RCWL-0516 Microwave Radar Sensor Experiences?

Some updates on my setup, if someone is interested. It worked very well for about 3 weeks. No actual false triggers, except for the random foxes and neighbors cats running around at night :slightly_smiling_face:

That is, until yesterday. That is when it started to fire false triggers like crazy, every couple of minutes over many hours. What happened ? Well, it started to rain…

image

So that thing is not only over sensitive to Wifi, but also to rain. It kinda makes sense if you think about it, rain is a big aggregated body of moving water after all. But it also makes this sensor totally useless for my use case. Even an inside sensor was triggered due to heavy rain from outside of the window. I will still try to make the module less sensitive by adding a resistor to the R-GN pads. We’ll see how that goes.

I also tried the HFS-DC06. This one was even worse (and more expensive) and pretty much unusable outside.

In the end, this is all cheap Chinese stuff. They’re designed to cost a few cents to manufacture, not to perform well. They are fun, work surprisingly well in a controlled environment. But as soon as you actually start relying on them for some automation, you will run into their failure modes. I started looking around for higher quality alternatives. Bosch is doing some heavy duty outside rated microwave sensors, I’ll look into those. Of course they cost like 50 times more than an RCWL, but reliability has a price I guess :slightly_smiling_face:

Kinda talking to myself here, but it’s been over a month now that my RCWL based motion sensor is in testing, so I thought I might write down my final observations running the module in an actual every day use scenario. For future reference, maybe someone will find it useful :slightly_smiling_face:

So I’m very happy with the module after all. After more than a month of testing, I’m now going to start using it in actual live HA automations.

  • Adding a 1MΩ resistor to R-GN has helped tremendously. No more false triggers due to natural elements (stress tested in two storms with very heavy rain), blowing wet leaves or similar. The module is way too sensitive without it, they should have added a small trimpot on those pads by default. Interestingly, adding the resistor will not notably affect the detection range, which stayed at around 6-7m outdoors through a window, but it will affect the threshold for the trigger. So a larger object needs to move for it to trigger, rain and blowing leaves won’t affect it anymore. The neighbor’s cat still does though :wink:

  • The module is affected by high temperatures. I originally left the wallbox containing it (see my image above) open for testing purposes. It sits on a window sill inside. The problem was when the sun was shining through the window directly on the module, heating it up. It would not detect anything farther than a few cm anymore. Cooling it down brought it back to normal. Closing to box and adding a larger plastic cover as sun screen above it solved the problem. Professional microwave sensors have an internal temperature correction btw.

  • RF propagation patterns are weird, completely unpredictable and counterintuitive. That module isn’t a ‘real’ radar and does not have any kind of ranging and digital signal postprocessing. It simply mixes the transmitted and reflected signals and triggers when the phase shift variance over time goes above a threshold. The more reflected signals you have (multi path propagation usually due to metal objects around the sensor), the more likely it is that moving bodies will induce a high phase variance. Moving or turning the sensor a little, only a few cm sometimes, will completely change the detection pattern. Metal objects around the sensor will reflect the signal and create new paths, which will increase sensitivity (sometimes too much).

  • Shielding the sensor works really well. Originally I put in a small metal angle to shield the sensor from possible interference from the Arduino (see pic above). This had an interesting side effect of shielding the sensor from basically anything behind it. The sensor box is in a stable building and the horses had a tendency to trigger it from time to time when moving around. That little metal piece solved that problem entirely. Note that the shield is not grounded. Grounding it would make it not work anymore :thinking: When not grounded it probably acts as a mirror for the transmitted signal.

  • These boards were produced using the cheapest process available. There’s a lot of variance between individual boards, not one board behaves like another, the manufacturing tolerances are huge. If you adjusted your detector to work well in a certain location, don’t replace the sensor board or you will have to start all over. Some RCWL boards simply don’t work at all. Two out of the five I had are just unusable. They kinda work in the lab, but they’re totally unreliable in a real world situation.

  • There’s a lot of strange information about this device online. Just to clarify a few of the misconceptions floating around on various forums. You don’t need some over the top decoupling with ridiculously large caps, pi filters and whatnot. The board itself already has plenty of small decoupling caps soldered onto it. A single 22µF cap over Vcc is enough. Some people suggest the modules works better when powered with 12v or more rather than 5v. That’s nonsense. The board has a 3.3v voltage regulator onboard which powers the entirely internal circuitry including the RF part.

So yeah, great little sensor. Keep it away from wifi and bring a little time tweaking it (position, shielding, R-GN) and it ends up working surprisingly well.

End of report haha :grinning:

2 Likes