KWM-2A Clinic: When the Preselector Peak and Grid Drive Peak Don’t Agree
A good question landed on the reflector recently from Lenny, WB6QAZ, who is bringing a round-emblem KWM-2A back to life. The radio is working well on both receive and transmit, but two things had him puzzled — and they’re both things nearly every KWM-2 owner runs into sooner or later. Here’s the question and my take on it.
⚠ Danger — Lethal Voltages Inside
The KWM-2/2A and its power supplies contain voltages that can kill you instantly. The 516F-2 and PM-2 power supplies develop roughly 800 volts DC on the high-voltage B+ line feeding the 6146 final amplifier plates, plus 275-volt low B+ and negative bias supplies. These voltages are present on exposed terminals, tube caps, the plate circuit, and throughout the underside of the chassis whenever the radio is energized. Contact with the HV line at these current levels is very likely to be fatal — this is not an exaggeration, and experienced technicians have died working on equipment exactly like this.
Filter capacitors store a lethal charge after power is removed. Never assume a de-energized radio is a safe radio. Before touching anything under the chassis, unplug the supply, wait several minutes, then verify zero volts with a meter and discharge the filter capacitors through a suitable bleeder resistor on an insulated probe — every time, even if you “just had it apart.”
Alignment is performed with the radio live. The transmit alignment steps described below require operating the radio with covers off and high voltage present. Use only insulated, non-metallic alignment tools. Keep one hand in your pocket when probing a live chassis. Remove rings, watches, and metal jewelry. Stand on a dry, insulated surface. Never work on live HV equipment alone — have someone within earshot who knows how to kill the power and perform CPR.
If you are not experienced and comfortable working around high-voltage tube equipment, stop here and put the radio in the hands of a technician who is. No alignment touch-up is worth your life.
“I am finding that where the preselector peaks and where the exciter tune gives max grid drive are not the same. This is true on all bands. If RX is peaked the exciter tune must be moved more to the clockwise for grid current peak. I’ve gone through the alignment procedure, but the issue remains.
The second thing has to do with the grid current reading in that at peak the mic gain is at around the first tick on the dial. Around 7 o’clock. On my other M-2’s and on my 32S-3 the peaks are around the third tick about 9 o’clock on the mic gain. I’m wondering if the two issues are related. Should I be concerned over the grid drive? This radio does have signs of previous work.”
— Lenny, WB6QAZ
The short answer: the two symptoms are almost certainly not directly related. The first is a transmit-chain tracking/alignment issue. The second is a gain and drive-level question that, depending on what the meter actually reads, may not be a problem at all. Let’s take them one at a time.
Issue 1 — RX Preselector Peak ≠ TX Grid Drive Peak
The first thing to remember about the KWM-2/2A is that the EXCITER TUNING knob is peaking different combinations of tuned circuits depending on mode. On receive, you’re peaking the RF amplifier and receive mixer circuits. On transmit, that same ganged slug rack is also tuning the transmit mixer and 6CL6 driver stages — circuits that are simply not in the receive signal path.[1]
The fact that Lenny’s offset is consistent in direction on all bands fits that story: a previous tech worked through the bandswitched transmit coils with the same (wrong) reference each time.
The practical fix
⚠ Before You Begin
The steps below are performed on a live chassis with full high voltage present. Re-read the safety warning at the top of this post. Discharge the filter capacitors before any under-chassis inspection, use insulated alignment tools only, keep your free hand clear of the chassis, and always run transmit tests into a 50-ohm dummy load — never an antenna — at reduced drive.
Rather than fighting the book procedure in isolation, anchor the transmit alignment to the receive peak. This is how most experienced Collins techs do it:
Step 1. Peak EXCITER TUNING on receive (band noise or a weak steady signal) at the alignment frequency for the band, and leave the knob exactly there.
Step 2. Without moving the knob, key the radio in LOCK at reduced drive and adjust the transmit mixer and driver coils for maximum grid current — slugs at the low-frequency alignment point, trimmers at the high end, per the instruction book sequence.[2]
Step 3. Repeat per band as required, then verify the RX peak and the grid-current peak now fall together across the dial.
Before trusting any alignment on a radio with prior work, inspect the usual culprits in the slug rack and coil deck: cracked or incorrect slugs, a slug screwed clear through its winding, swapped coils between positions, or a replaced trimmer or padder capacitor of the wrong value in the driver plate circuits. Any one of these will make the alignment chase its own tail.
Issue 2 — Full Grid Drive at 7 O’Clock on the Mic Gain
Here’s the part worth internalizing: the mic gain knob position is not the specification — the actual grid current reading is. If the radio reaches rated 6146 grid current with the knob at the first tick instead of the third, but the peak value is correct, the ALC is functioning, and the transmitted signal is clean, then all it means is that this particular transmit chain has more gain than his other rigs. A strong 6CL6 or 6U8 will do exactly that, and tube-to-tube gain variation in these stages is substantial.
That said, on a radio with previous work, a few things are worth verifying before shrugging it off:
| Check | Why It Matters |
|---|---|
Carrier null / balanced modulator |
If the balanced modulator is poorly nulled, residual carrier is contributing drive before the mic gain ever comes up — making drive appear “hot” at low knob settings and degrading carrier suppression on SSB. Check carrier suppression per the manual and touch up the balance pot and capacitor.[3] |
ALC operation |
Switch the meter to ALC and confirm deflection on speech peaks. A dead ALC line — open resistor, failed diode — lets drive run away and is one of the few ways this symptom becomes a genuine problem. |
Metering circuit resistors |
If a previous repair substituted off-value meter shunt or multiplier resistors, the indicated grid current may simply read high. A quick resistance check against the schematic settles it. |
6CL6 driver circuit components |
Confirm cathode and screen resistors are at correct values, and check for a leaky coupling capacitor into the driver grid — the same fault that cooks 6CL6s in these radios. |
So should Lenny be concerned about the grid drive? Only if the actual grid current exceeds spec, or the ALC isn’t holding it down. Genuinely excessive drive on the 6146s means splatter, IMD, and shortened tube life — that’s worth chasing. But if grid current peaks at the rated value and the ALC behaves, the early knob position is cosmetic. Set drive by the meter, not by where the knob points.
Root causes of incorrect grid current
When the grid current reading is genuinely wrong — not just arriving at an unfamiliar knob position — the cause falls into one of these buckets. The symptom pattern usually tells you which side of the table to start on:
| Symptom | Potential Root Causes |
|---|---|
Grid current too high, or full drive reached at very low mic gain |
• Poor carrier null — residual carrier from an unbalanced modulator contributes drive before the mic gain comes up • Inoperative or misadjusted ALC — open ALC line, failed ALC rectifier (V17A), or ALC ZERO adjustment off, letting drive run unchecked • High-gain tube substitutions in the 6CL6 driver (V8), 6U8 transmit mixer stages, or 6AZ8/6U8 RF amplifier (V7) • Off-value meter shunt or multiplier resistors from previous repairs, making indicated grid current read high • Wrong-value cathode or screen resistors in the driver stage raising stage gain • Leaky coupling capacitor into the 6CL6 grid pulling the grid positive and increasing conduction |
Grid current too low, absent, or band-dependent |
• Mistracked transmit mixer/driver coils — the tracking issue discussed above (low on all bands if consistently off, or band-specific if individual coils are off) • Weak or worn 6CL6 driver, 6U8 mixer tubes, or 6146 finals • Improper PA, driver, or feedback neutralization — note the known KWM-2/2A issue where the original neutralization circuit values are marginal for 6146B and later-production 6146W tubes; check the service bulletins and verify which 6146 variant is installed • Open, cracked, or shorted slug/coil in the transmit chain, or a broken slug rack coupling • Low B+ or screen voltage, or bias misadjusted at the power supply (verify resting plate current at 40 mA first) • Dirty or pitted contacts on the T/R relays (K3/K4) interrupting drive or metering paths • ALC line fault holding gain down — a stuck-on ALC voltage will make grid current sag toward zero after keying |
Grid current unstable, drifting, or sagging after key-down |
• ALC circuit fault (as above) bleeding gain down over seconds • Overheating or drifting resistors in the driver screen/plate circuits • Failing tube developing grid emission or heater-cathode leakage • Intermittent slug rack, coil, or bandswitch contact |
Work the list in order of effort: tube substitution and the carrier null/ALC checks cost minutes; metering resistor verification takes a schematic and an ohmmeter; neutralization and coil work come last because they’re the most invasive — and remember that every one of these checks beyond tube swapping puts your hands near a live or recently-live chassis. Discharge first, every time.
Recommended Test Equipment
Knob positions and ear tests will only take you so far. Worth noting up front: Collins designed the field alignment to be done with almost nothing — the instruction book lists the required equipment as just a 50-ohm, 100-watt dummy load and a receiver with a 100-kc calibrator and S-meter. So Lenny can execute the tracking fix in Issue 1 with the bench he almost certainly already has. The rest of this list is what turns “I think it’s fixed” into “I measured it, and it’s fixed” — here’s what each piece does against his two specific problems:
| Instrument | How It’s Used on Lenny’s Problems | Suggested Equipment |
|---|---|---|
50Ω dummy load (non-negotiable) |
Every transmit-side step in this post happens into this load: peaking grid current during the re-anchored alignment, the carrier null procedure (which couples a receiver to the dummy load per the manual), and all drive-level checks. It also guarantees the PA sees a true 50Ω load, so the grid current and plate dip readings Lenny compares between his radios actually mean the same thing on each. Chasing a tracking offset into a reactive antenna load is chasing a moving target. |
Heathkit Cantenna HN-31, Bird 8201, or any quality 50Ω load rated 100W+ continuous. Verify it actually measures near 50Ω before trusting it. |
RF wattmeter |
The after-the-fix proof for Issue 1. Once the transmit coils are realigned anchored to the receive peak, power output should peak at the same exciter knob position where the receiver peaks, with rated output at 1/3-scale grid current. If Lenny sees full power exactly at the RX peak position on every band, the tracking problem is solved — the wattmeter is what confirms it band by band. |
Bird 43 with appropriate slugs (the classic), or a modern Telewave/Coaxial Dynamics equivalent. Peak-reading capability is a plus for SSB. |
High-impedance voltmeter with HV probe |
Directly attacks Issue 2’s root-cause list on a radio with previous work. Use it to verify the meter shunt and multiplier resistors against the schematic (an off-value metering resistor would make grid current read high — explaining the 7 o’clock symptom with nothing actually wrong in the RF chain), to check driver cathode/screen resistor values, to confirm B+, screen, and bias voltages against the manual’s Table 4-3 voltage chart, and to set resting plate current via BIAS ADJUST. A 20KΩ/V VOM will load these circuits and lie to you; the high-impedance input matters. |
HP-410B or RCA Senior VoltOhmyst VTVM for period authenticity, or any modern 10MΩ-input DMM. Use a rated high-voltage probe for the 800V line — never bare meter leads. |
RF signal generator |
Two jobs. First, it establishes the receive-side anchor precisely: instead of peaking the exciter knob on fuzzy band noise, inject a calibrated signal at the alignment frequency and peak on the S-meter — that knob position becomes the reference the transmit coils get aligned to. Second, it localizes the mistracking and the excess gain: the manual’s Table 4-1 signal-tracing procedure injects known levels at the driver grid (V8), RF amp (V7), and mixer grids (V6, V5) and tells you what drive each stage should need. The stage that produces grid current at well below the table’s listed level is the stage with the extra gain — that’s how Lenny finds out where his 7 o’clock sensitivity is coming from. |
HP 8640B (the gold standard for this work), HP 606A, or a military URM-25. A modern synthesized generator works fine if you have one. |
Oscilloscope with RF sampler |
Answers the “should I be concerned?” half of Issue 2 visually. Watching the envelope at the dummy load shows whether the hot drive is actually doing harm: flat-topped peaks mean the drive/ALC relationship is wrong, clean rounded peaks mean it isn’t. It also shows ALC behavior in real time — the sag-after-keying pattern from a faulty ALC line is unmistakable on a scope and easy to miss on the panel meter. With a calibrated scope it doubles as the measuring instrument for the manual’s signal-tracing voltage levels. |
Any 20MHz+ scope (Tektronix 2213/2235 class or modern digital) with a proper RF sampler or high-impedance probe at the dummy load — never direct connection to the RF line. |
Two-tone audio generator |
The realistic-conditions test once the static checks pass. Driving the mic input with two tones and watching the envelope on the scope shows whether drive level and ALC are right under actual SSB conditions, not just in LOCK. It also exposes the carrier-leak suspect behind Lenny’s early grid current: residual carrier from a poor null shows up as a “fuzzy” or asymmetric two-tone pattern that won’t null out at the crossover points. Clean crossovers = good null = the early mic gain position is just stage gain, not leakage. |
A classic two-tone test set, a PC sound card generating non-harmonically-related tones (e.g., 700Hz and 1900Hz), or a function generator pair. |
Spectrum analyzer or SDR |
The definitive ruling on the carrier-null question. If residual carrier is contributing drive (making grid current appear at 7 o’clock), it shows up directly as a carrier spur on the display — measure suppression before and after touching up the null, and the question is settled with a number instead of a hunch. It also reveals spurious drive components, which are a less common but documented cause of transmit peaks not landing where they should. For Lenny, this is the single instrument that most directly separates “hot tube” from “carrier leak” as the explanation for his symptom. |
A TinySA Ultra or RTL/Airspy/SDRplay SDR with a high-attenuation sampler tap off the dummy load is remarkably capable for the money; a bench analyzer (Rigol DSA815 class or better) if budget allows. Never connect an SDR or analyzer to transmitter output without 40–60dB of proper attenuation — you will destroy the front end instantly. |
Frequency counter |
The rule-out instrument for Issue 1. Before blaming coil tracking, verify the HF crystal oscillator injection and BFO crystals are on frequency during signal tracing — an oscillator running off frequency shifts where the tuned circuits need to peak. Since the KWM-2 shares its oscillators between receive and transmit, a frequency problem can’t fully explain an RX/TX peak split, but confirming the oscillators are correct lets Lenny eliminate that variable in minutes and commit to the alignment fix with confidence. |
Any 8-digit counter good to 30MHz; HP 5328A class vintage units or inexpensive modern counters both serve. |
Tube tester |
The fastest, cheapest answer to Issue 2 — and Lenny is uniquely equipped here because he owns multiple M-2’s. Before any instrument touches the bench: swap the 6CL6 driver (and then the 6U8 mixers) from one of his 9-o’clock radios into this one. If the grid current peak moves from 7 o’clock toward 9, the mystery is solved — it was a hot tube all along, and nothing is wrong. A mutual-conductance tester quantifies the same answer by showing the gain spread between his tubes. This is the first thing to try, because it can close the entire second issue in ten minutes with the power off. |
Hickok 539/600 class mutual-conductance tester preferred; an emissions tester is better than nothing. Substitution with known-good tubes is the most honest test of all. |
The efficient order for Lenny specifically: tube substitution first (power off, ten minutes, may end the whole grid-drive question), then carrier null check with the SDR or receiver coupling, then ALC and metering-resistor verification with the VTVM, and the re-anchored alignment with signal generator and wattmeter to close out the tracking issue. With that done, “is the radio right?” stops being a judgment call: grid current at 1/3 scale in LOCK, rated power into the load at the RX peak knob position, carrier suppression measured and healthy, a clean two-tone envelope on the scope, and supply voltages matching the manual’s charts. That’s a verified KWM-2A.
Bottom Line
Redo the transmit-chain alignment anchored to the receive peak position, verify carrier null and ALC operation, and judge drive by the grid current reading rather than the mic gain knob. Two puzzles, two separate causes — and both very fixable on a round-emblem rig that’s clearly worth the effort. And once more, because it bears repeating: respect the high voltage, discharge those filter caps, and work as if your life depends on it — because it does.
Resources
The Collins Collectors Association hosts an excellent archive of KWM-2/2A documentation, including the 9th edition instruction book with the full alignment procedure: CCA Collins Radio Equipment Manuals. A complete scanned KWM-2/2A instruction book is also available at World Radio History, and the BAMA boatanchor manual archive remains a reliable mirror. For Collins-specific service knowledge, the service bulletins and technical bulletins in the CCA Technical Archives are well worth bookmarking.
- In the KWM-2/2A, the receive path through the shared tuned circuits uses the RF amplifier and receive mixer; the transmit path adds the transmit mixer and 6CL6 driver stages on the same ganged slug rack. The two modes therefore peak different circuit combinations on one knob. ↩
- See the transmitter alignment section of the Collins KWM-2/2A Instruction Book (9th edition, Collins Radio Company), available via the CCA manuals archive. Follow the slugs-at-low-end, trimmers-at-high-end convention throughout. ↩
- Carrier suppression adjustment is covered in the same instruction book; both the resistive balance control and the balance capacitor interact, so work them alternately for deepest null. ↩
- Collins KWM-2/2A Instruction Book, Section 2 (Operation), Transmitter Tuning procedure: with EMISSION in LOCK, “advance MIC GAIN to provide a grid current reading of approximately 1/3 scale,” then dip plate current to 230 mA (200 mA with a linear amplifier). Resting plate current of approximately 40 mA is set per the Initial Checks procedure in Section 1. See the complete instruction book. ↩