Module 14 of 16
Diagnostic Lab
A repeatable method that replaces guessing with evidence
Learning objectives
- Work the five-stage method in order on any appliance or commercial unit
- Decide at each gate whether to move on or narrow further
- Recognise the shortcuts that produce misdiagnosis and callbacks
- Document evidence so the repair is defensible and the confirmation is real
- Apply one repeatable diagnostic order on any machine, including equipment you have never seen
- Choose the next test on the basis of what it eliminates rather than what it confirms
- Recognise the four shortcuts that produce most callbacks, and what each one costs
Why method beats experience alone
Experienced technicians misdiagnose for a predictable reason: pattern matching. A symptom looks like the last three calls, so the part that fixed those calls gets replaced. It works often enough to feel reliable and fails often enough to cost real money in returned parts, second trips, and lost trust.
A method does not slow you down. It changes what you are doing between arriving and deciding. Every stage produces a fact, and each fact eliminates a whole family of causes. By the time you reach for a part, the part is not a guess — it is the only remaining explanation for what you measured.
- Pattern matching is a hypothesis generator, not a diagnosis
- Each stage should eliminate a family of causes, not a single part
- If you cannot say what measurement condemned the part, you have not finished diagnosing
The order is the whole point
Verify comes first because a third of calls are not the fault the customer described — they are usage, installation, or an intermittent that needs conditions recreated. Narrow comes second because testing components before you have chosen a system is how afternoons disappear.
Power and inputs precede outputs and loads because a controller can only act on what it receives. A machine that will not start because a pressure switch reports full is a healthy machine responding correctly to a false input, and no amount of load testing will find it. Confirm comes last, always, and under real operating conditions — not by 'it started, so it's fixed'.
When a diagnosis stalls, you almost always skipped a stage. Go back to the last stage that produced a hard measurement and continue from there.
Evidence discipline
Write down measurements as you take them, with the conditions attached: the value, where you measured it, and what the machine was doing at the time. A voltage reading with no cycle context is nearly worthless an hour later, and it is worthless to the next technician entirely.
Good evidence also protects you commercially. When a customer questions a board replacement, 'the board commanded the valve and there was no output voltage at the terminal while the input said fill' is a different conversation from 'it seemed like the board'.
- Record value, location, and machine state together
- Photograph nameplates, fault histories, and burnt or corroded findings
- Note what you ruled out, not only what you found
Confirmation is a stage, not a formality
Run the full cycle that failed, under the load and conditions that produced the complaint. A washer that drains on a service-mode pump test but was failing on a heavy load has not been confirmed. A cooler that pulls down empty at 10 pm has not been confirmed against a stocked box during a lunch rush.
Confirmation is also where you catch the second fault. Many callbacks are not misdiagnoses — they are correct repairs on machines that had two problems, one of which was masked until the first was fixed.
Why order beats knowledge
Two technicians with identical product knowledge produce very different results, and the difference is almost always testing order. Knowledge tells you what could be wrong. Order tells you what to eliminate first so that the remaining possibilities shrink with every measurement. A technician who tests in a good order can diagnose equipment they have never seen; a technician who tests by hunch struggles even on familiar machines.
The principle is simple: prefer the test that eliminates the most possibilities for the least cost and risk. Verifying the supply eliminates every electrical fault downstream of it and takes a minute. Opening a sealed system eliminates almost nothing and forecloses further evidence. Between those two extremes, every decision is a value judgement about information gained per unit of cost.
Order also protects you commercially. Every measurement recorded in the right sequence is a defensible record of why the part was replaced. When a repair is questioned — by a customer, a manager, or a warranty administrator — the sequence is the answer.
- Pick the test that eliminates the most for the least cost and risk
- Verify the supply and the permissives before evaluating anything that depends on them
- Irreversible tests come last, never first
Think in eliminations, not in suspicions
A suspicion is a hypothesis about one component. An elimination removes a whole branch of possibilities. Practically, that means asking of every candidate test: if the result is normal, what have I ruled out? If the answer is 'one part', the test is weak. If the answer is 'the entire supply path', 'every input the board depends on', or 'the mechanical side of this system', the test is strong.
This reframing changes behaviour immediately. Instead of measuring the part you suspect, you measure the thing that decides whether the fault is upstream or downstream of it. Instead of checking a valve, you command it and read at the board terminal and at the valve — two readings that split the whole circuit rather than test one item.
It also makes you comfortable with a normal result. A test that returns normal is not a wasted test; it is a branch removed. Technicians who only value tests that find faults are the ones who end up guessing.
- Strong test: a normal result removes a branch, not a part
- Split the circuit at the midpoint rather than probing at one end
- A normal result is progress — record it and move on
The four shortcuts that cause callbacks
First, code chasing: treating a fault code as a part number. A code describes a condition the control observed; the condition has several possible causes and the code names none of them. Second, symptom matching: recognising the complaint from a previous call and fitting the same part. It works often enough to be dangerous. Third, part swapping to test: fitting a part to see whether it helps, which produces no information when it does not and no diagnosis when it does. Fourth, stopping at the first finding: replacing the burned relay without asking what burned it, so the new one burns too.
Each has a characteristic cost. Code chasing wastes parts. Symptom matching produces confident wrong answers. Part swapping destroys the evidence trail. Stopping at the first finding produces the return visit that turns a profitable repair into a loss.
If you cannot state what you eliminated and how, you have not diagnosed the machine — you have made an educated purchase.
Working on equipment you have never seen
The method matters most when the product knowledge is absent. On unfamiliar equipment, spend the first minutes establishing three things: what the machine is supposed to do at this point in its cycle, what supplies and permissives it needs to do that, and what feedback the control uses to decide it worked. Those three answers come from the wiring diagram and the sequence chart, and they convert an unfamiliar machine into a familiar structure.
From there the order is the same as always: supply, permissives, command, feedback, load, mechanical. Every machine ever built fails inside that list.
- Establish expected behaviour before measuring anything
- Read the sequence chart and the diagram before opening panels
- Supply, permissives, command, feedback, load, mechanical — in that order
The diagnostic method
Stage 1
Verify
Establish what actually happens, in the customer's real conditions, before forming any theory.
- Interview: when did it start, what changed, does it happen every time or under specific conditions
- Run the failing cycle yourself and watch it fail
- Read fault history and service-mode input states before clearing anything
- Check installation basics: supply, levelling, ventilation, water, drainage
Gate: Can you describe the failure in measurable terms and reproduce it on demand?
Shortcut to avoid: Accepting the customer's diagnosis and starting on the part they named.
Stage 2
Narrow
Choose the system that owns the failure so you stop testing at random.
- Ask which system must be involved for this exact symptom to occur
- Use the machine's sequence: identify the last step that completed and the first that did not
- Split the machine in half with a single test where possible (interface unplugged, forced output, isolated load)
- Eliminate whole systems out loud, and note why
Gate: Have you named one system, and can you say why the others are out?
Shortcut to avoid: Testing individual components before choosing a system — the classic afternoon-killer.
Stage 3
Test power and inputs
Confirm the controller has correct supply and is receiving true information.
- Verify supply voltage and ground integrity at the control, under load
- Read live input states in service mode and compare against physical reality
- Test switches, sensors, and safety devices at their expected values, not just for continuity
- Confirm every interlock the machine requires to proceed
Gate: Does the controller's view of the machine match the physical machine?
Shortcut to avoid: Assuming inputs are fine because the display looks normal.
Stage 4
Test outputs and loads
Separate a controller that will not command from a load path that will not respond.
- Force the output in service mode and measure at the control terminal
- If the command is present, measure at the load: voltage arriving, current drawn
- Command present and no current means an open path or an open load
- No command with correct inputs means the control or its logic
Gate: Can you state whether the command, the path, or the load failed?
Shortcut to avoid: Replacing the load because it did not run, without proving it was commanded.
Stage 5
Confirm
Prove the repair under the conditions that produced the complaint, and check for a second fault.
- Run the complete failing cycle end to end, loaded and in real conditions
- Measure the outcome against spec: cycle times, temperatures, currents, pressures
- Re-read fault history for anything new
- Record the confirmed cause and the evidence in the service note
Gate: Would you stake the callback on this, with measurements to back it?
Shortcut to avoid: Powering up, seeing it start, and packing the van.
Failure modes and what confirms them
| Symptom | Mechanism | The tell |
|---|---|---|
| Repeat callback on the same machine | Confirmation skipped or done off-load | No recorded post-repair measurements under real conditions |
| Correct part fitted, fault remains | System never narrowed; the symptom belonged elsewhere | No stated reason why other systems were eliminated |
| Board replaced, fault returns | A false input the board was correctly obeying | Service-mode input state disagreed with physical reality |
| Intermittent never found | Complaint never reproduced | Verify stage ended without seeing the failure |
| Second fault appears days later | Masked fault revealed after the first repair | Full-cycle confirmation would have surfaced it on site |
| Repeat visit for the same complaint after a part was replaced | The first finding was fixed without identifying its cause | The replaced part shows the same damage pattern as the original |
| Diagnosis stalls with no clear next step | Testing by suspicion instead of by elimination | You cannot state what the last three measurements ruled out |
| Several parts replaced with no resolution | Part swapping used as a test method | No recorded measurement supports any of the replacements |
| Confident diagnosis contradicted at the end of the job | Symptom matched to a previous call rather than measured | The readings were never taken because the answer felt known |
Test procedures
Ranking a candidate test
Before you take a reading, ask what a normal result would eliminate.
| Candidate test | Eliminates if normal | Cost / risk | Strength |
|---|---|---|---|
| Verify supply voltage at the machine under load | Every fault upstream of the machine, plus sag-driven resets | Minutes, no risk | Very strong |
| Read stored fault history in order | Whole families of causes, and identifies the initiating fault | Minutes, no risk | Very strong |
| Command an output and read at the board terminal and the load | Splits board, harness and load in one step | Minutes, low risk | Strong |
| Compare a sensor against a calibrated reference | Sensor drift as an explanation for the whole complaint | Minutes, no risk | Strong |
| Check one suspected component's resistance | That one component only | Minutes, no risk | Weak but cheap |
| Fit a replacement part to see if it helps | Nothing, and it removes evidence | Cost of the part | Avoid |
| Open the sealed system | Almost nothing, and it forecloses further testing | High, irreversible | Last resort only |
Shortcut, cost, and the correction
| Shortcut | What it costs | The correction |
|---|---|---|
| Treating a fault code as a part number | Wasted parts, unresolved fault | Treat the code as a condition and verify the condition physically |
| Matching this call to a previous one | Confident wrong diagnosis | Run the order anyway; let the readings agree or disagree with memory |
| Fitting a part to test a theory | No information, no evidence trail | Take the measurement that would have proven the theory |
| Stopping at the first finding | A return visit within weeks | Ask what caused the finding before you fix it |
Safety and professional boundaries
- Never defeat a safety device to progress a diagnosis, not even momentarily.
- Irreversible tests — opening a sealed system, cutting a harness, dismantling a bearing — come after every reversible test has been exhausted.
- If the correct next test is outside your certification or your equipment, stop and escalate rather than approximating it.
Split the machine in half
- 1.Choose a single test that divides the machine into two testable halves — unplug the interface, force an output, or isolate the load.
- 2.Predict what each outcome would mean before you run it.
- 3.Run the test and record which half is implicated.
- 4.Repeat within the implicated half until one component remains.
Command, path, load
- 1.Force the output and measure at the control terminal: is the command present?
- 2.Measure at the load terminals: does the voltage arrive?
- 3.Clamp the supply lead: does current flow?
- 4.Command absent = control or inputs. Command present, no arrival = wiring. Arrival, no current = open load.
Live testing only where required, with rated PPE and a rated meter. Isolate for resistance measurements.
Run one elimination pass
- 1.State the complaint in terms of machine behaviour, not customer language.
- 2.State what the machine should be doing at the point where behaviour departs from expected.
- 3.List the supplies and permissives required for that behaviour, and verify them.
- 4.Verify the command from the control, at the control's own terminal.
- 5.Verify the feedback the control uses to confirm the action.
- 6.Verify the load, then the mechanical path.
- 7.Write down what each step eliminated before moving to the next.
Establish the cause of every finding
- 1.When you find a failed component, record its failure mode: open, shorted, burned, seized, drifted.
- 2.Ask which upstream or downstream condition produces that mode.
- 3.Measure that condition — current draw, restriction, supply quality, mechanical drag.
- 4.Correct the cause and the component together.
- 5.Note the cause on the job record so the next technician inherits the reasoning.
Expected readings and what they mean
| Measurement | Expected | Meaning |
|---|---|---|
| Supply at the control under load | Nominal ±10% | Sag under load points at supply or connections, not the control |
| Ground integrity at the control | Under 1 ohm to chassis | Poor ground produces false inputs and erratic logic |
| Commanded output at the terminal | Line or control voltage while forced | Absent means control or inputs; present moves you downstream |
| Load current while commanded | Per nameplate | Zero with voltage present means an open load or path |
| Post-repair cycle outcome | Within spec for time, temperature, pressure | This is what turns a repair into a confirmed repair |
| Supply voltage at the machine under load | Nominal, stable as loads engage | One measurement eliminates every upstream electrical fault and most reset complaints |
| Current draw of a load that destroyed its own control output | At or below the documented rating | Above rating identifies the cause of the failure and predicts the next one |
| Elapsed time to complete a stage versus the documented duration | Within the documented window | A long stage identifies where to look before any component is measured |
Field scenarios
Mini-scenario: dryer no heat, three plausible causes
- Electric dryer tumbles normally but produces no heat
- Customer reports the clothes have been drying slowly for weeks
- Exhaust run is long with two elbows and has never been cleaned
- No fault code capability on this platform
You could test the element, the thermal cut-out, or the airflow. Which first, and why?
Mini-scenario: the code names a part
- Front-load washer stores a drain fault
- The parts catalogue lists a drain pump as the associated part
- The machine drains slowly but does drain
- Sump has not been inspected
The code maps to a pump in the catalogue. What do you do with that information?
Mini-scenario: equipment you have never worked on
- Light commercial machine, unfamiliar brand, will not advance past its fill stage
- Documentation is available on the machine and online
- No visible damage, no unusual noise, supply confirmed at the machine
- The customer wants an answer today
You do not know this platform. What is the productive first move?
The dishwasher that will not fill
- A dishwasher starts, runs the drain, then sits with no fill and eventually faults. The customer already replaced the inlet valve based on an internet video.
- Service mode shows the machine calling for fill. Water pressure at the tap is good.
Work the method. Where does it land?
Verify: reproduce it — the machine calls for fill and no water enters. Narrow: the water system, since drain and control both function. Power and inputs: read the float or pressure input — if it reports full, the machine is refusing to fill correctly and the fault is the sensing path, not the valve. Outputs and loads: if the input is correct, force the valve and measure the command at the terminal, then voltage and current at the valve. Command absent with correct inputs means control; command present with no current means the coil or wiring. Confirm on a full cycle. The already-replaced valve is exactly what the method prevents.
Takeaway: The customer's part swap is data too: it tells you the valve is new, not that the valve was the fault.
Knowledge check
Why do power and inputs get tested before outputs and loads?
You force an output; the command is present at the control terminal but no current flows at the load. What does that prove?
What makes a repair confirmed?
You have two candidate tests. One would confirm your leading suspicion; the other would prove whether the fault is upstream or downstream of it. Which do you take first?
You find an open thermal cut-out on a no-heat call. What does professional practice require next?
Practise it in the labs
Apply this module on a live service call in the interactive diagnostic labs.
Key takeaways
- Verify, Narrow, Test power and inputs, Test outputs and loads, Confirm — in that order, every time.
- Each stage should eliminate a family of causes and produce a recorded fact.
- Command, path, load is the fastest way to place an electrical fault.
- A repair is confirmed by a completed cycle under real conditions, not by a successful start.
- Prefer the test that eliminates the most for the least cost and risk.
- A normal result is progress; record what it ruled out.
- Every finding needs a cause before it gets a part.
- The order — supply, permissives, command, feedback, load, mechanical — lets you diagnose machines you have never seen.
Finished this module?
Mark it complete to track your progress toward the Technical Systems Certificate.