Collins 32S-3 PA Neutralisation — Step by Step

The Procedure Most Likely to Be Skipped — And Most Consequential When Wrong

Definitive guide to 6146 PA neutralisation in the 32S-3/3A, KWM-2/2A, and 32S-1 — covering both the ceramic trimmer and air-variable circuits, the 6146B compatibility issue, oscilloscope technique, and the driver shield problem that kills finals

Why Neutralisation is Non-Negotiable

PA neutralisation is the most consequential alignment procedure in the Collins S-Line transmitter family — and the one most frequently skipped. The 6146 final amplifier tubes in the 32S-3 operate in a push-pull circuit where internal plate-to-grid capacitance provides a feedback path that can sustain oscillation, particularly at VHF frequencies well above the operating band. The neutralisation capacitor (C184) introduces a counter-phase signal that cancels this internal feedback path. When C184 is correctly adjusted, the PA is stable; when it is not, the PA can break into parasitic oscillation that is invisible on an HF wattmeter but devastating to the tubes and surrounding components.[1]

An un-neutralised or poorly neutralised Collins transmitter will exhibit one or more of the following: plate current dip and maximum output occurring at different PA TUNING settings; unexplained high plate current; erratic ALC behaviour; glowing 6146 plates at idle; excessive grid current; and in severe cases, destruction of the neutralising capacitor, the coupling capacitor, and the 6146 tubes themselves.[2]

⚠ The 32S-3 has a marginal neutralising circuit from the factory. Collins originally used poorly chosen component values that were later revised via circuit updates. If you are running 6146B or 6146W tubes in an early-circuit 32S-3 or KWM-2, the original ceramic trimmer neutralisation circuit may not have sufficient range to accommodate the higher plate-to-grid capacitance of these later tube variants. The air-variable upgrade is strongly recommended.[1]

Neutralisation Circuit Variants

Collins produced the 32S-3 and KWM-2 with two distinctly different neutralisation circuits over the production run. Identifying which circuit your transmitter has is the essential first step before attempting neutralisation.[3]

Neutralisation Circuit Comparison
Feature Early Circuit (Ceramic Trimmer) Late Circuit (Air Variable)
C184 Type
Erie ceramic trimmer, 1.8–8.7 pF
Johnson air variable, ~8.1 pF max
Series Capacitor
2 kV rated
6 kV rated (0.001 µF)
Series Resistor
1 watt
2 watt (470 kΩ)
6146B/W Compatible
Marginal — may not have sufficient range
Yes — adequate range for all 6146 variants
Adjustment Tool
Non-metallic alignment tool (ceramic trimmer)
Non-metallic tool from below chassis
Identification
Ceramic disc visible on phenolic board in PA
Small air variable mounted through phenolic board
Applies To
Early 32S-3, early KWM-2
Late 32S-3A, 9th ed. KWM-2/2A

Upgrade Recommendation: The air-variable neutralisation upgrade is available as a kit (W7KSG and others have produced them) and is strongly recommended for all early-circuit units, particularly if 6146B or 6146W tubes are being used or may be used in the future. The upgrade replaces the ceramic trimmer, series capacitor, and series resistor with higher-rated components that provide more stable, wider-range neutralisation adjustment.[4]

The 6146 Tube Variant Issue

The original 6146 tube was designed for HF service and has relatively low plate-to-grid capacitance. The later 6146B and 6146W variants, designed by RCA for improved performance, have slightly different internal geometry that results in higher stray capacitance — particularly from the larger anode structure coupling to external elements. While these tubes work perfectly in most transmitters, Collins’ original ceramic trimmer neutralisation circuit in early 32S and KWM-2 units had insufficient adjustment range to compensate for the increased capacitance, leading to the widespread reports of PA instability and “hot” 6146B operation that circulate in the Collins community.[3]

The 6146B and 6146W are fully compatible with the Collins transmitters when the neutralisation circuit has been upgraded to the air-variable configuration. With the upgraded circuit, any member of the 6146 family can be used without concern.[5]

Neutralisation Procedure — Step by Step

This procedure applies to the 32S-3/3A transmitter, the KWM-2/2A transceiver, and with minor differences to the 32S-1. The principle is the same for both circuit variants: remove plate and screen voltage from the 6146 PA tubes, apply drive signal, and adjust C184 for minimum RF feedthrough from the grid circuit to the plate circuit. A properly neutralised PA will show zero (or near-zero) RF output with the plates dead.[6]

⚠ HIGH VOLTAGE WARNING: The PA plate supply carries approximately 800V DC in the 32S-3 (higher in the KWM-2 with some power supplies). Always discharge the PA before touching anything inside the PA compartment. Short the plate supply to chassis ground through the plate caps before removing tubes. Verify with a voltmeter before proceeding.

Phase 1 — Setup and Safety
1Remove the cabinet. Remove the PA shield cover (5 screws — the cover slides out sideways).[4]
2Discharge the plate supply by shorting from the PA cage to the plate connector strip with a metal screwdriver or grounding rod. Verify zero volts with a meter.[4]
3Remove the 6146 plate caps, then remove both 6146 tubes. Set them aside safely. This removes both plate and screen voltage from the PA — the tubes must be physically absent for a valid neutralisation measurement.
4Verify the 6CL6 driver shield (V8): Remove the shield and inspect it. The shield must make perfect ground contact when reinstalled. A poorly grounded 6CL6 shield will cause the driver to oscillate independently, applying excessive drive to the PA sockets and producing misleading neutralisation readings. Do not substitute a generic tube shield — the 6CL6 runs near maximum rated dissipation in class A and needs the original Collins shield for proper thermal management.[5]
5Connect a dummy load to the antenna output. Connect the oscilloscope probe (10× or higher) to the PA plate circuit — specifically, to the plate pin of either 6146 socket or to the plate tank circuit at a convenient test point.
Phase 2 — Applying Drive Signal
6Reinstall the PA shield cover to maintain normal shielding conditions during the measurement.
7Power up the transmitter. Select the highest frequency band available (10 meters / 28 MHz is ideal because neutralisation is most critical at the highest frequencies where parasitic tendency is greatest).[5]
8Set EMISSION to TUNE, then switch to LOCK KEY. Advance MIC GAIN to produce drive to the PA. With the 6146 tubes removed, the exciter stages are driving into the empty PA tube sockets — any RF appearing at the plate circuit is feedthrough through the plate-to-grid capacitance path that C184 is supposed to cancel.
9On the oscilloscope, you should see a residual RF signal at the plate pins. This is the un-neutralised feedthrough. The goal is to adjust C184 until this signal disappears to zero (or the absolute minimum achievable).

Alternative Method — RF Voltmeter: If an oscilloscope is not available, connect an RF voltmeter (HP 410B or equivalent VTVM with RF probe) to the antenna output jack instead of using a scope at the plate pin. Adjust C184 for minimum RF voltage at the antenna output. The scope method is superior because it allows you to distinguish between feedthrough at the operating frequency and spurious VHF parasitic energy, but the RF voltmeter method works well for the basic adjustment.

Phase 3 — Adjusting C184
10Using a non-metallic alignment tool only, adjust C184 slowly while watching the oscilloscope (or RF voltmeter). A metal tool will couple hand capacitance into the circuit and produce a false null that shifts when you remove the tool.[4]
11For ceramic trimmer circuits: You should see two nulls (dips to minimum) as you sweep the trimmer across its range — the feedthrough goes from maximum to minimum, back to maximum, then to a second minimum. Choose whichever null is deepest. If you see only one null, or no null at all, the trimmer may need replacement or the circuit has drifted outside its adjustment range.[7]
12For air variable circuits: Adjust for the single deepest null. The air variable has a more linear characteristic than the ceramic trimmer — you should see a smooth, well-defined minimum. If using the DJ7HS procedure, start with rotor and stator approximately half-meshed and adjust from there.[4]

What You Should See on the Oscilloscope: Before adjustment, the scope shows a clear RF sinewave at the plate pin — this is the feedthrough. As C184 approaches the correct value, the sinewave amplitude decreases sharply. At the null point, the trace should drop to the noise floor of the oscilloscope. Any residual signal at the null should be examined for frequency — if it is at the operating frequency, neutralisation is close but not perfect; if it is at VHF, you may have a parasitic oscillation issue independent of the neutralisation adjustment.

Phase 4 — Verification and Reassembly
13Power down. Remove PA shield cover. Reinstall both 6146 tubes and their plate caps. Replace the 6CL6 shield — verify good ground contact. Replace PA shield cover.
14Power up. Set MIC GAIN to OFF. Switch to TUNE, then LOCK KEY. Verify proper PA idle current (per manual — typically ~50 mA with bias correctly set).
15Advance MIC GAIN and tune the PA for normal operation: peak the grid current with EXCITER TUNING, dip the plate current with PA TUNING, and load for rated output with LOADING. The plate current dip and maximum output should now occur at the same PA TUNING setting. If they do not coincide, the neutralisation needs further adjustment.[5]
16Check neutralisation on all bands. If the primary adjustment was performed on 10 meters (recommended), lower bands should be satisfactory. If parasitic tendencies are noted on any band, re-check with the grid-dip meter in wavemeter mode to identify VHF energy.[2]

⚠ Parasitic Oscillation Test: VHF parasitic oscillation may not be visible on a standard HF wattmeter. Use a grid-dip meter in diode (wavemeter) mode or a spectrum analyzer to sniff for VHF energy near the PA compartment. Parasitic oscillation at VHF produces high plate current, excessive tube heating, and can damage the neutralising components — all symptoms that are easily mistaken for a neutralisation problem when they are actually a parasitic suppression problem (L15/L16 spacing).[2]

The 6CL6 Driver Shield — The Hidden Killer

The 6CL6 driver tube (V7 in the 32S-3, V8 in the KWM-2) operates in class A at near maximum rated plate dissipation. If its shield does not make perfect ground contact, the driver will oscillate at VHF, applying full — actually, greatly excessive — drive to the 6146 final amplifier tubes. This condition will destroy 6146 tubes in minutes and can also damage the neutralising circuit components.[5]

Do not substitute a generic tube shield for the 6CL6. Standard replacement shields reflect heat back into the tube instead of conducting it away, leading to premature tube failure from overheating. The original Collins shield is designed for thermal management of this high-dissipation application. A shield that merely “fits” is not adequate — it must make reliable, low-resistance contact to chassis ground around its entire base.[5]

Diagnostic Tip: If the 6146 tubes glow red at idle with MIC GAIN at OFF and the PA correctly biased, suspect the 6CL6 driver shield before suspecting the neutralisation adjustment. Remove the 6CL6 shield, clean both the shield base and the tube socket ground ring, and reinstall. Verify that idle plate current returns to the specified value.

Applicability — 32S-3, KWM-2/2A, 32S-1

The neutralisation procedure described here applies to all Collins transmitters and transceivers using push-pull 6146 PA tubes with the Collins neutralisation circuit. The circuit component values differ between early and late production but the procedure is identical. The air-variable upgrade kit is interchangeable between the 32S-3/3A, KWM-2/2A, and 32S-1 — the mounting hardware and phenolic board dimensions are the same across the family.[4]

Neutralisation must be rechecked whenever 6146 tubes are replaced, as each tube pair has slightly different internal capacitance characteristics. This is one of the strongest arguments for the air-variable upgrade — ceramic trimmers may not survive repeated adjustment cycles, while air variables are inherently robust.[3]

References & Citations
  1. Antique Radio Forums. Collins 32S-3 Grid Current — KB6GM analysis. 32S-3 marginal neutralising circuit from factory; 6146B/W parasitic oscillation at VHF; grid-dip meter wavemeter diagnostic technique. March 2019. antiqueradios.com — 32S-3 Grid Current
  2. Antique Radio Forums. Neutralizing Final Tubes — WQ9E and KB6GM analysis. Parasitic oscillation diagnosis, plate current dip vs. maximum output coincidence test, VHF wavemeter technique, L15/L16 parasitic suppressor spacing. January 2020. antiqueradios.com — Neutralizing Final Tubes
  3. VK3KCM. 6146 Variants and Neutralization. Ceramic trimmer (1.8–8.7 pF) vs. air variable (8–50 pF) circuit comparison; KWM-2 Section 7 schematic differences; compatibility analysis. angelfire.com/de/vk3kcm — 6146 Variants
  4. Borowski, Gerd DJ7HS. Collins KWM-2: Installing New Parts for PA Neutralization. Complete photographic step-by-step of air variable upgrade kit installation; W7KSG kit; Johnson air variable capacitor; phenolic board mounting. qsl.net/dj7hs — KWM-2 Neutralisation
  5. Antique Radio Forums. Neutralizing Final Tubes — 6CL6 Driver Shield. 6CL6 class A dissipation, shield grounding requirement, shield substitution warning, driver oscillation as root cause of 6146 failure. WQ9E analysis. antiqueradios.com — Driver Shield
  6. Groups.io / Heathkit. 6146 Tubes — Neutralisation Technique. Screen voltage removal method; feedthrough null procedure; rough neutralisation on 20 meters, touch-up on 10 meters. January 2020. groups.io/heathkit — 6146 Neutralisation
  7. Collins Reflector. KWM-2A Problems — Neutralisation Dual Null. Erie ceramic trimmer behaviour: two nulls expected in sweep; single null indicates degraded trimmer; C123 feedthrough capacitor replacement. October 2004. Collins Reflector — Neut Circuit
  8. Antique Radio Forums. Collins PA Neutralizing Set Up. Air variable upgrade parts sourcing (RF Parts Johnson capacitor); space constraints; phenolic mounting board fabrication. December 2015. antiqueradios.com — PA Neutralizing
  9. Collins Collectors Association — RX For Your Collins. Master index including: “Repairing the 32S-3 — Carns and Bud, K7RMT,” “32S-3 Trouble Shooting Voltage Table,” and related PA maintenance articles. collinsradio.org — RX For Your Collins
  10. Collins 32S-3 Instruction Book, 7th Edition (June 1969). PA feedback neutralizing procedure, tube complement, circuit description. Collins Radio Company / Rockwell International. collinsradio.org — 32S-3 Manual (PDF)
  11. Collins KWM-2/2A Instruction Book, 9th Edition (January 1978). Section 4: Feedback Neutralizing procedure; Section 7: Equipment Differences and schematic changes documenting neutralisation circuit evolution. collinsradio.org — KWM-2/2A Manual Section 4 (PDF)
Credits & Acknowledgments

Gerd Borowski, DJ7HS — For the definitive photographic step-by-step guide to installing the air-variable neutralisation upgrade kit in the KWM-2, which applies equally to the 32S-3 family.

J A Call, W7KSG — For producing the air-variable neutralisation upgrade kit with the correct Johnson capacitor and mounting hardware.

Jim T., KB6GM — For the analysis of the original Collins neutralisation circuit’s marginal design and its incompatibility with later 6146 variants.

Rodger WQ9E — For the detailed 6CL6 driver shield analysis explaining how a poorly grounded shield produces symptoms easily mistaken for neutralisation failure.

VK3KCM — For compiling the 6146 variant and neutralisation circuit comparison from Collins service documentation.

Collins Collectors Association (CCA) — For maintaining the “RX For Your Collins” technical article library and the archived instruction manuals that document the neutralisation circuit evolution.