The Repair LibraryRead · Learn · Master

Diagnosing with the Oscilloscope

The chapter has built five skills one at a time: understanding the instrument, locking a waveform with the trigger, making the probe tell the truth, turning the trace into numbers, and capturing events that will not repeat. This closing section assembles them into a single working method — the thing you actually do when a faulty board is on the bench and the scope is beside it. It starts with where to put the probe first, because a scope session is a sequence of chosen nodes, not a wander: power rails before anything else, then the clock and reset that let logic run, then the signal path the symptom points to. It installs the setup ritual — verify the probe chain, lock the trigger, frame the trace — run identically at every node, so that a mis-set scope can never masquerade as a faulty circuit. It teaches judgment by waveform signature: every node has an expected shape, level, and timing, and diagnosis is the comparison of what is seen against that expectation, supplied by a datasheet, a schematic, or a known-good comparison. It adds the escalation ladder that chooses the capture method by how often the fault shows itself — measure the steady, accumulate the suspicious with persistence, trap the rare with a tripwire single-shot. And it closes with two-channel correlation, the step that turns a captured effect into a cause by recording the suspect and its neighbour in the same sweep, then walks a complete diagnosis from symptom to the waveform that names the fault. By the end, the oscilloscope stops being five separate skills and becomes one instrument in your hands — the most revealing one on the bench, pointed by method instead of luck.

IntermediateMedium Risk22 min read

What You Will Learn

  • You will learn to choose the first probe point from the symptom — rails first, then clock and reset, then the signal path.
  • You will learn to run the same setup ritual at every node so a mis-set scope never masquerades as a fault.
  • You will learn to judge a node by its waveform signature against a known-good comparison or documented expectation.
  • You will learn to escalate the capture method to match how often the fault shows itself.
  • You will learn to correlate two channels in one record to tie a captured effect to its cause.

What You Will Be Able To Do

  • You will be able to choose the first probe point from the symptom — rails first, then clock and reset, then the signal path.
  • You will be able to run the same setup ritual at every node so a mis-set scope never masquerades as a fault.
  • You will be able to judge a node by its waveform signature against a known-good comparison or documented expectation.
  • You will be able to escalate the capture method to match how often the fault shows itself.
  • You will be able to correlate two channels in one record to tie a captured effect to its cause.

Required Tools

  • A digital storage oscilloscope with two channels
  • A compensated 10x probe per channel with short ground paths
  • A faulty or fault-injected board to diagnose
  • A known-good board, channel, or documentation for comparison
  • A notebook for signatures, expectations, and findings

Section Overview

Five skills — instrument, trigger, probe, measurement, capture — become one method in this closing section: the sequence a technician actually runs when a faulty board meets the scope (the-oscilloscope-as-the-diagnostic-instrument). The probe goes to chosen nodes in a chosen order. Power rails are checked first because nothing works without them, then the clock and reset that let logic run, then the signal path the symptom points to — a scope session is a route, not a wander. The same ritual runs at every node. Verify the probe chain, lock the trigger, frame the trace — run identically each time, so a difference on the screen can only mean the circuit, never the setup (probing-compensation-grounding-and-loading). Each node is judged by its signature. A waveform signature is the shape, levels, and timing a healthy node shows, and diagnosis is the comparison of the trace against that expectation — from a datasheet, a schematic, or a known-good comparison. The capture method escalates with the fault's rarity. Capture escalation matches technique to how often the fault appears: a steady fault is measured, a suspicious signal is accumulated with persistence, and a rare event is trapped with a tripwire single-shot (capturing-transients-and-single-shot-events). And two channels turn an effect into a cause. Two-channel correlation records the suspect and a related signal in the same sweep, so "the rail sagged" becomes "the rail sagged when the load switched" (reading-and-measuring-waveforms). Route, ritual, signature, escalation, correlation — and the oscilloscope becomes one pointed instrument instead of five separate skills.

Why This Matters

Each skill in this chapter is necessary and none alone is sufficient — diagnosis happens when they run in order, on the right nodes, against the right expectations (the-oscilloscope-as-the-diagnostic-instrument). This matters because the probe order finds faults early: most dead and misbehaving boards fall at the rails or the clock, so checking them first resolves many diagnoses in minutes and scopes the rest (reading-and-measuring-waveforms). This matters because the ritual protects the verdict: a wrong probe factor or a drifting trigger imitates a fault exactly, and only a setup run identically at every node keeps the instrument off the suspect list (probing-compensation-grounding-and-loading). It matters because a trace without an expectation is just a picture: the signature — expected shape, levels, timing — is what turns looking into judging, and the known-good comparison supplies it when documentation runs out. It matters because fault rarity decides technique: measuring a steady fault, accumulating a rare one, and trapping a one-shot are different jobs, and choosing wrong wastes hours watching or erases the evidence (capturing-transients-and-single-shot-events). And it matters because correlation closes the case: a captured symptom names a fault; the second channel showing what moved first names the cause. Run the method end to end, and the scope takes a fault from symptom to the waveform that reveals it — which is what this chapter promised.

Required Prerequisites

  • A compensated 10x probe per channel — to run the setup ritual and correlate two nodes at once (probing-compensation-grounding-and-loading)
  • A notebook or worksheet — to record signatures, expectations, and each node's verdict as the route is walked
  • Service documentation, schematics, or datasheets — to supply expected signatures before probing
  • A known-good board or channel — to capture reference waveforms where documentation runs out
  • Probe accessories — ground springs and clips — to keep the ritual fast and the edges honest at every node
  • A two-channel digital storage oscilloscope — to practise correlation and the full method (the-oscilloscope-as-the-diagnostic-instrument)
  • A board with a known, steady fault — to walk the route and find it by signature
  • A board with an induced intermittent — to practise escalating from watching to trapping
  • A matching known-good board — to practise known-good comparison node by node
  • A rail with a switchable load and a visible enable or control line — to practise two-channel cause-and-effect captures (capturing-transients-and-single-shot-events)
  • A notebook of captured signatures — to build the reference library that speeds every future diagnosis

Real-World Applications

The assembled method is what working repair looks like at the bench. A technician handed a dead router probes the rails first, finds the 3.3 V rail flat, and never needs to look at a data line — the route ended the session at its first stop (reading-and-measuring-waveforms). A repairer facing a console that boots erratically verifies rails and clock clean, then walks the signal path and finds a reset line whose signature — a crisp release after power-good — is instead a slow, drooping ramp. Someone chasing a "sometimes crashes" laptop escalates: the rail measures clean, persistence shows an occasional sag ghost, and a tripwire single-shot finally traps it — depth, duration, and timestamp (capturing-transients-and-single-shot-events). A technician with the sag captured adds the charger's enable line on channel two, re-arms, and the correlated record shows the enable dropping first — the cause, not just the symptom. And a repairer without a schematic captures each suspect node on a known-good unit of the same model and diagnoses by difference (the-oscilloscope-as-the-diagnostic-instrument). The failures this prevents: hours on data lines while a rail was down, a verdict issued off a mis-set scope, and a captured symptom filed as a cause because no second channel asked what moved first.

Common Challenges

  • Knowing what the waveform should be. The scope shows what is there; diagnosis needs what should be there — the difficulty is assembling expectations from datasheets, schematics, markings, and known-good captures before judging, which is slower than probing but is the actual work (reading-and-measuring-waveforms).
  • Resisting the interesting detour. A busy screen offers endless curiosities, and an unstructured session dissolves into wandering — the difficulty is holding the route — rails, clock and reset, signal path — and parking oddities in the notebook until their turn.
  • Choosing the wrong rung of the ladder. Watching a rare fault wastes hours; trapping a steady one wastes setup — the difficulty is reading the fault's rarity from the symptom history first, so the capture method matches the quarry (capturing-transients-and-single-shot-events).

Safety Notes

Risk Level: Medium. A full diagnostic session is sustained live-board work — multiple nodes, two probes, and sometimes long unattended captures — so every discipline from the chapter applies at once.

Professional Tips Before Starting

  • Write the route before powering up. Rails, clock and reset, then the symptom's signal path — a listed route keeps the session moving and the detours parked (the-oscilloscope-as-the-diagnostic-instrument).
  • Capture the healthy version first. When a known-good board is available, record its signatures before touching the patient — comparison is fastest when the reference is already on screen.
  • Log every node's verdict. "Rail clean, clock clean, reset slow" is a diagnosis taking shape — a one-line verdict per node turns probing into progress.

The Oscilloscope Method — From Symptom to Waveform

Recap and Frame

This chapter opened by calling the oscilloscope the most revealing instrument on the bench; this section is the method that makes the revelation reliable (the-oscilloscope-as-the-diagnostic-instrument). The five skills are stages of one act. Understanding the instrument, locking the trigger, proving the probe, measuring against expectation, and capturing the rare are not five sessions — they are the standing skills one diagnosis draws on in sequence. The method has a shape. Choose the node from the symptom, run the setup ritual, judge the trace against its signature, escalate the capture if the fault hides, and correlate a second channel when the cause is the question. The route replaces wandering. Rails first, clock and reset second, signal path third — the order that finds most faults earliest and scopes the rest (reading-and-measuring-waveforms). The ritual replaces doubt. An identical setup at every node means a strange trace indicts the circuit, not the instrument (probing-compensation-grounding-and-loading). And the ladder replaces patience-as-strategy. Steady faults are measured, suspicious ones accumulated, rare ones trapped — technique chosen by rarity instead of hoping the fault performs on cue (capturing-transients-and-single-shot-events). Hold the frame — route, ritual, signature, escalation, correlation — and a fault becomes a short walk from symptom to waveform.

The Route — Where the Probe Goes First

A scope session is a sequence of chosen nodes, and the order is chosen for how electronics actually fail (the-oscilloscope-as-the-diagnostic-instrument). Rails come first. Nothing downstream works without its supply, and rail faults — missing, low, sagging, noisy — cause the majority of dead and erratic boards, so the rails are probed before any signal is considered. The rail check is a scope check. A meter calls a rail present; the scope adds what the meter averages away — ripple, sag under load, and noise riding the level — so even a "good" rail earns a moment under the trace (reading-and-measuring-waveforms). Clock and reset come second. Logic that has power still needs its heartbeat and its release: a clock at the right frequency with healthy edges, and a reset that releases cleanly after power stabilises — two quick signatures that let everything downstream run. The signal path comes third. With supplies and heartbeat cleared, the symptom picks the path — the chain that carries the failing function — walked node by node with the trace judged at each stop. The symptom can shortcut the route. A symptom that names its subsystem — no charge, no backlight, no audio — starts the signal-path walk in that block, but rails and clock are still glanced first because their faults imitate everything else. Rails, then heartbeat, then the path the symptom names, with shortcuts earned only after the first two are cleared — and the route is set. Walk the route in order, and most faults surrender at the first or second stop.

The Setup Ritual — The Same Scope at Every Node

Every node on the route gets the same setup, because a difference on screen must have only one possible source: the circuit (probing-compensation-grounding-and-loading). The probe chain is verified once and guarded after. The session opens on the calibration output — amplitude proving the probe factor, corners proving compensation, a cable flex proving the probe — and any probe or channel change repeats the check. The ground stays short and true. Every stop grounds the probe to verified circuit ground, with a ground spring where edges matter, so ringing on the screen is the circuit's ringing and not the lead's. The trigger is locked before judging. A stable trace — level in the swing, mode chosen, source on the signal that defines the event — comes before any conclusion, because a drifting or shimmering trace misreads as a fault. The frame fits the trace. The waveform fills the screen unclipped with at least one full cycle visible, so measurements read the signal rather than the screen's limits (reading-and-measuring-waveforms). Deviations from ritual are logged, not improvised. When a node forces a nonstandard setup — a 1x probe, an unusual coupling — the notebook records it beside the verdict, so the reading is never later mistaken for a like-for-like comparison. Chain verified, ground short, trigger locked, frame fitted, exceptions logged — the ritual runs the same at every stop. Make the scope the constant, and every difference on screen belongs to the board.

The Signature — Judging Against Expectation

A locked, honest trace still needs a verdict, and the verdict is a comparison against what the node should show (reading-and-measuring-waveforms). A signature is shape, levels, and timing. A healthy node's waveform has an expected form — square, sine, stepped — sitting at expected levels with expected timing, and a fault is a departure in any of the three: the right shape at the wrong level, the right level with sick edges, the right everything at the wrong rate. Documentation supplies the expectation first. Datasheets give clock frequencies and logic levels, schematics give rail values and signal directions, and board markings name the crystal — the expectation is written in the notebook before the probe lands. The known-good comparison supplies the rest. Where documentation runs out, the same node on a known-good board or the same signal on a working channel is captured as the reference, and diagnosis becomes a difference test between two traces. The library compounds. Every signature captured and filed — this model's rail ripple, this bus at idle, this reset release — makes the next diagnosis of the same family faster, which is how experienced technicians seem to glance and know. Judgment stays honest about tolerance. Real boards vary, so a signature match is a match within reason — nominal levels within their stated tolerance, edges clean rather than identical — and only departures that exceed the expected spread earn suspicion. Shape, levels, and timing, expected before probing, referenced against documentation or a known-good, filed for next time, and judged with tolerance — and the trace gets its verdict. Judge every trace against an expectation, and looking becomes diagnosing.

The Ladder — Escalating the Capture

Faults differ in how often they show themselves, and the capture method is chosen to match — escalating only as far as the fault demands (capturing-transients-and-single-shot-events). A steady fault is measured. A fault present on every sweep — a low rail, a slow edge, a wrong frequency — needs no hunting: the ritual, the signature, and the cursors close it at the first stop. A suspicious signal is accumulated. A node that looks right but is accused by the symptom goes under persistence — infinite for a standing watch — and a rare deviation builds into a visible ghost while a clean accumulation moves suspicion elsewhere. A rare event is trapped. A fault that strikes occasionally gets a tripwire single-shot: the healthy envelope measured, the level staked just beyond it on the fault's first move, the frame rehearsed, and the scope left to wait. Narrow quarry changes the acquisition. A hunt for spikes and dropouts runs in peak-detect so the event survives the record, and never in averaging, which erases exactly what is being hunted. Each rung reuses the last. The trap is aimed by the ghost persistence showed, and the trapped record is measured with the same cursors as a steady fault — the ladder is cumulative, not alternative. De-escalation is information too. A tripwire that never fires and a persistence watch that stays clean acquit the node within the watch's limits — a persistence display is blind between acquisitions, and a watch is only as long as it ran — and the route moves on with the notebook updated. Measure the steady, accumulate the suspicious, trap the rare — in peak-detect when the quarry is narrow, each rung feeding the next, with silence counted as evidence. Match the method to the fault's rarity, and no fault outwaits the bench.

The Correlation — From Effect to Cause

A captured fault names the symptom precisely; the second channel is what names the cause (capturing-transients-and-single-shot-events). Two channels share one record. Both channels are sampled into the same capture around the same trigger, so the timing between them is real — what moved first, what followed, and by how long, read directly off the stored trace. The pairing is chosen by hypothesis. Channel one holds the symptom — the sagging rail, the glitching line — and channel two holds the suspected cause: the load that switches, the enable that gates, the clock that times, chosen from the schematic's short list of what could move that node. The trigger stays on the symptom. The tripwire remains staked on the abnormal event, so every capture contains one fault and, beside it, what the suspect was doing at that instant. First mover carries the blame. In the correlated record, the trace that moved before the symptom sits upstream of it — or shares an upstream cause the next capture must find — so an enable dropping before the rail sags indicts the enable's source, while a rail sagging before the enable drops exonerates it. A wrong suspect is progress. A second channel that sits innocent through the fault crosses one cause off the schematic's list, and the next candidate takes its place on channel two — hypothesis by hypothesis until the mover is found. One record, a hypothesis on channel two, the trigger on the symptom, the first mover blamed, and each innocent suspect narrowing the list — and the capture becomes a cause. Ask the second channel what moved first, and the diagnosis names a component instead of a symptom.

The Walk — A Complete Diagnosis

Assembled, the method reads as one continuous walk, and a worked example fixes its shape (the-oscilloscope-as-the-diagnostic-instrument). The symptom frames the job. A device restarts at random — no pattern the owner can name — so the fault is rare, the suspect list starts at the supply, and the session begins with the route and the ritual: probe chain verified on the calibration output, ground spring fitted. The route clears the steady possibilities. The main rail measures nominal with healthy ripple, the clock sits on frequency with clean edges, and reset holds high — no steady fault, so the ladder must climb (reading-and-measuring-waveforms). Persistence finds the ghost. Infinite persistence on the rail during normal operation slowly accumulates a faint downward ghost — a sag is visiting — and its depth on screen tells where the tripwire belongs. The trap catches the event. A single-shot arms just below the healthy minimum on a falling slope, mid-record; twenty minutes later the capture holds a deep sag with the processor's reset visible in the aftermath — symptom confirmed, cause still open. Correlation names the mover. Channel two takes the schematic's first suspect — the charging stage's enable — and the re-armed trap catches the next sag with the enable dropping first: the restart is the charger's fault, not the rail's (capturing-transients-and-single-shot-events). The notebook closes the loop. Signatures, captures, and the verdict are filed under the model, so the next board with this complaint starts at the charger with the tripwire recipe ready. Symptom framed, steady faults cleared, ghost found, event trapped, mover named, and the finding filed — one walk, five skills, one diagnosis. That walk — not any single control — is what diagnosing with the oscilloscope means.

Common Mistakes

  • Probing the symptom before the supplies. Data lines get hours while a sagging rail goes unchecked — walk the route: rails, clock and reset, then the signal path (the-oscilloscope-as-the-diagnostic-instrument).
  • Skipping the ritual at "quick" stops. One unverified probe factor and the verdict is worthless — the setup ritual runs at every node, especially the fast ones (probing-compensation-grounding-and-loading).
  • Judging a trace with no expectation. A waveform cannot be wrong until "right" is defined — write the expected signature from documentation or a known-good before the probe lands.
  • Camping on the wrong rung. Staring at a live screen for a rare fault, or building a trap for a steady one, wastes the bench — read the fault's rarity from the symptom history and pick the rung to match (capturing-transients-and-single-shot-events).
  • Filing the symptom as the cause. A captured sag is an effect until something is shown moving first — put the suspect on channel two and let the record assign blame.

Troubleshooting Guidance

The method comes down to route, ritual, signature, escalation, correlation. If the board is dead or erratic and nothing is known: start the route — rails under the trace first, then clock and reset, then the symptom's path (reading-and-measuring-waveforms). If a trace looks wrong at any stop: re-run the ritual before judging — chain, ground, trigger, frame — so the setup cannot be the fault (probing-compensation-grounding-and-loading). If the trace is honest but the verdict is unclear: the expectation is missing — pull the signature from the datasheet, the schematic, or a known-good capture. If every steady check clears but the symptom persists: climb the ladder — persistence on the accused node, then a tripwire single-shot where the ghost points (capturing-transients-and-single-shot-events). If the hunt is for something narrow: switch the acquisition to peak-detect before trusting any clean trace. If the event is captured but the cause is unknown: re-arm with the schematic's first suspect on channel two and read which trace moved first. If the second channel sits innocent: cross that suspect off and cycle the next candidate onto channel two — each acquittal narrows the list. If the diagnosis closes: file the signatures and the tripwire recipe under the model, because this board has siblings (the-oscilloscope-as-the-diagnostic-instrument). The throughline: walk the route, run the ritual, judge by signature, escalate by rarity, and let the second channel assign the blame.

Verification & Testing Methods

Confirm the method ran as a method, not a wander:

  • [ ] I walked the route in order — rails under the trace, then clock and reset, then the symptom's signal path — and logged a one-line verdict at every node.
  • [ ] I ran the setup ritual identically at each stop — probe chain verified, ground short and true, trigger locked, frame fitted — and logged any forced exception beside its reading.
  • [ ] I judged each trace against a written waveform signature — shape, levels, and timing from documentation or a known-good comparison — with tolerance, not against memory or hope.
  • [ ] I practised capture escalation deliberately: measured the steady, accumulated the suspicious with persistence, trapped the rare with a tripwire single-shot in peak-detect, and counted a silent watch as evidence.
  • [ ] I used two-channel correlation to move from effect to cause — trigger on the symptom, suspect on channel two, blame assigned to the first mover — and filed the signatures and the verdict for the next board.

Then try the practice exercises below — hands-on method work; scenarios differ from the quiz.

Practice Exercises

  1. Walk the route on a healthy board (5 minutes, hands-on). Run the full sequence — ritual on the calibration output, then rails, clock, and reset under the trace — recording each node's signature and a one-line verdict, building the reference the fault-finding exercises compare against (the-oscilloscope-as-the-diagnostic-instrument).
  2. Diagnose by known-good difference (5 minutes, hands-on). On a board with a known or induced steady fault, walk the same route and judge each node against the healthy signatures from Exercise 1, stopping at the first departure and naming it — shape, level, or timing (reading-and-measuring-waveforms).
  3. Climb the ladder on an intermittent (5 minutes, hands-on). Induce a rare fault — a flaky connection or a switched load — then escalate deliberately: confirm the steady checks pass, accumulate the ghost with persistence, and trap one event with a tripwire single-shot in peak-detect (capturing-transients-and-single-shot-events).
  4. Assign blame with channel two (3 minutes, hands-on). Re-arm the Exercise 3 trap with the cause — the switched load or flexed joint's supply — on the second channel, and read the correlated record to state which trace moved first and what that verdict means.

These core steps — the route, the ritual, the signature, the escalation, and the correlation — are tested in the Chapter Quiz at the end of this chapter, where a score of 80% is required to continue.

Key Takeaways

  • A scope session is a route, not a wander: rails under the trace first, then clock and reset, then the signal path the symptom names — the order that finds most faults at the first or second stop (the-oscilloscope-as-the-diagnostic-instrument).
  • The setup ritual — probe chain verified, ground short, trigger locked, frame fitted — runs identically at every node, so a difference on screen can only belong to the circuit (probing-compensation-grounding-and-loading).
  • Every trace is judged against a waveform signature — the shape, levels, and timing a healthy node shows — written down from documentation or captured from a known-good comparison before the probe lands, and filed afterward so the library compounds.
  • Capture escalation matches technique to the fault's rarity — measure the steady, accumulate the suspicious with persistence, trap the rare with a tripwire single-shot in peak-detect — and a silent watch is an acquittal within its limits, not a failure.
  • Two-channel correlation turns a captured effect into a cause: the trigger stays on the symptom, the schematic's suspect rides channel two, and the trace that moved first carries the blame — with each innocent suspect narrowing the list.

Skills Learned

  • You can now choose the first probe point from the symptom — rails first, then clock and reset, then the signal path.
  • You can now run the same setup ritual at every node so a mis-set scope never masquerades as a fault.
  • You can now judge a node by its waveform signature against a known-good comparison or documented expectation.
  • You can now escalate the capture method to match how often the fault shows itself.
  • You can now correlate two channels in one record to tie a captured effect to its cause.

Glossary Additions

  • waveform signature — the expected appearance of a healthy node's waveform, expressed as three things together: its shape (square, sine, stepped, pulsed), its levels (amplitude, logic highs and lows, DC value), and its timing (frequency, duty cycle, edge speed). A fault announces itself as a departure in any of the three, so diagnosis with an oscilloscope is the comparison of the captured trace against the node's signature — assembled from datasheets, schematics, and board markings before probing, or captured from the same node on a known-good board when documentation runs out. Signatures are judged with tolerance, since real boards vary within their ratings, and filed after each diagnosis: a library of captured signatures for a device family is what lets an experienced technician glance at a trace and know.
  • capture escalation — the discipline of matching oscilloscope capture technique to how often a fault shows itself, climbing only as far as the fault demands. A steady fault, present on every sweep, is simply measured and judged against its signature; a suspicious node accused by the symptom but clean on casual watching goes under persistence — infinite for a standing watch — until a rare deviation accumulates into a visible ghost; and a rare or one-time event is trapped with a tripwire single-shot, armed just outside the healthy envelope, in peak-detect when the quarry is narrow. The rungs are cumulative — the ghost aims the trap, and the trapped record is measured like any steady trace — and a watch that stays silent is evidence too, acquitting the node within the watch's limits and moving the route onward.
  • two-channel correlation — capturing a suspected cause and its effect in the same oscilloscope record, so their relative timing is real and readable. Channel one carries the symptom — the sagging rail, the glitching line — with the trigger staked on the abnormal event; channel two carries a suspect chosen from the schematic: the load that switches, the enable that gates, the clock that times. Because both channels share one capture around one trigger, the stored record shows which trace moved first, and the first mover is upstream of the fault — an enable dropping before a rail sags indicts the enable's source, while a suspect that sits innocent through the event is crossed off and replaced by the next candidate. Correlation is the step that turns a captured symptom into a named cause, one hypothesis per capture.

Suggested Next Sections

Must read next:

  • Why Intermittents Are the Hardest Faults — Chapter 9 turns to intermittent and environmental faults as a subject of their own: why they resist ordinary diagnosis, and how heat, cold, vibration, and time are used to force them into the open.

Recommended: