Flame sensing on a gas furnace and what a weak signal means
Flame rectification, documented reasons a signal goes weak, and why cleaning the rod is a repair only when deposits are the reason.
What this means
A flame sensor does not detect heat or light. It passes a small alternating current through the flame, which conducts in one direction more readily than the other, and the control reads the resulting direct current in microamps as proof that a flame exists. Anything that weakens that current, including polarity, grounding, gas pressure, and deposits on the rod, produces the same shutdown.
Equipment and model context
Cross-brand explanation; product figures come from the model documentation
- Residential gas furnaces with hot surface or spark ignition and a flame rectification sensor
Acceptable flame current is model-specific and published in the service literature for the furnace in question, so a figure taken from one manufacturer does not qualify another. The measurement itself is taken in series with the sensor lead on a live furnace and belongs to a qualified technician.
What this guide covers
- The burner lights and shuts down within seconds, repeating several times before the furnace stops trying.
- The furnace locks out and reports a no-flame-sensed code after a fixed number of ignition attempts.
- Heating works intermittently and improves for a few weeks after a service visit before returning.
What changes the answer
- Rheem names improper phasing or polarity as a cause of failed flame sensing.
- Rheem states that a dirty flame sense rod can prevent flame rectification.
- Rheem attaches incorrect manifold gas pressure to failed flame sensing, because the flame must reach and envelop the rod.
- Trane describes the flame sensor as ensuring the gas valve opens only while flames are burning, so any interruption in the signal closes the valve.
Rectification, not detection
The sensor is a single metal rod sitting in the burner flame with one wire attached. There is no thermocouple, no optical element, and no moving part. The control board applies an alternating voltage to the rod, and the flame acts as a conductor between the rod and the burner assembly, which is grounded.
What makes the arrangement work is that a flame conducts unequally in the two directions. The surface area of the burner is far larger than the surface area of the rod, so current flows more readily one way than the other, and the alternating signal comes back partly converted to direct current. The board measures that direct current in microamps, and its presence is what proves a flame exists rather than merely that a gas valve opened.
The consequence is that the flame sensing circuit depends on the ground path as much as on the rod. Rheem lists improper phasing or polarity among the causes of failed flame sensing, which is a statement about the electrical supply to the furnace rather than about the sensor at all.
Documented ways the signal goes weak
Rheem's published list separates deposits on the rod, incorrect polarity or phasing, and incorrect manifold gas pressure. Those are three different systems, and each produces the same symptom of a burner that lights and is shut down.
Deposits are the familiar one. A coating on the rod can insulate it, and the current can fall until the board no longer accepts the signal as proof of flame. The appearance or timing of the symptom does not identify deposits on its own.
Gas pressure is a separate technician check. Rheem includes incorrect manifold pressure in its failed-flame-sense causes and separately links excessive pressure to overheating and a main limit opening. The specified pressure and flame-current thresholds come from the exact furnace service data.
Grounding sits underneath all of it. The rectified current has to return through the burner assembly and the chassis, so a poor ground connection reduces the signal in a way that looks identical to a dirty rod. Cleaning the rod on a furnace with a grounding problem produces a repair that lasts until the next inspection.
What the lockout sequence tells you
Rheem describes four ignition trials, with a one-blink code reported when no flame sense is detected after four attempts or when four flame failures occur during a single call for heat. Reset is by removing the 115 volt supply or switching the thermostat off for five to ten seconds.
That distinction between four failed ignitions and four flame failures is worth holding onto. In the first the burner never lit; in the second it lit four times and the signal was lost each time. Both end in the same lockout code, and they point at different parts of the system.
So the observation to make before resetting is whether you can hear or see the burner light. A burner that lights and dies makes flame sensing or flame stability relevant. A burner that never lights did not produce a flame for the sensor to prove, so the exact code and model sequence, not the sensor page, must direct the diagnosis.
Why the popular repair is only sometimes the repair
Cleaning the rod addresses one documented cause. If deposits were the cause, restoring the rod surface may restore the flame signal; that result does not eliminate the other electrical and combustion checks when the symptom returns.
A recurring failure needs the flame current, grounding, polarity, manifold pressure, and burner condition checked against the model service literature. Recurrence alone does not prove which one is wrong.
There is a further reason to leave the work alone. Reaching the sensor means opening the burner compartment on a gas appliance, and measuring the signal means breaking into the sensor lead with the furnace running. Neither is owner-level work, and the useful thing an owner can supply instead is the observation of whether the burner lights before it stops.
| Cause | What it does to the signal | What points at it |
|---|---|---|
| Deposits on the rod | Insulating layer reduces rectified current | Rod condition and measured flame current |
| Polarity or phasing | Rectification path disrupted at the supply | Appears after electrical work or a new board |
| Manifold gas pressure | Flame does not properly envelop the rod | Flame appearance and burner sound change |
| Poor grounding | Return path for the rectified current is resistive | Signal improves when the ground is corrected |
The current path runs from the control board through the sensor lead, across the flame, into the grounded burner, and back to the board. A fault anywhere on that loop reads as an absent flame.
- Control boardApplies alternating voltage and reads microamps of direct current
- Sensor rodSmall surface area, insulated from the burner
- FlameConducts unequally in the two directions, which produces the signal
- Burner assemblyLarge grounded surface that completes the return path
| Observation to make | What it separates |
|---|---|
| Does the burner light at all | Ignition failure from flame sensing failure |
| How many seconds does the flame last | Sensing loss from an over-temperature trip |
| How many attempts before it gives up | Whether the control reached its trial limit |
| Does it recur after a repair | Whether the reported cause was actually removed |
| Has any electrical work been done recently | A polarity or grounding cause from a fouling one |
Questions people ask about this
How does a furnace flame sensor work?
It relies on flame rectification. The board applies alternating voltage to a rod in the flame, the flame conducts more readily toward the large grounded burner than back to the small rod, and the board reads the resulting direct current in microamps as proof a flame is present.
What causes a weak flame signal?
Rheem names a dirty flame sense rod, improper phasing or polarity, and incorrect manifold gas pressure. A resistive ground path belongs on the same list, because the rectified current has to return through the burner assembly and chassis.
Why does my furnace light and then shut off?
Trane describes the flame sensor as ensuring the gas valve opens only while flames are burning. If the control does not receive a satisfactory signal it closes the valve within seconds, then retries, which produces the pattern of repeated short bursts of heat.
How many times will a furnace try to light?
Rheem publishes four ignition trials for its gas furnaces, with a one-blink lockout code after four failed trials or four flame failures during a single call for heat. Other manufacturers publish their own counts in their service literature.
Should I clean the flame sensor myself?
Reaching it means opening the burner compartment of a gas appliance, and testing the circuit requires a live-furnace measurement. The owner-level contribution is observing whether the burner lights before it stops.
What is a normal flame current reading?
It is model-specific and published in the service literature for the individual furnace, so a value quoted for one manufacturer does not qualify another. Measuring it also requires breaking into the sensor lead on a running furnace.
Evidence recordHow this page was checked
official manufacturer support article · checked 2026-09-08
Every technical claim above was written from primary documentation held in the HVAC Bench evidence record: Rheem and Trane technical literature. Where a source limits a definition to certain models, test conditions, or product classes, that limit is repeated here rather than generalised.
- Documentation class
- official manufacturer support article
- Scope of the definition
- Confirm against the exact model manual
- Last checked
- 2026-09-08