How to setup multiple X-Band motion detectors

meesterstump

For my project, I plan to use two sensors (arranged about 120 degrees to each other) pointed outwards through a window so as to detect people walking by the sidewalk outside. I assume that I will encounter interference if they both are pulsing at the same time.

Is there a recommended approach to this? I have been unable to find any reference of anyone using multiple X-Band sensors in the same project. A reference like this one for the Maxsonar sensors would be great. My first thought is to stagger operation of each by enabling and disabling them in rapid sequence, but I'm not sure at what frequency to best use or how that will effect the readings. I'll just have to experiment, but any advice is much appreciated!

ElectricAye

How close together will the two sensor units be? If they are almost next to each other, you might get interference through their rears, so I would shield the two units from each other that way. If they are aiming in directions 120 degrees from each other, and their beams are not hitting the glass at a glancing angle that could reflect radar back toward the other X-band unit, then I'm guessing you might be able to make it work without needing to stagger their operations. If the radar hits the glass or any other object at a glancing angle, the reflections might create interference, so keep that in mind.

If you decide you need to stagger their operations, then use the ENABLE pin to do so. Do not stagger their operations by powering them up and down. The Xband units are modulated at 2KHz, so you definitely would not want to stagger operations at a rate near that. Maybe you could get away with staggering every 1.0 second or so, but I'm just guessing on that.

meesterstump

Thanks for the comment. Yes, the sensors will be only a few inches apart, pointing outwards through the same pane of glass. I doubt I can shield them well enough from each other.

I am just now testing the results of staggering them with an interval of 100 milliseconds (using the ENABLE pin.) Ideally I'd like to lower that frequency so that it's more responsive, but I'm worried that I'll stop getting a meaningful pulse count if it goes much lower. Preliminary results seem promising enough, although I've got some fine-tuning of the frequency, pulse count threshold, and sensor sensitivity to get reliable person detection working...

ElectricAye

meesterstump wrote: »

... the sensors will be only a few inches apart, pointing outwards through the same pane of glass. .....

Maybe you can shield their rears with some foil or something. But what concerns me most of all is that you have their beams separated by 120 degrees through the same pane of glass. Have you tested this arrangement with people on the sidewalk? I'm concerned that angle might be so great that you are losing a lot of your power via reflections off the glass. Even though radar will travel through glass, such an angle will reflect some of the power - I just don't know how much.

meesterstump

I thought I'd post a short follow-up. The goal of my project is to quickly and accurately detect people walking by a big window which will trigger an action in the window that is visible to those people. Originally, I used a single X-Band sensor in the center of the window facing directly out, but the trouble was that by the time the sensor had triggered, the person was usually over half way past the window and would not notice the resulting action. So then I opted for a second sensor to extend the range in both directions. Here is a quick illustration of how I have the sensors positioned. Ignoring the effects of reflection and refraction from the glass, I also illustrated the sensitivity vs. angle as an outline (this was pulled from the documentation) and shaded in the arc of maximized sensitivity.

After a little testing, I determined that placing the sensors at about 45 degrees to the glass (at right angles to each other) seems to be the best compromise. I can't say exactly how much sensitivity is lost due to the glass and its angle, but it is not so much as to be a real concern in this particular application. Now, when a person walks by, they are mostly detected early enough to trigger an action that is visible to them as they are first in view of the window front.

I have the sensors' sensitivity potentiometers dialed to nearly the minimum. In code, each sensor is enabled for 100ms and the number of pulse cycles are counted. In this setup, the threshold for triggering action is 2 (this filters out a few minor movements that seem to always be present.)

ElectricAye

Interesting set-up. Glad to hear it's working. And thanks for providing feedback on your results - it's always good for all of us to learn something from what people do here.