Introduction

The Collins R-390A/URR is a receiver built to survive decades of hard military service, and by almost every measure it has succeeded. Yet even the most rugged designs contain components that time eventually defeats. The 26Z5W — a half-wave rectifier tube used in pairs within the PP-621/URR power supply — has become one of the receiver’s quiet vulnerabilities. New-old-stock examples are increasingly difficult to source, and the tubes that do surface vary widely in quality and provenance.^[1]

A well-engineered solid-state replacement removes this uncertainty entirely. Silicon rectifiers are essentially indestructible under normal operating conditions, generate no heater current demand, and — if the design is handled correctly — can be installed reversibly without any trace of modification to the original receiver chassis. This article covers the circuit context, the critical issue of B+ voltage rise, heater string compensation, recommended components, and a step-by-step construction approach suitable for any restorer with basic bench skills.

The 26Z5W in the R-390A Power Supply

The PP-621/URR power supply generates the receiver’s primary B+ rail using a conventional full-wave centre-tap rectifier. Two 26Z5W tubes — one on each half of the secondary winding — conduct on alternate half-cycles of the 115 V AC mains input, which is stepped up and centre-tapped at the power transformer secondary.^[2]

Each 26Z5W is an indirectly heated half-wave rectifier in a compact octal envelope. Key published ratings are:^[3]

Under normal load the B+ rail in the PP-621/URR measures approximately 225–235 V DC at the main filter capacitors. This voltage feeds the IF amplifier chain, the BFO, and the first RF stage through the regulated secondary supplies. It is not a tightly regulated rail — the original design accepts a moderate tolerance band — but sustained over-voltage shortens capacitor life and increases dissipation in downstream regulators.

Why Replace? Failure Modes and Supply Concerns

The 26Z5W was produced in quantity through the 1960s and early 1970s for military and commercial use, but production ended long ago. The tubes now encountered in the surplus market fall into three broad categories: genuine JAN-spec NOS in original packaging, pulled tubes of unknown history, and counterfeit or relabelled commercial types presented as military grade. The last category is increasingly common as genuine stock shrinks.^[4]

Common failure modes observed in the field include:

• Open heater: The tube passes no current; the corresponding half-cycle is lost, producing severe B+ ripple and half-wave hum on received audio. Often mistaken for a filter capacitor failure on first inspection.
• Cathode-to-heater short: Injects heater AC directly onto the B+ rail. Creates a characteristic 60 Hz buzz and can damage downstream components if not caught quickly.
• High internal resistance (weak emission): B+ sags under load; the receiver loses sensitivity and the automatic gain control misbehaves. Can be subtle and difficult to diagnose without a tube tester capable of measuring rectifiers under realistic load current.
• Leaky plate-to-cathode: Results in elevated B+ at idle that collapses under load — the exact inverse of weak emission behaviour.

Given that a well-maintained R-390A will outlast any available tube stock, replacing the 26Z5W pair with silicon is straightforward preventive engineering. The restoration community — including the r-390a.net reflector and the Collins Collectors Association — has broadly accepted this modification as both reversible and sound.^[5]

The Critical Issue: B+ Voltage Rise

This is the single design element that separates a successful conversion from a harmful one. A silicon diode exhibits a forward voltage drop of approximately 0.7 V at typical rectifier currents — compared to the 26Z5W’s 12–18 V drop. Removing the tube drop without compensation raises the B+ rail by roughly 20–30 V depending on actual load current.^[6]

Compensation is straightforward. Two approaches are used in the restoration community:

Approach A — Series Dropping Resistor (Recommended)

A non-inductive wire-wound resistor placed in series with the B+ output — after the rectifiers but before the first filter capacitor — drops the excess voltage. The value is calculated from the voltage differential and the load current:

A value between 150 Ω and 220 Ω will cover the range of actual B+ rise encountered across different PP-621/URR units. Measure your B+ after installation at normal operating temperature and trim if necessary. The target is 225–232 V DC under load. Suitable parts include Ohmite 10 W wire-wound chassis-mount resistors (series L or 20J) or equivalent Dale/Vishay types — all readily available from Mouser or Digi-Key.

Approach B — Inrush Limiting with NTC Thermistor

A Ametherm (formerly ICL) SL22 10003 or equivalent NTC power thermistor (10 Ω cold, drops to approximately 1–2 Ω at operating temperature) provides a degree of inrush protection at power-on while contributing minimal steady-state drop. This does not adequately compensate the full voltage rise on its own and should be paired with a fixed dropping resistor of reduced value (~100 Ω) if used. It adds complexity without proportionate benefit for the R-390A specifically, since the PP-621/URR already has adequate filter capacitance for a controlled start. Approach A alone is preferred.^[7]

Heater String Compensation

The R-390A’s tube complement is heated from a series string running at 26 V AC supplied by a dedicated winding on the power transformer. The two 26Z5W tubes each draw 0.30 A, for a combined contribution of 0.60 A to the string. When the tubes are removed, this load disappears — but the heater winding was designed to drive the complete string, and the transformer’s secondary regulation means that removing load will raise the string voltage.^[8]

In practice, the impact is modest but should not be ignored. The remaining tubes in the heater string will see a slightly elevated filament voltage, which shortens their life over long periods. Compensation options:

• Dummy load resistors (recommended): Two resistors, each matching the 26Z5W heater impedance (26 V ÷ 0.30 A = 86.7 Ω), wired between the heater string pins on the octal replacement sockets. Use 5 W wire-wound non-inductive types. Standard value: 82 Ω, 5 W per socket. These can be mounted directly inside the octal plug body or on a small tagboard clipped to the socket.
• Measure and accept: Many experienced restorers measure the heater string voltage before and after and accept the result if the rise is less than 0.5 V. At 26.5 V the remaining tubes are well within tolerance. This is a defensible position for receivers used intermittently.

Component Selection

Rectifier Diodes

The 1N4007 is the correct choice. It is rated 1000 V PIV and 1.0 A average forward current — both comfortably exceeding the 26Z5W’s operating envelope in this circuit. Its 1000 V PIV rating provides a large margin against peak inverse voltage spikes. Ultra-fast recovery types (UF4007 and similar) offer no benefit at 60 Hz line frequency and should not be used in preference — the 1N4007’s switching characteristics are entirely appropriate.^[9]

Install two diodes per socket position — stacked in series — to provide a combined PIV of 2000 V. This is belt-and-suspenders engineering that adds measurable protection against the voltage transients that appear on aged power transformer secondaries at switch-on. Add a 0.01 µF / 1 kV ceramic disc capacitor across each diode pair to suppress high-frequency transients that the 1N4007’s junction capacitance does not catch.

RFI Considerations

This section is not optional reading. It addresses a failure mode that is invisible on a DC voltmeter and easy to miss on initial bench testing, but which can measurably degrade the R-390A’s noise floor — particularly on the lower HF bands where the receiver’s performance is most impressive and where the interference mechanism is most effective.

The Mechanism: Reverse Recovery Transients

The 26Z5W rectifies by thermionic emission — a fundamentally smooth process. As the anode voltage swings negative at the end of each conduction half-cycle, the tube simply stops conducting. There is no stored charge to recover, no switching edge, and no transient. The transition from conduction to cutoff is governed by the AC waveform itself.^[12]

A silicon p-n junction diode behaves differently. During forward conduction, minority carriers are injected across the junction and stored in the bulk semiconductor. When the voltage reverses, these carriers must be swept out before the junction can support reverse voltage — a process known as reverse recovery. During this brief interval (the reverse recovery time, t_rr) the diode conducts in the reverse direction, drawing a pulse of current that terminates abruptly when the stored charge is exhausted. This abrupt termination — the “snap” — drives a high-frequency current transient into whatever impedance it sees.^[13]

The 1N4007’s reverse recovery time is specified at approximately 30 µs maximum. At 60 Hz line frequency the diode is switching twice per cycle (once per half-wave), and the recovered charge per transition is modest — but the transient’s spectral content extends well into the RF spectrum. The PP-621/URR’s B+ filter capacitors attenuate low-frequency ripple effectively; they are far less effective against the sharp high-frequency content of a snapping diode.

Coupling Paths into the Receiver

The noise generated at the rectifier junction finds three primary routes into the receiver’s sensitive circuits:

• B+ rail conduction. The transient propagates along the B+ wiring from the rectifier octal sockets to the filter capacitors, and from there through the B+ distribution network into the IF amplifier decoupling chain. The R-390A’s IF stages are decoupled individually, but the decoupling capacitors are sized for audio-frequency ripple. Their impedance at 1–30 MHz is finite and non-trivial.
• Mains-conducted emission. The transformer secondary and primary windings are coupled capacitively. Transient current on the secondary side is partially reflected onto the mains wiring. If the receiver’s AC feed shares a ground path or common-mode coupling with the antenna system — as is often the case in a single-operator station — this conducted emission can re-enter via the antenna port. The effect is most pronounced on 160 and 80 metres.
• Magnetic radiation from the power transformer. The transformer’s core is already a source of low-level 60 Hz and harmonic field radiation. The addition of asymmetric transient currents on the secondary modulates this radiation envelope. Proximity to the first RF stage is a risk factor — in the PP-621/URR, the power transformer is well separated from the RF deck, which mitigates but does not eliminate this path.

Suppression Measures — Implemented in the Plug Module

All three suppression elements are fitted inside or immediately behind the octal plug body, as close as physically possible to the diode junction. Distance matters: every centimetre of wire between the diode snap and a suppression component is an unterminated transmission line segment at HF frequencies.

1. RC Snubber across the diode pair

A series combination of 100 Ω (1 W carbon film) and 10 nF (1 kV ceramic disc) is connected directly across the diode pair — pin 2 to pin 6 — in parallel with the existing 0.01 µF bypass capacitor. The resistor damps the resonance between the diode junction capacitance and the transformer leakage inductance that would otherwise ring at a frequency determined by those parasitics. The capacitor provides a low-impedance path for the high-frequency component of the switching transient. Together they convert the sharp snap into a controlled exponential decay.^[14]

2. Ferrite bead on the B+ output lead (pin 6)

A Fair-Rite 2643625002 bead (0.25 inch OD, #43 nickel-zinc material) is threaded onto the output wire between pin 6 of the octal plug and the B+ filter capacitor. #43 material has its peak impedance in the 25–300 MHz range, providing resistive (lossy) suppression across the HF spectrum. At 10 MHz the bead contributes approximately 100 Ω of series impedance with negligible reactance — it absorbs rather than reflects, which avoids creating a resonant stub. DC resistance is less than 0.05 Ω; the voltage drop at 150 mA is under 8 mV — operationally irrelevant.^[15]

3. AC mains filtering

The conversion should be paired with a quality mains line filter at the station AC entry point. A filter with common-mode and differential-mode attenuation rated at ≥40 dB across 150 kHz–30 MHz (IEC CISPR 16 Class B equivalent or better) will substantially eliminate the conducted emissions path. Schaffner FN2010, Corcom 6EQ1, or equivalent single-phase EMI filter modules are appropriate. If the station already uses a filtered distribution strip or a Timewave ANC series noise canceller, verify that its filter stage covers the frequency range of concern.

Noise Floor Verification Procedure

Perform this check before and after the conversion to confirm that noise floor has not degraded. The R-390A’s own S-meter and the calibrated attenuator provide a convenient self-contained measurement.

1. Terminate the antenna input with a 50 Ω BNC termination (a good-quality non-inductive type; carbon composition resistors are preferred over wire-wound for this purpose).
2. Allow the receiver to warm up fully — minimum 30 minutes with the original tubes; silicon rectifiers reach thermal equilibrium in under 5 minutes but the remaining tube complement still needs time.
3. Set the receiver to USB mode, 4 kHz bandwidth, RF gain fully clockwise, audio gain to mid-position. Note the S-meter reading and listen for any 60 Hz or 120 Hz related hum, buzzing, or broadband noise increase.
4. Repeat on each band segment from 0.5 MHz through 30 MHz. Pay particular attention to 1.8 MHz (160 m) and 3.5 MHz (80 m) where low-level conducted noise is most audible.
5. Any increase in S-meter reading above the pre-conversion baseline, or any new audio artefact correlated with the mains frequency or its harmonics, indicates insufficient RFI suppression. Verify snubber and ferrite bead installation before investigating further.

Construction: The Octal Plug-In Module

The goal is a self-contained plug-in module that is indistinguishable from the outside of the receiver from a standard tube installation, and fully reversible by simply pulling the plugs and reinserting the original tubes. All components are fitted inside or directly behind the octal plug body.

Octal Pin Assignment (26Z5W)

Assembly Sequence

1. Obtain two Amphenol 86CP8 octal plug bodies (or equivalent ceramic-body octal connectors). Identify pin 2 (plate input) and pin 6 (cathode output) on each.
2. Solder two 1N4007 diodes in series, anode-to-cathode, between pin 2 and pin 6. Orient so that the cathode end (banded) connects to pin 6. Insulate with heat-shrink tubing — 3 mm over individual leads, 6 mm over the assembly.
3. Solder a 0.01 µF / 1 kV ceramic capacitor across the diode pair (pin 2 to pin 6). Keep leads short.
4. Solder the 82 Ω, 5 W resistor between pins 7 and 8. This resistor will run warm — ensure clearance from the heat-shrink on the diode assembly. Wire-wound types are adequate; non-inductive types are ideal.
5. Repeat for the second plug.
6. Mount the 200 Ω / 10 W dropping resistor on the power supply chassis near an existing standoff or terminal strip — not inside the octal plug. Run a short lead from the B+ filter capacitor positive terminal, through the dropping resistor, to the existing B+ distribution point. Use high-temperature wire (PTFE insulated, 600 V rated).

Installation and Verification

Before installation, verify your work at the bench with a variable AC supply and a current-limited B+ test load (a 2 kΩ, 50 W resistor approximates the R-390A B+ load reasonably well). Confirm rectification is occurring on both plug assemblies and that there are no shorts between pin 2 and pin 6 in the reverse direction.

Installation procedure in the receiver:

1. Remove the original 26Z5W tubes and set aside.
2. Insert the two solid-state plug modules into the vacated sockets. The key on the octal plug aligns with pin 1 — ensure correct orientation.
3. Install the B+ dropping resistor on the power supply chassis. Allow 25 mm clearance from adjacent components and wiring; it will run at approximately 60–80 °C under load.
4. Power up through a Variac or series lamp limiter. Bring AC mains voltage up slowly to 115 V while monitoring B+.
5. Measure B+ under normal operating conditions (receiver fully warmed up, receiving a signal). Target: 225–232 V DC. If B+ is high, increase the dropping resistor value in 22 Ω increments.
6. Measure the 26 V heater string with both plug modules installed. Nominal is 26.0 V; accept up to 26.8 V. If the rise is greater, verify that both 82 Ω dummy load resistors are properly connected.

Commercial Drop-In Alternatives

Several cottage-industry suppliers have produced plug-in solid-state replacements for the 26Z5W over the years. Availability varies — always verify current stock before specifying any of these in a restoration plan:

• Tubular Electronics (tubularelectronics.com): Has offered a range of solid-state rectifier replacements in octal form factors. Quality is generally well-regarded on the reflectors, but verify that the 26Z5W-specific version includes voltage compensation, as not all offerings do.^[10]
• Antique Radio Supply (antiqueradiosupply.com): Has stocked solid-state rectifier adapters at various times. Check current catalogue.
• Hayseed Hamfest (hayseedhamfest.com): Known for well-documented replacement modules for vintage receivers. The DIY approach above gives equivalent or superior results at lower cost, but their pre-built units are a legitimate option for restorers who prefer not to construct their own.

Expected Results and Long-Term Outlook

A correctly installed conversion produces a B+ rail that is more stable than the original tube implementation. Silicon diodes have no warm-up drift, no emission decay curve, and no cathode poisoning mechanism. The B+ will be at its operating voltage within the first mains cycle after switch-on — which actually requires some care. The original tube rectifiers provided a natural soft-start as the cathodes warmed; silicon does not. This makes it important that the filter capacitors are in good condition before this conversion is carried out. If the capacitor bank has not been reformed or replaced, do that first.^[11]

Ripple voltage on the B+ rail should measure below 5 mV RMS with good filter capacitors in circuit. Hum on received audio at S0 background noise — with the antenna input terminated — should be inaudible. If hum is present after the conversion, the filter capacitors are the first suspect, not the rectifier modules.

Reversibility and Originality

The plug-in construction approach means that the receiver’s original wiring is untouched. The only semi-permanent change is the addition of the B+ dropping resistor, which is mounted on a terminal strip or standoff — not soldered into the original circuit. A future owner wishing to restore full tube operation need only:

1. Remove the two solid-state plug modules
2. Remove the dropping resistor and its associated wiring
3. Insert original 26Z5W tubes

No desoldering of original wiring is required. The conversion is fully transparent to any subsequent collector or restorer.

Conclusion

The 26Z5W solid-state conversion is one of the most justifiable modifications available to the R-390A restorer. It eliminates a genuine obsolescence risk, improves B+ stability, reduces the thermal burden inside the power supply, and can be executed with basic workshop skills and a handful of readily available components. Done correctly — with proper B+ compensation and heater string dummy loads — it is invisible to the receiver’s operating characteristics and entirely reversible.

The R-390A was designed to work reliably for decades. With this conversion in place, the power supply’s most vulnerable component is no longer the limiting factor in that timeline.

---