We have made a mistake. We created this pcb with accelerometer , a very good one (ADXL345) datasheet here and it sleeps ,and waking up on acceleration event. in our case we set it to almost the minimum possible of 124mg thresh .

What happens is that it works great almost for any touch or move that you apply to it by hands, means if its on a table even a tiny move will fire it on .

But, when its mounted to a door, or window, if you open the door not quickly enough it just won't wake up .

We have tried everything, including taking sensitivity to maximum.

Seems that it only needs a change that is not occur on an average door opening .

  1. Is there anything you can do to improve this not including migrating to other sensors ?

  2. We have a free GPIO connected to interrupt, we can add some part there, that will sense this tiny movement, is there anything like tilt sensor that will do it?

Thanks .

  • 2
    \$\begingroup\$ Have you worked out how the power budget would hold up if you used a sleep mode on the microcontroller and periodically polled it instead of using the threshold detection? \$\endgroup\$ – PeterJ Nov 30 '15 at 12:06
  • \$\begingroup\$ @PeterJ great idea, You mean to wake up and read acceleration many times? I will have to do that at least every second, so in sleep mode right now i am on 80uA, will it be much more current ? How you calculate it? \$\endgroup\$ – Curnelious Nov 30 '15 at 12:55
  • 1
    \$\begingroup\$ Start by NOT SLEEPING, and using the accelerometer to tell you what your acceleration profile is. \$\endgroup\$ – Scott Seidman Nov 30 '15 at 14:15
  • \$\begingroup\$ I first have to know what would be the current for every half second wake up because it can't be too much.. \$\endgroup\$ – Curnelious Nov 30 '15 at 14:19
  • 1
    \$\begingroup\$ @Curnelious - "I first have to know what would be the current for every half second wake up because it can't be too much" Have you tried reading the data sheet? It's listed. \$\endgroup\$ – WhatRoughBeast Nov 30 '15 at 16:50

The first thing to do is to mount the accelerometer as far out on the swinging end (not the hinge end) of the door as possible. Then, of course, the accelerometer axis must be aligned perpendicular to the door.

High pass filtering may help, but that requires leaving power on.

Ultimately, you may not be able to do get what you want within the constraints you have imposed on yourself. The obvious answer is to use a device appropriate for this application. Not all things are possible just because you can write a spec for them.

"I bought a bushel of tomatoes instead of a lug wrench, and now I'm on the side of the road with a flat tire. How do I use these tomatoes to get the tire off. No, I insist on not walking to the hardware store around the corner to get a lug wrench."

|improve this answer|||||
  • \$\begingroup\$ Not sure about this answer " normally you cant do that", i was told this so many times that turned to be wrong. If you have a real explanation on why (physically ) you cant do that, please let me know . There are no constrains, only to detect a door open with a very low current consumption. \$\endgroup\$ – Curnelious Nov 30 '15 at 14:09
  • 3
    \$\begingroup\$ @Curnelious I'm with Olin on this one. Why did you chose the accelerometer to detect a door opening? A Hall sensor on the frame plus a magnet on the door would be able to detect any opening, no matter how slowly you perform it. \$\endgroup\$ – Dmitry Grigoryev Nov 30 '15 at 15:30
  • 1
    \$\begingroup\$ @Curnelious physical explanation? Please. The door can be opened with low accelerations indistinguishable from the noise specific to this device. \$\endgroup\$ – Eugene Sh. Nov 30 '15 at 19:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.