage
Harmonics appear → headroom
Field hooks (fast isolation test)
Quick check: measure depth with multiple RL conditions (probe/ADC/fixture) and record sensitivity; large sensitivity indicates a missing isolation plan.
Fix: add buffering at the most sensitive interface (often the load side) and define controlled impedances for repeatable cancellation.
Pass criteria: depth and f0 stay within limits across the defined source/load set and measurement configurations.
Fix: add buffering at the most sensitive interface (often the load side) and define controlled impedances for repeatable cancellation.
Pass criteria: depth and f0 stay within limits across the defined source/load set and measurement configurations.
H2-9. Trimming & Production Strategy (One-Knob vs Two-Knob)
Production-grade notch performance is defined by repeatability: stable f0 alignment and stable residual depth across units. The trimming strategy depends on whether the dominant error looks like center shift or cancellation mismatch.
Strategy decision: choose knob count by the observed error shape
If units mainly differ by f0 shift
Observation: the notch is “deep enough” once centered, but the center is off unit-to-unit.
Interpretation: center error dominates; mismatch is secondary.
Action: prefer One-knob trim to align f0; rely on matching/tracking to hold depth.
Interpretation: center error dominates; mismatch is secondary.
Action: prefer One-knob trim to align f0; rely on matching/tracking to hold depth.
If depth varies widely even after centering
Observation: f0 can be aligned, yet residual varies a lot across units (shallow notches persist).
Interpretation: cancellation mismatch (ratio tracking / tolerance / loading) dominates.
Action: improve matching first; if required, use Two-knob to stabilize depth under tight instrument-grade targets.
Interpretation: cancellation mismatch (ratio tracking / tolerance / loading) dominates.
Action: improve matching first; if required, use Two-knob to stabilize depth under tight instrument-grade targets.
One-knob
Align f0 with a single adjustable element
- Goal: center f0 to the line frequency window; depth is guarded by matching/tracking.
- Fast: one adjustment axis reduces iteration count and production time.
- Failure signature: if f0 passes but residual fails, mismatch (not center) is the primary problem.
Pass template
Center: |f0 − f_line| ≤ X (Hz or ppm) → then
Residual: residual@f_line ≤ Z (mVrms or dBFS).
Residual: residual@f_line ≤ Z (mVrms or dBFS).
Two-knob
Center + depth stabilization under tight limits
- Knob A: align f0 (coarse).
- Knob B: fine trim residual depth (limited range to avoid passband damage and overfitting).
- Best fit: high-end instruments where depth must hold across unit spread and corners.
- Control rule: cap iteration count; define a clear non-convergence fail/bin path.
Cycle control template
Max iterations: N trims total. If not converged → Fail/bin (root-cause: mismatch, loading, or leakage).
Minimal production loop (6 steps; repeatable and fast)
- Inject: inject a controlled 50/60 Hz tone at the defined node and amplitude.
- Measure: measure residual at the line bin (FFT bin or narrowband method).
- Adjust: trim knob A (and knob B if enabled) under the defined iteration cap.
- Store: store coefficients/trim codes to EEPROM (when applicable).
- Verify: re-measure residual and center error; optionally spot-check passband impact.
- Bin & log: pass/fail binning + log trim codes for traceability.
H2-10. Measurement & Debug (Bench Methods That Don’t Lie)
Deep notches are often limited by the measurement chain rather than the DUT. When the target depth is high, source distortion, spectral leakage, and grounding can dominate the “notch bottom.”
What to measure (define acceptance, then choose methods)
Depth / residual
Residual@50/60 (mVrms or dBFS) using a defined reading method (FFT bin or narrowband).
Center error
|f0 − f_line| ≤ X (Hz or ppm) under the defined load/source set.
Passband damage
Insertion loss / ripple / group-delay impact (spot-check template; limits defined by the system budget).
Temperature points
Worst-corner verification across Tmin/Tmax when temperature stability is required.
Choose the method: sweep vs single-tone injection
Single-tone injection (50/60)
Closest to real hum and best for production repeatability. The key risk is the test chain: grounding and source THD can set the floor.
Frequency sweep
Useful for verifying notch shape and passband impact, but the deepest point can be dominated by analyzer noise floor and spectral leakage.
Hard rule
If the measured notch bottom equals the instrument floor (or changes when swapping the source), the limitation is the test chain—not the DUT.
FFT depth reading (operational checklist)
1) Resolution (RBW)
Use enough record length so the 50/60 bin is narrow compared to the desired reading accuracy.
2) Leakage control
Prefer coherent sampling (integer cycles). If not possible, select a window that suppresses leakage for deep-notch readings.
3) Bin reading
Read the line bin (or a narrow band around it) and compare against a baseline measured with a known short/termination.
4) Averaging
Average to stabilize the noise floor estimate; do not hide a stable tone by over-smoothing the display.
Measurement traps (notch-relevant only)
Probe ground / loop pickup
Symptom: 50/60 appears even with a shorted input.
Quick check: shorten the ground path or switch to differential probing; compare line-bin changes.
Fix: minimize loop area; prefer differential measurement for deep-notch verification.
Quick check: shorten the ground path or switch to differential probing; compare line-bin changes.
Fix: minimize loop area; prefer differential measurement for deep-notch verification.
Instrument ground / ground loops
Symptom: readings change when adding/removing equipment (PSU, source, PC interface).
Quick check: isolate one connection at a time; a large step indicates loop-driven hum.
Fix: enforce a controlled grounding strategy; avoid multiple return paths in the test setup.
Quick check: isolate one connection at a time; a large step indicates loop-driven hum.
Fix: enforce a controlled grounding strategy; avoid multiple return paths in the test setup.
Fixture/cable injection
Symptom: notch bottom changes with cable routing or fixture swaps.
Quick check: reroute cables away from power bricks/transformers; compare bin stability.
Fix: standardize fixture routing and shielding; keep test geometry repeatable.
Quick check: reroute cables away from power bricks/transformers; compare bin stability.
Fix: standardize fixture routing and shielding; keep test geometry repeatable.
Acceptance template (method must be specified)
Method: “FFT bin residual” (or a defined narrowband method) with fixed record length and window.
Depth: residual@f_line ≤ Z (mVrms or dBFS).
Center: |f0 − f_line| ≤ X (Hz or ppm).
Passband: insertion loss/ripple limits ≤ A/B (placeholders set by the system budget).
Temperature: verify at Tmin/Tmax when required; use the same method and fixture.
Depth: residual@f_line ≤ Z (mVrms or dBFS).
Center: |f0 − f_line| ≤ X (Hz or ppm).
Passband: insertion loss/ripple limits ≤ A/B (placeholders set by the system budget).
Temperature: verify at Tmin/Tmax when required; use the same method and fixture.
H2-11. Engineering Checklist (Layout, Parts, Guarding, Reliability)
Precision mains notches fail in predictable ways: mismatch, leakage, and reference contamination flatten the notch bottom, while poor maintainability slows debug and production closure. The checklist below is notch-specific and prioritizes repeatable depth and f0 stability.
Priority checklist (P0 must-do)
P0
Mandatory for depth repeatability
- Use stable dielectrics: C0G/NP0 or stable film capacitors for ratio-critical nodes (avoid high-leakage dielectrics).
- Prefer resistor networks/arrays: matching and tracking outperform absolute precision for cancellation depth.
- Compact, symmetric “notch island” placement: keep ratio network tight; minimize loop area and parasitics.
- Guard high-impedance nodes: guard ring to the appropriate reference potential to suppress surface leakage.
- Cleanliness requirement: flux residues and contamination can lift the notch floor; enforce cleaning + verification for critical nodes.
- Keep the notch reference clean: do not route large return currents through the notch reference region.
P1
Strongly recommended for faster debug & yield
- Thermal symmetry: keep ratio elements in the same thermal zone to preserve tracking across ambient gradients.
- Define keep-out around the notch island: separate from digital switching and power conversion hot zones.
- Provide a bypass jumper: one-step isolation between “notch issue” and “system coupling issue.”
- Add test points: injection node, notch output, and key references for repeatable H2-10 verification.
- Fix the load definition: verify notch depth with the intended load/next-stage input impedance.
P2
Optional for instrument-grade corners
- Conformal coating strategy: consider only after validating it does not introduce leakage or dielectric absorption side effects.
- Dedicated shield can/guard copper: for extreme environments where surface coupling dominates.
- Redundant trim pads: multiple trim footprints to cover supply chain substitutions without respins.
Review-template fields (copy/paste)
Check item → Why it matters → How to verify → Pass criteria (limits set by the system budget).
H2-12. Applications (Where Precision Notch Wins)
Precision 50/60 Hz notches add the most value when hum is a narrowband dominant interferer and the signal chain needs a repeatable residual limit. Each use case below includes a chain, one common pitfall, and a pass-criteria template.
Bridge / strain
Low-level sensors with strong mains pickup
Chain: Sensor → INA/PGA → (LP) → 50/60 notch → ADC
Pitfall: overload or rectification upstream generates harmonics; a notch at 50/60 does not remove distortion products.
Pass criteria: residual@50/60 < Z and no new dominant harmonics at 100/120 beyond H (limits set by budget).
Pitfall: overload or rectification upstream generates harmonics; a notch at 50/60 does not remove distortion products.
Pass criteria: residual@50/60 < Z and no new dominant harmonics at 100/120 beyond H (limits set by budget).
Bio-potential
Narrowband hum suppression without reshaping the signal
Chain: Electrodes → INA/BP → 50/60 notch → ADC
Pitfall: placing the notch where the signal is already near rails can fold interference into harmonics; ensure headroom before filtering.
Pass criteria: residual@50/60 < Z while baseline stability and waveform integrity remain within the defined clinical/instrument limits.
Pitfall: placing the notch where the signal is already near rails can fold interference into harmonics; ensure headroom before filtering.
Pass criteria: residual@50/60 < Z while baseline stability and waveform integrity remain within the defined clinical/instrument limits.
Audio / measurement
Deep notch with controlled passband damage
Chain: Source → buffer → 50/60 notch → gain stage → ADC/recorder
Pitfall: notch depth is easy to chase while phase/group-delay ripple is ignored; audible/measurement artifacts can remain.
Pass criteria: residual@50/60 < Z and passband insertion/phase ripple within A/B (templates set by system goals).
Pitfall: notch depth is easy to chase while phase/group-delay ripple is ignored; audible/measurement artifacts can remain.
Pass criteria: residual@50/60 < Z and passband insertion/phase ripple within A/B (templates set by system goals).
Industrial DAQ
Long cables + protection + notch coordination
Chain: Long cable → protection (RC/TVS) → buffer/PGA → 50/60 notch → ADC
Pitfall: changing the protection network or cable impedance can change notch behavior if the notch is not buffered/isolated.
Pass criteria: residual@50/60 < Z across the defined cable/fixture set, with no instability or unexpected insertion loss.
Pitfall: changing the protection network or cable impedance can change notch behavior if the notch is not buffered/isolated.
Pass criteria: residual@50/60 < Z across the defined cable/fixture set, with no instability or unexpected insertion loss.
H2-13. IC Selection Logic (Op-Amp, Networks, Trims, Switches)
Precision 50/60 Hz notch performance is limited by three practical ceilings: (1) ratio-mismatch (depth), (2) drift/leakage (depth and f0 stability), and (3) non-ideal active elements (distortion, stability, and repeatability). This section turns those ceilings into a selection flow and a vendor-ready inquiry checklist.
Selection Flow (Targets → Topology → Parts → Verification → Production Fit)
- Start from acceptance targets: residual hum at the measurement/ADC node, allowed passband damage, and temperature range.
- Choose buffering/regeneration only as needed to meet load sensitivity and depth targets (avoid complexity if mismatch dominates).
- Select parts by the dominant error term: mismatch → matched networks; drift/leakage → dielectric/cleanliness; non-idealities → op-amp/switch behavior.
A) Op-Amp / FDA Selection (Noise Floor, Distortion, Stability, Leakage Interaction)
Use-case rule: if the measured notch bottom stops improving when matching is tightened, the limit is often the low-frequency noise floor (1/f region) or leakage-induced imbalance, not “center frequency error”.
Reference op-amp / FDA part numbers (starting points only)
What to ask / what to verify (notch-specific)
- Low-frequency noise evidence: 0.1–10 Hz or near-DC noise data; confirm the notch bottom is not noise-limited.
- Large-signal behavior at 50/60 Hz: verify that distortion does not create hum harmonics that bypass a single-notch solution.
- Stability margin in regenerated variants: verify phase margin with the intended Twin-T + feedback network and worst-case load.
- Bias/leakage interaction: confirm input bias current and its temperature behavior do not dominate mismatch at chosen resistor values.
B) Matched Networks & Capacitors (Depth = Ratio Tracking + Low Leakage)
Depth is set by cancellation; cancellation is set by ratio matching (ε). Absolute tolerance helps f0 placement, but matching/tracking sets the achievable notch floor across time and temperature.
Reference matched resistor solutions (starting points only)
- Matched quad network (ratio-critical blocks): LT5400
- Precision thin-film resistor array (moisture/stability focused): Vishay ACAS 0606/0612 AT (example ordering code shown in datasheet: ACASA1100S2200P5AT)
Reference capacitor part numbers (starting points only)
Choose the dielectric by leakage and stability first; then scale R/C to hit f0.
- C0G/NP0 MLCC (stable, low loss): Murata GRM31C5C1H104JA01K (0.1 µF, 50 V, C0G/NP0, 1206), TDK C3225C0G1H104J250AA (0.1 µF, C0G, 1210), KEMET C1210C104J5GACTU (0.1 µF, 50 V, C0G/NP0, 1210)
- Film/foil (ultra-low leakage/DA option via higher R, smaller C): WIMA FKP2C021001G00ESSD (0.01 µF, polypropylene film/foil, 63 V)
Quick decision rules
- Depth target high: prioritize tracking/matching (networks, paired capacitors) over “tight absolute tolerance”.
- Humidity/contamination risk: prefer lower leakage dielectric and lower impedance nodes; design for guard/cleanliness.
- Temp range wide: tracking specs and long-term drift matter more than room-temp datasheet numbers.
C) Trims & Switches (Noise, Linearity, Charge Injection, Repeatability)
Trimming is only “precision” if the trim element is stable, low-leakage, and repeatable. Otherwise, the notch appears deep on day-1 and drifts shallow in field conditions.
Reference trim elements (starting points only)
Reference low-leakage switches / multiplexers (starting points only)
“One-knob” vs “Two-knob” trim mapping
- One-knob (fast production): tune f0 with a single stable element; rely on matched networks to hold depth.
- Two-knob (instrument-grade): separately tune f0 and depth; requires repeatable trim elements and a clearly bounded iteration count.
Vendor Inquiry Fields (Copy/Paste Checklist)
Ask vendors to provide data that directly maps to notch depth, drift, and repeatability:
- Matched resistors: tracking tempco, ratio drift over temperature, long-term drift, moisture sensitivity.
- Capacitors: dielectric type, leakage/IR, aging behavior, tolerance distribution (pairing feasibility).
- Op-amps/FDAs: 0.1–10 Hz noise (or low-frequency noise), drift specs, large-signal distortion at low frequency, stability guidance for feedback networks.
- Switches/digipots: leakage current vs temperature, charge injection, resistance linearity, code repeatability and memory retention (if NVM).
- Qualification: operating temp range, humidity exposure limits, cleaning/handling recommendations for high-impedance nodes.
Pass Criteria Templates (Replace X/Y/Z With System Budget)
- Residual hum: at the measurement/ADC node, 50/60 Hz bin amplitude < X dBFS (or < X mVrms).
- Notch depth at room: depth ≥ Y dB with defined source/load impedance.
- Center frequency stability: |Δf0| / f0 ≤ Z% across temperature range.
- Depth over temperature: worst-case depth degradation < A dB from room-temp value.
- Passband damage: insertion loss < B dB and phase/group-delay ripple within the system allowance.
- Production fit: calibration converges within ≤ N iterations and total trim time ≤ T seconds per unit.
Note: The part numbers above are reference starting points for datasheet lookup. Final selection must be driven by the flow in Figure 13 and verified under the defined source/load, temperature, and production constraints.
Request a Quote
H2-13. FAQs (Precision 50/60 Hz Notch)
Short, actionable answers only. Each item uses a fixed 4-line format: Likely cause / Quick check / Fix / Pass criteria. Replace placeholders (X/Y/Z/…) with limits from the system budget.
