Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Why doesn't the alarm trigger? Sentry useless?

This site may earn commission on affiliate links.
I put a 100+ dB alarm module on my back triangular windows,
with a threatening sticker. It flashes an LED. The intent was, in
addition, for the (loud) shriek to set off the car's alarm if someone
bangs on the window to 1) break it, 2) reach in and drop the back
seat and 3) access the trunk. A common Tesla break & enter in
California.

After a lot of (painful) testing I got the sensitivity just right so it
doesn't go off when slamming doors etc. The next step was to
test triggering the M3's main alarm in Sentry Mode. No luck.
Nicki, my M3, ignores the screaming alarm. So, in a loss of
faith crisis, I forgot all about the window alarm and started
banging on the windows, the hood, the side panels. No
reaction. The one time I managed to set it off by slamming
the windshield I nearly broke my hand, and the windshield,
in the process.

This alarm is nearly useless, need a fix.

It's nice to have a video clip of the break-in, but that's not
really much of a deterrent for thieves with limited cranial
capacity. Does engaging Sentry Mode make the default
alarm less sensitive? I'll test it some more when my arm
recovers. WTF?

What is supposed to set the alarm off? Is it supposed
to react to body odor, or what? Does it listen? for what?
With what microphone? The manual is thin on details, and
that's an understatement. Is there a known huge variation
in sensitivity for different cars, since some people seem to
be plagued with false alarms? is it possible that the mic
turns off somehow, e.g. by Bluetooth pairing to any media
device in a 100 ft radius? That could be one clever thief
hack, just sayin'

Is there any way, short of wiring in a claymore to blow the
doors off, to make my car react to non-consensual
penetration?
 
I had the phone out of Bluetooth range, inside the house, though it wasn't in airplane mode.
I did try with airplane mode on today, just the basic Alarm, and the screamer on the back
window didn't set it off. I'll try again with the Sentry Mode on.

But what's the deal with "cabin overheat protection" (limitation to 105 deg F)
being incompatible with the Alarm? I saw that somewhere, but in the controls
they don't mutually exclude. I always have Alarm on.

But in my failed Sentry Mode tests I did manage to set off the alarm once.
The doctor said my wrist isn't broken, to wear a brace and take pills. LOL.
 
We have no Tesla engineers hang out here?
We probably do, but AFAIK none have come out of the closet yet. :p

It's possible that the Sentry mode alarm works similar to other glass break sensors in that it doesn't react to just any noise, but listens for a specific acoustic spectrum that is typical for breaking glass. That said, when I tested it I was able to trigger the alarm by banging on the little triangle window.
 
A rubber ball? Is that the latest "out of Compton" hoop B&E?

Or are we protecting the car from kids playing in the street?

It beats using my hands for tests, I'll give you that for sure.
But FWIW what I'm looking for is a response to real threats.
A lot of broken back windows reported in the Bay Area.
 
A rubber ball? Is that the latest "out of Compton" hoop B&E?

Or are we protecting the car from kids playing in the street?

It beats using my hands for tests, I'll give you that for sure.
But FWIW what I'm looking for is a response to real threats.
A lot of broken back windows reported in the Bay Area.


My sentry alarm has triggered when large and loud trucks drive by my parked car. I assume it’s due to vibration of some sort.
 
  • Like
Reactions: qwertzy
I think you're working with too many (incorrect) assumptions about what ought to set off the alarm, as opposed to what actually does set off the alarm.

For example, beating the windshield with soft hands, which you've pointed out damages you but not the window, did not set off the alarm. You cite this as a fault... to me, it sounds like the car got it right! No damage to car, no real threat of breaking in (proof: you failed), and the situation was recorded and the warning was on the screen. I see no need to harass all the neighbors with the alarm, as well.

I don't know what sort of sound analysis is being done to determine what triggers the alarm - but I think it's a mistake to think it's just volume level. It's more than likely a neural net looking at the audio spectrum, so that sounds that look like they come from speakers (like another car with loud radio or your screecher) won't cause a false positive. A no-damage thump from a hand does not set it off, but the different spectrum of sound from a ball's impact does set it off - that's good discrimination on the TYPE of sound, as opposed to the volume of the sound.

You can't go into analyzing a neural net's behavior with a decades-old mindset about how "detection" should work with simple analog metering.
 
  • Like
Reactions: C141medic
That's nice. But that's kind of old school and polite.
I'm more concerned with someone opening a window
... with a brick, wrapped in a (microfiber) towel ;-)

Either way your window gets broken. The alarm is there to scare them off before they steal your stuff.

The only way to avoid this is to have an alarm that sounds BEFORE they touch your car, which would be annoying for everyone.
 
  • Like
Reactions: C141medic
A rubber ball? Is that the latest "out of Compton" hoop B&E?

Or are we protecting the car from kids playing in the street?

It beats using my hands for tests, I'll give you that for sure.
But FWIW what I'm looking for is a response to real threats.
A lot of broken back windows reported in the Bay Area.
To properly test the alert state (horn honking and fugue playing), you need to don a furry blue costume, munch on a chocolate chip cookie and pound vigorously on the windshield with your front paws. At least that's how I saw it working on a YouTube video...
 
  • Funny
Reactions: C141medic
Either way your window gets broken. The alarm is there to scare them off before they steal your stuff.

The only way to avoid this is to have an alarm that sounds BEFORE they touch your car, which would be annoying for everyone.

Or to have it become common knowledge that breaking windows on a Tesla will result in loud noises. Which will, in turn, reduce the number of broken windows.
 
I think you're working with too many (incorrect) assumptions ...
You can't go into analyzing a neural net's behavior with a decades-old mindset about how "detection" should work with simple analog metering.

Wait a minute, good points, but who's assuming what here? Neural net vs decades old analog what? Really? Are you getting it from the Tesla developers? If so, or if you truly know the details, great, then please help, engineer to engineer.

My objectives: 1) find out if my car is working normally, 2) how to couple additional detection into the existing triggering, or 3) at least fully understand it to make best use of it. Because as it is, unlike the majority of the Model 3's design, the alarm is not fully meeting my (mistaken and antiquated) psychological needs.