New Board Didn't Fix It: The Temperature Sensor Was at Fault

A new board did not fix it, so the compressor is next in line. What a repair that changed nothing has already told you.

By Team Snowflake | Updated 24 Aug 2026

Teaching case based on recurring Snowflake service patterns; identifying details are combined.

How this presents

Running for two or three hours, then stopping. No blinking light and no code to look up. A new board has already been fitted and the behaviour is identical, down to the same interval, which is the part worth paying attention to.

Setup: Mitsubishi Electric Wall-mounted10-12 yearsCondoBukit Panjang
Misread as: The compressor next, because the board has been eliminated and that is what remains.
Check first: What the replaced part depended on, and whether that was replaced too

What else produces this

A null result has three readings, and two of them are cheap. The replaced part was not the cause. The replacement is itself faulty. Or the part was never independent, because it was acting on something else that was left in place. On anything that decides from a reading, the third is both the most common and the quickest to test.

A behaviour identical after replacement, down to the same interval, argues against a coincidental second fault.
A compressor holding steady output through the first hour rules out anything failing under sustained load.
A stop arriving while the room is still warmer than the set point means the decision was wrong, not the machinery.
A sensor reporting lower than a thermometer at the same position is the input the new board inherited.

Why it happens

A replaced component that makes decisions carries none of its own information. It reads inputs and acts, so replacing it swaps the thing that was obeying and leaves the thing that was wrong. The experiment was sound; the conclusion drawn from it was not. Eliminating the decider does not eliminate what it was told, and everything downstream of a bad measurement inherits the bad measurement.

What resolution involves

Replace the input, not the next component up. What proves it is a full cycle watched to the end, with the compressor staying on until the room genuinely reaches the set point. There was never a code, so no code clearing means nothing here. The board fitted earlier stays, because the earlier experiment already established it was sound.

How to tell it worked

Longer than the fault took to appear, which was two or three hours. Anything shorter tests the machine at a point the old fault never reached, and passes for the same reason the old board passed.

Before you approve a quote

A repair that changed nothing is a completed experiment

A null result is information. It eliminates the part that was replaced, and nothing beyond it.
Ask what the replaced part depended on. A component that decides from a reading inherits whatever that reading gets wrong.
Elimination pointing at the next most expensive thing is a habit, not a deduction. Ask which inputs were checked.

Questions this case answers

Aircon shuts off every few hours with no error code: is that a fault?

Not necessarily a failure. A fault trips a code; a clean silent stop means something told the unit its job was done. That makes the instruction the suspect rather than the machinery, and instructions are cheaper to check than machinery is to replace.

A new control board did not fix it. What does that tell you?

That the board was not the cause, which is real information and worth having. It does not follow that something larger is failing. A board acts on what it is told, so an unchanged fault after a new one points at what it is being told rather than at the next component along.

How do I avoid paying for parts one at a time?

Ask what each candidate part depends on before agreeing to it. Swapping in dependency order rather than in price order tends to end the sequence early, because the components that decide things are usually downstream of the components that measure them.

Is the compressor next if a board swap did not work?

Not on this evidence, and elimination is not the same as deduction. A compressor running steadily for an hour with consistent output is not the thing stopping the unit an hour later. Ask which inputs were verified before anything sealed gets priced.

Ready to get started?

Tell us what’s going on. Symptoms, setup, photos, anything we should know. We’ll assess and come back with the right next step.

WhatsApp us