Detecting door that's barely open


(Nelson Pardee) #1

I have a door that “wants” to be barely open, but not enough to trip the sensor. Adjusting the door is not an option just now. On one of the doors I probably should buy a wireless deadbolt, and the combination of the sensor and the deadbolt should detect the problem. However, I’m not ready to sink that much money into all my exterior doors just yet. Any suggestions? Yes, I can adjust the sensor position to mitigate the problem, but probably not enough to make it go away. Any suggestions would be appreciated.


(ActionTiles.com co-founder Terry @ActionTiles; GitHub: @cosmicpuppy) #2

You can definitely adjust the Contact Sensor physically to mitigate the problem nearly 100%.

Whatever portion of the Contact Sensor is on the door, you need to shim it out from the door (such as stacking sticky pads underneath it), moving it further and further out of alignment with the portion on the door frame.

Eventually you will have moved the two parts of the Contact Sensor until they are “just barely” registering the “closed” state – i.e., you will find the extreme range of the distance between the magnet sensor and the magnet.

Now since you are at the “most extreme point” only when the door is “fully closed”, therefore even the tiniest extra separation will now indicate that your door is not “fully closed”.


This same technique can be done with other sorts of sensors: My garage door Echolink Tilt Sensor, is not perpendicular (straight up) when the garage door is closed; rather it is angled significantly towards the horizontal, but just not quite enough to be tilted to the “open” amount (NB: This sensor triggers at about 45 degrees angle, it does not require a 90 degree turn to fully horizontal). So if I tilt it at 40 degrees … then you see only 5 degrees more horizontal and I know my garage door is at least a tiny bit cracked open.


(Gary D) #3

One warning about increasing the distance: Some “contact” sensors (that are really magnetic proximity sensors) become less sensitive as the battery ages (and provides less voltage.)


(ActionTiles.com co-founder Terry @ActionTiles; GitHub: @cosmicpuppy) #4

Good point.

It emphasizes that the best solution for @nrpardee situation might be a good old fashioned physical lever switch (I’m too lazy to look one up…)… The kind with a short spring loaded reed with a little roller bearing on it that pushes in when the door opens over it. It also is “tuned” by placement choice.


(Eric) #5

The SmartSense MultiSensors angular sensors are incredibly sensitive. I bet those would detect the door not fully closed.

The tight limits required make the monitoring of this door unreliable. I would not use it in an alarm scheme, just notification.

ooo I got it- close the door!


(Nelson Pardee) #6

I’m coming to the same conclusion- can’t set something up to trust for alarm. But I do have a question- for a normal door that swings on a hinge, how would it detect angular motion. I thought that was detecting orientation compared to level. The description of this product on the SM web site is extremely limited, and the documentation isn’t any better.


(Eric) #7

MultiSensor knows its orientation to earth’s magnetic field “threeAxis”. I don’t think it responds to gravity.

e.g.

Current States
temperature: 45 F
battery: 63 %
contact: closed
threeAxis: 2,1044,4
acceleration: inactive
lqi: 100 %
rssi: -37 dBm
status: closed
door: closed

from IDE login
https://graph.api.smartthings.com/login/auth


(ActionTiles.com co-founder Terry @ActionTiles; GitHub: @cosmicpuppy) #8

No… the threeAxis is an accelerometer (only the open/close Contact Sensor is magnetic).

You can verify this by shaking the sensor and see that it registers a much high “g-force” value than if it were just reporting a value relative to “earth’s magnetic field”.

And measuring shaking is exactly why this sensor is useful to detect door knocking, glass breaking, washing machine activity, etc…

But acceleration also serves orientation, at least while on a planetary body, because “down” has an acceleration value of roughly 1G on Earth (I think it is reported as +/- 1000).


(Eric) #9

ah so it does know gravity’s direction. Are the units then 0.001*gravity?

It never quite zeros out then. You’d expect all the vectors to add up to 1 at rest.

I wonder where I got the magnetic orientation from - musta made it up.