Diagnostic Flowchart Mindset Reference

Why this matters

Two techs walk into the same job - one diagnoses in 15 minutes, the other spends 90 minutes swapping parts before stumbling on the cause. The difference isn't tool quality or product knowledge; it's process. Good diagnosis is structured: divide the system into testable parts, eliminate possibilities by measurement, test before swap, and document what you learn. Bad diagnosis is "throw parts at it until it works."

The five rules of diagnosis

1. Symptom is not the cause

A water heater that "won't make hot water" has many possible causes: gas valve, igniter, dip tube, thermostat, anode, sediment, undersizing. The symptom narrows the possibilities; it doesn't identify the cause. Don't replace the first part that "might be it."

2. Test before you swap

Every reasonably-priced part can be tested:

• Resistance with a multimeter (heating elements, motors, sensors)
• Voltage at the device with the system running (relays, switches, control circuits)
• Visual inspection (cracks, burns, oxidation, water damage)
• Manual operation (push the float up, jump the switch, lift the relay armature)

Test takes 30-90 seconds. Replacement costs real money and 30 minutes. Test always wins on time and customer cost.

3. Isolate variables (divide and conquer)

When a system has multiple components in series (gas furnace's ignition train: thermostat → board → inducer → pressure switch → igniter → gas valve → flame sensor), test the dividing midpoint to halve the possibilities.

Example: furnace won't fire.

• First test: does the board receive the heat call? (Measure R-W with thermostat calling)
• If yes, the thermostat-to-board path is fine; focus downstream
• If no, focus on thermostat, wiring, transformer

Each test eliminates half the system from suspicion.

4. Change one thing at a time

When you make a change to test a theory, change one thing only. Replaced the capacitor AND tightened the contactor AND cleaned the coil - system now works - which did the fix? You don't know.

If you must change multiple things (because they all need it), test the system between each change. Then you know which fixed it.

5. Document what you find

Every test result is data. Write it down - on the work order, in your service notes app, on a sticker on the equipment. Next year's tech (often you again) starts the diagnosis with the data already in hand.

Common diagnostic flaws

Confirmation bias: you've decided it's the capacitor. Every test you run, you interpret in favor of "capacitor." You miss the actual blower motor signal because you weren't looking. To avoid: explicitly state alternative hypotheses ("could be capacitor, could be motor, could be control board") and test for ALL of them.

Parts cannon ("throw parts at it"): replace the first plausible part. If it works, declare victory. If it doesn't, replace the next plausible part. Burns time, money, and customer trust. Switch to measurement-first.

Dismissing intermittent symptoms: "couldn't reproduce" is not a diagnosis. Most intermittent failures have a pattern - heat, cold, load, time of day. Ask the customer for specifics; correlate.

Trusting customer reports verbatim: "the furnace won't turn on" might mean: thermostat is dead, fuse is blown, gas is off, ignition has been failing for weeks but customer just noticed today. Restate the symptom: "so when you set the thermostat to heat, you hear nothing at the furnace, and the room stays cold?" - yes/no answers narrow the actual symptom.

Forgetting the obvious: the breaker, the disconnect, the gas valve handle, the customer's recent activity ("I had some work done last week"). Walk the obvious-cause checklist before going deep.

Ignoring the equipment's diagnostic feedback: modern equipment has LED codes, error codes, history logs. Read the code first; it's the manufacturer telling you what the equipment thinks is wrong.

A general diagnostic framework

For any "doesn't work" call, work through:

1. Define the symptom precisely. "Doesn't cool" vs "cools but not enough" vs "cools then stops" vs "cools but compressor runs constantly" - each is a different problem.

2. Reproduce the symptom. Verify what the customer describes. Sometimes the customer's description differs from what's actually happening.

3. Check the obvious. Power, fuel, water, breakers, valves - the components that fail loudly and cheaply.

4. Read the equipment's self-diagnosis. LED codes, error codes, status indicators.

5. Form a hypothesis. Based on symptom + equipment feedback, what part of the system is most likely failing?

6. Test the hypothesis with measurement. Voltage, resistance, pressure, temperature - whichever applies.

7. Confirm the cause before replacing. If the test confirms, replace. If the test denies, refine the hypothesis.

8. Replace, then verify the fix. System runs end-to-end without recurrence of the symptom. Don't accept "seemed to work."

9. Document. Notes on what you found, what you replaced, and any conditions that may recur.

Hypothesis quality

Good hypotheses are testable. "It's the capacitor" is a testable hypothesis (resistance, capacitance, voltage drop). "It's just bad" is not.

Good hypotheses are specific. "Something's wrong with the gas valve" is vague. "The gas valve solenoid is open-circuit because I measure infinite resistance" is specific and supportable.

When you're stuck, the prompt is: "what specific measurement, if I took it, would tell me whether my current theory is right?" If you can't answer that, your theory isn't well-formed.

When to call for help

• 60+ minutes into a diagnosis with no clear path forward
• Symptom doesn't match anything in your training
• Multiple symptoms across multiple components - root cause is elsewhere (water leak destroying multiple things, voltage problem affecting whole panel)
• Customer needs the system back fast and you're not making progress

Calling another tech, looking up the manufacturer's service bulletin, posting in a trade forum - none are admissions of failure. They're appropriate use of resources to keep the diagnosis moving.

When the diagnosis is wrong

It happens. Customer reports the system failed again. Two questions:

1. Did the original test confirm the hypothesis, or did you skip the confirmation step?
2. Did the new failure match the original symptom, or is it different?

If you confirmed and the same failure recurred - it's a recurring root cause that you didn't address (a chronically overloaded circuit causing motor burnout, for example). Diagnose the root cause this time.

References

• "The Fifth Discipline" (Senge) - systems thinking foundations
• "How Doctors Think" (Groopman) - diagnostic reasoning patterns (medical, but transfers directly to trade troubleshooting)
• Trade-specific service manuals - manufacturer documentation is the authoritative diagnostic flow for the specific equipment
• Manuall internal references for specific equipment types