Reading an Error or Fault Indication
Why this matters
A fault code or a blinking indicator is the equipment trying to tell you something, but it speaks in shorthand and it is easy to misread. The tech who takes the code at face value swaps the part the code names and gets a callback when the real cause was upstream. A fault indication is a starting point, not a diagnosis. Reading one well, knowing what it actually claims, what it cannot know, and how to confirm it, is a core cross-trade skill that works the same on a furnace control, a pump panel, an appliance board, or a pool controller.
What a fault indication is, and is not
A fault code reports the symptom the control detected, not always the root cause.
- A code that says "sensor out of range" means the control saw a bad signal. That could be a failed sensor, a wiring fault, or a real condition the sensor correctly reported. The code cannot tell which.
- A code that says "no flame" or "no flow" means the control did not get the proof it expected. The cause could be anywhere in the chain that produces that proof.
Treat the code as "here is what I noticed," not "here is the part to replace." The control knows what it measured; it rarely knows why.
The three layers of any fault indication
Every fault display, however it is shown, carries up to three pieces of information. Read all three.
- What it monitors: the function or value the fault relates to (ignition, pressure, communication, a specific sensor).
- The nature of the fault: too high, too low, missing, lost signal, timeout, hardware fault.
- The history: whether it is active now, or a stored record of a past event. A stored code from last week is a clue, not a current condition.
A code you only half-read sends you the wrong way. "It is faulting on ignition" is incomplete until you know whether ignition is failing now or once did, and whether the control saw no spark, no flame proof, or a lockout after retries.
Active vs stored: do not chase a ghost
Controls keep two kinds of fault information and they demand different responses.
- Active fault: the condition is present right now. The equipment is locked out or limited. This is what you diagnose today.
- Stored or historical fault: a log of past events. Useful for spotting a recurring intermittent, but it does not mean the condition is happening now.
Clearing stored codes before you read them throws away the history. Read first, record them, then clear if you need a clean slate to see what is active.
Indicator patterns when there is no display
Plenty of equipment has no code readout, only a light. The pattern is the message.
| Pattern | Typical meaning |
|---|---|
| Steady on | Normal, or a single steady-state status |
| Slow, even blink | A specific coded condition; count the blinks |
| Fast blink | Often a lockout or a more serious fault |
| Blink-pause-blink groups | A two-part code: count each group |
| Off when it should be on | No power to the control, or the control is dead |
Count the blinks deliberately and watch for the repeat, the group of three then a pause then a group of two is a different code than five straight. Misreading the count is misreading the whole message.
Confirm before you condemn
The code points; your instrument confirms. Never replace what a code names without verifying.
- A sensor fault: measure the sensor's actual signal and compare against its expected range before condemning it. A wiring fault throws the same code.
- A "no proof" fault (no flame, no flow, no rotation): trace the chain that produces that proof and find where it breaks. The code names the missing result, not the broken link.
- A communication fault: check the connection and the device at each end, not just the device the code mentions.
The code narrows the search from the whole machine to one subsystem. That is its value. Closing the search to one part without measuring is how good codes lead to wrong repairs.
Record the code before you clear it
The fault indication is evidence. Write down the exact code or blink pattern, whether it was active or stored, and the conditions when it appeared, before you reset or clear anything. If you cannot finish the diagnosis, that record is what the next tech, or you on the return trip, starts from. A fault you cleared without recording is information you destroyed.
The mental model to keep
The control is an honest but limited witness. It tells you truthfully what it measured and when. It does not know the cause, it cannot see past its own sensors, and it will report a wiring fault and a dead sensor with the same code. Read what it actually says, confirm with your own meter, and you turn a vague complaint into a narrowed, confirmable diagnosis.
References
- Manufacturer documentation for fault-code definitions and blink-pattern keys (read the legend for the specific equipment)
- Trade-standard practice for fault-code interpretation and verification before replacement
- See related: Nuisance Alarm vs Real Alarm (decision tree); A Reading in Range but Still Wrong (decision tree)