Radar Sensors Put To The Test

[Andreas Spiess] picked up a few inexpensive radar sensors. He decided to compare the devices and test them and–lucky for us–he collected his results in a video you can see below.

The questions he wanted to answer were:

  • Are they 3.3 V-compatible?
  • How much current do they draw?
  • How long to they show a detection?
  • How far away can they detect the motion of a typical adult?
  • What is the angle of detection?
  • Can they see through certain materials?
  • Can the devices coexist with other devices in the same area? What about WiFi networks?

Good list of questions, and if you want to know the answers, you should watch the video.

The devices he examines are the RCWL-0512, HW-MS03, WB3-12, XYC-WB-D1, and the HFS-DC06. The RCWL module is the least expensive, and we found several places selling them for anywhere from fifty cents to a dollar each. The most expensive module–the HFS-DC06–is about $5.

If you are interested in these, this video will save you a lot of experimentation time. The boards are all somewhat similar, but [Andreas] covers the differences between them early in the video.

We’ve seen cheap radar detectors before, but not this cheap. We’d love to revisit some of the other radar projects we’ve seen in the past and see if they could use these very cheap devices.

68 thoughts on “Radar Sensors Put To The Test

  1. Not wanting to be pedantic or anything but these are not radars. More like mixers with lousy local oscillator isolation using audio beat detectors. Radar sounds better, but just not so, IMHO.

    1. “Radar is an object-detection system that uses radio waves to determine the range, angle, or velocity of objects.”
      It sounds like they are radar to me. Sure, very simple and ingeniously made ones, unlike the big things we are used to.

      1. RAdio Detection (or Direction) And Ranging
        This can only detect movement, they are just what’s sometimes called a doppler device…

        In order to become a radar, the carrier freq. needs to be freq. modulated, then you’d have a real radar (an FMCW one), capable of actually ranging something (albeit not very far).

    2. I was thinking also that these were not truly radar. I saw a video where the Loosely coupled oscillator was affected by capacitance changes in the room such as moving arms and legs. I don’t recall where I saw the video but these circuits seem to be something similar.

  2. Any chance these are in fact passive sensors (excepting maybe the most expensive unit)? Something like how a SAT TV 10GHz LNB would detect thermal noise from a human, ie all that’s on these boards (rf side) is an antenna and a preamp/LNA…

  3. I wonder if the antenna is this one:
    http://ieeexplore.ieee.org/document/7175452/

    Could you please report the exact length it runs?

    Since such an antenna has a spherical TEM far field, I think one can exploit reflection phase shift to count the ripples in the current to calculate the distance. That would be super cheap. (The more expensive device probably uses the backplate to enhance the forward TEM field by superimposing the back one on top of it)

    I am curious, if it uses this principle, in principle by replacing the antenna with a different shape will convert it to different purpose, i.e. a super precise point radar, energy transfer device, etc.

    Please google Fermi tapered antennas
    i.e. http://users.encs.concordia.ca/~latams/realizations/mm-wave-antennas/

    (reminder: I am a theoretician, and practical considerations is beyond me)

      1. The antenna in the IEEE paper looks pretty much the same as the various designs on the concordia.ca page.

        Specifically, the one on the IEEE paper has a cut-out as seen on the concordia yellow background photo, bottom left, labelled “horizontal element”, and copper shaped like those at the bottom-right of the same image (also labelled horizontal element). The IEEE design doesn’t have the perpendicular piece mounted in the notch though.

    1. No. The antennas in the linked articles look nothing like the ones used in these modules. These module antennas are just simple quarter wave monopole fed by microstrip. They are meandered to reduce board space. Except the HFS-DC06 uses a patch antenna.

  4. The way these work is ingenious. There is a 1 tranzistor oscillator feeding an antenna. When some reflective thing moves in front of them, the reflected wave has a slightly different frequency due to Doppler effect. Because the antenna is tuned, it will capture this and feed it back into the circuit, causing interference which produces a wave at the frequency of the difference between transmitted and reflected (~ Hz). This wave is taken out of the circuit through a passive filter and fed into a typical PIR motion sensor detector IC.
    Smart no 2: because the operation frequency depends on some passives with high tolerances, the chances of 2 sensors interfering with each other are minimal.

    1. Yep! Single transistor radar!
      I think Elektor had a project like that for 2.4GHz in the decades past, maybe also the russian “Radio” magazine.
      One amusing fact about the construction is that it’s a really old idea.
      The original WW2 proximity fuses on anti-air shells was similar, except with a vacuum tube.

      The sad thing is that these sensors are really dirty (in therms of RF) and likely illegal everywhere as there’s no harmonics suppression and no stability. But they are low power and cheap so none likely cares.

      1. Mainly the low power works in their/everyone else favor. With that being said someone probably already has radar ICs with all that. It’s not going to be $5.

      1. A quick google would have answered your question “The language appears in writing in the Superman comics at first as random squiggles. Then during the 1970s E. Nelson Bridwell created an alphabet of 118 letters, Kryptonese, from the squiggles. In 2000 DC Comics introduced a transliteration alphabet containing symbols for each of the letters of the English alphabet.”

  5. “Good list of questions, and if you want to know the answers, you should watch the video.”
    Sorry this is a pet peeve of mine. Watch a 16 minute video to get answers that I could have read in 16 seconds? Pass.

    1. I agree with your logic. Sometimes a video is useful. Sometimes not. I prefer to have text whenever there is no video necessary because text is possible to skim through and get an overview, then stop for details. Kind of like a progressively encoded jpeg.
      However … people who make content tend to skip to one format. Nowadays video brings in more money and up to a certain quality level it is easier to do.

    2. Sometimes taking the time to watch a quality video provides you with many more ideas and directions you can go with things. If you want quick answers to the questions, read the datasheet for the devices – sometimes the answers are not black and white, and require demonstrations to explain properly – this guys videos are a gold mine of information.

      1. This is distributed circuit design as opposed to a lumped element design, therefore there is nothing trivial or obvious about it. What you are seeing are radial stubs. You can read about them here:

        https://www.microwaves101.com/encyclopedias/quarter-wave-tricks

        The radial stub in the lower left is part of a bias tee connected to a discrete transistor in an oscillator circuit. The thin trace that feeds the tip should be ~1/4 wavelength. The radial stub is sized large to act as a capacitor or high frequency shunt.

    1. We use these where I work as discrete motion detectors along with cameras looking for motion. The angle is about 180, distance is 20m. They have no micro so you need to provide your own that has a decent ADC(ours is 20 bit), along with a preamp. All of the rooms we put them in are RF shielded entranceways/man traps so they won’t overshoot and pick up random traffic outside. With RF shielding, using a fourier transform, you can also get a volumetric measurement which we use to alert on threshold.

      1. I am looking for a microwave sensor that can do 180 degree motion detection of 15m or more like you said you were able to do with the HB100. Is it possible? would like to know more if you have such sensor? thank you very much.

  6. Thanks so much for this review. I have an application that needs a longer range so I chose the HFS-DC06 which you tested to have an 8m range vs. 4m for the other sensors. However, I’m having a bit of trouble. When I first got it, it seemed non functional as it was ‘triggering’ 100% of the time. It took a while before I realized I had to turn down the sensitivity potentiometer. However, with it turned town to about 75%, it periodically triggers with no apparent motion. Leaving it running all night in an empty room, it triggered every couple minutes. Turning it down further to the point where this phantom triggering stops (at about 60%) the range seems to be limited to about 1-2m. Do you think I have a defective unit? Any other suggestions?

    1. The HFS-DC06 needs a very stable and clean power supply. In my case, the device did not work well on 3.3 volt.
      It then displays the behaviour you described. I use it on 5 volt, and it works very well. It is in fact the only radar sensor that works in a reliable way with an esp8266. I use it with a Wemos d1 mini. This device tolerates the 5 volt output of the sensor. Btw, the HFS-DC06 is the only sensor, of the 4 types radar sensors i have tested, that works reliable and without picking up “noise” from the ESP8266 wifi radiation.
      Beside al of these good qualities it is also direction sensitive, so you can “aim” it.
      My HFS-DC06’s shortest ON time is less than 2 seconds! so it is ideal in using it in Mqtt.
      It’s true, it is more expensive than most of the other sensors but 2,50 Euro still isn’t very much!

      1. Thank you, Djkahn, for your reply. I will try using 5v, though I think I may have tested that already. I will also try making sure I’m using an extra powerful power supply for my Pi Zero W. To be safe, I’ve also picked up an extra HFS-DC06 to ensure it wasn’t a bad unit.

      2. thx for your comment, i am planning of using the device also with an esp8266 for sensining if anybody is in the room and swithcing lights accoredingly. My question is how sensitive the device is?
        What is if anybody is sitting on a desk and typing (just moving fingers and hand slightly) or lies on the couch and just moves every now and then.

        Does the HFS-DC06 get those movements too?

  7. Hey all,

    With regards to the RCWL-0512…

    Just trying to figure out if this sensor can detect distance as well as movement? I’d like it to act as a proximity sensor, where a response is triggered based on the detection of an object between distance value x and y.

    Additionally, I’m wondering if there is any way to limit the angle of detection? I need something with less than 30-45 degrees.

    If the RCWL-0512 isn’t the best sensor for this job, do any of you have a better recommendation?

      1. At these frequencies thin metal acts as a beam-stop and/or reflector. Putting the board near the bottom of an empty tin can should, I believe, give a simple but crude approximation of a parabolic reflector – detecting the radiation out of one end and only accepting radiation from that direction.

        The question that I would like to ask is: How quickly do they respond to a ‘valid’ movement? Andreas gives values for how long they stay ‘on’ after a detection, but not how quick that detection takes. For example, if one of these was set up in a tightly colimated holder ‘looking’ through a small hole, would it see a bee flying close to that hole (hopefully it would since the bee contains water) but how quickly would it respond?

  8. I just bought an HFS-DC06, and was hoping to use it as doppler radar to measure vehicle speed. It only has 3 pins – gnd, Vcc and Out. Do you know if any of the other PCB traces can connect to the amplified analog signal so I can measure frequency?

    1. I look for a distance control to the backside of my car – trough the back-side screen (glas) with one of the radar modules. Can anyone recommend me a solution? If a car comes from behind inside the radar range of 5 – 8 meter i want to get a signal.

  9. Do you think it would be possible to use one of these devices to verify if an object is a person? I am using a VL53L1X Time Of Flight sensor to detect if an object is <2m but want to verify if the object is human. Current best attempt uses a PIR sensor but they aren't great, especially outside.

  10. I just tested RCWL-0512, and it seems to work well (or even better) if powered with 3.3V. This can be really useful when we only have a 3.3V supply. I noticed that the only chip on the module contains a LDO which provides 3.3V for both the high frequency circuit (which includes the transistor) and everything else inside the chip. So by shunting Vin and 3.3V module’s pin, this LDO will be bypassed, and powering it with 3.3V should be equivalent with the normal operation. I haven’t tried powering it on the 3.3V pin directly without shunting to Vin, as it seemed less safe.

  11. Hello everyone, can you advise me how to reduce the switching time below 2 seconds on HFS-DC06? ideally, I only need a short pulse in the range of a few milliseconds. It is clear to me that I will probably have to replace some resistance, so I am wondering if anyone has experience. well thank you

Leave a Reply to whiskCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.