The Numbers Only Make Sense if Two Things Are Wrong: Decision Tree

Why this matters

You take your readings, and no single fault explains all of them. Fix the one cause that fits half the numbers and the other half still make no sense. This is the trap that sends techs in circles: they assume one root cause, keep re-testing, and never accept the answer the data is handing them, which is that two things are wrong at once. Two concurrent faults are less common than one, but when the numbers refuse to reconcile to a single cause, the numbers are usually right and the assumption is wrong.

Start here: rule out the measurement before the machine

Before you accept two faults, make sure you are not chasing one bad reading. A faulty meter or a bad test point invents contradictions that are not there.

• Verify your instrument on a known-good source or a fresh battery.
• Re-take the odd reading at a second point on the same circuit or line.
• Confirm the customer or a control did not change a setting between readings.

If the contradiction survives a clean re-test, it is real. Move on.

Branch 1: do the readings even fit one fault

Lay every reading next to what a single suspected fault would predict. A single fault produces a consistent signature. When readings pull in opposite directions, one cause cannot own both.

• Electrical: correct supply voltage but low voltage at the load, and correct current. A single break would drop current too. Correct current with a voltage drop points to a resistance somewhere carrying load, plus something else masking it.
• HVAC: high head pressure that says overcharge or poor heat rejection, together with low suction that says undercharge or restriction. One charge level cannot be both. Suspect a restriction plus a separate charge or airflow issue.
• Plumbing: good static pressure but poor flow, and no visible restriction at the fixture. Static-versus-flow disagreement is often two things: a supply-side restriction and a fixture-side one stacking up.
• Pool: normal filter pressure but poor return flow. If the filter is not loaded and the pump is fine, look for a suction-side air leak plus a separate return restriction.

Branch 2: separate the primary from the secondary

Two faults are rarely equal partners. One usually drives the machine's behavior; the other distorts your reading of it.

• Primary fault is the one causing the customer's complaint.
• Secondary fault is often what makes the numbers lie: a drifted sensor, a partly blocked path, a weak connection that adds resistance.

Fix nothing yet. Predict what each reading should become once you clear the secondary distortion. If clearing it would make the remaining numbers point cleanly at the primary, you have the pair.

Branch 3: confirm the pair before you commit parts

Do not present or replace two parts on a theory alone.

• Clear or isolate the smaller fault first if you safely can (bypass a suspect sensor with a known value, open a suspected restriction), then re-read.
• If the remaining readings now reconcile to the primary fault, the pair is confirmed.
• If they still do not add up, you may have three, or your primary is wrong. Go back to Branch 1 rather than swapping harder.

Field key

The recap

1. Verify the instrument and re-take the odd reading first - kill the false contradiction.
2. Check whether any single fault can produce all the readings. If not, accept two.
3. Name the primary (drives the complaint) and the secondary (distorts the numbers).
4. Clear or isolate the secondary, re-read, and confirm the rest reconcile before committing parts.

References

• Trade-standard practice for reconciling instrument readings against expected fault signatures
• Manufacturer service data for normal operating values by system
• See related: Two Symptoms One Cause vs Two Faults; The Data Log and the Symptom Don't Line Up