The Alarm That Tripped on Windy Nights

Why this matters

A family left for a week and the house dispatched police three nights running while it stood empty. Each response found a locked house and nothing disturbed, and the third brought a letter from the city about the false alarm count on the permit. The fault was one sliding door that moved further in a gust than its contact could tolerate. Nobody on site saw it happen, so the whole diagnosis had to be rebuilt from the signal history and a night of weather data, which is the part of this case worth learning.

The record the tech started from

By the time the service call was booked the family was home and the door was shut, latched and reading closed. The tech had no live fault to chase. What existed was the central station's history for the account, the keypad's event log, and a homeowner who could say only that the neighbor had heard the siren.

The station history, trimmed to the three events:

Three things in that table carry the diagnosis, and none of them needs a theory to read.

Same zone every time. Zone 4 was the family room slider and nothing else. No motion detector in the family room tripped before or after, and the house had one covering that room. A person who opened the slider and stepped inside would have walked through the motion pattern within a few seconds. A door that opened and closed by itself would not.

Restores measured in seconds. Two to six seconds open, then closed again. That is not someone forcing a door and leaving it ajar. It is a contact that separated briefly and came back.

The account was armed away. The house was empty, so every perimeter and interior zone was armed. Zone 4 was programmed as a perimeter instant zone because the family used the front door as their entry, so any open on zone 4 was an alarm with no entry delay to absorb it.

What the record could not say

The history said where and how long. It did not say why, and there were three live candidates for a perimeter contact that opens for a few seconds on its own.

A failing contact or a loose termination. A reed switch with a cracked capsule or a wire nicked under a staple can open on vibration. That would show as an open on the zone, but it would tend to show at random times, on calm nights as well, and the keypad log for the weeks before showed nothing on zone 4 at all.

A wireless contact dropping out. Ruled out on inspection: this slider was hardwired.

Real movement of the door. The only candidate that explains why the three nights were the three nights.

The tech pulled the hourly record from the nearest public weather station for the week the family was away. The three alarm nights were the three nights with gusts recorded above 30 mph. The four calm nights in between had gusts under 15 mph. A correlation over seven nights is not proof, but it turned the visit from "find the fault" into "make the door move the way the wind moved it and watch the contact."

Before touching anything: the account goes on test

Testing this door meant opening zone 4 repeatedly with the system armed and disarmed. The tech called the central station first, gave the account number and his dealer ID, and asked for the account to be placed on test for two hours, with the station confirming the test window back to him. See related: How to Put an Alarm Account on Test and Take It Off. Without that call, the reproduction itself would have been the fourth dispatch.

How the slider moved

The contact was a surface-mount reed switch on the fixed jamb at the latch side, near the top, with its magnet on the top corner of the sliding panel's latch stile. That position is common because it is out of reach of furniture and pets, and it is the worst place on the door for movement.

 PLAN, LOOKING DOWN AT THE HEAD TRACK (latch side)

  fixed jamb       head track channel
  +--------+   +---------------------------+
  | REED   |   |  top guide clearance      |
  | SWITCH |   |  lets the panel top move  |
  | at top |   |  in and out of the frame  |
  +--------+   |   <-- rest: 3/8 in gap -->|
       |       |   <-- gust: about 1 in -->|
       |       |        +-------------+    |
       |       |        | MAGNET on   |    |
       |       |        | panel top,  |    |
       |       |        | latch stile |    |
       |       |        +-------------+    |
       |       +---------------------------+
  exterior wind pushes the glass -> panel top
  swings away from the jamb; bottom held by
  the rollers, middle by the latch, moves less

A sliding panel is held at the bottom by its rollers riding the sill track and at the latch stile by the latch. At the top it is held only loosely by the head track, which has clearance so the panel can be lifted in and out. When wind loads a large sheet of glass, the panel acts like a sail pivoting near the bottom, and the top corner moves most.

The tech had the homeowner stand outside and push on the glass with an open hand, in pulses, while he held a tape on the jamb. At rest the gap between switch and magnet was 3/8 in. Pushing moved the top of the panel until the gap measured about 1 in at the peak. The rollers were worn enough that the panel also sat low and loose in the head track, which made the play worse than a new door would show.

The contact's datasheet listed its gap for a nonferrous frame at about 3/4 in, and this door was aluminum. On a steel frame the same contact would have a shorter gap, which is why contact ratings always come with the mounting surface named. See related: Door and Window Contact Types and Gap Ratings. A rest gap of 3/8 in against a rating of about 3/4 in left 3/8 in of margin, so any gust that swung the panel top more than 3/8 in opened the circuit, and the pushes were swinging it about 5/8 in. The homeowner's pushes did it every few tries, and each one showed as a zone 4 open on the keypad.

Why a few seconds was enough

A zone does not report the instant a reed opens. The panel samples the loop and requires the change to last for its loop response time before it acts, which the panel's documentation sets and which some panels let you lengthen per zone. Gusts at 30 mph and above come in pulses lasting seconds, long enough to beat any normal loop response.

Lengthening the loop response on zone 4 would have hidden the symptom and created a real weakness: a door that can be pulled open and shut again inside the longer window without reporting. The tech did not consider it. The fault was the door and the contact placement, and that is where the fix went.

Three changes, and why all three

Any one of these might have stopped the alarms. The tech made all three because each one removed a different part of the margin problem, and a door that faults only in weather is expensive to come back to.

Tighten the slider. He adjusted both rollers up until the panel rode properly in the head track, and adjusted the latch keeper so the latch pulled the panel firmly against the jamb weatherstrip. After adjustment the same hand-push moved the top corner noticeably less. This also fixed a draft the family had noticed and a latch that needed a lift to engage, which is a security problem of its own. Where the rollers are too worn to adjust or the sill track is damaged, the door needs a door repair visit, and the contact work waits for it.

Move the magnet to where the door does not move. The new position was the latch stile at about mid-height, with the switch on the jamb directly opposite. That point is close to the latch, which holds the panel against the jamb, so wind load produces very little separation there. It is also the point a burglar has to move to open the door, so detection did not suffer. The old screw holes in the aluminum were filled with a color-matched sealant.

Use a wide-gap contact. The replacement was a surface contact rated by its maker for about 1.5 in on a nonferrous frame, roughly twice the original. Wide-gap contacts exist for exactly this: doors and windows with movement that is normal for the hardware, including sliders, overhead doors and loose wood sashes. The tradeoff is that a wide-gap contact also tolerates a door opened a small amount, so it goes only where the hardware genuinely needs it.

The zone wiring was re-terminated at the new switch location with the end-of-line resistor kept at the device, where it had been. See related: End-of-Line Resistor Supervision: Open, Short and Normal.

The finish test

With the account still on test, the tech repeated the reproduction, this time harder.

• Rest gap at the new location: 1/4 in.
• Peak gap with the homeowner pushing as hard as she could: 1/2 in.
• Margin to the new contact's rated gap: 1.5 minus 1/2 leaves about 1 in.

He also had her pull the panel against the latch from inside, the direction an intruder would work it, and the zone stayed closed until the latch released. Then he opened the door 2 in and confirmed the zone faulted, which is the check that a wide-gap contact still detects an actual opening.

He armed the system, opened the slider, let the alarm report, and called the station. The operator confirmed a burglary alarm on zone 4 received under the test window. Only then did he ask the station to take the account off test, and he confirmed with the operator that it was back in normal service before leaving. A system that leaves the visit still on test is unmonitored, and nobody notices until something happens.

What went on the ticket and to the customer

The ticket recorded the history rows by date and time, the weather correlation, the measured gaps before and after, the contact replaced and its rated gap, the door adjustment, and the confirmed test signal. That record matters if the city disputes the false alarm count, because some jurisdictions waive or reduce a fine when the dealer documents a repair, and some do not. See related: Alarm Permits, False Alarm Fines and the Dealer's Part.

The conversation with the homeowner was short:

 TECH: The three alarms were the slider moving in
       the wind. The sensor sat where the door
       moves most. I moved it, adjusted the door,
       and fitted a sensor that tolerates more
       movement. I tested it with you pushing on
       the glass.
 CUSTOMER: Could someone have opened it?
 TECH: Not from what we saw. Each one closed
       again in seconds and the family room
       motion detector never tripped.
 CUSTOMER: The city sent a letter.
 TECH: The ticket has the dates and the repair.
       Send them a copy with your reply.

Where this case generalizes, and where it does not

The method is the useful part: when a record shows one zone, short restores and a pattern that matches outside conditions, reproduce the condition before replacing parts. Wind is the common one, but the same reading of a history finds a garage door that faults when the cold stiffens the weatherstrip and a window that faults when the HVAC blower pressurizes the house.

It does not generalize to a zone with random, long or unrestored opens. Those point at the device or the wiring first, and a wide-gap contact would only hide them.

References

• See related: Door and Window Contact Types and Gap Ratings
• See related: How to Put an Alarm Account on Test and Take It Off
• See related: End-of-Line Resistor Supervision: Open, Short and Normal
• See related: Alarm Permits, False Alarm Fines and the Dealer's Part
• Contact manufacturer datasheets for rated gap by mounting surface