Analog Board Troubleshooting

Analog circuits don't have the clean 1/0 verdicts of digital — signals are degraded, not just dead. The compensation: analog circuits are chains of recognizable stages, each with predictable DC conditions you can check with a DMM before any signal even matters. Learn the stages, learn their DC fingerprints.

1. Power supplies — where most repairs live

Every board has one; most board failures involve it. Two families:

Linear (the simple chain)

Input → transformer (if AC) → rectifier → bulk filter cap → linear regulator → output caps → load

DC fingerprints, in order:

1. Input present? (Fuse, connector, switch)
2. After rectifier: DC ≈ peak of AC input (×1.41 minus diode drops). Hum/ripple here is normal — big ripple means the bulk cap is dying.
3. Regulator input ≥ output + dropout (~2V classic, ~0.1–0.5V LDO). A regulator with insufficient input headroom isn't broken — it's starved.
4. Output at nominal ±tolerance, nearly ripple-free.

Classic faults: open fuse (find why), shorted rectifier diode (often takes the fuse), dried bulk cap (hum, sag under load), shorted regulator (output = input voltage — dangerous downstream!), regulator in thermal shutdown (cycles on/off as it heats — feel it / freeze it).

Switching (SMPS) — efficient, busier

Input → input filter → switching element (MOSFET) at tens–hundreds of kHz → inductor/transformer → rectifier (diode or synchronous FET) → output caps → feedback divider → controller IC

Bench approach:

1. Output dead: check input voltage AND controller enable pin (sequencing! — see 07 — Digital Board Troubleshooting §3). Then is the switch node switching (scope: a hard square waveform)? No switching with good input+enable = controller or its supply. Switching but no output = inductor/rectifier/shorted output.
2. Output wrong value: feedback path — the divider sets the output; a drifted/cracked divider resistor literally re-programs the supply.
3. Output present but ripply/noisy: output caps (ESR) — the #1 SMPS wear-out.
4. "Hiccup" (output pulses up and collapses repeatedly): the supply is current-limiting into a downstream short — the supply is often fine; find the short (04 — DMM Mastery §4).
5. Shorted main MOSFET: usually obvious — blown fuse, burnt part, near-0Ω across it. Replace its driver-circuit companions too; the killer often lurks there (and check the rectifier — cascade failures are the norm in SMPS).

⚠️ Off-line (mains-fed) SMPS primaries are at lethal, earth-referenced-scope-destroying potentials and bulk caps hold charge after unplugging. Discharge and verify before touching; differential probe on the primary side.

2. Op-amp stages — the analog workhorse

DC fingerprint of a healthy op-amp with negative feedback: the two inputs sit at (very nearly) the same voltage (the "virtual short"), and the output sits wherever the feedback math puts it.

The three-measurement op-amp check:

1. Supply pins (V+ and V−; could be +15/−15, +5/0, etc.) — right there at the pins, not just at the rail TP.
2. Both inputs — equal? If the inputs differ by more than millivolts, either the feedback loop is broken (open feedback resistor/trace → output slams to a rail) or the op-amp is dead.
3. Output — at the predicted DC level, and not parked at a rail. Output stuck at a rail: inputs explain it? (Input difference says it should be there → fault is upstream or in feedback.) Inputs say it shouldn't? → dead op-amp.

Then signal: inject/follow AC through the stage — gain ≈ resistor ratio (inverting: −Rf/Rin; non-inverting: 1+Rf/Rg). Distorted peaks = stage running out of supply headroom or bias shifted.

Comparators look like op-amps but run open-loop (inputs NOT equal — that's normal for them) and often have open-drain outputs needing a pull-up.

3. The rest of the analog zoo — quick fingerprints

• Voltage references (e.g., 2.500V, 4.096V parts): one DMM reading. A drifted reference skews every measurement the board makes — on precision/instrumentation CCAs check it early; ATE often flags it as many simultaneous "out of tolerance" failures.
• Filters (RC/LC): passives — verify with the in-circuit resistance rules and capacitor checks (03 — Component Identification & How Each Part Fails, 04 — DMM Mastery). Symptom of a dead filter cap: noise/garbage passing where it shouldn't.
• Analog switches/multiplexers: verify control logic levels, then continuity through the selected channel (powered, signal-level test). A blown channel (overvoltage) passes nothing or leaks.
• ADCs/DACs: boundary parts — verify reference, rails, clock, then known-input vs reported-code (functional test territory).
• Sensor front-ends (instrumentation amps, bridges): check excitation voltage, then the amp with the op-amp ritual.
• Trimmers/pots: mechanical, vibration-sensitive — wipers go intermittent/open with age. Don't re-adjust sealed/calibrated trimmers casually: in aerospace, calibration adjustments are controlled operations — a paint-sealed trimmer is sealed for a reason; breaking the seal without authorization is a process violation.

4. Analog failure signatures

Multichannel comparison is the analog superpower: instrumentation and aerospace boards often have 4/8/16 identical channels. The good channels are your schematic-free reference — measure the same node on each; the outlier is your fault.

5. Grounding subtleties worth respecting

Precision analog boards separate analog and digital grounds, joined at one point. Two repair implications:

• A measurement referenced to the wrong ground can be millivolts-to-volts off — clip your black lead/scope ground to the section's own ground.
• A cracked joint in the ground tie (or a corroded chassis standoff) creates offsets, hum, and crosstalk that look like component failures. When a board has multiple inexplicable analog symptoms at once, measure ohms across the ground system before chasing each symptom individually.

6. Self-check

1. Linear regulator: 28V at input, 4.96V at output (5V nominal). Verdict? Reveal answer
2. Op-amp inputs read +2.1V and +2.1V; output is mid-range. Op-amp health? Reveal answer
3. Op-amp output slammed at +14.7V (15V rail); inputs read 3.0V and 1.2V. Reveal answer
4. SMPS output pulses up to 1V repeatedly, never reaches 5V. Reveal answer
5. 8-channel board, channel 5 reads offset by 0.4V vs siblings at the second op-amp stage. Next move? Reveal answer

Next: 09 — Teradyne Machines & Automated Test (ATE)