The Condensing Furnace That Locked Out on Cold Nights

The 5 am call

A 96 percent two-pipe furnace, installed two winters earlier, locked out at about 3 am with the outdoor temperature at 6 F. The homeowner reset it at the switch twice; it ran for an hour and locked out again. The fault history on the board showed the same code every time: pressure switch failed to close. It was the fourth lockout that winter, and all four had happened on the coldest nights of the year. Two daytime visits had found nothing.

This case is in two parts because the call was. First the house had to be made safe and kept warm without making anything worse. Only then could anyone diagnose a fault that disappears every time the sun comes up.

Part one: stop and contain

CO before anything. The tech walked in with a personal CO monitor already running at breathing height and checked the basement, the furnace closet and the attached garage the vent passes through. A pressure switch lockout means the furnace refused to fire, which is the safety doing its job, but a vent that has been freezing and thawing all winter can also have a joint pulled apart. All three spaces read 0 ppm, and the homeowner's CO alarms were present and tested. Had any space read above zero, the furnace would have stayed off, the space ventilated, and the occupants sent outside while the source was found.

The first two minutes outside. Snow drifted over a sidewall termination is the most common cause of a cold-night pressure switch lockout and the cheapest to rule out, so the tech checked the intake and exhaust terminations before touching the furnace. Both were clear, with a normal frost collar on the exhaust and nothing restricting either.

Containment options, weighed.

• Reset and walk away. It had already been reset twice and run an hour each time. That buys an hour, not a night, and the pipes in the house were the real exposure with the house at 59 F and falling.
• Jumper the pressure switch to get heat. Never. The switch is what proves the vent can carry flue gas before the burner lights. Bypassing it on a vent that is partly blocked fires the furnace into a vent that may not clear its products. The tech did not carry a jumper into the house.
• Find and clear the restriction, then run under watch. This needs the restriction found, which is the diagnosis. In the meantime the house needed heat.
• Temporary electric heat. The tech set two of the homeowner's portable electric heaters plugged directly into wall receptacles, not extension cords, three feet clear of anything that burns, and told the homeowner not to use the oven or a gas range for heat, which is how families get poisoned during a heating outage.

Isolation for the work. Furnace power off at the service switch and proven dead with a tester checked on a known live source before and after, per NFPA 70E Article 120 in the current edition, and the furnace gas cock closed. The inducer and board are inches from the condensate trap the tech was about to open.

Part two: diagnose, in the order the evidence allowed

The pressure switch itself. A switch that drifts can fail to close on a perfectly good vent. With the furnace thawed and running later that morning, a manometer teed into the switch hose read negative 1.35 in wc at the inducer port. The switch label gave its close point as negative 0.90 in wc. A switch that closes at 0.90 with 1.35 available has margin and closes with room to spare, and on the bench it closed at its rated point. The switch was reporting the vent honestly. Replacing it would have put a part number on the invoice and the same lockout on the next cold night. See related: Inducer and Pressure Switch Diagnosis.

The inducer. Amps on the inducer read inside its nameplate figure, the wheel spun freely with no rub, and its housing drain was clear. A failing inducer produces low draft all the time, not only on the coldest nights. Not this.

The condensate trap. A blocked or frozen trap backs condensate into the collector box, and water in the collector box blocks the pressure switch tap and drops the reading. The trap was clean and drained a poured cup of water in seconds. The furnace sat in the heated basement, so the trap was not freezing. But the tech followed the condensate line from the trap and found it ran out of the basement through the rim joist into the garage, along the vent route, to a garage floor drain. It was a quarter full of ice, still passing water but on its way to blocking, and a frozen drain line backs the trap up exactly the way a blocked trap does. That was a second problem and a clue: something in the garage was cold enough to freeze water.

The vent run. The intake and exhaust left the basement through the rim joist and crossed 12 ft of unheated garage ceiling before the sidewall termination.

BASEMENT (heated)  | GARAGE CEILING (unheated)  | WALL
                   |                            |
furnace intake <===|============================|<= in
furnace exhaust ==>|=====\_________/============|=> out
                   | H          H           H   |
                   |       sag 3/4 in deep      |
                   |       ice ring here        |
hangers 8 ft apart, manual calls for 4 ft
exhaust must fall back toward the furnace

A level on the exhaust showed the problem. The furnace manual requires horizontal exhaust to slope back toward the furnace at 1/4 in per foot, so condensate formed in the pipe runs back to the furnace's drain. Over the 12 ft garage run that is 12 times 1/4 in, or 3 in of fall toward the furnace. The pipe had been hung roughly level, and between two hangers 8 ft apart, against the manual's 4 ft spacing, it had sagged into a belly about 3/4 in deep.

On a mild night, condensate pooled in that belly and the exhaust flow carried enough heat to keep it liquid. On a single-digit night in an unheated garage, the pooled water froze from the pipe wall inward. Each cycle added more condensate and more ice. The ring of ice narrowed the exhaust until the inducer could no longer pull enough negative pressure to close the switch. Every daytime visit came after the garage warmed and the ice melted, so the manometer read healthy and the vent looked clear.

The tech confirmed it the direct way: he opened the coupling nearest the belly before thawing and found a ring of ice still frozen to the bottom third of the pipe.

The fix

Re-pitch the whole horizontal run. The fix was not to lift the belly. Every hanger was reset so the exhaust fell continuously back toward the furnace at the manual's 1/4 in per foot, measured with a level at each hanger, and the intake run was set to the slope its own section of the manual calls for. Re-pitching moved the pipe at the rim joist penetration, so the tech checked that the joints there had not been stressed.

Hangers at the maker's spacing. Hangers added to bring spacing to 4 ft along both pipes, using hangers that support the pipe without pinching it. PVC is flexible when warm, and a vent that carries warm, wet flue gas sags between hangers that were adequate the day it was installed.

Joints re-made where cut. Two couplings had to be cut out to reset the run. New couplings were primed and cemented with the product the vent material's listing requires. Primer and cement vapor is flammable and heavier than air, so the garage door was open, the furnace was off and its gas closed, the water heater in the garage was off, and nothing that sparks ran until the vapor cleared.

Insulation where the manual calls for it. The manual for this furnace calls for insulating exhaust that runs through an unconditioned space beyond a stated length, and this run met that condition. The tech insulated the garage section of the exhaust with the type and thickness the manual names.

Condensate line protected. Rerouted inside the heated basement to a condensate pump discharging into the laundry standpipe, so no part of it passes through the garage. A heat cable listed for that pipe and use is the other option where the route cannot change, but it adds an electrical device that can fail silently; a route that stays warm does not.

Code basis. The fuel gas code requires a listed appliance to be vented according to its manufacturer's installation instructions (IFGC Section 503, or the IRC's fuel gas chapter, as adopted by your jurisdiction). So the manual's slope, hanger spacing and insulation requirements were not suggestions; they were what the install had to meet.

Restore and confirm

Restore. Gas cock opened and checked at the union with leak solution, panels on, power restored, the tech standing clear of the burner compartment door on the first ignition. The furnace lit on the first try. The garage water heater, shut off for the cementing, was relit per its label and seen firing with its vent drafting. Combustion readings were taken and recorded against the start-up sheet (see related: The Gas Furnace Start-Up Sign-Off SOP), ambient CO stayed at 0 ppm in the basement and the garage, and the manometer at the switch port read negative 1.40 in wc on a running furnace.

Confirm under the condition that caused it. A fix to a cold-night fault cannot be proven at noon. The tech set a follow-up for the next cold snap and asked the homeowner to note any lockout. Nine days later the overnight low reached 4 F, colder than the night of the call. The board's fault history showed no pressure switch events, and a manometer reading the next morning, before the garage warmed, read negative 1.38 in wc, essentially unchanged from the running reading after the fix.

What made this one slow

Three things combined to send two competent daytime visits home with nothing. The fault existed only below a temperature the visits never saw. The device reporting it was working perfectly, so testing it proved nothing. And the cause sat in a room nobody associated with the furnace. The board's fault history, read for when the lockouts happened rather than just what code they carried, was the record that pointed at the cold. The first question on any cold-night callback is now "what was the outdoor temperature at each lockout", and the answer goes on the ticket.

References

• See related: Inducer and Pressure Switch Diagnosis
• See related: Furnace Vent Categories and What Each Allows
• See related: Condensate Drain Code and Trap Field Card
• IFGC Section 503 (or IRC fuel gas chapter), as adopted by your jurisdiction
• NFPA 70E Article 120, current edition