The Collins R‑389, R‑390 & R‑390A

For seventy years the Collins R‑390A has worn the unofficial title of “the best vacuum-tube shortwave receiver ever built.” That reputation is earned, not nostalgic. This article traces the R‑389/R‑390/R‑390A family from its 1949 Signal Corps origins through three decades of production, then turns to what matters most to a listener or ham who wants to actually own and operate one today: which version to buy, how to drive it, what breaks, and how to bolt a modern SDR panadapter onto a 1950s boatanchor.

1. Why these receivers still matter to operators

The R‑390A is not a museum piece you switch on twice a year. On a quiet rural antenna it will out-listen a great many modern receivers on weak-signal AM and CW, and its mechanical filters give a passband shape factor approaching 2:1 — selectivity that crystal-filter designers spent two decades chasing.¹ Three things make it special for the listener:

• Frequency readout you can trust. The Veeder-Root mechanical counter reads directly to 100 Hz — an odometer-style digital display decades before the digital age — and a calibrated unit lands within a few hundred Hz of a logged frequency. No guessing where a slide-rule dial “really” points.
• Selectivity by mechanical filter. The R‑390A’s 2, 4, 8 and 16 kHz mechanical filters are the heart of its weak-signal performance, letting you dig a signal out of a crowded broadcast band that would swamp a wide-open receiver.
• A 455 kHz IF output on the back panel. This single connector — designed in 1950 to feed military RTTY converters — is the reason a 1950s receiver can drive a 2020s SDR panadapter. More on that in §11.

The trade-offs are real and worth stating up front: it weighs about 85 lb (39 kg), draws roughly 200–250 W with the crystal/PTO ovens warm, runs hot, and demands periodic maintenance. This is a receiver you commit to, not one you casually shelve.

2. Origins & the Signal Corps specification (1949–1953)

In 1949 the U.S. Army Signal Corps Engineering Laboratories at Fort Monmouth, New Jersey, placed development contract W36‑039‑sc‑44552 with Collins Radio Company of Cedar Rapids, Iowa. The work fell under Signal Corps specification SCL‑1134‑B, Department of the Army Project No. 3‑24‑01‑051, and Signal Corps Project No. 15‑805G‑2.² The objective was two receivers within a planned suite spanning 15 kHz to 3,000 MHz:

• R‑389/URR: 15 kHz – 1,500 kHz (VLF/LF/MW)
• R‑390/URR: 500 kHz – 32 MHz (HF general coverage)

Both were required to copy AM, CW, FSK and MCW. The contract document, classified at the time, carries the standard Espionage-Act caveat — a reminder that this was Cold War strategic-intercept equipment, not a consumer radio.² The project engineer at Collins was L. W. (Lou) Couillard, who signed the Final Engineering Report, CER No. 297. The development effort consumed roughly 30,000 engineering man-hours.³

From the 51J series

Collins did not start from a blank sheet. The 51J‑series commercial receivers — the 51J‑4 entered military service as the R‑388/URR — had already proven the company’s permeability-tuned oscillator (PTO) architecture. Signal Corps evaluation found the 51J approach sound but lacking in spurious rejection, ruggedness, modularity and the dynamic range needed in dense military RF environments. The R‑390 kept the best 51J ideas and re-engineered the rest:

• Permeability-Tuned Oscillator (PTO). A moving ferrite slug varies inductance for linear, backlash-free tuning. The R‑390’s oven-stabilised PTO holds drift to roughly 100 Hz across a wide temperature range after warm-up.
• Veeder-Root mechanical counter. Direct digital readout to 100 Hz — the feature every operator notices first.
• Crystal-referenced injection. A 1 MHz calibration oscillator provides 100 kHz check points so you can re-zero the dial against a known marker.
• Mechanical filters (R‑390A). See the correction note below — these are metal-disc mechanical resonators, not quartz.

Conversion architecture

The R‑390 frequency plan splits at 8 MHz:³

• Above 8 MHz: double conversion, crystal-controlled first injection, final IF at 455 kHz.
• Below 8 MHz: triple conversion — an extra variable first-IF stage (~17–25 MHz) keeps images and spurious responses outside the tuning range where they would otherwise fall in-band.

The R‑389 used a comparable double-conversion scheme down to 15 kHz, with the same crystal oscillator participating in both conversions out of phase so internal drift largely cancels — a Wadley-loop-like trick that leaves only the PTO contribution. Crucially, both receivers share the same 455 kHz final IF and several common modules (IF, audio/regulator, power supply), a deliberate Signal Corps logistics requirement.

Engineering-model delivery schedule

Both designs exceeded requirements, and production contracts were awarded before the engineering program was even complete — Collins reportedly took an order for some 3,000 receivers by the end of 1950.⁵

3. Production & variants (1951–1984)

R-390 and R-389 (1951–1953)

Collins received a production contract for both receivers, with Motorola directed as a subcontractor for a large quantity of R‑390s (data plates read “Manufactured by Motorola Inc. for Collins Radio Co.”). Only Collins built the R‑389 and the autotune R‑391; Motorola built only the R‑390. Total R‑390 production is usually given as roughly 15,000–16,000 units; R‑389 production was just 856, which is why a working R‑389 is one of the rarest Collins military artifacts.⁶ A first-contract R‑390 cost the government roughly $2,500 — on the order of $30,000 in today’s money.

The cost-reduction study → R-390A (1954–1955)

The R‑390 had done its job so well that in 1954 the Signal Corps asked Collins for a cheaper version that met the same specification.⁷ Collins identified per-unit savings (the figure usually cited is about $233) by:⁵

• Replacing the R‑390’s six-stage LC/crystal IF strip with a four-stage mechanical-filter IF strip, eliminating the expensive precision-wound IF transformers;
• Reducing the tube count and simplifying the RF deck to a single RF amplifier stage while keeping an acceptable noise figure across most of the range.

The R‑390A kept identical coverage (0.5–32 MHz in 32 one-MHz bands), the same PTO/crystal injection, the same Veeder-Root readout, and the same module-bay chassis. Production acceptance was governed by MIL‑R‑13947, revised to MIL‑R‑13947B effective 26 October 1960 (with later amendments and notices).⁸ Its four mechanical filters (nominally 2, 4, 8 and 16 kHz at −6 dB) reach shape factors approaching 2:1 on the 2 kHz position — performance no contemporary LC design could match.

R-390A production by contractor

R‑390A production ran from 1955 through about 1970, then briefly restarted in 1984. The two community datasets differ slightly: the navy-radio.com table records high serial numbers, while Chuck Rippel’s often-cited summary gives approximate delivered quantities. Both put total production above 55,000 across roughly nine prime contractors.¹⁰

Variants worth knowing

• 1967 EAC contract — Teflon-insulated harnesses, IERC black tube shields (better cooling), improved meters and paint. The collector’s favourite.
• Microdial BFO — fitted for RTTY-intercept duty, extending BFO range and enabling dual-FSK copy.
• Black-panel “CIA/NSA” sets — black-anodised front panels associated with SIGINT sites such as Clark AB; the finish’s exact significance is not firmly documented and the “rare spy radio” framing in sale listings should be treated with caution.
• R‑725/URR — an EAC R‑390A re-fitted by Arvin with a Series 500 (R‑390-style LC) IF module for the AN/TRD‑15 DF set.
• R‑1247 — Navy SHIPALT designation for R‑390As in shipboard SSB systems.
• Fowler Industries (1984) — five units under Navy contract N00024‑84‑C‑2027 at a reported $34,000–$38,000 each; reportedly a Helena Rubinstein subsidiary. The final data point of a 30-year run.

4. Performance benchmarks vs. the 51J series

The table below consolidates the CER No. 297 development data, MIL‑R‑13947B production limits, the 51J‑4 spec sheet, and community measurements. Where development-era and production figures differ, both are noted.

No production tube-era receiver surpassed the R‑390A’s combination of frequency accuracy, selectivity and dynamic range — which is exactly why the military kept stocks of it long after solid-state successors arrived.

5. What it did in service

From the mid-1950s through the late 1960s the R‑390 family was the backbone of U.S. strategic HF signals intelligence. The Army Security Agency, Air Force Security Service and Naval Security Group equipped fixed intercept stations with banks of R‑390As — often dozens per operations building — fed from rhombic, Beverage and broadband antennas through multicouplers.¹¹ The same 455 kHz IF output that today drives an SDR then drove FSK converters and diversity selectors, letting one receiver serve many signal-processing chains.

Purpose-built diversity systems (AN/FRR‑34, AN/MRR‑8, AN/FRR‑39 and others) used banks of R‑390/R‑391 receivers. For voice and point-to-point work, the rear IF fed CV‑591/591A SSB demodulators as single sideband displaced AM. Afloat, R‑390As (Navy R‑1247 configuration) served as primary HF receivers, their balanced 125 Ω input cutting common-mode noise in the electrically filthy environment of a warship.

The receivers turned up wherever the U.S. listened: Vint Hill Farms (Virginia), Two Rock Ranch (California), Kagnew Station (Eritrea), San Vito (Italy), RAF Chicksands (UK), Misawa (Japan) and Clark AB (Philippines), many feeding huge AN/FLR‑9 “Wullenweber” circular antenna arrays.¹¹ When sets were finally surplused — many through St. Julian’s Creek Annex in Virginia — they received blue paint stripes on the front panel to mark them as disposed of, the origin of the collector term “blue striper.”¹²

6. For the operator: which one should you own?

If you are buying your first set, buy an R‑390A. It is the most common, the best supported, and the easiest to keep running. The non-A R‑390 is a wonderful receiver but its LC IF transformers are scarce and expensive; choose it only if you specifically want that flavour or find a clean example. The R‑389 is for the dedicated longwave specialist with deep pockets — treat it as a collector item, not a daily driver.

7. Operating guide — driving the R-390A

The front panel intimidates newcomers but the logic is straightforward once you map the controls.

Tuning and bandspread

The left counter sets the MHz band (32 one-MHz segments); the main tuning knob moves you across that megahertz with the kHz digits reading directly. The KILOCYCLE CHANGE and MEGACYCLE CHANGE knobs are mechanically coupled to the slug racks — turn them gently and never force a band change. Use the built-in crystal calibrator (CAL position) against the nearest 100 kHz marker to zero the dial before logging a frequency.

Bandwidth and AGC

Start in the 4 kHz filter for general listening; drop to 2 kHz to fight adjacent-channel interference on crowded bands or for CW, and open to 8 or 16 kHz for relaxed broadcast-band AM fidelity. Use MEDIUM AGC for most work; SLOW for steady carriers and SSB, and FAST only for fast-fading signals. The carrier-level meter is your tuning ally — peak it.

CW and SSB

For CW, switch the function to CW, enable the BFO, and use the 2 kHz (or narrower, if fitted) filter. For SSB, the stock BFO works but is fiddly: the R‑390A was built for AM/CW/FSK, and clean single-sideband copy benefits enormously from an external product detector or a period CV‑591A SSB converter fed from the rear IF jack. Set the BFO to zero-beat a known carrier, then offset for the sideband in use. Run the RF gain high and the AF gain to taste; back off RF gain on very strong signals rather than riding the AF.

Audio outputs

Two audio paths exist: the LOCAL 600 Ω output (front-panel phone jack / speaker) and a LINE 600 Ω balanced output with its own level control on the rear — handy for feeding a recorder, a decoder, or a sound-card interface for digital modes. The line output is transformer-isolated, which helps avoid ground loops.

8. Living with a boatanchor — common faults & restoration essentials

These receivers are robust but they are 50–70 years old. The community has documented every failure mode in exhaustive detail; the items below are the ones that catch nearly every new owner. Work from the Pearls of Wisdom and Chuck Rippel’s owner’s-manual material before you start.

• Hardened PTO grease. The single most common complaint: stiff or seized tuning. The PTO end-point grease dries out; the cure is a careful clean-and-re-lube of the PTO and the slug-rack lubrication points. Do not over-grease.
• Killer capacitors and the 26Z5W. Aging paper/electrolytic capacitors and weak 26Z5W rectifiers can take out the power supply. Many restorers fit solid-state rectifier replacements and re-cap the supply early. Inspect for prior over-voltage damage to the audio and IF decks.
• 3TF7 ballast tube. The current-regulator ballast is a wear item; modern solid-state drop-in replacements are readily available and remove a failure point.
• Meter and dial lamps. Dead panel illumination is near-universal and trivial to fix.
• Slipping/notchy gear train. Worn gears, a loose oldham coupler, or a maladjusted zero-adjust make the counter lag the actual frequency. Mechanical, not electrical — budget time.
• Crystal calibrator and BFO drift. Usually a tube or alignment issue; both are well covered in the alignment notes.
• IF/RF alignment. A proper alignment per the TM restores the last few dB of sensitivity and the textbook filter shapes. Do it last, after the set is electrically sound.

9. The mechanical-filter legacy

The R‑390A’s production-scale use of metal-disc mechanical filters — with ~2:1 shape factors at 2 kHz — set the selectivity benchmark that crystal-filter makers chased for two decades. It was the first military communications receiver to show that near-rectangular passband selectivity was achievable in production hardware, directly influencing later Collins designs such as the 51S‑1 and R‑1051.¹ Those same Collins mechanical filters live on in the amateur 75A‑4 and S‑Line receivers, which is why a well-aligned R‑390A still holds its own against far newer equipment on a crowded band.

10. The R-389 and the new LF/MF amateur bands

Here is a point rarely made in the technical write-ups but directly relevant to today’s ham: the R‑389’s 15–1,500 kHz coverage makes it the family member built for the 2200 m (135.7–137.8 kHz) and 630 m (472–479 kHz) amateur allocations. The R‑390 and R‑390A start at 500 kHz, so the 630 m band sits just below their tuning range — and 2200 m is far out of reach. If you are chasing WSPR beacons or weak-signal CW on the LF/MF bands, or DXing the longwave broadcast band, the R‑389 is the natural (if rare and pricey) Collins choice. Its ±50 Hz calibration accuracy below 950 kHz is remarkable for a 1950s LF receiver and is genuinely useful for narrow-band digital work.

For the vast majority of operators who will never find an R‑389, the practical answer is to tap an R‑390A’s IF or pair any of these receivers with a modern LF-capable SDR — which brings us to integration.

11. Modern integration — SDR panadapters & solid-state upgrades

The R‑390A’s 1950 design decision to bring out a 455 kHz IF on the rear panel is a gift to the modern operator. That port lets you add a panadapter — a spectrum/waterfall display — without modifying the receiver: tap the IF, feed a cheap SDR (an RTL‑SDR with a downconverter, or a direct-sampling SDR that reaches 455 kHz), and run SDR software for a visual band scope alongside the analog set. You keep the R‑390A’s superb front end and mechanical filters while gaining a waterfall to spot signals before you tune them.

Done with care, the result is the best of both worlds: a tube receiver that sounds and performs like a Collins, with a 21st-century waterfall riding on top. Several how-to write-ups for exactly this conversion live in the resources below and on this site’s SDR-integration articles.

12. Documentation & resources

The R‑390 community is one of the best-documented in all of vintage radio. Start here:

• r-390a.net — the definitive community reference: the Pearls of Wisdom (Wei-i Li, ed.), Chuck Rippel’s owner’s-manual material, IF/RF deck notes, sensitivity & alignment, systems FAQ, and Tom Marcotte N5OFF’s contract/manufacturer data.
• Radio Boulevard (Henry Rogers) — the most comprehensive multi-part R‑390A and R‑389 rebuild logs on the open web.
• navy-radio.com (Nick England) — Navy NAVSHIPS manuals, field changes, and the NAVSHIPS 0967-063-2010 manual.
• Primary engineering source: Collins CER No. 297, the Final Engineering Report on the R‑389/R‑390 (Sept 1953), donated to the CCA archive by Bill Hawkins.
• Technical manuals: TM 11-856 (R‑390/URR); TM 11-5820-358-35 (R‑390A field/depot maintenance); TM 11-855 (R‑389/URR).
• Specification: MIL‑R‑13947B, catalogued at r-390a.net references (copies courtesy Tom Marcotte N5OFF and Craig McCartney).
• Museums: the National Cryptologic Museum (Fort Meade, MD) and the U.S. Army Communications-Electronics collection (legacy Fort Monmouth holdings, now at Aberdeen Proving Ground).

Footnotes

1. Mechanical-filter shape-factor and legacy claims: r-390a.net Historical Notes (Pearls of Wisdom), r-390a.net/Pearls/history.pdf. ↩ ↩
2. Contract W36-039-sc-44552, SCL-1134-B, DA Project 3-24-01-051, Signal Corps Project 15-805G-2, and the classified caveat: Collins CER No. 297 cover pages, collinsradio.org; corroborated in the CQ “Refurbishing a Collins R-390-A/URR” series. ↩ ↩
3. Project engineer L. W. Couillard, ~30,000 engineering man-hours, and the 8 MHz double/triple-conversion split: CER No. 297. ↩ ↩
4. Mechanical-filter construction (metal nickel-alloy discs, coupling wires, magnetostrictive/ferrite transducers): Rockwell/Collins mechanical-filter technical description, wa3key.com/filters.html; corroborated by Collins filter datasheets and U.S. patents on disc-resonator/ferrite-transducer filters. ↩
5. Cost-reduction analysis (~$233/unit) and early ~3,000-unit order: “Refurbishing a Collins R-390-A/URR Series Receiver,” CQ Amateur Radio, and the radiomanual.info reprint. ↩ ↩
6. R-390 production ~15,000–16,000; R-389 production 856: Radio Boulevard and R-390A Rebuild Part 2. ↩
7. R-390A project undertaken 1954 as a reduced-cost replacement: R-390A, Wikipedia, citing Chuck Rippel. ↩
8. MIL-R-13947B effective 26 Oct 1960, with Amendment 7 (1966), Notice 1 (1986), Notice 2 (1996): r-390a.net references. ↩
9. Tube-count discrepancy: r-390a.net Tubes FAQ (26 tubes, nine types) vs. Wikipedia’s envelope enumeration (23). The 3TF7 is a current-regulator “ballast” tube. ↩ ↩
10. Total >55,000; per-contractor figures (Rippel summary): Wikipedia and high-serial data at navy-radio.com; “nearly 7,000” Collins units per The Arthur Collins Story. ↩
11. SIGINT roles, diversity systems, intercept sites and CDAA integration: r-390a.net systems FAQ; site histories at Vint Hill Farms (Wikipedia) and the Nautilus Institute Japan SIGINT report. ↩ ↩
12. “Blue striper” surplus marking and St. Julian’s Creek Annex; EMP-survivability retention folklore: Radio Boulevard R-390A Rebuild Part 2; The Arthur Collins Story. ↩