Collins KWM-2 / KWM-2A Field Service

Some KWM-2 / 2A Tricks

A re-issue and enrichment of the January 2016 CCAE field-service article by Georges Ricaud, F6CER (CCAE #098), with primary-source cross-references for the Collins Collector community.

Collins KWM-2A front panel showing exciter tuning, S-meter, band switch and tuning knob - photograph by Georges F6CER from the original CCAE article

Front panel of the Collins KWM-2A — photograph from the original CCAE article by Georges F6CER.

Most of the KWM‑2 transceivers that turn up in Europe and Australasia belong to the first production generations of the early 1960s. They are often in poor cosmetic condition, electrically tired, and quite tricky to bring back to life. Repair takes time, money, and is not always successful — but, almost sixty years on, what a pleasure to rediscover that the engineers at Collins Radio were, quite simply, the best.

There are at least three distinct schematic revisions of the KWM‑2 in circulation. The earliest appears in Bill Orr W6SAI’s Single Sideband for the Radio Amateur,[1] showing the two-diode balanced modulator and first-generation product detector; two later revisions are preserved in the Collins Collector Association archives.[2] One of those drawings will match “your” particular serial number; the differences are easy to spot once you have all three to hand.

Before you begin, confirm the equipment is complete, has been cleaned, that all tubes are healthy and located in the correct sockets, and that the 9-pin external PTO connector is in place on the rear apron. Check continuity between J5−J6, J20−J21, and J22−J23; these are the bridging jumpers that complete the antenna, sidetone, and noise-blanker bypass paths.

DANGER — HIGH VOLTAGE. The KWM‑2/2A and the 516F‑2 power supply contain lethal voltages on both the AC and DC sides. The +800 V plate supply for V9/V10 is present at the PA cage and at PM‑2 connector pins; the −70 V bias rail and the +275 V B+ are present at multiple test points. One hand in the pocket. Never work alone. Bleed every electrolytic before touching anything.

Contents

At switch-on — cold dead heaters
Wrong start — meter behaviour at turn-on
Panel-meter tricks — S-meter, ALC, RCVR GAIN
VOX and PTT commands with a mind of their own
Audio madness — phone jack and 4-ohm output
Audio take-off and motor-boating — C106
Permanent whistle in the speaker — V2 (6U8)
Transmitter take-off on 80 m — alumina and shielding
Quick-reference fault summary
References and further reading

1. At switch-on — some heaters glow, others stay dark

The KWM‑2 receives its 6.3 V and 12.6 V heater voltages, and its low-voltage DC rails, from the PM‑2 or 516F‑2 power supply through the multi-pin Cinch−Jones P331-style power cable.[3] If the heater string is broken anywhere in that path, some tubes will glow and others will not. Inspect the following points in order:

Inside the 516F‑2 chassis, inspect the power cable going to the KWM‑2. Make sure the cable shield and the big white wire are properly soldered together and connected to chassis ground.
Take a look at the other side of the power cable as well — gently lift the cover of the power plug to verify everything is still properly soldered.
Inside the KWM‑2, under the chassis near the power socket J13, make sure the big white wire coming from J13 is still well soldered to the chassis ground lugs of J1 (RF OUT) and J22 (NB ANT).

Cross-reference: Surplus Sales of Nebraska, ke9pq.com (Bob Sullivan) and DH Distributors all carry replacement KR2565‑1 (Potter & Brumfield) coil relays, replacement filter cans, and chassis-mount capacitor cans should the power-cable inspection reveal that one of the supply rails has been dragged down by a failed filter element. The original chassis-mounted multi-section can is the part most commonly under suspicion when heater behaviour is erratic in conjunction with B+ collapse.[4]

2. Wrong start — meter behaviour at turn-on

Normal start-up: the meter needle swings hard right, then gently returns to the left as the time-delay relay closes and the high-voltage rails come up.[5] The receiver then comes alive and band noise is heard in the speaker. Two failure modes are common.

A) The needle stays at full scale and never returns

Make sure J20 and J21 are connected together (the noise-blanker bypass jumper).
Switch the equipment off, open the PA cage and inspect the antenna relay K3 contacts. Clean if necessary — relay contacts that have not been exercised in years develop tarnish films that prevent the receive path from being completed at start-up.
Near K3, verify L32 (the receiver-path RF choke) is not open-circuit.

B) The meter takes a very long time to fall back to zero

Check that R83 on the E30 turret is not open-circuit.
Check that the RF GAIN control R84 is not open and does not have erratic contact while rotating.

Enrichment — meter pinning in receive: If the meter pins hard during normal receive (full-scale on band noise with no signal), David Harmon K6XYZ writes on the CCA listserve that a common cause is accidental disturbance of the RCVR GAIN ADJ sub-panel pot (next to the ALC ADJ). It is unfortunately located within finger-reach of any operator who decides to “peak up” receive sensitivity. With the calibrator on and the meter switch in S-METER, the pot should be set so that a moderate calibrator signal on 80 m reads roughly 15 dB over S9; a daub of nail polish or glyptal after adjustment will tell you next time whether somebody has moved it.[6]

3. Panel-meter tricks — S-meter, ALC, RCVR GAIN

A) The S-meter zero adjustment is impossible

Verify the health of V1 (6AZ8). Swap it with V3 or V4 — some 6AZ8 tubes (even NOS examples) develop enough grid current to make zeroing impossible.
Check that the RCVR GAIN ADJ variable resistor R132 is not open-circuit.

The 6AZ8 grid-current problem — the diagnostic test: Per Collins Service Bulletin information collated by Yves WN4I and circulated on the CCA listserve, the diagnostic test is to measure the DC voltage at the V1 or V3 control grid with no signal present. The nominal AVC line voltage is in the vicinity of −1.0 V; the ALC line should be approximately −1.3 to −1.8 V. If the no-signal grid voltage differs from the AVC or ALC line by more than 0.1 V, the offending 6AZ8 should be replaced.[7]

B) S-meter OK on receive but dead on transmit (no plate, no grid, no ALC)

A classic of older KWM‑2 examples fitted with open-frame relays. The chassis-side contacts (3, 4, 5 and 6, 7, 8) of relay K4 are responsible for switching the meter between the transmit and receive sense circuits. When those contacts are dirty or pitted, there is no transmit-side indication at all.

Remedy: Burnish the K4 contacts with a contact-cleaning blade or a strip of clean white paper moistened with DeoxIT D5, then apply a micro-drop of DeoxIT GOLD. Severe cases require relay replacement — the original is a Potter & Brumfield KR2565‑1 (DPDT, 115 V DC coil), still in stock at Surplus Sales of Nebraska as of writing.[4]

C) The ALC zero cannot be set properly in the ALC meter position

V4 (6AZ8) is prone to developing grid current and must be replaced.

Enrichment — if a fresh 6AZ8 still will not let the ALC zero: Two factory-published changes will extend the range of the ALC zero adjustment.

Range-extension mod: R170 (220 Ω across the ALC zero pot) can be lifted from circuit. This extends the available adjustment range without affecting normal operation.[7]
R38 update (factory): In later units, R38 was increased from 150 Ω ½ W to 220 Ω ½ W (Collins CPN 745‑1324‑000) for improved ALC zeroing.[7]
60 Hz ripple mod (Yves WN4I): Add a 20 µF electrolytic across the −70 V bias line at the input of R189 (close to the rear of the emission switch). This stops mains-ripple modulation of the ALC zero.[7]

4. VOX and PTT commands with a mind of their own

A) PTT and VOX have no effect at all

Verify the coil of relay K2 is not open-circuit.
Suspect V4 (6AZ8) — swap it with a known good 6AZ8 (or cross-swap it with one of the other 6AZ8s in the rig).

B) You can go from receive to transmit, but never back again

There is a short in C225 (a 0.1 µF ceramic).
R20 and R47 (the two 68 kΩ / 2 W resistors on top of E50) have drifted high.

Enrichment — R20/R47 cathode resistor “classic”: This is one of the best-known KWM‑2 failures, well documented on the CCA archive. The two 68 kΩ 2 W parallel cathode resistors in the V4b (6AZ8) VOX-relay-amplifier cathode circuit were under-rated for the ambient temperature on the chassis. As they age and drift up, the rig refuses to drop back to receive. The recommended fix on the CCA archive is to replace the pair with a single 34 kΩ 7 W resistor; restoring the original 2×68 k/2 W parallel is electrically correct but the problem will return in a few years.[8]

C) Relays rattle like a machine-gun

C225 is almost-but-not-quite shorted.
The VOX TIME CONSTANT zero is drifting — turn the control slightly CW until the system becomes stable.
V14 is developing instability — try adjusting the VOX GAIN control R39.

The most recent KWM‑2 and all KWM‑2A units shipped with two important factory modifications around V14 (6BN8):

A 100 kΩ or 220 kΩ resistor R201 added between pin 8 of V14 and the red/white shielded wire going to the VOX GAIN control R39.
A 2.2 MΩ resistor R199 added from pin 3 of V14 to the green/white wire going to the VOX TIME CONSTANT pot R43.

Enrichment — further VOX-stability mods: The Collins Amateur Service Bulletin programme also recommends installing a 0.01 µF capacitor from V1 pin 7 to ground to suppress relay transients, and 0.1 µF across contacts 12 and 13 of relay K2 to reduce VOX instability. Both are quick, reversible, and almost always worthwhile on early-serial units.[7] For a deeper treatment of the keying-circuit interactions, refer to the Carns KWM‑2/2A keying-circuit tutorial published through the CCA.[9]

5. Audio madness — phone jack and 4-ohm output

A) Audio in headphones only; nothing at the rear-panel 4-ohm AUDIO jack J10

The PHONES jack on the front panel is faulty — the extra set of contacts that mute the loudspeaker when a plug is inserted is dirty and needs cleaning.
L27, the RF choke soldered between J22 (NB ANT) and J10 (4 OHMS AUDIO), has a short to chassis ground. (Bear in mind L27 is a low-pass filter element protecting the audio output from RF pickup; if it shorts to chassis it will dump the audio output to ground at the same time.)

B) No audio anywhere except in CW mode

Audio is absent both at the headphone jack and the rear loudspeaker output J10, but appears the moment the emission switch is set to CW. Cause: The contacts 13 and 14 of relay K2 — the pair that returns one side of audio transformer T6 to ground in SSB/AM — are dirty or damaged.

Remedy: Burnish the K2−13/14 contacts with paper and DeoxIT, or replace K2. The original K2 is also a Potter & Brumfield KR2565‑1 (115 V DC, DPDT) — the same part used elsewhere in the rig and still in stock at Surplus Sales of Nebraska.[4]

6. Audio take-off and motor-boating with the AF gain advanced

Push the AF gain up and the receiver begins motor-boating, putt-putt-putt, or breaks into a low-frequency oscillation. C106 is defective. C106 is the audio-decoupling section in the chassis-mount multi-section electrolytic can — the very same can that filters the receive B+ — and when one section opens or loses capacitance, the audio amplifier sees positive feedback through the shared B+ bus.

Enrichment — sourcing a new C106: Newly-manufactured drop-in replacements for the chassis-mounted KWM‑2 multi-section can are available from DH Distributors and from Bob Sullivan KØNR (ke9pq.com). The ke9pq part is a nickel-plated custom can with tabs that drop into the original chassis cut-out and uses modern radial electrolytics inside.[10] While the can is out, replace the bias-supply electrolytic C113/C137 and the B+ bypass C124 (the schematic’s listed value for C124 is widely known to be incorrect on some early printings — verify against the part fitted before ordering).[7]

If C106 measures within tolerance on an ESR meter and the symptom persists, check that the audio-stage cathode bypass C102 has not been replaced with a wrong value at some point in the rig’s life — a common find on units that have passed through multiple owners is a non-Collins electrolytic of the wrong capacitance in this position.[11]

7. Permanent whistle in the speaker — V2 (6U8)

The whistle is generated by V2 (6U8). This tube is the dual-purpose CW carrier oscillator and sidetone source: it generates the CW transmit signal, the audible sidetone during CW transmit, and the test tone used in TUNE and LOCK positions during transmitter alignment.

V2 is fussy. If the contacts on the mode switch S9F are dirty, the sidetone runs continuously through the wiring harness into the audio amplifier even when the rig is in SSB receive.

Two-step remedy:

Clean S9F with DeoxIT and exercise the switch through all positions a dozen times.
Solder a 47 Ω or 100 Ω ¼ W resistor between pin 3 of V2 (screen grid) and the 0.1 µF decoupling capacitor C49. This damps the screen circuit and prevents spurious oscillation when the cathode return is momentarily open during switching.

Enrichment — a related factory mod: Collins also issued a factory bulletin adding C261, a 100 pF capacitor (CPN 912‑2817‑000), from V2 pin 2 to ground, to eliminate a parasitic-oscillation tendency in the tone oscillator. If the screen-grid damping resistor above does not entirely cure the whistle, install C261 as well.[7]

8. Transmitter take-off on 80 m — alumina and shielding

This is a classic problem in older aluminum-chassis Collins gear, and the same failure mode appears in the 32S‑1/2/3 transmitters. It is most prominent on the 80 m band but can also affect the receiver. To reproduce:

Set the band switch to 3.4 − 3.5, the VFO to 100, and turn the calibrator on.
Tune the EXCITER TUNING control across its range. On a healthy rig there is one position where the calibrator peaks — the legitimate calibration point. On a sick rig you will find two or three positions where the noise increases, but only one gives a correct calibration peak.
In TUNE, the grid current goes full scale at two or three different exciter-tuning positions. In LOCK, the plate current shoots to a high value and is no longer controllable by the MIC GAIN control.

All of these symptoms point to one root cause: the aluminum chassis has developed an oxide layer (alumina) where the coil shields and their finger-stock contacts mate to the chassis. Alumina is an excellent insulator, so the shielding is no longer ground-referenced and the tuned circuits parasitically intermodulate.

Restoration procedure

Take the coil shields apart, together with all the silvered finger-stock contacts.
Gently clean every mating surface with fine steel wool. To reach tight places, wrap a small wad of steel wool around the end of a pencil with cellophane tape.
Be very careful with the finger-stock contacts — the silver-plated phosphor-bronze stock is brittle and will snap if over-bent.
Protect the cleaned mating surfaces with a thin coating of conductive grease. Suitable greases include those sold for antenna mast-section joints or for automotive exhaust-pipe fasteners.
Re-open the PA cage, remove V8 (6CL6) from its socket, and back the two socket-retaining screws out and back in two or three times to break up oxide on the threads. These screws not only hold the socket; they also ground the under-chassis shield that isolates the V8 input and output pins. Loss of this shield ground is a direct path to driver-stage instability.

A warning about the “quick access” coil-shield modification: Some previous owners cut a slot in the coil shields so they could reach the tuned-circuit trimmers without removing the fibre switch shaft. If your rig has had this done, you are almost guaranteed unstable receive and transmit on all bands. The only proper fixes are to replace the cut shields with intact originals (Surplus Sales of Nebraska sometimes has them), or to seal the slot with self-adhesive aluminium tape pressed firmly enough to make good RF contact along its full length.

When the job is done, clean everything. Belt and braces is the right philosophy on a 60-year-old transceiver.

Quick-reference fault summary

Symptom	Prime suspect	Section
Some tube heaters dark at switch-on	Power cable shield / J13 / chassis-ground lugs at J1, J22	§1
Meter never returns from full-scale at start	J20-J21 jumper / K3 contacts / L32 open	§2A
Meter very slow to fall back to zero	R83 open / RF GAIN R84 open or noisy	§2B
S-meter zero will not adjust	V1 (6AZ8) grid current / R132 open	§3A
No meter on transmit (no Ip, Ig, ALC)	K4 contacts 3-4-5 / 6-7-8 dirty	§3B
ALC zero impossible	V4 (6AZ8) grid current; lift R170; check R38	§3C
PTT / VOX inert	K2 coil open / V4 (6AZ8)	§4A
Stuck in transmit (no return to receive)	C225 shorted / R20+R47 drifted high	§4B
Relays rattle / VOX chatter	C225 leaky / V14 unstable / VOX TC drift	§4C
No audio at rear J10 (only in headphones)	PHONES jack switch dirty / L27 shorted	§5A
No audio except in CW	K2 contacts 13/14 dirty (T6 ground)	§5B
Motor-boating at high AF gain	C106 chassis-mount can defective	§6
Permanent whistle in speaker	V2 (6U8) takeoff via dirty S9F; add 47-100Ω on V2 pin 3	§7
Instability / multiple peaks on 80 m	Alumina on coil-shield ground points; V8 socket shield	§8

Original article credit

The original Some KWM-2/2A Tricks field-service article was written by Georges Ricaud, F6CER — CCAE member #098 — and published in January 2016 by the Collins Collector Association Europe (CCAE) at www.ccae.tm6cca.com. Georges is a long-standing CCAE contributor and a fixture of the European Collins community.[12]

This enriched re-issue preserves Georges’ original eight-section structure and original cover photograph, adds cross-references to the Collins Collector Association (CCA) archive of Collins Amateur Service Bulletins and to the CCA collins listserve, and inserts a quick-reference fault summary. All factual additions are individually footnoted.

If you find these notes useful, please consider joining the CCAE (membership is free) at www.ccae.tm6cca.com, or the parent CCA at collinsradio.org. The combined CCA / CCAE archives are the single most important source of working-knowledge on these radios outside the original Collins factory documents.

References and further reading

Orr, W. I. (W6SAI). Single Sideband for the Radio Amateur. ARRL, multiple editions through the 1960s. The KWM‑2 first-generation schematic appears in the SSB transceiver chapter.
Collins Collector Association archive of KWM‑2 / KWM‑2A manuals and schematics, including the 3rd edition (March 1962) and 9th edition (January 1978): collinsradio.org/archives/manuals/. The CCA archive distinguishes between the early two-diode balanced-modulator revision and the later four-diode ring revisions.
Collins Radio Company. Instruction Book, KWM‑2 and KWM‑2A Transceivers, multiple editions. Power-cabling, J13 pinout and interconnection diagrams: see Section 1 (Installation) and Figure 1−2 (Traveling Station Interconnections). Mirror at worldradiohistory.com.
Surplus Sales of Nebraska, KWM‑2 / KWM‑2A parts catalogue: surplussales.com/categories/2242-collins-kwm-2-kwm-2a-parts/. Active stock includes the KR2565‑1 (Potter & Brumfield DPDT, 115 V DC) relay, original-spec MIC GAIN/On-Off pots, S-meter, T1 transformer, 6146B matched pairs (Roe Tester-curve-matched), and the 6AZ8 (Philips/ECG).
Collins Radio Company. KWM‑2/2A Instruction Book, Section 1 paragraph 1.4 et seq.: “When the S-meter falls back to zero, the circuits will have switched to…” describes the normal turn-on meter behaviour.
Harmon, D. (K6XYZ). “Re: KWM2‑A — New AGC Problem”, posted to the CCA [email protected] mailing list, 3 May 2015. Archived at mailman.listserve.com/pipermail/collins/2015-May/msg00052.html. The advice about painting the RCVR GAIN ADJ pot after alignment is from this post.
Compiled KWM‑2/2A Collins Amateur Service Bulletin information, drawing on Collins Service Information Letters and Agency Bulletins (Section 7 of the Collins KWM‑2/2A service manual, identifiers 1−39), with contributions from Yves WN4I, Jim N4BE, Ken N6TZV and others on the CCA collins listserve. Useful index pages: angelfire.com/de/vk3kcm/kwm2.html and /kwm2a.html. Includes the R170 lift, the R38 update to 220 Ω (CPN 745‑1324‑000), C261 100 pF tone-oscillator parasitic mod (CPN 912‑2817‑000), C264 4 µF PSU-ripple bypass, R175/C262 IF cathode mod, and the 20 µF −70 V ripple bypass at R189.
Documented R20/R47 cathode-resistor “classic” failure in the V4b VOX-relay-amplifier cathode circuit and the recommended 34 kΩ / 7 W single-resistor replacement, from the Angelfire KWM‑2/2A information index maintained from CCA mailing-list traffic: angelfire.com/de/vk3kcm/kwm2a.html.
Carns, J. KWM‑2/2A Keying Circuit Tutorial, Collins Collector Association reprint. The Carns tutorial is the standard reference on the interaction between PTT, TUNE/LOCK KEY and VOX keying paths inside the KWM‑2.
Bob Sullivan KE9PQ, “Collins KWM‑2 / KWM‑2A NEW Capacitor Replacement Kit Including Chassis Mount Multi Can”: ke9pq.com. Drop-in nickel-plated replacement can with modern radial electrolytics inside. DH Distributors offers an equivalent custom-built can.
Carruthers, K. (Gamma-Five), “My Collins KWM-2” restoration notes: gamma-five.com/KWM-2/. Documents finding a non-Collins, wrong-value electrolytic at C102 in a previously-owned chassis — a useful reminder that not every “original-looking” cap is original.
Ricaud, G. (F6CER). Some KWM‑2/2A Tricks. Collins Collector Association Europe, January 2016. Source PDF: ccae.tm6cca.com/DOWN/CCAE-SOMEKWM2TRICKS.pdf. CCAE home: ccae.tm6cca.com.
Schroeder, E. F. (DJ7HS). Collins KWM‑2: Curing a Strange Problem (2013): qsl.net/dj7hs/kwm-2-strg.htm. An exemplary methodical-fault-finding write-up; the AVC-line approach described there is a useful complement to the techniques in this article.
EB5AGV & José Gavila. KWM‑2A Repair Notes. jvgavila.com/kwm2a_1.htm. European restorer’s narrative of the same kind of issues described in this article, from a different rig and a different vantage point.