FT707 Repair – M0ICR – Radio and Electronics

Status at a glance

Diagnosed, not yet repaired. A jumper between RX13.5V and 13.5V OUT on the AVR board terminal strip brought receive back to life instantly. That puts the fault in the RX supply switching path of the AVR unit (PB-2099-A), with the RL01 relay contact RL1-1 as the one switching element in that path.

Symptom	Transmits at full power on the correct frequency. On receive: buzzing/rasping, S-meter stuck near S1, all controls otherwise work. Buzz level follows the AF gain control; its pitch shifts slightly as you retune.
History	Radio stored unused for months. Three good QSOs about 48 hours before the fault. Next day signals faded to S1 after a few minutes; keying PTT rapidly (~10 times) sometimes brought it back. The day after, constant buzz on receive.
Cause	RL01 (parts list RL4601, original Fujitsu FBR321D012), contact RL1-1 on the RX13.5V (normally-closed) side. Probably oxidised or pitted. The relay still clicks cleanly both ways, so the coil and mechanism are fine.
Fix	Replace RL01 with a NOS Takamisawa/Fujitsu RY12W-OH-K (12 V coil, DPDT, gold-overlay contacts) wired on flying leads. Replace C01 (1000 µF) as sensible maintenance with the tested 25 V NOS part.
Time	About 2–3 hours at the bench, including testing. Skill needed is below what building three QCX-mini kits required.
Not part of this fix	The VFO frequency jumping (currently stable, cause unknown), the quieter buzz on AM transmit (recheck after the fix), and the hand-built board in a plastic bag (see other issues).

Read this before soldering — two things need checking first

The jumper test proves the fault is somewhere between those two terminal-strip pins. RL1-1 is the only switching contact in that path and the best suspect, but nobody has measured across it directly. Step 3 below adds a quick measurement that settles it before you commit to the swap.
The idea of wiring the new relay to the 8 harness wires on the terminal strip is only part of the answer. See the access options.

Background

1. What the AVR unit does

The FT-707 has a small voltage-regulator board, the AVR unit PB-2099-A (the number is silkscreened on it; there are other revisions, PB-2201 and PB-2094B, with different relay/snubber arrangements). It connects to the harness through terminal pins and through the plug PL4301, which mates with PL4302 on the main harness.

Two relays sit side by side on it:

Board ref	Parts-list ref	Original part	Job
RL01 (contacts RL1-1, RL1-2)	RL4601	Fujitsu FBR321D012, 12 V coil	Switches the supply rails between RX and TX: RX13.5V ↔ TX13.5V on one pole, RX8V ↔ TX8V on the other. This is the faulty one.
RL02 (contacts RL2-1, RL2-2)	RL4602	Fujitsu FBR221D012, 12 V coil	Lower-stakes switching: clarifier reference voltage and the meter's RX/TX source. Not involved in this fault.

Do not confuse the two "RL1" names The large antenna changeover relay near the antenna socket is labelled just "RL1" on the chassis. It is a different component from RL01 on the AVR board. In the AVR schematic the contacts are RL1-1 and RL1-2, which belong to RL01.

2. How the switching works

The coil is driven by the RL CONT line through transistor Q02, with diode D01 across the coils as a flyback diode. The relay is energised only in transmit. At rest (receive) each pole sits on its normally-closed contact:

Pole 1: COM is the switched output. N.C. = RX13.5V (rest/receive), N.O. = TX13.5V (energised/transmit).
Pole 2: COM is the switched output. N.C. = RX8V (rest/receive), N.O. = TX8V (energised/transmit).

On the full schematic the receive path runs RX13.5V → C11 → contact RL1-1 → R07/C07 filtering → the shared node feeding the Q02/Q03 regulator stage. No diode sits anywhere in this path. (D01 is the coil flyback diode and D02 is a 6.2 V zener in the clarifier reference.)

Other things that hang off the same board:

The COUNTER (frequency display) supply taps the RX13.5V line through RO2, upstream of RL1-1. So the counter's supply quality depends on the incoming 13.5 V and C01, not on the relay contact.
The VFO unit PB-2097 takes its power from the board's 8 V line (connector J4301, pin 2).
OH2FFY/OH2DX describe Q01 as looking like a zener-referenced regulator but actually being capacitor-driven, so closer to an active ripple filter than a true regulator. Useful background, not the fault.

3. Why this relay is a known weak point

The low-profile relays in the FT-707 are reported to fail early. Yaesu tried to suppress arcing with a capacitor across the contacts; on early units these were reportedly soldered to the underside of the board almost as an afterthought.
OH2FFY/OH2DX (the same project mirrored on two sites) argue the right snubber is a series resistor–capacitor to ground, not a capacitor straight across the contacts, which can cause its own problems in the presence of RF. Their rebuild uses one better-suited relay type (silver-plated contacts, adequate current rating) in place of Yaesu's two different types, and replaces the cap-across-contacts snubbers with series R-C snubbers.
Your fault pattern also fits contact oxidation after storage. Low-current "signal level" contacts do not carry enough current to clean themselves. After months idle a thin oxide film builds up. Rapid PTT cycling mechanically wipes it for a while, which is why keying 10 times sometimes revived the receiver. Each cycle of the relay also adds a little arc damage, which is why it can go from fine to dead in a couple of days.

I never read the OH2FFY/OH2DX pages themselves The site blocked automated fetching in the original chat (it did again today), so the points above come from search-result summaries. The exact relay type and snubber values they use are not in this guide. Read the write-ups yourself if you want the permanent fix; links are in references.

4. Why a like-for-like swap may not last forever

A new relay with the same factory snubbing could eventually arc again, but the replacement is a gold-overlay type and RX-side contacts see little current. For a radio used occasionally that is a reasonable trade. If you want the permanent version, add series R-C snubbers per the OH2FFY/OH2DX write-up as a later upgrade.

5. Reading a relay pinout

The RY-series and Omron G2V/G2VN/G5V-2 relays share a standard 16-position grid with only 8 pins fitted. That is why the pins are numbered 1, 4, 6, 8, 9, 11, 13, 16 and not 1–8. Pins 1 and 16 are the coil; the rest are two groups of three (COM, N.C., N.O.) — one per pole.

Coil data for the RY12W-K (12 V) from the datasheet you supplied: 960 Ω, 150 mW, about 12.5 mA. The tolerance is ±10 %. The suffix -OH means gold-overlay contacts (it is stated in the ordering information); it is not an electrical difference.

Evidence — what is proven and what is inferred

Observation	What it tells you	Weight
Relay clicks cleanly on key-down and key-up	Coil, armature and return spring are fine. Fault is in the contact surfaces, not the mechanism.	supporting
Buzz level follows the AF gain control	The noise travels through the normal audio chain rather than being picked up at the speaker.	supporting
Fade to S1 over minutes; ~10 rapid PTT cycles sometimes restored it	Textbook marginal relay contact. A failed component such as a diode would not fade or respond to PTT cycling.	strong
Three QSOs 48 hours earlier, then fault	Fast onset fits a contact tipped over by use, not slow capacitor ageing.	supporting
Jumper RX13.5V → 13.5V OUT restored clean receive instantly	Fault is in the path between those two pins; RL1-1 is the only switching contact in it.	direct test
Full schematic: no diode in that path	Removes the diode theory raised earlier in the chat (see corrections).	from schematic
Contact resistance across RL1-1 (N.C.→COM) measured directly	Would remove the last bit of doubt before soldering.	not done yet

Timeline of the diagnosis (condensed)

Symptom described; first suspects were T/R relay contacts and supply electrolytics.
You suspected the relay because you had keyed several times before it failed. Searching pointed to RL01 on the AVR board and the OH2FFY/OH2DX write-ups.
Repairing the old relay in place was ruled out as too fiddly; replacing it on flying leads was chosen.
Service-manual pages gave the parts list (RL4601 = FBR321D012, RL4602 = FBR221D012), the AVR schematic, Yaesu's relay diagram and the PB-2099-A board layout.
You went through your NOS relay stock (see replacement relay) and settled on the Takamisawa RY12W-OH-K after reading its datasheet.
Side findings: warm-up drift (normal), erratic VFO and counter behaviour (see other issues), quieter buzz on AM transmit, a hand-built board with an MC3357, and a possibly leaking C01 that turned out to be glue-like residue.
Terminal strip photo identified the pin order; the jumper test then confirmed the fault location.

Replacement relay — what was considered

Requirements: 12 V DC coil, two changeover poles, low-current signal duty (around 12.5 mA coil, small supply currents), good contact material. Physical size does not matter because it goes on flying leads.

Candidate	Source	Coil	Contacts	Verdict
Takamisawa/Fujitsu RY12W-OH-K	Your NOS box	12 V, 960 Ω, 150 mW	DPDT, ~1 A, gold overlay	chosen Same manufacturer family as the original FBR321D012; gold contacts; pinout confirmed from the datasheet you supplied.
Omron G2V-282P (12 V)	Your NOS box (~50)	12 V printed on case	DPDT	Usable backup. No datasheet found online; pinout borrowed from the sibling G2VN, so check with a meter. The "47/6", "47/9" suffix is not a date code and not a reliable voltage code; trust the printed "12VDC".
STC 47SB12T (cross-referenced to NTE R40-11D2-12)	Your box	12 V, 960 Ω, 150 mW	DPDT, 2 A @ 28 V DC, gold-clad	Good alternative.
Finder 30.22 (12 V)	Your box	12 V, ~720 Ω	DPDT, 2 A	Good, well-documented; larger.
Omron G5V-2-DC12	Would need buying (in stock at TME)	12 V	DPDT, 2 A, silver-gold-clad	Only if you run out of stock.
RS PRO 12 V DPDT signal relays	RS	12 V, 720 Ω	DPDT, 3 A	Out of stock when checked.
Omron MKS3PI-5 (3PDT, 10 A power relay)	RS	12 V	3PDT, 10 A	rejected Far too large and over-specified for signal-level switching.

If you ever want an exact-style part, Cephos Relays Ltd (Kent, UK) specialise in cross-referencing obsolete relays, including other Fujitsu FBR parts. Your FBR321D012 and FBR221D012 were not in their published table, so you would have to email them.

A relay-coil check you asked about Coil resistance does not need to match the original's. The coil is driven through Q02 as a switch, and 960 Ω (about 12.5 mA) is a light load. Real trouble would only come from a coil that is very low (a near-short stresses Q02) or open (it simply will not pull in). If a candidate reads wildly off, bin it and take another.

Pinout and wiring map

RY12W-OH-K pins (as drawn in the Fujitsu/Takamisawa RY-series datasheet, and the same layout as the Omron G2VN diagram):

Pin	Function
1, 16	Coil (non-polarised)
6	COM — pole 1
4	N.C. — pole 1 (rest / RX)
8	N.O. — pole 1 (energised / TX)
11	COM — pole 2
13	N.C. — pole 2 (rest / RX)
9	N.O. — pole 2 (energised / TX)

New relay pin	Wire to (original RL01 pad / signal)
1 and 16	The two original coil pads (RL CONT side and the other coil side — trace which is which on the board)
6	Pole 1 output → regulator side (13.5 V)
4	RX13.5V
8	TX13.5V
11	Pole 2 output → 8 V side
13	RX8V
9	TX8V

The drawing shows the relay's connections, not the PCB layout. Trace the original pads by continuity before you desolder (step 2).

One error in the original chat that you should know about At one point the chat read the pinout with COM and N.C. swapped (pin 4 = COM). It was corrected once you sent the clearer diagram: 6 and 11 are COM; 4 and 13 are N.C.; 8 and 9 are N.O. The table above is the corrected one, and the Fujitsu datasheet agrees. Verify it on the bench anyway (step 4).

Repair — step by step

Tools and materials

Tools

Fine-tip iron, solder, flux
Desoldering braid or solder sucker
Multimeter (continuity, resistance, diode range)
Screwdrivers for the case
Wire cutters/strippers, tweezers
Heat-shrink and a heat source
Hot glue or double-sided foam tape
Phone/camera for reference photos
12 V supply or battery pack for bench-testing the coil
Inspection mirror and head torch (access may be awkward)
Kapton tape or folded card to shield neighbouring parts

Materials

1 × RY12W-OH-K plus 1–2 spares from your NOS stock
1 × 1000 µF 25 V electrolytic from the NOS box (tested: 941 µF, ESR 0.28 Ω, brand "LL")
8 lengths of stranded hook-up wire, about 3 cm each, in several colours
Small cable ties, tape flags for labelling

Colour code suggested in the chat: one colour per function (coil, COM, N.C., N.O.), with two shades or a tape flag to separate pole 1 from pole 2. Write your key down before starting.

Phase 0 — Safety and set-up (about 15 min)

Disconnect DC power, antenna, mic and key. Wait a few minutes.

Remove the cover that gives access to the AVR board (see access). Photograph the whole area before touching anything.

Discharge C01 (and any other large electrolytic nearby) through a few hundred ohms. Do not short it directly.

Phase 1 — Find your way around the board

The AVR board has a terminal strip (left to right in your photo): GND, GND, 13.5V OUT, 13.5V IN, C01, RX8, TX8, TX13.5V, RX13.5V. Other pads are labelled CLARI, RL RX, RL TX, VR2 and so on. Identifying features: two relays side by side (RL01 and RL02), a row of large electrolytics (C01, C02, C03, C05), transistors Q01 (2SD880Y), Q02, Q03 (2SK19), and the PL4301 plug. Do not mistake the large filter-can board (the RF/IF unit) or the board marked PB-2086 for it.

Phase 2 — Map the old relay's pads (about 20 min)

Photograph RL01 from several angles, including orientation against RL02 and C01.

Power off. With the multimeter, find the coil pins of RL01 (fixed resistance, no continuity elsewhere). Measure and write the value down for comparison with the new relay's 960 Ω.

Trace each of the remaining RL01 pads to the named signal: RX13.5V, TX13.5V, RX8V, TX8V, the two switched outputs, and RL CONT. Write the names on a printed photo or a sticky note.

The PB-2099-A layout scan was too small to read pin by pin, so rely on continuity tracing, not the scan.

Phase 3 — Settle the last doubt: measure across RL1-1 (about 10 min)

This is new, and it takes two minutes. The jumper test bridged the whole path. Now check the contact itself:

Rig off. Measure resistance from RL01's RX13.5V N.C. pad to its pole-1 output (COM) pad. A healthy relay contact reads close to 0 Ω (a few tenths at most). A high reading, or erratic readings when you tap the relay, confirms the contact.

Repeat for the RX8V N.C. → pole-2 COM pair.

If the contact measures fine Stop and think rather than carry on. The remaining parts in that path are R07, C07 and C11 and the solder joints. A dry joint or cracked track would also be cured by the jumper.

Phase 4 — Bench-test the new relay first (about 10 min)

Measure coil resistance across pins 1 and 16. Expect about 960 Ω (±10 %). Reject any that read far off or open.

Apply 12 V to pins 1/16. Check for a clean click and that it does not warm up in a few seconds.

Unpowered: pin 6 to pin 4 and pin 11 to pin 13 should be closed. Powered: pin 6 to pin 8 and pin 11 to pin 9 should be closed. Mark the verified relay and put it aside.

Phase 5 — Replace C01 (about 15 min)

C01 (1000 µF) sits at 13.5V IN, before Q01 and before the relay splits the supply. It is not the cause of the buzz. It is worth replacing at this age because the counter feed taps the same rail upstream of RL01. The residue you saw on top of the old can looked hard and old, and glue is a very common way of securing large caps, so leakage was never confirmed.

Photograph the old C01 and mark its polarity stripe/lead on the board.

Desolder the old C01. Check the pads and board underneath for real corrosion or residue; that settles the glue-or-leak question.

Fit the new 25 V part with the correct polarity, solder, trim. Check polarity again before power-up. This is the easiest mistake to make.

The old C01 was never measured. If you want a record, measure it out of circuit once removed.

Phase 6 — Getting at RL01: three options

You found the board hard to reach and could not simply pull it out. The wires on its terminal strip are individually soldered, so freeing the board means desoldering them (check first whether PL4301/PL4302 is a genuine plug that would be easier). The options:

Option	How	Notes
A. Work in place	Move nearby parts aside, use mirror, torch and a fine iron. Shield neighbours with Kapton or card.	Simplest if you can see and reach the relay pads.
B. Remove the board	Label and desolder the terminal-strip wires one at a time (tape flags, photo first), then unplug PL4301 if it is a plug. Work on the bench; refit and check every wire against the photo.	Most comfortable for the fiddly part. More work at each end.
C. Wire the new relay to the harness	Solder the new relay's leads to the terminal-strip wires instead of the board pads.	incomplete See the warning below.

Why option C is not the whole job The terminal strip carries the four inputs (RX13.5V, TX13.5V, RX8, TX8) plus supply and ground. It does not carry the relay's two switched outputs (the COM nodes that feed the regulator stage) or the coil drive from RL CONT, and those are internal board nodes. An external relay would still need wires to those pads on the board. The chat said it "changes nothing electrically", which was too neat. Treat option C as a variant of A, and lift the old relay's pins completely rather than leaving it half-connected. If you did leave the old coil connected in parallel, you would double the load on the Q02 driver (two coils, about 25 mA together instead of 12.5 mA). It would probably cope, but it is not worth finding out.

Phase 7 — Remove RL01 and fit the new relay (about 60–75 min)

Add flux and heat each pin of RL01 in turn, clearing it with braid or a sucker. Do not force the part out; old pads lift. Keep the old relay.

Cut 8 stranded wires (about 3 cm each): 2 for the coil, 3 for pole 1 (COM, N.C., N.O.) and 3 for pole 2. (An earlier version of the guide said seven; there are eight.)

Solder all 8 to the new relay first, on the bench, using your wiring map. Heat-shrink each joint.

Solder the free ends to the original pads one at a time, ticking each off against your map and the colour key. Insulate the board-end joints.

Recheck every connection against the map. Do a continuity check from each pad to the matching relay pin.

Mount the relay with hot glue or foam tape so the joints carry no strain. Keep wires away from hot parts (PA transistors) and moving parts (VFO dial).

The 3 cm wire length and the cramped-chassis question

3 cm is a sensible length: enough to place the relay sensibly, short enough to stay tidy. Early in the chat the chassis was described as "cramped" without any basis; your photos show reasonable clearance and the real constraint is only reaching the AVR board. Soldering the relay end first is a convenience, not a requirement.

Testing

First power-up (case still open)

Use a dummy load on the antenna socket and power on.

Receive: buzz gone, S-meter responds to a signal, noise level follows the RF/AF controls normally.

Key PTT: a clean click on key-down and on release, full TX power as before.

Cycle PTT 15–20 times, listening for the buzz to return.

Recheck the AM transmit buzz (it was quieter than the receive buzz but present). If it vanishes it shared the cause; if not, it is a separate fault.

Leave it on for 30+ minutes and watch frequency stability across bands, including 10 m.

Before closing up

Route all wiring clear of pinch points, refit the cover and repeat a quick RX/TX check with the case closed.

Your working FT-707S is a reference You can measure RX13.5V, TX13.5V, RX8V and TX8V on it in receive and transmit, look at where its AVR board sits and how it is mounted, and compare VFO and 10 m behaviour. It will also tell you whether a small AM-transmit buzz is normal on this model. Swapping boards between the two radios is a legitimate technique but only once you are sure how the connections release.

Troubleshooting

Symptom	Check
No click from the new relay	Coil wiring to pins 1/16 and the two coil pads. The coil is non-polarised, so reversal is harmless, but a wrong pad means no drive.
Click but buzz still there	N.C./N.O. mapping. If pole 1 and pole 2 are swapped, or N.C. and N.O. are reversed on one pole, RX may still get the wrong supply. Re-test continuity in circuit.
TX behaves oddly	TX-side (N.O.) wiring on one or both poles.
Nothing works	Solder bridge or dry joint on a new connection. Raking light and a magnifier find these quickly.
Buzz gone but frequency still jumps	Not the relay. See the VFO notes below.
Still unsure the relay is faulty	Swap in a second NOS relay rather than debugging one unit forever.

Never key the PTT while a test jumper is fitted The jumper test is "power on, receive only, touch, listen, remove". It bypasses the relay so TX switching cannot do its job.

Other issues found along the way

None of these is cured by the relay swap, though the AM buzz might be. They are kept apart so they do not muddy the main job.

Issue	What was found	Status
Warm-up drift on 40 m	Drifts for the first ~30 minutes, then settles.	normal Typical of an analogue VFO. No action.
VFO / counter instability	Started as ±10 kHz jumping on 10 m, spread to lower bands, and at its worst 10 m landed at 23–25 MHz. It persisted on a 10-second AM transmit, so it is not simply "RX supply through RL01". It later stopped after you opened the radio and handled it; it was stable again on 40 m after a 5-minute power-off.	unresolved Probably a disturbed connection that handling cleared, and may return. Test with long on-times. Before trusting a wild reading, check against WWV (10 or 15 MHz) or the FT-707S to see whether the display or the RF is wrong.
Quieter buzz on AM transmit	Still present when transmitting in AM, much quieter.	recheck after fix
Hand-built board in a plastic bag	Perfboard wired into several chassis points. Visible parts: Motorola MC3357 (FM IF chip, date code 8021) and a Murata CFU455 455 kHz ceramic filter. A rear-panel toggle: on AM it replaces the buzz with clean hiss; buzz on other modes only gets quieter.	Most likely an FM detector add-on, not a "huff-puff" VFO stabiliser as you were told (those use counter/divider ICs, not an FM IF chip). Unrelated to the relay fault. It does show the radio has been modified before. The chat also suspected a poorly done earlier repair.
C01 residue	Orange-brown crust on the top of C01.	probably glue Settle by looking under the cap when it is out.
Speech compressor	The rig has none. An external inline mic processor was recommended over an internal mod, as a separate project.	Parked.

Corrections and open points

The original chat changed its mind several times. These are the places where an earlier statement was wrong or is still unverified, so nothing misleads you when you work from this page.

Earlier statement	Where it stands now
"Diodes in series with each RX/TX line could be the fault"	wrong Based on a partial schematic. The full schematic shows no series diode in the RX13.5V path. The diode-check phase was removed.
Pin table with pin 4 as COM, 6 as N.C.	wrong Corrected: 6/11 COM, 4/13 N.C., 8/9 N.O.
"7 wires" (and the guide's Phase 5 wording)	wrong Eight: 2 coil + 6 contacts.
"47/6 is a date code" then "47/9 means 9 V"	both wrong Do not decode the suffix; trust the printed 12VDC. Moot now you are using the Takamisawa part.
FBR321D012 described as 3-pole	wrong Yaesu's relay diagram shows a 2-pole changeover.
Everything (buzz, counter, VFO) is one RL01 fault	dropped Frequency instability persisted on transmit, which uses the other contact.
C01 as the likely cause of the sudden fault	dropped The 48-hour onset and contact-cycling behaviour fit the relay; C01 is maintenance only.
Bridging the terminal strip "only bypasses RL1-1"	to verify It bypasses the whole path between those pins. Phase 3 measures the contact directly.
Which coil pad is RL CONT and which is the other side	to verify Never confirmed on the real board. Trace it in Phase 2.
Exact RL01 pad positions on PB-2099-A	to verify The layout scan was too small to read.
OH2FFY/OH2DX relay type and snubber values	not read The site could not be fetched by the assistant.

References and what to have to hand

Documents and links from the chat

OH2FFY write-up: oh2ffy.altervista.org/ft-707_avr/ft707-02.html (and the OH2DX mirror: oh2dx.altervista.org/ft-707_avr/ft707-02.html) — the AVR board rebuild with better relay and series R-C snubbers. Not read by the assistant.
FT-707 service manual scan: qsl.net/pa2efr/manuals/Yaesu/FT707_serv.pdf (a scanned PDF; you also own your own copy).
Cephos Relays Ltd, relay cross-reference: cephosrelays.com/relay-cross-reference.
Fujitsu/Takamisawa RY-series datasheet (the one you sent, showing the -OH suffix and the 960 Ω / 150 mW coil data).

Pages from the service manual worth printing

AVR unit PB-2099 schematic (the clean one you called "the best circuit I have").
PB-2099-A component-side layout.
Parts list page with RL4601/RL4602.
Yaesu's relay internal-contact diagram for RL4601.
Your own photos: the labelled terminal strip, the relay area, and the hand-built board.

Images from the original chat (schematics, board photos) are not reproduced here because they were not retrievable from the saved conversation. Source: the claude.ai chat "Yaesu FT707 receive buzzing noise", 8 September 2026.