Circuit Design

Full engineering specification for the Ghost Spring Tank — transformer coupling rationale, complete signal flow, component selection, output buffer, enclosure, and build rules.

At a Glance

Ghost Spring Tank

One spring. One voice. Transformer-coupled for the organic bloom that no direct-drive circuit can replicate — the same mechanism that made the Fender 6G15 Reverb Unit sound like nothing else.

Spring TankAccutronics 9AB3C1B — 3-spring, long decay, 8Ω input / 2550Ω output
Driver XfmrAccutronics REB3S — dedicated spring reverb driver transformer, 8Ω secondary
Driver StageBD139 NPN transistor (TO-126) → REB3S primary. Class A, ~18mA quiescent
Op-AmpsOPA2134PA throughout — input buffer, recovery preamp, output buffer
Signal Path100% analog. Post-recovery 300Hz HPF on wet signal only
ControlsDwell (10kΩ lin) · Mix (100kΩ audio taper + 47pF silver mica bright cap) · Tone (100kΩ audio taper)
PowerInternal ±15V linear — Triad F-219X toroidal 30VA dual-primary, LM7815/LM7915. IEC mains inlet
Form Factor2U aluminum rackmount. Hammond 1455T2201 chassis, rack ears included
Tank MountHorizontal, open-side down, 4× Shore 30A rubber isolation grommets
PCBVector T44 FR4 perfboard, 0.1" pitch — point-to-point construction, no custom PCB

Signal Flow

Complete Path

Every stage in the signal chain with component values inline. Left to right: input from line source, through the spring and back out to the power amp.

Line In (from Lehle Parallel M)C_in             1µF film · DC-blocking coupling cap at input jackInput Buffer     U1 OPA2134 · R1 1MΩ from U1(+) to GND (bias return + input Z) · unity gain · R2 100Ω series outputDwell Pot       RV1 10kΩ linear · C1 1µF film coupling cap · controls drive levelBD139 Driver    Q1 Class A · R3b 6.8k + R4 1k bias divider · R5 68Ω emitter degen
                   C2 100µF bypass · ~18mA quiescent · D3 1N4148 flyback clampREB3S Xfmr     T2 primary (from Q1 collector) → 8Ω secondary9AB3C1B Tank   RT1 · 3 springs · long decay · 8Ω input / 2550Ω outputC3 Coupling     470nF film · blocks DC from tank output terminalsRecovery Preamp U2 OPA2134 · non-inverting · tank → (+) input · Ri=470Ω / Rf=100kΩ → gain=214× (~46dB)
                   Rbias=100kΩ at (+) input to GND · 1–5mV → ~1Vrms300Hz Wet HPF  C4 100nF film + R6 5.6kΩ · design f = 1/(2π×5600×100n) ≈ 284Hz; measured ≈312Hz in full sim
                   wet signal only — dry path bypasses this stageTone Pot        RV3 100kΩ · high-shelf on wet signal onlyMix Pot         RV2 100kΩ audio taper · 3-terminal passive blend (NOT a volume knob)
                   dry (U1 via Rdry 10kΩ) → CCW lug 1 · wet (Tone wiper) → CW lug 3 · wiper lug 2 → U3
                   C_bright=47pF silver mica across the FULL pot (lug 1 ↔ lug 3)Output Buffer  U3 OPA2134 · voltage follower · R7 100Ω series · <100Ω output ZLine Out (to McIntosh MC100)

The Key Mechanism

Why Transformer Coupling

This is the part that makes the Ghost Spring sound like a spring reverb and not a plate sim. Most DIY spring reverb designs drive the tank directly from a transistor or op-amp output. That works — but it doesn't sound like a Fender 6G15.

The REB3S transformer's inductance forms a resonant circuit with the tank's input impedance (8Ω). This resonance creates:

The Accutronics REB3S is designed specifically for this application. It's used in boutique spring reverb units and is correctly matched to 8Ω tanks. Using a generic transformer (or no transformer) changes the resonant character fundamentally.

The 9AB3C1B — Why This Tank

Three springs give a denser, more uniform reverb tail than two-spring tanks. Two-spring designs can produce a metallic "ping" on hard attacks because the two resonant modes are more distinct. Three springs blend into a closer-to-continuous reverb tail — the same reason the original Fender 6G15 and most boutique spring reverbs use 3-spring tanks.

Model code9AB3C1B — 3-spring long-decay tank, 8Ω input / 2550Ω output (refer to Accutronics tank catalog for full code decode)
Input Z8Ω — matches REB3S secondary directly, no impedance mismatch
Output Z2550Ω — requires high-impedance input at U2 recovery stage (OPA2134 FET input: 10¹³Ω)
DecayLong — the "A" type. This is the sustained ambient tail Jerry Garcia's clean tone requires at slow tempos.

Circuit Decisions

What Makes This Design Different

Post-Recovery High-Pass Filter

Most DIY reverb designs filter before the driver — a HPF or shelving filter before the spring to prevent low-end boom. This unit takes the opposite approach: the full-bandwidth signal enters the spring, and the HPF is applied to the wet output after the recovery preamp.

The result: the physical attack transient (the full-frequency "thump") enters the spring intact, creating a realistic reverb bloom. The reverb tail is then filtered. Low-end boom clears up as the note decays, which is more natural than a pre-filtered attack.

HPF values: C4 100nF film + R6 5.6kΩ → first-order corner 284Hz (~300Hz); the corner measured in the full simulation lands at ≈312Hz (the small shift is loading on the R6/C4 node). The bench pass window is 250–320Hz — expect ~312Hz on the meter. Use 5.6kΩ exactly — 4.7kΩ pushes the corner to 338Hz (too aggressive), 6.8kΩ drops it to 234Hz (insufficient mud rejection).

Bright Cap on Mix Pot

The Mix pot acts as a voltage divider at audio frequencies. At low mix settings, it attenuates high frequencies more than low frequencies — the reverb tails lose their shimmer before they lose their body. A 47pF silver mica capacitor bridges the full pot, end to end (CCW lug 1 to CW lug 3), bypassing it for high frequencies and keeping the reverb "glassy" as the wiper approaches full-wet. Note: the bright cap goes across the two end lugs, not from a lug to the wiper.

Mix pot is a 3-terminal passive blend, not a volume knob. Dry signal (from U1 via Rdry 10kΩ) enters the CCW end (lug 1); wet signal (from the Tone wiper) enters the CW end (lug 3); the wiper (lug 2) taps the blend and feeds U3. Wiring both signals to one lug — or the wet source onto the wiper — collapses the blend (this was a real netlist bug). See the build sequence and the builder guide for per-lug wiring.

Why silver mica: Silver mica is stable to ±1% over temperature. A ceramic disc capacitor will add high-frequency distortion that's audible in reverb tails. Do not substitute.

Emitter Bypass Capacitor (C2)

R5 (68Ω emitter resistor) provides thermal stability and sets quiescent current. Without C2, R5 degenerates the BD139's gain at audio frequencies — the spring is underdriven and loses high-frequency drive. C2 (100µF low-ESR Nichicon) bypasses R5 at audio frequencies, restoring full gain while keeping the DC stability of R5 intact.

Recovery Gain — 214× (46dB)

The spring tank's output is 1–5mV at low-to-moderate playing levels. The recovery preamp (U2 OPA2134 in non-inverting configuration) brings the signal up to line level: tank signal enters the (+) input via C3 (series coupling cap); Rbias (100kΩ) shunts the (+) input to GND for DC stability; Ri (470Ω) and Rf (100kΩ) form the feedback divider at the (−) pin → gain = 1 + (100k/470) = 214×. This is not unusual — spring tanks are low-output transducers, and this gain is normal for the stage.

Headroom note: 214× gain is sized for the low end of tank output. Hard pick transients can produce 10–20mV at the tank output, giving 2–4V at U2's output — well within the OPA2134's ±13V swing. Set Dwell conservatively on first use and verify U2 output stays clean on peaks. If clipping is heard at moderate playing levels, Ri can be raised slightly (e.g. 680Ω gives gain 148×) to add headroom.

DC offset at U2 output (bench reality): a real OPA2134's input offset (Vos up to ~500µV), multiplied by 214×, leaves a steady 20–150mV DC at U2's output. This is normal and is blocked by C4 before the output — do not chase it as a fault. The final output (after C4, at U3/J2) settles to <5mV.

Cf — strongly recommended on perfboard: fit a 10–22pF capacitor across Rf. On point-to-point construction, stray capacitance at U2's inverting input forms a pole with the 100kΩ Rf that can cause HF gain peaking or ringing in this 214× stage; Cf (22pF → ~72kHz corner, above the audio band) rolls the loop off cleanly. It is a cheap optional build-time part, not a BOM line — but recommended insurance against layout-dependent instability.

Component Rationale

Why Each Part

Every component was chosen for a specific engineering reason. These notes are the difference between a build that sounds right and one that doesn't.

Op-Amps — OPA2134PA

StageRefWhy OPA2134
Input BufferU1FET input (10¹³Ω) places virtually no load on the upstream device. Unity-gain stable.
Recovery PreampU2FET input is critical here — U2 must load the tank's 2550Ω output without signal attenuation. THD+N=0.00008%. Slew rate 20V/µs eliminates slew-limiting distortion on pick transients.
Output BufferU3Voltage follower — drives the MC100's RCA input and any cable capacitance without treble loss. <100Ω closed-loop output impedance.
Do not substitute

Do not replace OPA2134 with NE5532 or TL072. Both have significantly higher noise floors and lower input impedance — the NE5532 is bipolar-input (low impedance) and will attenuate the tank's output signal at the recovery stage. The OPA2134 is specified throughout for consistency; you need only one part number and the builder can swap a section if one fails.

Transistor — BD139 (TO-126)

High-current NPN bipolar rated 1.5A / 80V. At 18mA quiescent current it runs well within ratings with negligible heat. The TO-126 package mounts flat against the chassis for passive heatsinking if needed. BD139 has excellent hFE linearity at low currents — critical for low distortion in the driver stage.

Do not substitute

Do not use a TO-92 small-signal transistor (2N3904, BC547, etc.). Insufficient current capacity will clip drive transients and harden the reverb attack character. The BD139 TO-126 is specifically sized for this application.

Resistors — Yageo MFR 1% 250mW Metal Film

All resistors are metal film, 1% tolerance. Carbon film resistors have higher noise (current noise 10–100× higher) and temperature drift that degrades the signal floor, especially in the 214× gain recovery stage. Order 5 of each value — metal film resistors are cheap and you want spares for the build.

Critical Resistor Values

RefValueFunctionWhy This Value
R11MΩU1(+) → GND — bias return and input impedanceShunt to GND, placed after C_in. Provides the DC bias return path for U1's FET input — without it, U1(+) has no reference and drifts to a rail. Also sets 1MΩ input impedance (negligible load on upstream device).
R3b6.8kΩUpper bias dividerSets base voltage to ~1.92V → Ve~1.22V → Ic~18mA
R41kΩLower bias dividerDivider current ~1.92mA. R5 emitter degeneration is the actual bias-stabilizing element — verify by measuring Ve ≈ 1.22V, not the divider math.
R568ΩEmitter degenerationIc = Ve/R5 = 1.22V/68Ω ≈ 18mA. Primary bias stability and thermal runaway protection.
Ri470ΩGain set — (−) pin to GNDGain = 1+(Rf/Ri) = 1+(100k/470) = 214× ≈ 46dB. Tank signal enters (+) pin; Ri forms the lower leg of the feedback divider at the (−) pin.
Rf100kΩRecovery gain (feedback)100kΩ keeps thermal noise below OPA2134's own input noise
R65.6kΩHPF cornerFirst-order f = 1/(2π×5600×100n) = 284Hz; measured ≈312Hz in full sim (bench window 250–320Hz). Use exactly 5.6kΩ.
Rbias100kΩU2 non-inv (+) input → GNDSets recovery stage input impedance. 100kΩ ensures 97.5% signal transfer from the 2550Ω tank (470Ω would lose 84%). Sources from same 100kΩ stock as Rf — no extra BOM line.

Capacitors — Film Required in Signal Path

No ceramic caps in the signal path

Ceramic capacitors have piezoelectric microphonics and voltage-dependent capacitance (distortion) at audio frequencies. All signal-path capacitors must be WIMA MKS2 film. This is non-negotiable in a hi-fi circuit with 214× gain in the recovery stage.

RefValueTypeWhy
C11µF/63VWIMA MKS2 filmCoupling before driver. HPF corner = 1/(2π×1µ×10k) = 16Hz — well below guitar fundamentals.
C3470nF/63VWIMA MKS2 filmTank output coupling. Blocks DC offset from tank terminals. Corner at ~3Hz with Rbias = 100kΩ — purely DC-blocking, no audio attenuation.
C4100nF/63VWIMA MKS2 film300Hz HPF with R6. Must be film — ceramic drifts with temperature, shifting the cutoff.
C_bright47pFSilver micaBright cap across Mix pot. Silver mica ±1% stability, lowest HF distortion. Do not use ceramic disc.
C2100µF/25VNichicon UKW audio gradeEmitter bypass. Low-ESR is critical — high-ESR cap won't bypass R5 at high audio frequencies.
C5–C8100nF/63VWIMA MKS2 filmOp-amp supply decoupling — one per supply pin, placed as close as physically possible to IC. 4 total: 2 supply pins per DIP-8 package × 2 packages.

Potentiometers — Vishay/Spectrol 296 (MIL-PRF-39023)

Military-grade cermet element, gold-plated wiper, stainless shaft, rated 10,000+ cycles. These are overkill for a guitar rig but they will outlast everything else in the build and will never develop contact noise.

Note on taper: Mix (RV2) and Tone (RV3) are specified as audio taper — Bourns PDB181-GTR01-104A0. Audio taper matches how ears perceive level changes and gives the blend and tone controls an even feel across the full sweep. A 100kΩ linear pot in a passive blend against a 10kΩ dry resistor compresses most of the audible range into a narrow arc near minimum. Dwell (RV1) is correctly linear.

Builder's note — taper is a preference call: If MIL-PRF-39023 certification is required throughout, substitute Vishay/Spectrol 296UAL104B2 (100kΩ linear) for RV2/RV3. The circuit is electrically identical; only the feel of the controls changes. Some builders prefer linear for more predictable feel in studio use. Order what suits the intended application.

Output Stage

Low-Impedance Output Buffer

The output buffer (U3) drives the McIntosh MC100's RCA input and any cable between the unit and the power amp. Without a buffer, the Mix pot's output impedance varies with rotation — at center position it's at maximum (half of 100kΩ = 50kΩ), which would roll off high frequencies through cable capacitance and interact unpredictably with the MC100's input.

TopologyNon-inverting voltage follower (gain = 1×). Output connected to inverting input.
Input ZFET input (10¹³Ω) — negligible load on the Mix pot wiper
Output Z<1Ω at audio frequencies (op-amp closed-loop). Effectively zero.
Series RR7 100Ω — short-circuit protection; prevents oscillation into capacitive cable loads
Bandwidth>100kHz — well beyond audio
THD<0.001% (OPA2134 spec)

MC100 Interface

Power Supply

Internal ±15V Linear

The power supply is internal to the unit. No wall wart, no switching regulator — a toroidal transformer, bridge rectifier, and pair of linear regulators. This is the correct approach for a hi-fi reverb unit and the reason the circuit doesn't need external power management.

StagePartWhy
TransformerTriad F-219X, 30VA dual-primary toroidal (2×115VAC in / 2×15VAC out)Toroidal has ~10× lower magnetic field leakage than E-I laminate. Critical — the spring tank is a sensitive magnetic transducer and will pick up 60Hz hum from a nearby transformer. Keep transformer as far from the tank as the chassis allows. Primary wiring: Wire the two 115VAC primary windings in parallel for 120VAC mains (see build sequence). Secondary wiring: Wire the two 15VAC secondary windings in series — the junction of the two windings is the 0V/GND reference (forming 15-0-15), and the outer ends connect to BR1's AC inputs. Wiring the secondaries in parallel gives a single rail with no negative supply — the op-amps and LM7915 will not receive −15V.
RectifierW04G bridge, 2A/400V400V is derated 10× from the 42V peak. 2A rating gives 4× headroom over typical load — reliable even with inrush.
Filter2× 1000µF/50V Nichicon UKW low-ESR~350mVpp bus ripple at the ~40mA worst-case load — the regulators' ~68dB 120Hz rejection turns that into <1mV on the rails. Downsized from 2200µF (#95): low-mains trough still clears the regulator dropout floor by ~0.75V. 50V rated for headroom; low-ESR minimizes heat.
+15V RegLM7815CT (TO-220)Linear regulation = zero switching noise. Switching regulators inject kHz noise that passes through op-amp power supply rejection. LM7815 is proven and inexpensive.
−15V RegLM7915CT (TO-220)Same rationale. Provides negative rail for op-amp negative supply pins.
Reg output caps3× 25µF per rail (parallel, 75µF) + 2× 100nF filmElectrolytic for stability per LM78xx/79xx datasheet (LM7915 floor ≥25µF aluminum); parallel wiring, no balancing resistors (#94). Film in parallel for HF suppression above ~100kHz.
Insulating pads required on TO-220 regulators

The TO-220 tab is electrically connected to the output pin. Without mica insulating pads between each regulator and the chassis, mounting both regulators to the same chassis creates a short between +15V and −15V through chassis ground. Use TO-220 mica pads + M3 nylon screws or an insulated shoulder washer.

Enclosure

2U Rackmount Chassis

The 2U form factor is required — 1U is too tight to mount the 9AB3C1B tank horizontally with spring clearance. 2U gives comfortable room for the tank, PCB, power supply, and ventilation.

Chassis

Hammond 1455T2201 — 2U aluminum rackmount. Aluminum is mandatory: steel would interact magnetically with the toroidal transformer. Rack ears are included with the Hammond chassis.

Internal Layout

Rear Panel ─────────────────────────────────── Front Panel
│                                                  │
│  ┌──────────────────────────┐                   │
│  │                          │                   │
│  │  Spring Tank (9AB3C1B)   │     ┌──────────┐  │
│  │  horizontal, open down   │     │ Audio PCB │  │
│  │  grommet-isolated        │     │ (Vector   │  │
│  │                          │     │  T44)     │  │
│  └──────────────────────────┘     └──────────┘  │
│                                                  │
│  ┌──────────┐                                    │
│  │ PSU PCB  │ ← toroidal transformer here        │
│  │ (Triad   │   as far from tank as possible     │
│  │  F-219X) │                                    │
│  └──────────┘                                    │
│                                                  │
────────────────────────────────────────────────────
       ↑ IEC C14 inlet · Fuse · Rocker switch ↑

Front Panel Controls

ControlTypeFunction
DWELL10kΩ linear pot, ¼" D-shaftDrive level into transformer — spring saturation and reverb density
MIX100kΩ audio taper pot + 47pF bright cap (Bourns PDB181-GTR01-104A0)Dry/wet blend
TONE100kΩ audio taper pot (Bourns PDB181-GTR01-104A0)High-shelf EQ on wet signal only

Rear Panel Connectors

ConnectorTypeNotes
InputSwitchcraft 112A ¼" TSFrom Lehle Parallel M output (future state) or directly from Alembic FX-1
OutputSwitchcraft 112A ¼" TSTo McIntosh MC100 RCA input (with 1/4" TS → RCA adapter cable)
IEC PowerSchurter 5110.1052 (EMI-filtered IEC C14 + fuse holder)500mA slow-blow fuse. Integrated EMI filter suppresses mains-borne noise before it reaches the transformer. Slow-blow essential — inrush current at power-on blows fast-blow fuses.
Power SwitchTE 1825232-1 SPST rocker, 6A/250VInterrupts mains before transformer primary

Tank Mounting

Front Panel — Custom via Front Panel Express

Custom 2U aluminum panel with laser-drilled holes and engraved labels. Full hole list:

Use Front Panel Express designer software to spec all hole positions to fit the Hammond 1455T2201 chassis dimensions.

Build Notes

Critical Rules for the Builder

These five rules are the difference between a working build and one that hums, oscillates, or drifts. Non-negotiable.

  1. Star grounding. All grounds — op-amp ground pins, pot grounds, jack grounds, PSU output — connect to a single point on the chassis, near the IEC inlet. No daisy-chaining. Any second chassis ground connection creates a ground loop and 60Hz hum.
  2. Decoupling caps physically close. C5–C8 (100nF film, one per op-amp supply pin — 4 total: 2 pins per package × 2 packages) must be placed within 1" of the OPA2134 supply pins on the perfboard. Further than that and they won't suppress high-frequency noise — the circuit will likely oscillate or have a high noise floor.
  3. Tank orientation: open-side down. Mount the 9AB3C1B horizontally with the open side (where you can see the springs) facing down. Springs hang below the transducers. If mounted open-side up, springs shift over years and reverb character drifts unpredictably.
  4. Transformer placement and orientation. Mount the Triad F-219X as far from the spring tank as the chassis allows. Orient it so the transformer's toroid axis is perpendicular to the tank's spring axis. This minimizes inductive coupling (60Hz hum in the reverb tail).
  5. Shield the recovery wire. The wire from the tank's output (2550Ω side, RCA connector) to U2's non-inverting input is the most sensitive signal in the circuit — 1–5mV. Use Belden 8451 shielded cable. Ground the shield at the U2 end only (not the tank end) — one-end grounding prevents a shield ground loop that would induce 60Hz hum directly into the recovery stage.
  6. Protective earth: green/yellow wire, star washer. The IEC C14 inlet earth pin must be bonded directly to the chassis with green/yellow wire. Use a star washer under the lug to cut through any anodizing on the aluminum and ensure a solid metal-to-metal connection. This is a safety earth, not a signal ground — it connects to the chassis, not the audio star ground point. Do not omit it.

PCB — Vector T44 FR4 Perfboard

FR4 fibreglass is mandatory. The cheaper brown phenolic perfboard absorbs moisture and increases leakage current between pads — at the recovery stage gain of 214× this manifests as audible noise. Use 4× M3 nylon hex standoffs (not metal) to mount the board — metal standoffs can accidentally create a second chassis ground connection.

Build Sequence

Build in this order to make each stage testable before adding the next:

  1. PSU first. Before installing T1, make both wiring connections: (1) Primaries in parallel for 120VAC — connect both start leads together and both finish leads together (observe the phasing dots on the datasheet); series wiring gives 230VAC and incorrect secondary voltages. (2) Secondaries in series for 15-0-15 bipolar supply — connect one end of winding A to one end of winding B; that junction is the 0V/GND reference. The two remaining outer ends connect to BR1's AC input pins. Wiring the secondaries in parallel gives a single ~15VAC rail with no negative supply — the LM7915 and all op-amp negative supply pins will be unbiased. Once wired, install T1, BR1, NTC1, C11/C12, U4, U5, reg output caps (C13/C14 3×25µF parallel per rail + C17/C18 100nF), C15/C16 bulk decoupling. Power up with a variac or current-limited bench supply if available. Measure: +15V at U4 output pin, −15V at U5 output pin, <50mV ripple on each rail. Do not proceed until both rails are correct.
  2. Input stage. Install C_in (1µF — from input jack to U1(+) node), then R1 (1MΩ — from the same U1(+) node to GND, as a shunt, not in series), D_clamp+/D_clamp− (1N4148 pair — clamp the U1(+) node to ±15V rails), TVS1 (SMBJ15CA — across input jack tip-to-sleeve), U1 (input buffer OPA2134). R1 must be on the downstream (U1) side of C_in to provide a DC bias return path for U1's FET input. Inject a 1kHz sine at −20dBV. Scope at U1 output: clean sine, no clipping, gain ≈1.
  3. Driver stage. Install Q1 (BD139), R3/R3b/R4 (bias network), R5 (emitter), C1 (1µF coupling), C2 (100µF bypass — positive lead to Q1 emitter, negative to ground), D3 (flyback clamp), T2 (REB3S driver transformer). Connect the tank's 8Ω input side. Scope at T2 primary: sinusoidal drive signal, no ringing or DC offset. Measure Q1 emitter DC: expect 1.0–1.4V at zero signal (verified sim 1.09V → Ic ≈ 16mA; first-order 1.22V → 18mA — both pass). The collector will read ≈ +15V — T2's transformer primary is a near-DC short to the rail, so Vce ≈ 13.8V drops across the transistor itself, not a load resistor. Do not expect a mid-rail reading at the collector.
  4. Tank output and recovery. Install C3 (470nF), Rbias (100kΩ), Ri (470Ω), Rf (100kΩ), U2 (recovery OPA2134). Connect the tank's 2550Ω output side. Tap the spring with a finger — you should hear a reverb impulse at U2's output. Scope gain at U2 output vs. U2 (+) input: should be ≈214×.
  5. HPF and tone. Install R6 (5.6kΩ), C4 (100nF), RV3 (Tone pot). Verify HPF corner is ~300Hz by sweeping a function generator — signal below 300Hz should attenuate; above 300Hz passes flat.
  6. Mix and output buffer. Install Rdry (10kΩ), C_bright (47pF), RV2 (Mix pot), U3 (output buffer OPA2134), R7 (100Ω isolation). Wire RV2 as a 3-terminal passive blend: dry from U1 via Rdry → lug 1 (CCW); wet from the Tone wiper → lug 3 (CW); wiper lug 2 → U3(+); C_bright (47pF) across lug 1 ↔ lug 3 (the full pot). Do not tie both signals to one lug or the wet source to the wiper — that collapses the blend. Connect input jack J1 and output jack J2. Sweep Mix pot: CCW = dry only, CW = wet only, mid = blend.
  7. Front-panel pots and wiring. Install RV1 (Dwell), connect Molex KK housings to all panel pots and jacks. Retest full signal chain end-to-end with guitar signal.
  8. Chassis install, grounding, and final checks. Mount PCB on standoffs, connect star ground, install IEC inlet, fuse, rocker switch, ground lift switch. Check: no hum at idle, Dwell changes reverb density, Mix blends correctly, Tone brightens wet signal. Run for 30 minutes; check that regulators are warm but not hot (<60°C case temp).

Test Points Summary

Test PointExpectedFault if wrong
U4 output pin+15.0V ±0.5V DCRegulator or rectifier issue
U5 output pin−15.0V ±0.5V DCRegulator or rectifier issue
PSU rails ripple<50mV p-p at idleFilter cap ESR or wiring issue
Q1 emitter (no signal)1.0–1.4V DC (sim 1.09V)Bias resistors (R3b/R4/R5) or BD139 fault — this is the correct node to measure for bias verification
Q1 collector (no signal)≈ +15V DCT2 primary is near-DC short to rail — a reading well below +15V means T2 is open or Q1 is saturated
U2 (+) input (signal)1–5mV at normal playing levelTank not driven, C3 open, or Rbias fault
U2 output (signal)~200–1000mV at normal playing levelOp-amp fault, Ri/Rf values wrong
U3 output (idle)<5mV DC offsetDC offset from upstream; check Rbias and Ri
Chassis ground to star pointGround lift switch in wrong position, or open connection

Phase Alignment — First Power-Up Check

Spring tank polarity varies between Accutronics batches. If the reverb sounds hollow, thin, or "phasey" when mixed in at 50/50, the tank output is out of phase with the dry signal. Fix: on the output Molex KK connector at the tank's 2550Ω side (the RCA that goes to U2's input), swap the two wires. The input connector (8Ω driver side) is not involved. No design change needed — this is a 10-second wire swap on first commissioning.

Conformal Coating

After all testing is complete and the build is verified, apply MG Chemicals 422B acrylic conformal coating spray to the PCB to protect against humidity and oxidation over time. Mask or skip the following before spraying — coating these will cause problems:

Apply in two thin coats, allowing full dry time between coats. Hold the can 8–12" away and use short strokes — thick single-pass coats pool and crack.

Solder

Kester 44, 63/37 tin/lead, 0.031". Kester 44 is the standard for hand-soldered audio. 63/37 eutectic alloy — snaps solid with no mushy semi-solid phase, reducing cold joints. 0.031" diameter is correct for through-hole perfboard work. Do not use lead-free — higher melting point increases heat stress on sensitive components and produces higher-resistance joints.