Motion sensor

The reason why I disassembled this cheap motion sensor & light switch was, that I needed a cheap but sensitive and rugged motion sensor for my - ehm - home automation (ok, light control system at its present state, but it will grow, I am sure... :-) ). The sensor is placed above the garage door, and when it senses movement, the central control box switches on the exterior lights for a certain time (10 seconds currently, although I will make this time longer soon - there are also some interactions with the switches controlling the individual exterior lights, e.g. pushing them the lights go off immediately).

I chose this type of sensor because sensors used for alarms (of the weatherproof type) are at least an order of magnitude more expensive than this (they are more sensitive and directional and reject false "alarms" better, but nothing of this is necessary here), are usually bigger and uglier than this, don't provide the day/night sensor, and cannot be simply bought in the shop around the corner (I asked and was told that they have to be ordered with a waiting time of several weeks).

On the other hand, this cheap type of sensors comes exclusively with mains supply and power output switch, intended to be directly wired on the lamp. So, I had to open it up and tweek it to suit my needs.

But first, let's have a look at the sensor as it comes out of the box.

motion sensor - untouched yet

It has a quite nice design, can be fixed easily to the wall, and can be easily adjusted along one (supposedly horisontal - but I mounted it slightly tilted) axis. The case is a grey plastic with silver paint, but as long as I proceed carefully and won't scrape down the paint, it looks cool. There are two knobs on the upper part of the "ball" - we will "spoil" both later, but originally one determines the time until the light is on ("TIME"), and the other determines how early will start the sensor work at dusk ("LUX"). There is no tweak for "sensitivity" of the motion sensor itself.

motion sensor - first step of disassembly

The disassembly starts with removing the "ball" from its holder. The leads to the sensor itself can be seen - the "holder" contains no active electronics (this picture shows the sensor AFTER modification, and the extra thin grey wire carries the DAY/LIGHT signal).

After disassembling the two halfs of "ball" (which are glued together, providing the IP44 cover), the electronics on two interconnected PCBs is revealed.

motion sensor - the two boards

The first PCB holds the "power source", which is of "transformer-less" design plus a voltage regulator (78L05); and also the "power output", which is basically a relay.

motion sensor - the 'power board'   motion sensor - schematics of the 'power board'

The second PCB is the sensor itself.

motion sensor - the 'sensor board'

The signal from the IR sensor (which is in the round glass-covered package in the middle; with the mirrors (a metalised piece of plastic) above to widen its sensing area to the sides) is heavily amplified (and filtered) in a two-stage AC-coupled amplifier, then the third opamp forms a detector together with two diodes, and its output is fed to a RC-network with large (and adjustable) time constant, which determines the ON time of the detector after having spotted a movement.

There is a second sensor, which is sensitive to visible light (it is positioned on the right side of PCB, in a LED-like clear package), which can via a transistor (Q1) block the output of the movement sensor - this is the DAY/NIGHT sensor. On the other hand, the DAY/NIGHT sensor itself can be blocked via a diode from the output of the whole sensor, so that if once the sensor switches a lamp on, the light from this lamp impinging onto the sensor would not cause "DAY" condition which would in turn cause switching off of the lamp... If this "final block" would be omitted, the whole setup might have lead to a wildly flashing lamp...

motion sensor - schematics of the 'power board'

The tweak I made is denoted by red on the schematics. On the supply/output board, the voltage drop resistor/capacitor network was bypassed, but with the rectifier bridge and voltage regulator remaining in place, thus enabling 12V supply (of any polarity, error-proof). Similarly was bypassed the output relay (which I removed completely to reduce power consumption).

On the sensor board, the capacitor of RC network was replaced by a 3 orders of magnitude smaller, so that the sensor output was not delayed by minutes but only fractions of a second. Also, the DAY/NIGHT sensor was completely separated from the rest, and its output was brought out of the sensor using an extra wire.

Measured consumption of the modified sensor was around 20 mA at 12 V, and the sensitivity of both the motion sensor and the DAY/NIGHT sensor were more than enough. I have two of such tweaked sensors, but found use only for one of them at the moment, the second providing me a backup.

Just out of shear curiosity I disassembled an another similar sensor, this time one which can be commonly found on strong outdoor halogene lamps. This is less suitable for standalone use, as it does not have a "holder" of its own (it is mounted directly onto the lamp). It had a similar two-board design, but there was no stabilised power source on the "power board", and the power output was solid state, featuring a triac (BT13x). The "power board" had also traces for the "toggle-switch-rapidly-for-lamp-permanently-on" circuit, but that part was not populated.

The following is the schematics of the "sensor" part of this device (no other pictures of this one, sorry):

motion sensor - schematics of the 'power board'

This is a much less clean design than the first one and it took me considerable time to figure out its function. The input amplifier is only single-stage, the second opamp provides the rectification and at the same place the the DAY/NIGHT sensor's blocking output is fed in. The "timer" RC-network follows with the third opamp as the output amplifier; while the fourth opamp serves as a voltage regulator mainly for the infrared sensor (although the opamp itself is supplied from unregulated 12V supply).

This sort of tweak can be done mainly because the design of the sensors is very simple and they are similar to each other in concept. Note that even I might have not re-closed the sensor's case perfectly, the tweak does not compromise safety, as the sensor will not be connected to mains circuit anymore.