Collins 136B-2 Noise Blanker
Engineering Changes & Circuit Modification Reference
Covering factory-issued changes, community-developed improvements, alignment notes, antenna adaptation, and advanced modifications for contemporary fixed-station use.
1. Background & Design Overview
The Collins 136B-2 is a stand-alone noise blanker accessory designed to operate with the KWM-2 and KWM-2A transceivers, and compatible with the 75S-series receivers and 32S-series transmitters of the S-Line. Introduced circa 1959–1962 (manual fourth edition dated 1 August 1962, fifth edition dated November 1966), it represents Collins Radio’s production implementation of a noise blanking concept developed by the company for impulse noise suppression in SSB communications.
Unlike audio-stage clippers or series limiters — which operate on already-filtered signals and produce characteristically ‘raspy’ distortion on SSB — the 136B-2 operates at the wideband first (variable) IF of the double-superheterodyne receiver chain, well ahead of the narrow mechanical filter selectivity. This is the critical architectural advantage: impulse noise, having not yet been stretched by selective filtering, can be cleanly gated out in a window of only 5–30 microseconds without severe signal degradation.
The unit weighs approximately 1¼ lbs and requires a dedicated 40 MHz sensing antenna, making it primarily intended for mobile automotive use with a standard 55-inch AM broadcast whip. It draws power from the KWM-2/2A accessory connector.
1.1 Circuit Block Diagram Summary
The 136B-2 divides into four functional blocks, each with distinct performance requirements that influence every known modification or improvement:
| Block | Function | Key Performance Parameter |
|---|---|---|
| 40 MHz TRF RF Amplifier (~70 dB gain, 1 MHz BW) |
Amplifies the noise channel signal at 40 MHz with approximately 1 MHz bandwidth. Limiting prevents saturation. Tubes: 6BJ6 pentode stages. | Must respond in <0.25 µs rise time; bandwidth must exceed receiver IF BW by at least 10:1 |
| Envelope Detector | Half-wave diode rectifier with RC load. Converts noise pulses to sawtooth waveform. | RC time constant directly governs blanking pulse proportionality and maximum effective PRF |
| Pulse Amplifier/Limiter | Amplifies and clips the sawtooth to produce rectangular blanking pulses proportional to noise amplitude. Maximum blanking pulse width ~30 µs. | Pulse rise time <1 µs to full amplitude; must drive gate with sufficient voltage swing |
| Balanced Diode Gate (in KWM-2 variable IF) |
Push-pull IF gate using matched diodes. Balanced with R and C trimmers to prevent blanking-pulse harmonics appearing at output. | Gate balance is the single most critical alignment point; imbalance causes residual pulse breakthrough |
2. Factory-Issued Engineering Changes
A review of the complete Collins Service Bulletin index for amateur equipment (KWM-1/2/2A, S-Line series) reveals that no dedicated service bulletins were issued specifically for the 136B-2 noise blanker. The CCA service bulletin index (archived at wa3key.com and the CCA technical archives) lists SBs and SILs for the KWM-2/2A (SB-1 through SB-10, SIL 1-75, SIL 2-75, SIL 2-3-60), but none specifically address the 136B-2 accessory.
However, three factory-level engineering changes are documented within the manual revision history itself (4th edition August 1962 → 5th edition November 1966), and two KWM-2 service bulletins have indirect bearing on noise blanker performance:
| Change Reference | Description | Impact on 136B-2 Performance |
|---|---|---|
| 136B-2 Manual Edition 4→5 (1962→1966) |
Unspecified internal revisions between the 4th and 5th editions. Early production units (pre-serial ~T4000) may differ from later units in component values and board layout. | Verify your unit against the 5th edition schematic. Pre-revision units may have different detector time-constant values affecting pulse width at low noise levels. |
| KWM-2/2A SB-8A (AGC Overshoot) |
Eliminates AGC overshoot on noise pulses and provides dual time-constant AGC action. | Directly complements the 136B-2: if AGC is pumping on noise transients, the blanker gate and AGC are fighting each other. SB-8A must be applied before 136B-2 can perform optimally. SB-8B and SB-8C extend AGC hang time. |
| KWM-2/2A SB-8B (AGC Source Change) |
Changes source of delay bias to AGC rectifier. | Stabilises the AGC reference point, reducing interaction between AGC action and blanker gate switching transients at the variable IF. |
| KWM-2/2A SB-8C (Hang AGC) |
Adds delayed-decay (hang) AGC to RF amplifier stage. | Hang AGC prevents rapid AGC recovery between blanked noise pulses from causing audible audio level fluctuation on the desired signal. |
3. Community-Developed Circuit Modifications
The following modifications are drawn from documented community discussions (CCA reflector, jptronics Collins archives, diyAudio vintage radio forum, eham, and the KA7OEI noise blanker design notes), the original Collins internal technical document Discussion of Collins Developed Noise Blankers for the KWM-1, 75A-4, and 75S-1 Amateur Receivers (Tom Anderson WW5L archive), and analysis of the 136B-2 schematic.
3.1 Diode Gate Balance Optimisation (Priority 1)
The balanced diode gate — the element inserted into the KWM-2 variable IF path — is the most performance-critical element in the system, and the most commonly misaligned. The factory design uses matched diodes in a push-pull configuration with both a resistive balance trimmer and a capacitive balance trimmer. When the gate is not precisely balanced, harmonic components of the blanking pulse appear at the IF output, creating a characteristic ticking or buzzing on the received audio even in the absence of actual impulse noise — the very symptom the blanker is supposed to eliminate.
Modification procedure:
- With the 136B-2 installed and the KWM-2 operational, inject a small continuous carrier at the KWM-2 antenna connector — any signal near the operating frequency will do. Connect an oscilloscope to the variable IF test point (or monitor on audio).
- Enable the noise blanker. Apply a test pulse to the 136B-2 trigger input (a square wave generator at 1 kHz, 1–5 V amplitude, injected via a 50 Ω resistor at the noise channel input, approximates ignition noise). Alternatively, a handheld spark source briefly fired near the 40 MHz antenna is sufficient.
- Adjust the resistive balance trimmer first for minimum pulse breakthrough at audio output, then trim the capacitive balance trimmer for further null. The two adjustments interact — iterate both for best null.
- The target is less than 5 mV of blanking pulse transient appearing at the IF output for a full-amplitude gate drive. Some community practitioners report achieving 20–30 dB improvement in pulse suppression through careful iteration.
3.2 Gate Diode Replacement and Matching
The 136B-2 manual specifies matched germanium diodes in the IF gate. After 60+ years, original diodes often exhibit drift in forward voltage, leakage, and junction capacitance, degrading gate balance. The community-established approach, consistent with design principles used in more modern noise blanker designs, is diode replacement with matched modern parts:
| Option | Diode Type | Advantages | Considerations |
|---|---|---|---|
| Preferred (fast recovery) |
1N4148 / 1N914 (silicon, matched pair) |
Readily available, long-term stable, very fast reverse recovery (~4 ns). Consistent VF across temperature. | Higher VF than original germanium (~0.6 V vs ~0.25 V). Gate bias voltage may need minor adjustment to re-establish proper ON/OFF switching margins. |
| High performance (PIN diode gate) |
MPN3404 or equiv. PN4117 / BAR64-03W |
PIN diode action provides low distortion during partial conduction — critical for preserving IMD performance when blanker is active with strong adjacent signals. | MPN3404 is difficult to source (HP/Avago legacy part). BAR64-03W (Infineon) is a viable current-production substitute. Carrier lifetime issues limit PIN diodes at VLF/LF. |
| Avoid | 1N34A, 1N60 (aged germanium) |
Nominally similar to original type. | Very high leakage after aging, poor temperature stability, lot-to-lot variation makes matching difficult. Retain only if confirmed NOS and tested. |
Matching procedure: Select four diodes from a batch. Measure VF at a known test current (1 mA is convenient). Sort by VF. Use the two most closely matched for the gate pair. A difference of less than 2 mV in VF is achievable with 1N4148 parts and results in significantly easier balance trimmer adjustment.
3.3 Detector RC Time Constant Optimisation
The envelope detector that follows the 40 MHz RF amplifier produces a sawtooth output whose decay time constant (RC) determines how long a blanking pulse is generated per noise impulse. The original design targets a maximum blanking pulse of approximately 30 µs, which was appropriate for the noise environment Collins modelled — automotive ignition at highway speeds. Contemporary RF environments, especially in urban/suburban locations, include switch-mode power supply noise and DSL/cable gateway impulses at much higher repetition rates.
Problem symptom: At noise pulse repetition frequencies (PRF) above approximately 2 kHz, the detector is re-triggered before the previous blanking pulse decays. The result is that the gate is held open (blanked) continuously, destroying the received audio entirely rather than selectively suppressing noise peaks. Community analysis (diyAudio, 2024) confirms no existing retriggering-type blanker handles PRF above ~2.2 kHz without producing a continuous ~6 ms blanking event.
Modification — reduce RC for contemporary urban environments:
Reducing the detector load resistor value shortens the sawtooth decay and therefore the blanking pulse width, allowing the blanker to handle higher PRF. The tradeoff is reduced blanking pulse width for large amplitude noise events.
| Environment | Recommended RC Target | Typical Blanking Pulse Width | Best For |
|---|---|---|---|
| Original spec (automotive mobile) |
~10–15 µs | Up to 30 µs | Ignition noise, 50–500 Hz PRF |
| Suburban fixed station (SMPS noise) |
5–8 µs | 10–15 µs | Switch-mode PSU hash, 1–2 kHz PRF |
| Urban fixed station (heavy RFI) |
2–4 µs | 5–8 µs | DSL/cable gateway noise, LED dimmer hash |
3.4 RF Amplifier Tube Substitution (6BJ6 → 6BH6)
The 40 MHz TRF RF amplifier stages use 6BJ6 pentode tubes. While not scarce, the 6BJ6 was designed primarily for TV IF service and some sample variation in transconductance affects the blanker’s sensitivity threshold. Community experience indicates:
- The 6BH6 (sharp-cutoff pentode) is an electrically compatible substitute with typically higher and more consistent Gm, providing slightly better RF amplifier gain without circuit modification.
- EF95/6AK5 pentodes are noted in the general vintage radio community as high-performance equivalents for VHF service, though pin compatibility requires verification against the specific 136B-2 tube socket configuration.
- Select tubes for similar Gm within a ±10% window to preserve the RF amplifier’s gain flatness across the 39–41 MHz passband.
3.5 40 MHz Sense Antenna — Fixed Station Adaptation
The 136B-2 was designed for mobile use with a 55-inch broadcast-band whip, which is electrically large at 40 MHz and provides reasonable gain. For fixed-station use, the sense antenna is the most commonly cited performance bottleneck. Collins’s own documentation explicitly warns that resonant HF antennas (e.g., a 14 MHz beam, a 3.9 MHz vertical) perform very poorly as 136B-2 sense antennas because their 40 MHz efficiency is extremely low.
| Antenna Option | 40 MHz Efficiency | Practical Notes | Recommended? |
|---|---|---|---|
| 55″ AM broadcast whip (per factory spec) |
Good — near-resonant at ~40 MHz (~1.4m) | Ideal for mobile. For fixed station, mount on mast or nearby — keep coax run short. | YES (mobile) YES (fixed if practical) |
| Telescopic whip (adj. to ~37″) |
Good if tuned to 40 MHz | Set to approximately 37 inches for resonance. Trim for lowest VSWR using grid-dip or antenna analyser. A simple L-network can match 50 Ω coax. | YES |
| Indoor active whip (broadband, 1–50 MHz) |
Good — flat response through 40 MHz | E.g., Wellbrook ALA100, MFJ-1020C, or homebrew FET active antenna. Particularly useful in apartments where an outdoor antenna is impractical. Provides excellent 40 MHz coverage without resonance constraints. | YES (recommended for fixed station) |
| HF directive beam or loaded vertical | Poor — typically <1% efficiency at 40 MHz | Categorically not recommended per Collins documentation. The 136B-2 will have negligible noise channel sensitivity and will not fire reliably. | NO |
| End-fed HF wire (80–10 m) |
Variable — may have harmonic resonances near 40 MHz | Unpredictable. Depends on wire length. A 20-foot random wire with no tuner presents a random impedance at 40 MHz. Test empirically; performance is inconsistent. | MARGINAL |
Coax lead length warning: The Collins documentation specifically notes that using the KWM-2’s own HF communications antenna as the 136B-2 sense antenna is likely to fail because HF resonant structures reject 40 MHz. Additionally, a long coaxial run from the sense antenna introduces lead-in delay that could cause the blanking gate to fire after the noise pulse has already entered the IF — defeating the entire purpose. The blanker path must be faster than the receiver path. Keep sense antenna coax under 15 feet.
3.6 Pulse Amplifier Limiter — Sensitivity Threshold Adjustment
The pulse amplifier receives the sawtooth from the detector and clips it to produce rectangular blanking pulses. The threshold at which it fires determines the blanking sensitivity — how small a noise pulse will trigger the gate. If the threshold is too low, strong AM broadcast signals at or near 40 MHz will cause false triggering (continuous blanking). If too high, weak impulse noise is not suppressed.
Original design rationale: Collins chose 40 MHz for the noise channel precisely because it is a relatively protected frequency in the US, used mainly by low-power FM and Part 15 devices. At 40 MHz, AM-modulated signals (which would look identical to noise to the detector) are rare in most US urban environments. However, some industrial scientific and medical (ISM) band emissions, Part 90 land-mobile users, and amateur services near 40 MHz can cause intermittent false triggering in densely populated areas.
Modification — bias adjustment for false-trigger suppression:
Increasing the cathode bias on the first pulse amplifier tube raises the clipping threshold, reducing sensitivity to weak 40 MHz signals while retaining response to strong noise impulses. Increase the cathode resistor by 10–15% increments while monitoring blanker response to a test spark. Stop when the blanker is just reliably triggered by the spark but does not fire on ambient 40 MHz background.
3.7 40 MHz Input Bandpass Filter Enhancement
The stock 136B-2 relies on the TRF RF amplifier’s own tank circuit selectivity to reject out-of-band signals. Community members operating near broadcast or PMR activity have added an external cavity or LC bandpass filter ahead of the noise channel input to improve rejection of signals that could cause false triggering.
Practical implementation:
- A simple 2-pole Butterworth LC bandpass filter centered at 40 MHz with a Q of 20–30 (bandwidth ~1.3–2 MHz) placed in the coax lead from the sense antenna to the 136B-2 RF input provides 30–40 dB of rejection at ±5 MHz. This rejects the 37 MHz and 43 MHz adjacent regions while passing the full 39–41 MHz noise channel.
- Component values: For a 50 Ω shunt-C topology, L ≈ 250 nH (wound on T37-6 toroid, ~13 turns #26 AWG), C1 and C2 ≈ 47 pF silver mica for the series arms. Trim with a VNA or grid-dip oscillator.
- Alternatively, a commercial helical bandpass filter for 40–41 MHz amateur activity can be adapted. Keep insertion loss under 2 dB to preserve noise channel sensitivity.
3.8 Gate Intermodulation — Adjacent-Channel Strong Signal Issue
The Collins technical document identifies what it calls an inherent limitation of any gating-type noise blanker system: when the IF gate switches, it generates sidebands spaced at the gate (blanking) frequency around any signal present in the wideband IF passband. If a strong carrier exists within the variable IF passband — even a few kHz outside the final mechanical filter passband — those blanking-rate sidebands can fall on top of a weak desired signal at the IF output, degrading S/N rather than improving it.
This effect is not a malfunction; it is a fundamental property of time-gating. Community practice has established these mitigating steps:
- Reduce RF gain on the KWM-2 when strong adjacent signals are present. This reduces the IF signal level at the gate, reducing sideband generation amplitude.
- Ensure gate balance is perfect (§3.1). A balanced gate generates primarily odd-order sidebands at odd multiples of the blanking pulse repetition rate, which are typically further from the desired signal. An unbalanced gate generates even-order products that cluster more densely.
- Minimise blanking pulse duty cycle: the 136B-2 does not blank continuously but only on detected pulses. If the sense antenna is poorly matched or the threshold is too low, excessive false triggering inflates duty cycle and sideband energy. The fixes in §3.5 and §3.6 directly reduce this.
4. Alignment Procedure
The 136B-2 alignment involves three inter-related procedures that should be performed in sequence. Before beginning, ensure all electrolytic capacitors have been reformed or replaced (the unit is ≥58 years old), all tube sockets are clean, and the KWM-2/2A AGC service bulletins (SB-8A/B/C) are already applied.
| Step | Procedure | Instrument | Target |
|---|---|---|---|
| 1 | 40 MHz TRF RF Amplifier Alignment — Peak each tank circuit trimmer capacitor sequentially for maximum sensitivity at 40 MHz. Inject a –80 dBm 40 MHz signal at the noise channel antenna input. | Signal generator 40 MHz RF voltmeter at pulse amplifier input |
Maximum output at pulse amplifier; flat response ±0.5 MHz |
| 2 | Detector/Pulse Amplifier Threshold — Verify blanking pulse width vs. input amplitude. Inject a 40 MHz AM-modulated signal at 50% depth and increase level from –90 dBm upward. Note the threshold at which blanking pulses appear. Adjust cathode bias per §3.6 if needed. | Signal generator (AM mod) Oscilloscope at gate control line |
First blanking pulses appear at –75 to –65 dBm noise channel input |
| 3 | IF Gate Balance — With test carrier injected at KWM-2 antenna, enable blanking, apply test spark or pulsed 40 MHz stimulus. Alternate between R and C balance trimmers for minimum pulse breakthrough at KWM-2 audio output. | Audio monitor or oscilloscope at KWM-2 audio output | Pulse breakthrough <5 mV at audio; <2% of blanking pulse amplitude in IF |
5. Known Fundamental Limitations
The 136B-2 architecture, while clever for its era, has limitations that cannot be fully overcome by modification alone:
| Limitation | Root Cause | Workaround or Alternative |
|---|---|---|
| High PRF lock-up (>2 kHz repetition rate) |
Retriggering blanker architecture: each incoming noise pulse resets the sawtooth RC timing. At PRF > 1/RC, blanking becomes continuous. | Reduce RC (§3.3) to handle higher PRF. Above ~5 kHz PRF, the 136B-2 concept fails; consider a purpose-built synchronous or fast-reset blanker design. |
| Strong adjacent signal intermodulation | Gating creates sidebands at blanking rate × any signal in wideband IF. | Perfect gate balance (§3.1), reduced RF gain, minimal false triggering. |
| No fixed-station noise antenna included | Designed for mobile whip; no provision for HF antenna use. | Active whip or resonant 40 MHz dipole as described in §3.5. |
| Single noise channel at 40 MHz only | A noise source strong at HF/LF but weak at 40 MHz will not trigger the blanker regardless of how well the unit is aligned. | If noise source lacks 40 MHz component (e.g., certain switching regulators), the 136B-2 cannot help. A wideband IF-based or DSP noise blanker is required. |
| Tube-dependent reliability | Six 6BJ6 and associated tubes. Heater current, aging, and microphony affect threshold consistency. | Regular tube testing and matching (§3.4). Consider one-time solid-state replacement of the RF amplifier stages if available. |
6. Source References
[1] Collins Radio Company. Instruction Book — 136B-2 Noise Blanker, 5th Edition, November 1966. Manual P/N 523-0007-00-004311. Available: collinsradio.org/archives/manuals/136B-2_5th-ed-11-66_.pdf
[2] Collins Radio Company (internal document, author/date unknown, transcribed by K4OZY/jptronics). Discussion of Collins Developed Noise Blankers for the KWM-1, 75A-4, and 75S-1 Amateur Receivers. Provided to archive by Tom Anderson WW5L. Available: jptronics.org/Collins/bulletins/collins.noise_blanker.html
[3] WA3KEY. Collins Service Bulletin Index — KWM-1/2/2A and S-Line Series. wa3key.com/sbindex.html (accessed March 2026). Confirms no dedicated 136B-2 SBs in the amateur equipment series.
[4] Collins Collectors Association. CCA Service Bulletins Index. collinsradio.org. KWM-2/2A SB-8A, SB-8B, SB-8C (AGC modifications).
[5] KA7OEI (Clint Turner). A Line-Synchronous Noise Blanker. ka7oei.com/syn_blank.html. Detailed design notes on PIN diode gate selection (MPN3404, 1N914 comparison) and blanking gate architecture.
[6] diyAudio Community Forum. No RF Gear Here?, Page 42, March 2024. Discussion of 136B-2 architecture, PRF limitations, and gated-LO alternative concept. diyaudio.com/community/threads/no-rf-gear-here.309531/page-42
[7] Electric Radio Magazine. Collins S-Line, KWM-1, KWM-2, KWM-2A Service Modification Compendium, 260 pp. ermag.com. Comprehensive factory bulletin compilation; consult for any 136B-2 items not available online.
[8] DJ7HS (Ernst F. Schroeder). My Collins KWM-2. CCAE technical document. ccae.tm6cca.com. Practical KWM-2 restoration notes with NB switch repurposing when 136B-2 is not fitted.
This document compiles published factory data, community modification discussions, and engineering analysis of the Collins 136B-2 noise blanker system. It is intended as a reference for restoration, alignment, and performance improvement of vintage Collins equipment. All circuit modifications should be verified against the current 5th-edition manual schematic before implementation. The author accepts no responsibility for equipment damage arising from modifications performed without proper test equipment.
Mike Peace VK6ADA · r-390a.net Administrator