Collins Collectors Association — Technical Bulletin Commentary
CCA Technical Bulletin 1
Understanding and Mitigating S-Line Transmitter Spikes
BULLETIN REF: CCA-TB-001 · APPLIES TO: 32S-1 · 32S-3 · 32S-3A · KWM-2/2A · COMMENTARY: VK6ADA, April 2026
⚠ Priority Action Required — All 32S-3 / 32S-3A Owners
If you do not know the complete modification history of your 32S-3, treat it as suspect. An incorrectly modified C20 capacitor may be causing your transmitter to radiate out-of-band spurious signals exceeding FCC Part 97 limits — with no indication on any front-panel meter.
1. Introduction
The Collins S-Line — comprising the 32S-1, 32S-3, and 32S-3A transmitters paired with the 75S-series receivers — represents one of amateur radio’s most enduring technical achievements. Designed in the early 1960s and engineered to broadcast-grade standards,[1] these instruments remain in daily HF operation more than six decades after production ceased.
CCA Technical Bulletin 1 addresses three interrelated problem classes that can cause a nominally well-maintained S-Line transmitter to produce transient or sustained spurious RF energy:
• The C20 capacitor modification — a widely propagated “improvement” that causes the 32S-3 to fail its own −50 dBc spurious specification and produce out-of-band emissions.
• ALC system overshoot — a transient spike at the leading edge of every transmission caused by inherent loop delay in the Automatic Level Control circuit.
• VOX relay and keying transients — impulsive RF events generated by relay contact bounce, B+ line perturbation, and insufficient time-constant shaping.
This enriched commentary draws on the original CCA bulletin,[2] original Collins service documentation,[3] and contemporary RF engineering analysis to provide a complete technical treatment with mitigation procedures for the working restorer.
2. The C20 Capacitor Modification — A Spurious Emission Crisis
2.1 Background and Origin
At some point during the 32S-3’s production run, a modification began circulating recommending that capacitor C20 — which couples the BFO signal into the first mixer cathode (V4) during TUNE, LOCK KEY, and CW modes — be changed from its factory value of 0.01 µF to 0.001 µF. The stated rationale was increased grid drive and improved CW performance.[4]
The modification gained wide distribution through published media and was applied to an estimated hundreds of units. The CCA considers any 32S-3 with an unknown modification history to be a suspect unit.
Technical Context: The Role of C20
In SSB mode, the 455 kHz BFO output passes through the mechanical filter before reaching the first mixer — the filter attenuates all harmonics outside its passband. In CW, TUNE, and LOCK KEY modes, the BFO signal bypasses the filter and is applied directly to the first mixer cathode via C20. Without the mechanical filter acting as spectral gatekeeper, BFO harmonic content can mix with the VFO to produce spurious outputs at the transmitted RF frequency.
2.2 The Mechanism of Spur Generation
Don Jackson W5QN’s spectrum analyser investigation[5] identified the physical mechanism precisely. Changing C20 from 0.01 µF to 0.001 µF shifts the resonant peak of the LC network at the BFO plate from approximately 410 kHz to 445 kHz, raising BFO drive to the first mixer by approximately 5 dB at 455 kHz. This elevates harmonic content — particularly the 5th harmonic of the BFO (≈2.278 MHz) and the 7th harmonic (≈3.194 MHz) — which then enter the first mixer and produce intermodulation products within the 2.955–3.155 MHz IF passband.
Two dominant spurious outputs result on 20 metres when the transmitter is tuned to 14.000 MHz:
| Spur | Generation Mechanism | RF Output | Band Status |
|---|---|---|---|
| Spur #2 | 2×VFO − 5×BFO | 14.039 MHz | In-band |
| Spur #3 | 2×IF − Spur #2 (7th BFO harmonic) | 13.961 MHz ← OUT OF BAND | ⚠ Out-of-band |
Critically, Spur #3 is frequency-stationary — it does not move as the VFO is tuned. A 32S-3 with the modified C20 will continuously radiate at approximately 13.961 MHz whenever operated in CW mode, regardless of where the VFO is set. Jackson confirmed this on all four test units he examined.
Regulatory Compliance — FCC Part 97
Under 47 CFR §97.307(d),[6] all spurious emissions between 30–225 MHz must be attenuated to at least −43 dBc. The 32S-3’s own factory specification is the more stringent −50 dBc.
With a 0.001 µF C20 installed, Spur #3 typically measures −27 to −32 dBc — failing both the FCC limit and the Collins factory specification by a wide margin. This constitutes an illegal transmission under Part 97.
2.3 Inspection and Remediation
Identifying the modification: Remove the 32S-3 from its cabinet. Locate C20 under the chassis — it couples the BFO plate circuit to the V4 first mixer cathode. The correct factory value is 0.01 µF (EIA code “103”). A modified unit will carry a 0.001 µF capacitor marked “102”. Alternatively, connect a spectrum analyser, key the transmitter in CW or TUNE mode at the low end of any band, and inspect the output. Two prominent spurs flanking the carrier — with one remaining stationary as you tune the VFO — confirm the modification is present.[7]
The repair: Replace C20 with a 0.01 µF ceramic or film capacitor rated for the operating voltage. Where complete replacement is impractical, soldering a 0.01 µF capacitor directly across the existing 0.001 µF leads restores the correct net capacitance. Following installation, perform the First Mixer Balance Adjustment per Section 4.5.7 of the 32S-3 manual.
Note for 32S-1 owners: This problem does not apply to the 32S-1. In that transmitter, all CW-mode 455 kHz energy passes through the mechanical filter, providing spectral cleanup in all emission modes.
3. ALC System Overshoot — The Leading-Edge Spike
3.1 Physical Mechanism
Every ALC-equipped transmitter — including all S-Line models — is susceptible to a leading-edge power spike at the onset of each transmission. The root cause is a fundamental property of closed-loop gain control: the ALC circuit cannot act until RF power has appeared at the detector, been measured, and a corrective voltage has propagated back to the IF gain-control stages. This round-trip group delay, typically 2–10 milliseconds in analogue IF chains, creates a window during which the transmitter operates at full unregulated output.[8]
ALC Loop Signal Path — Source of Overshoot Window
AUDIO IN ──▶ IF STAGES (full gain) ──▶ DRIVER ──▶ PA STAGES
│
RF OUTPUT
│
SWR / Power Detector
│
ALC Rectifier / Filter
│
(loop delay 2–10 ms) ◀─── ALC Voltage Return
│
IF Gain Reduction ──▶ regulated output
⚠ During loop delay: PA operates at FULL UNREGULATED DRIVE = SPIKE
In the 32S-3, ALC is derived from detected voltage at the PA grid circuit (via CR5 and CR6), with a fast attack / slow release dual time-constant applied to V6 and V3 respectively.[9] The slow-release hang time means the spike recurs every time the ALC has been allowed to fall — at the beginning of a CW element following a pause, or at the onset of each spoken syllable in SSB. It is worst in slow-speed CW and conversational SSB with long pauses.
W8JI (Tom Rauch) has documented ALC overshoot in contemporary transceivers at levels of 150–200 watts from units nominally set for 50 watts output — a genuine risk of component damage in external amplifiers.[10] Vintage tube transmitters are generally less severe than modern solid-state exciters, but the mechanism is present and measurable in all S-Line equipment.
3.2 ALC Overshoot and the 30L-1 / 30S-1 Amplifier Chain
When a 32S-3 drives a 30L-1 or 30S-1 linear amplifier, the amplifier’s ALC line should be connected to the transmitter ALC bus. The 30L-1 can develop up to approximately −4 V DC on its ALC output under sustained overdrive, but this protection is inherently retrospective — it cannot suppress the initial spike that precedes ALC stabilisation. The 20.5-foot RF interconnect cable supplied by Collins (which must not be cut) introduces a deliberate electrical length chosen for low-distortion impedance matching.[11]
Mitigation: ALC Overshoot
• Operate with moderate ALC deflection — never drive the ALC meter hard right. Maximum average power is better achieved via the speech processor than through ALC compression.
• Verify ALC zero-adjust stability[12] — an unstable ALC zero allows the system to drift, exacerbating overshoot at every key-down.
• Connect the 30L-1 / 30S-1 anti-ALC feedback line to the transmitter ALC bus for secondary clamp protection against sustained overdrive.
• Consider a pre-ALC bias resistor — a low fixed negative bias on the ALC bus (from a voltage divider on the B− supply) pre-sets the ALC slightly above zero, reducing the unguarded window on key-down.
4. VOX Relay and Keying Transients
4.1 Sources of Keying Transients
The 32S-3 and KWM-2/2A employ three distinct keying pathways — PTT/mic button, VOX, and grid-block CW keying — each capable of generating characteristic transients. The CCA’s KWM-2 Keying Circuit analysis[13] provides the most detailed treatment, and its findings apply broadly across the S-Line family. Key transient sources include:
B+ sag on TX switching: When the VOX or PTT relay closes, the sudden current demand on the 516F-2 causes a transient B+ rail drop. In units where the 516F-2 has been solid-state converted with an improperly positioned dropping resistor, this sag is exaggerated and can cause relay oscillation (“clacking”). The documented fix is addition of a 20 µF / 350 V filter capacitor at the hot side of relay K2, plus a 2.2 MΩ resistor at R199.
Grid-block keying envelope shaping: Poorly adjusted or degraded shaping components produce fast-rise CW envelope edges with high-frequency click energy extending well outside the signal bandwidth. The 32S-3 manual Section 4.5 specifies the alignment procedure for keying characteristic from “soft” to “hard”.
Contact bounce in the VOX relay: Relay contacts on 60-year-old equipment may produce multiple make-break cycles over several milliseconds, each generating a brief burst of RF at uncontrolled amplitude and phase — detectable on a high-speed oscilloscope or fast peak-reading meter.[14]
VOX time constant mismatch: Very early KWM-2 units did not include the VOX TIME CONSTANT potentiometer (added by Service Bulletin 2). Without it, or with the pot at an extreme setting, the VOX can cycle rapidly between transmit and receive, producing repetitive spike events.
Power Supply Note: The 516F-2 and Transient Coupling
The 516F-2 supplies B+ to both transmitter and receiver simultaneously. Keying transients on the transmitter side couple directly into the receiver B+ rail unless the internal supply sequencing and filter capacitance are adequate. This is particularly relevant in solid-state-converted supplies, where the original resonant choke characteristic — which provided natural L-C transient filtering — may have been replaced with simple resistive dropping that lacks the same suppression properties.[15]
Before assuming a transient originates in the transmitter RF chain, verify the 516F-2 filter capacitors are within tolerance and the resonant choke is intact.
4.2 Hot-Switching and Amplifier Protection
Hot-switching — the condition where the transmitter begins delivering RF drive before the amplifier’s antenna relay has fully closed — exposes relay contacts to RF arcing, degrades contact life, and can introduce an uncontrolled amplitude spike into the amplifier input during the relay bounce period.[16]
Modern T/R sequencing practice requires a 20–30 ms delay between antenna relay closure and the onset of transmitter drive. In the original S-Line station this sequencing is provided by the relay hierarchy in the 516F-2 and the amplifier’s control relay. Where the S-Line is integrated with modern solid-state amplifiers via interface boxes, this sequencing must be explicitly verified.
Mitigation: VOX Relay and Keying Transients
• Add 20 µF / 350 V filter cap at K2 hot side plus 2.2 MΩ R199 (per CCA KWM-2 keying analysis) if B+ sag or relay clacking is observed.
• Verify VOX time constant — confirm SB 2 has been applied to KWM-2/2A units; check VOX HANG TIME pot is present and functional.
• Clean relay contacts with Caig DeoxIT (never abrasives) to reduce bounce on aged contacts.
• Verify CW envelope adjustment per 32S-3 manual Section 4.5 — set rise/decay times for clean spectral occupancy without clicks.
• Verify T/R sequencing ≥20 ms before drive when using external amplifiers — particularly when interfacing with modern solid-state linears.
5. Systematic Diagnostic Procedure
The three spike classes require different diagnostic tools and methods. The following sequence is recommended for a complete station audit:
| Step | Test | Instrument | Pass Criterion |
|---|---|---|---|
| 1 | Inspect C20 value | Visual / LCR meter | 0.01 µF (code “103”) |
| 2 | Spectrum scan in CW/TUNE at low band edge | Spectrum analyser or SDR panadapter | All spurs < −50 dBc; no stationary out-of-band spur |
| 3 | ALC overshoot — leading edge of first CW element or syllable | Fast peak-reading wattmeter or storage oscilloscope | Peak < 1.5× steady-state rated output |
| 4 | CW envelope rise and decay time | Oscilloscope (envelope detect: diode + 1 kΩ) | Rise 5–10 ms; no bounce or step discontinuities |
| 5 | VOX relay bounce check | Oscilloscope — monitor keying line during VOX switching | Single clean make; no multiple closures in first 5 ms |
| 6 | 516F-2 B+ sag on TX switching | Oscilloscope — 10:1 probe on B+ rail | < 20 V transient sag; settled within 20 ms |
Modern Tool: SDR Panadapter for Spur Hunting
A Software Defined Radio panadapter — such as a Web-888, RX-888 Mk II, or RTL-SDR V3 connected via directional coupler to the transmitter output — provides a low-cost substitute for a dedicated spectrum analyser. With 8–64 MHz of instantaneous bandwidth, both Spur #2 and Spur #3 from a C20-modified 32S-3 on 20 metres are visible simultaneously on one panadapter screen.
Use a 40–60 dB directional coupler (Bruene-style[17]) to protect the SDR front end from the 100 W transmitter. The RX-888 Mk II is particularly suited to this role, accepting up to 32 MHz instantaneous bandwidth in direct-sampling mode.
6. RF Engineering Context — Why Vintage Transmitters Deserve Scrutiny
The S-Line was engineered to meet the regulatory and performance standards of its era. The FCC amateur spurious emission limit then applicable was more permissive than the current Part 97 standard,[18] and spectrum congestion was a fraction of that encountered today. A transmitter that met its factory specification in 1966 may not comply with 2024 Part 97 requirements — particularly after modification, component drift, or neglected alignment.
Mixer intermodulation products are proportional to drive level. Third-order intermodulation products increase by 3 dB for each 1 dB increase in drive. Driving the first mixer harder — as the C20 modification does — accelerates spurious growth disproportionately. An 8 dB increase in BFO drive produces a 13+ dB rise in spur level.[19]
Mechanical filter selectivity is a spectral gatekeeper. Collins mechanical filter technology provides extremely steep skirts compared to LC designs of the era — attenuating BFO harmonics entering the first mixer by 60–70 dB in SSB mode. The absence of this filtering in CW and TUNE modes is precisely why C20’s loading effect matters only in those emission modes.
Out-of-band spurious signals are not benign. A stationary spur 39 kHz below the 20 m lower band edge falls on 13.961 MHz — within the international fixed and broadcast service allocations. A 100 W transmitter with Spur #3 at −32 dBc is continuously radiating approximately 63 milliwatts on that frequency whenever operated in CW mode. Regulatory consequences for the licensee and interference to international listeners are the direct result.
7. Resources and Further Reading
Collins Collectors Association
• CCA Technical Bulletins Listing — collinsradio.org (public listing; download members only)
• RX For Your Collins — cross-referenced technical article library by model
• Collins Radio Equipment Manuals Archive — digitised 32S-3, KWM-2, 516F-2 manuals
RF Engineering Reference
• ALC Exciter Power Overshoot — Tom Rauch W8JI, with measurement methodology
• 47 CFR §97.307 — Emission Standards — FCC Part 97 spurious emission limits (Cornell LII)
• Understanding the Bruene Coupler and Transmission Line (Bold) — CCA RX archive
Test Equipment
• RTL-SDR Blog — RTL-SDR V3, RX-888 Mk II hardware and software for station monitoring
• Ameritron AWM-30 / ATR-30 — among the fastest-responding commercial peak-reading meters for transient capture
Footnotes and Citations
[1] Collins Radio Company, Cedar Rapids, Iowa. For corporate and equipment history see CCA Collins Historical Archives — Equipment of Collins Radio. Gene Senti designed the KWM-2, 30L-1, and KWS-1; Warren Bruene W5OLY designed the 30S-1 and 30K-1, retiring from Collins in 1984.
[2] Collins Collectors Association, CCA Technical Bulletin 1: Understanding and Mitigating S-Line Transmitter Spikes. Public listing: collinsradio.org/listing-of-cca-technical-bulletins/. Member download: collinsradio.org/tech-bullettins/.
[3] Collins Radio Company, Instruction Book — Type 32S-3 Transmitter, 7th ed., June 1969. Digitised via CCA Collins Technical Archives.
[4] Don Jackson W5QN, The C20 Modification for the 32S-3/3A Transmitter. Jackson attributes the modification’s wide propagation to writings by Bud Whitney K7RMT. Full PDF: 32S3-C20-Modification-Rev-4-Pub.pdf.
[5] Jackson (W5QN) confirmed the spur problem on four separate 32S-3 test units. All exceeded the −50 dBc factory specification with 0.001 µF C20; all met specification with the original 0.01 µF restored.
[6] 47 CFR §97.307(d): spurious emissions on frequencies between 30–225 MHz must be attenuated to at least −43 dB below the mean power of the fundamental. See Cornell LII — 47 CFR 97.307.
[7] The stationary nature of Spur #3 is the key diagnostic indicator: its frequency is determined solely by crystal BFO harmonics, not the VFO, so it remains fixed as the VFO is tuned — unlike the desired signal and Spur #2, which both move.
[8] Tom Rauch W8JI, ALC Exciter Power Overshoot. Full text: www.w8ji.com/alc_exciter_power_overshoot.htm.
[9] Collins Radio Company, 32S-3 Instruction Book, Section 3.1 (ALC Circuit): ALC detected via CR5/CR6; fast time-constant applied to V6, slow time-constant to V3.
[10] Rauch (W8JI) measured an IC-706 producing approximately 150 W peak on initial key-down with the radio set for 50 W output. Tube transmitters are generally less vulnerable due to higher inherent damping in IF gain-control elements.
[11] See CCA RX archive: Do You Need That 20.5-Foot Driver to Amp Cable, listed under 30L-1 at collinsradio.org/rx/.
[12] Don Jackson W5QN, Dealing with 32S-3 ALC Zero Adjust Instability. Listed in CCA RX archive under S-Line Transmitters.
[13] Collins Collectors Association, KWM-2/2A Keying Circuit — Functional Description and Issues. Primary author: Marc Niebergall K7WXK; reviewed by Don Jackson W5QN. PDF: KWM-2-Keying-Functional-Description-and-Issues.pdf.
[14] Rauch (W8JI). The Ameritron AWM-30 peak-reading meter system is among the fastest-responding commercial instruments for detecting relay bounce transients, capable of capturing sub-millisecond spikes invisible on conventional wattmeters.
[15] Don Jackson W5QN, Understanding Your 516F-2 Resonant Choke. Listed under 516F-2 in the CCA RX archive.
[16] Hot-switching and T/R sequencing: see W8JI, and Antique Radio Forums thread Using Collins 30L-1 — Keying delay and relay protection (August 2020), which documents the 20–30 ms delay requirement.
[17] Warren Bruene W5OLY, originator of the Bruene directional coupler used in all Collins 302C series couplers. CCA reference: Understanding the Bruene Coupler and Transmission Line (Bold), listed in CCA RX archive under 302C-3.
[18] FCC Part 97 spurious emission requirements have been progressively tightened. Transmitters type-accepted under earlier standards may not comply with current §97.307 without modification or re-alignment.
[19] Third-order intermodulation product level relative to fundamental in a mixer: IM3 ≈ 2(P − IIP3) dBc, where P is drive level and IIP3 is the third-order intercept. Each 1 dB increase in drive raises the IM3 product by 3 dB. See Pozar, Microwave Engineering, 4th ed., §10.3.
Collins S-Line Technical Reference Series — vk6ada.com.au
This document is an enriched commentary on CCA Technical Bulletin 1, supplemented with RF engineering analysis and primary source references. The original CCA Technical Bulletin 1 is the property of the Collins Collectors Association and is available to CCA members via collinsradio.org. The CCA is licensed by Rockwell Collins to reproduce and disseminate Collins amateur radio technical documentation.
Technical enrichment and commentary: Mike Peace VK6ADA / r-390a.net Administrator. All reasonable care has been taken; verify critical measurements and modifications against primary sources before proceeding.