Fox-Tango Era • Yaesu Linear Amplifiers

Yaesu FL-2100 / FL-2100B / FL-2100Z
Tuned Input Circuit: SWR Expectations & the Mica Capacitor Problem

What input SWR should a healthy 572B grounded-grid amplifier show — and which of the two input capacitor styles is the one that fails?

Reader Question — Dave, AI6VX (via the ham-amplifiers group)

1. I understand that input SWR is affected by the impedance of the tubes, and that this changes with Plate and Loading adjustments — but what is typical for this amp? Should I expect all bands to tune up into a dummy load with an input SWR under 2.0?

2. I’ve read that the caps on the input pi network are a common failure point, where the termination inside the package disconnects from the lead over time and with heat. But I see two different cap styles used over the years — the FL-2100Z has modern-looking epoxy-dipped micas, while an early FL-2100 has large rectangular block caps. Which type are the “commonly fail” ones?

The Short Answer

Question 1: Yes. A healthy, factory-aligned FL-2100 series amplifier, properly loaded to rated plate current, should present an input SWR below 2:1 on every band into a dummy load — typically 1.2–1.7 near band centre, drifting upward toward the band edges. Any single band that cannot be brought under 2:1 with the amplifier correctly tuned is a fault, not a characteristic of the design — and the input network is the first suspect.

Question 2: Both styles are silver mica, and both generations fail in service — but the specific mechanism you describe (the internal foil-to-lead termination letting go under heat and thermal cycling) is the classic signature of the epoxy-dipped style. The older rectangular moulded blocks are not immune; they more often drift, crack, or develop silver-migration leakage. Treat the entire per-band input capacitor set as a wear item on any FL-2100 you restore, regardless of vintage.

Section 1 — How the FL-2100 Input Circuit Actually Works

The FL-2100 series runs a pair of 572B/T-160L zero-bias triodes in class-B grounded grid. Drive is applied to the cathodes (through the filament circuit), and the cathode of a grounded-grid triode is a low, dynamic, non-linear impedance — nominally in the neighbourhood of 50–75 Ω for the pair, but varying across the RF cycle, with drive level, with tube condition, and with the reflected load presented by the Plate and Loading settings of the output tank.

To keep the exciter happy, Yaesu placed a band-switched tuned input network between the input SO-239 and the cathodes: one slug-tuned coil plus fixed silver mica capacitors per band, selected by the rear wafer of the bandswitch. Each network does two jobs at once:

Impedance transformation — matching the ragged cathode impedance to a 50 Ω exciter, and

Flywheel action — the network’s stored energy smooths the half-cycle conduction of class-B cathode drive, which improves linearity (IMD) as well as the match.

Two practical consequences follow, and they answer the first half of Dave’s question:

First: the networks are fixed-tuned at the factory to the centre of each band. There is no front-panel input tuning. The slugs were set once in Tokyo, sealed with a dab of adhesive, and left. Fifty years later, an SWR that is fine at 14.175 MHz and climbs at 14.350 MHz is normal behaviour, not a defect.

Second: because the cathode impedance moves with plate tuning and loading, the input SWR reading is only meaningful with the amplifier keyed, driven, and correctly tuned to rated plate current. Checking input SWR into a cold or lightly-loaded amplifier will give you pessimistic, misleading numbers. Tune the output first — Plate dip, Loading advanced for rated current and clean peak power — then read the input SWR. Under-loading the output tank raises the apparent cathode impedance and the input SWR with it.

Section 2 — What Input SWR Is “Typical” for a Healthy Amplifier

Yaesu never published an input SWR specification — the manual simply lists input impedance as 50–75 Ω unbalanced. The figures below are what the restoration community consistently reports for a healthy, unmolested example driven at normal exciter power into a dummy load, measured with the amplifier tuned and loaded correctly:

Band	Healthy Example, Band Centre	Notes
80 / 40 m	1.2 – 1.5	Usually the best-behaved bands. A sudden high reading here is almost always a failed input capacitor, not drift.
20 m	1.3 – 1.7	Rises noticeably toward 14.350 on a network aligned at 14.175.
15 m	1.4 – 1.8	Network Q is higher; slug position matters more.
10 m	1.5 – 2.0	The worst band on nearly every example. Stray reactance dominates at 28 MHz and the network only covers part of the wide band well.

So the direct answer: yes, expect all bands under 2:1 into a dummy load, with real-world caveats at band edges and on 10 metres, where a properly working amplifier may brush 2:1. Community reports of misbehaving units are instructive by contrast — a documented FL-2000B fault case showed 80/40/15 all under 2:1 while 20 m sat at 3.5:1 and 10 m was effectively unmeasurable. That pattern — most bands fine, one or two bands wild — is the fingerprint of a per-band component failure, because each band has its own coil and capacitor set. Uniform mediocrity across all bands points instead at tubes, loading technique, or the shared cathode path.

Why this matters more in 2026 than in 1978

The FT-101’s pi-network output tuned into whatever the amplifier presented and never complained. A modern solid-state exciter with SWR foldback will silently cut its drive into a 2.5:1 input, and the operator sees “low output, tired tubes” when the tubes are fine and an input mica has failed. If your FL-2100 seems weak behind a modern rig, measure the input SWR before you shop for 572Bs.

Section 3 — The Two Capacitor Styles, and Which One Fails

Across the production run (FL-2100 → FL-2100B → FL-2100Z, roughly 1970–1985, plus the Sommerkamp FL-2277 badge variants) Yaesu fitted the input networks with silver mica capacitors in two package styles, and Dave has spotted the transition exactly:

Package	Where You’ll See It	Dominant Failure Mode
Rectangular moulded block (“postage stamp” / domino style, hard phenolic or epoxy case)	Early FL-2100 and FL-2100B production	Capacitance drift, case cracking, silver-migration leakage between plates. Failures tend to be gradual and value-related rather than a clean open.
Epoxy-dipped silver mica (the familiar brown/tan “blob”)	Later FL-2100B and FL-2100Z production	The mechanism Dave describes: the crimped/pressure termination between the lead and the internal electrode stack separates with heat and thermal cycling. The result is an intermittent or hard open — the band goes from working to unusable, sometimes literally between overs.

So which is the “commonly fails” type? The honest answer from the field is: the lead-termination disconnect story attaches to the dipped epoxy style — that construction relies on a mechanical bond between the lead and the plated mica stack, buried in epoxy where thermal expansion works on it every transmit cycle. The FL-2100Z restoration literature bears this out: a well-documented Z rebuild (M0KDS) traced a dead 80 m band with sky-high input SWR straight to failed silver mica input caps of the dipped type, cured by replacing C19/C402 with 3 kV ceramics.

But don’t read that as a pass for the rectangular blocks in your early FL-2100. Moulded micas of that era have their own well-known disease — silver migration and drift — and in this circuit any failure matters more than usual, because these are not small-signal parts in an IF can. Every one of them carries the full 100-watt exciter drive as circulating RF current in a high-Q network, in a poorly-ventilated corner of a chassis that runs hot. That duty cycle is precisely what accelerates both failure mechanisms. It’s also worth saying the counterpoint that experienced FL-2100 rebuilders make: silver micas that haven’t failed are excellent RF parts, and plenty of originals are still perfectly serviceable — test before you condemn.

Restoration Rule of Thumb

One dead or wild band = suspect that band’s input capacitor(s) first, coil slug second, bandswitch wafer contact third. All bands poor = look at loading technique, tube emission, the cathode/filament RF path, and the input relay contacts before touching the networks.

Section 4 — Diagnosing the Input Network

Step 1 — Band-by-band SWR survey. Amplifier into a dummy load, SWR meter between exciter and amplifier input. Tune each band properly (Plate dip, Loading to rated current), log the input SWR at band centre and both edges. The pattern tells you almost everything before you lift the lid.

Step 2 — Cold sweep with a VNA. With the amplifier off, unplugged, and the HV supply verified discharged, a NanoVNA on the input SO-239 (amplifier switched to OPERATE so the relay routes the input to the network) will show each network’s resonance. You won’t see the true operating match — the cold cathodes don’t load the network the way conducting tubes do — but a network resonant far outside its band, or showing no resonance at all, has a failed component. Compare bands against each other: the healthy ones calibrate your expectations for the sick one.

Step 3 — Provoke the intermittent. The dipped-mica lead disconnect is classically heat-triggered and mechanically sensitive. With the VNA connected, tap each capacitor with an insulated tool and warm each one gently (hair dryer, not heat gun) while watching the trace. A resonance that jumps or vanishes under tap or heat is your culprit — this catches failures that measure perfectly on a bench LCR meter at room temperature.

Step 4 — Verify out of circuit. Lift one lead and check capacitance and insulation resistance. A drifted moulded block reads off-value; a migrated one leaks; a lead-detached dipped mica may read open, or read fine until flexed. If a removed part misbehaves in any of these ways, replace its entire band set — its siblings have lived the same thermal life.

Section 5 — Replacement Parts and Realignment

Like-for-like: new CDE / Cornell Dubilier CM-series dipped silver micas (500 V minimum; 3 kV types where the schematic calls for them) from RF Parts, Mouser, or Surplus Sales. Modern production dipped micas do not share the termination weakness of 1970s Japanese stock in anything like the same numbers, and they preserve the network’s original Q and temperature behaviour.

Pragmatic alternative: HV ceramic disc (3 kV, NP0/C0G where obtainable in value). The documented M0KDS FL-2100Z rebuild used 3 kV ceramic discs for the 80 m pair (680 pF and 820 pF) and brought the band from unusable back under 2:1. Ceramics are cheap and rugged; the trade-off is looser tolerance and tempco, so verify final values and expect to touch up the slug.

Ordering from Yaesu part numbers: if you work from the original parts list, note the convention documented on foxtango.org — FL-2100 series ceramic capacitors carry part numbers beginning K24, while the micas begin K313. Parts desks have historically confused the two, so double-check what arrives against what you ordered. Values differ across production runs and serial ranges (the Z adds WARC coverage and extra networks), so pull values from the schematic matching your serial number, not from someone else’s photo.

Realignment: after replacement, set each band’s slug for minimum input SWR at the centre of your operating segment — not necessarily the geometric band centre. Do this keyed at modest drive into the dummy load with the amplifier properly loaded, adjusting in short transmissions. A non-metallic alignment tool only; the slugs are brittle and the factory adhesive should be cracked free gently before turning. Reseal with a small dab of nail varnish when done.

⚠ High Voltage Caution

The FL-2100 series carries 2,400 V DC on the plate circuit and stores lethal energy in the filter capacitors after switch-off. The cabinet safety interlock shorts the HV when the cover is removed — but never trust a fifty-year-old interlock with your life. Unplug, wait, verify zero volts at the filter stack with a meter rated for the job, and apply a grounding stick before any hand goes near the chassis. The input network compartment sits close to the filament/HV wiring; treat the whole underside as live until proven otherwise.

Yaesu FL-2100 / FL-2100B / FL-2100ZTuned Input Circuit: SWR Expectations & the Mica Capacitor Problem