Variable Output

[BiatuAutMiahn]

I was wondering if there were a way to modify this to give a variable or PWM output representing a distance, or speed?

I figure a number of these placed strategically might allow one to map and object in 3d space.

[barewires]

The device is omni-directional meaning it sees in all directions - XYZ, therefore making it unsuitable as a focused sensor - it is either on with motion or off. In my cluttered livingroom / lab I get 3 m response, any movement whatsoever triggers it. It sees through drywall and even people outside my front door when it is placed in my hallway. I have added a 1 meg resistor on the back at location R-GN to reduce the default detection range from 7m to 5m.

[BiatuAutMiahn]

ok, but if you were to put a few like 3 or 4 of these would you be able to detect the position of a moving object? like GPS (timing differences) or triangulation?

[barewires]

I have experimented with multiple units in close proximity 1 cm apart, pointing different directions and they all react to any motion at the same time. There is no possible way (without building microwave horns or wave guides) to get directional or distance information, as far as I can see.

[BiatuAutMiahn]

I see, however theoretically, I wonder if one were to tweak the board to that the output signal 'on' time is reduced from a couple of seconds to like 10ms or around there, and then hook the outputs of 3 or 4 to the GPIOs of a raspberry Pi, or arduino...and set the modules further apart, like 1ft or 4 inches, there should be a very small time difference in which each module is being triggered.

If they all still trigger too fast, then either they need to be set further apart...or the rPi or arduino needs to be replaced with something that has a faster timing

[barewires]

No, Not going to happen. They are analog devices and the timing is variable, controlled by op-amps, capacitors and resistors and physical placement and proximity to metal. While not suited for what you want they are incredibly amazing devices for detecting cats and lions coming in and out. Bon chance!

[barewires]

As an observation I put the device in a round metal 9 cm can with a 7 cm plastic window (fidget spinner) and get good directional response through the window.

[BiatuAutMiahn]

hmm nice. Ty

[nissimzur]

Does someone did PCB layout and can share or sell it to me?
nissim.zur@gmail.com

[Joerg-rw]

set the modules further apart, like 1ft or 4 inches, there should be a very small time difference in which each module is being triggered.

Those devices have operation principle immanent interference between them, you basically can't operate them in a meaningful non-interfering way within same range zone. Device A won't be able to tell apart reflections of own signal from those of device B, C...

I did tests and been able to trigger a sensor by simply powering up another one in 6m distance and that wasn't the distance limit but simply the max distance I could test at my scenario.
Considering how much stronger direct signal from antenna of B to antenna of A must be, compared to signal from A getting reflected and received again on A, I'd expect you need at least 10m distance between sensors to have sufficient attenuation of crosstalk

[xenoeng]

You could only power and check each one in turn, cycling through them? Trickier to do with a distributed system but not impossible. ESPs and MQTT could provide the messaging and sequencing needed.

[Joerg-rw]

I just an hour ago thought about some prerequisites for that, since the plain device as is needs some seconds to power-up, but I guess that's mainly the opamp and timer stuff in the chip. The RF-section probably could get enabled and stopped within microseconds, you just need to avoid triggering a false positive on the chip then from that.
I'll report if I actually look into that any closer

[xenoeng]

Any false triggering at power up could be ignored by the software. Could have a configurable settling time for different environments and conditions.

[Joerg-rw]

sure, no problem technically. Just it reduces the poll frequency so much that you can't do reasonable polling since all sensors of a RF-zone have to be down for too long when one of them just powers up. Long enough so somebody could walk through the complete range without getting detected.
Assuming a XY grid of sensors at ~max distance [±(root(2)-1)/2 ] and a settle time of 2s, you could poll one out of the 4 sensors 1:1, 1:2, 2:1, 2:2 at a time. To poll them all (plus all the other sensors of the grid you could do concurrently (1:3,1:4,2:3,2:4; 3:1,4:1,3:2,4:2; 3:3,3:4,4:3,4:4 ... ) you need to do 4 times powerup&settle = 8s, plus you want a certain active monitoring time per sensor after the power-up-settling. Let's assume this is 1s active monitoring then an intruder has 4*(2s+1s) = 12s in which he could walk across that 22 senor area with risk of getting detected only in 4s out of those 12s, since 42s=8s simply none of the sensors is active. Or more precisely, every sensor would be active for 1s out of 12s and you can't speed that up, only slow down to e.g. 2s out of 4*(2s+2s)=16s or 3s out of 4*(2s+3s)=20s
[edited to fix an oopsie in topology. Sorry]

[xenoeng]

You are right, I was thinking 3 or 4 sensors. As the numbers go up it becomes impractical.

[Joerg-rw]

You'd probably want to reduce startup&settle time to <0.2s which should be feasible with a few modifications/extensions to the circuit.
I never tested what the ENABLE input actually does, might do exactly what we need :-)

[xenoeng]

Interesting! Can you share?
Watching with interest.