False triggers #2
Comments
|
I've been struggling for WEEKS with a miniature DIY motion detector device built with RCWL-0516 + ESP8266 + mini 5V P/S. |
|
Maby an isolated DC-DC converter can help to remove the interferences that come from the ESP to the RCWL over the supply voltage. I suggest some ferrites on the data and power cables to! |
|
Good point. |
|
I think I found a solution. I just use the electrical source from the above link but without mounting a filter it does not work properly. I used a pi filter (1000 microfarad +10 ohm + 1000 microfarad) mounted on the sensor supply (Vin). Of course, if there are "parasite generators" around the sensor, such as mechanical relays or mechanical switches, false flashes may occur. I use it with a ESP 8266, and a SSR G3MB-202 that has its own snubbering circuit and at its command sometimes false flashes occur . I have solved these problems in the software by temporarily disabling the pin to which the sensor is bound. |
|
Mine is undertest since yesterday with a single 1000uF capacitor directly soldered on the Vin input. |
|
Yes this is the configuration I've used . |
|
A bit of update :: |
So far, soooooooooooooo good !!!!The pi-filter (thanks to Mariusb57) works like a charm : I never had any more false triggers since I installed it between Vcc (+5V) and the Vin input of RCWL-0516 sensor ! I wish I could now understand the exact way this 'resistor-only' pi-filter works as (normally) a pi-filter is made up of an inductor instead of a resistor ... any clue ? |
|
The pi filter is not my merit, I have read this : |
|
I was a bit mistaken as it does not work THAT good. Actually, as soon as I REcase the whole thing together (ESP + RCWL + P/S), it starts giving me again false triggers, particularly when the sensor stands very close to the ESP32E ... it's just disappointing for I wanted to make the whole device dense and compact. If I can't get rid of those interferences between the ESP and the sensor, I will definitely have to rethink my layout and keep the RCWL-0516 outside of the casing as I confirm that it works flawlessly whith the pi-filter and some room (5cm) between the ESP and the sensor ... |
|
Not only this I have other microwave sensors which works at different frequencies and all have problems if kept close to ESP chip. Even keeping little far from ESP once in a while I have false alarm, so I used 10uf + 0.1uf capacitors close to RCWL and this helped and I used software to determine whether its a false/real trigger. |
|
I have played with a HB100 (similar idea to RCWL-0516 but at 10 GHz). The HB100 has just a low-level analog output ("IF signal") so amplifying and detecting circuit must be added. Listening to that output (through a preamp, with headphones) you can clearly hear the pulsing 2.4 GHz packets coming from my wifi hub even from 10 feet away. So it's clear these devices will be sensitive to wifi signals and if it's in a case right next to a wifi transmitter like the ESP device, I'm sure that can interfere. I have played around with HB100 in a microwave horn (cardboard and aluminum foil) and with the RCWL-0516 in a cookie tin as a "cup antenna" like http://www.microwave.gr/content/view/175/1/ and this does give me somewhat more directivity and range. If the wifi module was on the other side of the metal reflector, that could help reduce interference, but obviously using an antenna structure also increases the size of the device a lot. |
|
By the way, Pin 12 of the IC on the RCWL-0516 device (it is an opamp output) gives you an analog signal somewhere between 0 and 3V that wiggles when you move your hand. So if you wanted to take that into an ADC input in a Arduino etc., you might be able to do some signal processing to detect or reject specific kinds of signals. |
|
Hi, |
|
I tested three boards in proximity placed XYZ 1 cm minimum apart and they all appear to work independently with no interaction or interference. Two on the same circuit RaspberryPi 3 (or Zero W), powered from 5v and connected to GPIO Pin7 and Pin11 work just fine. |
|
Thanks for the note. I want to use two RCWL radars in the same enclosure. Can they be placed one BEHIND the other ? Two modules kept side by side appear to work without problems. But my application has a support structure right through the device, so I need to keep the two radars one behind the other. The sensitivity seems to have reduced drastically after this. Is that normal ? |
|
The devices are omni-directional and it is only necessary to have one at each location maybe 2-3 m apart. This minute I placed a second device on top of the first, separated only by a piece of postit note as insulation and they respond almost at the same time. As the circuits are analog (op-amps, caps, resistors) the on/off timing is variable with each one being affected by local placement that changes the sensitivity. Both devices are connected by jumper wires approx 10 cm long. It is very important to keep all metal at least a cm apart. Stacked devices will have mutual interaction due to metal makeup of the boards. Once again, plugging directly into a breadboard with the mass of hidden metal contacts will seriously affect behaviour. |
|
Has anybody found a way to make real use of this sensor? I would like to put it in a comercial unit, but I also get some false triggers with it. I am using the ESP12F. Does anyone know any other sensors with s similar form factor than are more reliable? |
|
Have had the same problems. Resolved them by doing the following:
I also tried the following at the suggestion of others for the ESP: |
|
@syedamerali |
|
Actually the software tweaking does not make the slightest difference. Last night I ran it without the software tweaks but the capacitor in place - no false triggers. Therefore I would assume that there is very little significance to the statement that the 2.4 GHz of the WiFi interferes with the RCWL microwave frequency. |
|
@syedamerali |
|
images are attached. (I think) I have also ordered a PCB with a ground plane let us see how that goes. |
|
Thanks ! |
|
Some more pictures:
|
|
To answer your questions:
|
Here we go ... indeed, that's certainly what I should start with ... that is to get the P/S OUT OF the assembly !
|
|
As you say...1000 words, it certainly helps! The 'active' portion is pointing up, which is good. I am concerned about the mass of metal below the cat detector, power supply, cap and transformer; how about rotating the PS, cut off the central plastic web and get the PS well away. As the OUT is digital 3.3 v high, I wonder why you need the pull-down resistor? I found that the OUT is current limited and you can hang ANY LED without a dropping resistor, which may be useful. I would scope the DC supply and see if there is much ripple. A low value cap may be needed on the RCWL power in. Also the mass of metal on the big cap is so close to the circular capacitor trace, may be best to put it outside the active area, if at all. Great photos! The concentric circles form the C and the serpentine trace on the top is the inductor L, forming an LC tank circuit oscillator. This is the primary oscillator and all metal should be kept well away. https://en.wikipedia.org/wiki/LC_circuit I assume that the ESP is 3.3 volt compatible so I question the need for the logic level converter? It may be best to twist all power wires together, even though they are short. The RCWL has the 3V3 voltage regulator output so it might be tried to power it with 5 v and use the 3V3 OUTPUT to power other circuitry under 100 mA. Another thought is to power the whole thing with a mini or micro USB power supply and get rid of the cheap AC PS, remember you get what you pay for. :D Sorry if I TL;DR, I often (always) view the images and comment and now see others have concerns over the metal mass of the PS. |
|
This is my Cat Detector, a Raspberry Pi Zero W (for WiFI) sitting in a box near my front door, sees my neighbour and even me in my kitchen 2 rooms away grr! Any activity outside my door, cleaners, post (mail), visitors, neighbours cat; gets logged to a time and date file and it also tweets me and emails. Node-RED comes with the latest Rasbian. pi@pi0-whitebox:~ $ ls -l -t * pi@pi0-whitebox:~ $ cat whiteboxNov10 |
|
|
|
I've made a PCB design for the RCWL, ESP-01, step-up converter and some other stuff to use the radar sensor. It's my first PCB. Would anyone with some electrical knowledge like to look at my PCB design? If it's any good, we can maybe add it to the repo. |
|
It would seem that the easiest solution to the false triggering problem is to set up 2 RCWL-0516 aligned in the same direction and AND their outputs. A HIGH output from the AND indicates a real event. Run the AND output into a power MOSFET to run your device. |
|
And to add to the frustration no two RCWL's are alike. I used bastero's circuit given on Aug 18, 2019. Life was wonderful. Everything else remaining the same I plugged in a different RCWL and bang, false triggers all over again. Switched back to the original RCWL no false triggers. |
|
RCWLs "see" 360 degrees. That's not good for putting 2 in parallel nearby. In my setup, I put a strip of metal flashing between the 2 RCWLs. This seems to block any interaction between the 2. |
|
I have a Arduino Uno based project that uses a 433MHz receiver to get data from an Acurite weather station. I use an IR remote to do various functions and that all works fine. I wanted to use the RCWL to sense if someone was nearby and turn on the display and if no movement was detected after 10 minutes to turn off the display. I had the same problem with false triggers that you all have had. I tried adding 100uF and 100nF capacitors between Vcc and Gnd and also a ferrite bead on the +5V, and soldered a resistor to the R-GN pads which helped to some extent. But the suggestion of @jbeale1 above to use the pin 12 op-amp output saved the day. I soldered a wire from pin 12 of the IC and feed that into an Arduino anolog input. Now I can set the threshold anyway I want. You can see the motion waveform using the Arduino IDE serial plotter and use any kind of signal processing algorithm for your application. What worked for me was to average 8 analog readings compare that to a threshold (400 in my case) wait 200 milliseconds and read the analog pin again the same way and if the second reading is also above the threshold, then true motion has occurred. Works great with zero false alarms. |
|
@BernardRat What if you do not use Capacitors you mentioned ? what is the resistor for in the image ? Please provide your code if possible? Thanks. |
|
I'm not sure if I still need the capacitors but since they could only help I left them in place, you can try it without and let us know if it makes any difference. The resistor soldered to the RCWL board is around 1Meg Ohm and reduces the sensitivity to about five meters (didn't have an SMD resistor). It detected movement in an adjacent room so I wanted to limit the range, but that's not a false trigger. The code is trivial, here are some snippets `#define motionAnalogPin A0 // RCWL-0516 analog output pin // function to read and lightly smooth analog data // check for motion |
|
Thanks. |
|
Thank you. This is excellent input for my project that suffers unbearably
from false triggers.
Thanks for sharing it.
Rajaraman
Coimbatore,
India
…On Sun, Apr 26, 2020 at 3:09 PM BernardRat ***@***.***> wrote:
I have a Arduino Uno based project that uses a 433MHz receiver to get data
from an Acurite weather station. I use an IR remote to do various functions
and that all works fine. I wanted to use the RCWL to sense if someone was
nearby and turn on the display and if no movement was detected after 10
minutes to turn off the display. I had the same problem with false triggers
that you all have had. I tried adding 100uF and 100nF capacitors between
Vcc and Gnd and also a ferrite bead on the +5V, and soldered a resistor to
the R-GN pads which helped to some extent. But the suggestion of @jbeale1
<https://github.com/jbeale1> above to use the pin 12 op-amp output saved
the day. I soldered a wire from pin 12 of the IC and feed that into an
Arduino anolog input. Now I can set the threshold anyway I want. You can
see the motion waveform using the Arduino IDE serial plotter and use any
kind of signal processing algorithm for your application. What worked for
me was to average 8 analog readings compare that to a threshold (400 in my
case) wait 200 milliseconds and read the analog pin again the same way and
if the second reading is also above the threshold, then true motion has
occurred. Works great with zero false alarms.
[image: Image00002]
<https://user-images.githubusercontent.com/64352987/80303732-4f256280-87b2-11ea-8aff-8d339cb73b58.jpg>
[image: Image00003]
<https://user-images.githubusercontent.com/64352987/80303737-58163400-87b2-11ea-89de-3d097b89df29.jpg>
[image: Image00001]
<https://user-images.githubusercontent.com/64352987/80303741-5e0c1500-87b2-11ea-8250-8ff3b70a0f90.jpg>
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#2 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABXAAHYJCPOND4NEAINQHHLROP6MJANCNFSM4DHRNINA>
.
|
|
Hi, @BernardRat @bastero @nextgenman any updates in your solutions? What is your experience in last months ? Thanks a lot. |
Hi ? |
|
Hello, I changed my code to compute the derivative of the analog motion output (dy/dt) which so far seems to have increased sensitivity with less false alarms. Here is the updated code: int readAnalogRCWL(){ The above function returns the time rate of change of the analog output (dy/dt). In my application if the function returns a value greater than 70, then true motion has been detected. You can use the serial plot in your application to help determine what values work best. Have fun. |
My final solution is PCB 1.2 board (my own disign) with mt3608 (B6286 SOT23 chip) soldered on it. In first design I used 11v output (R1 100k and R2 5.6kOm in datasheet mt3608) but final decision is to use R2 as 10kOm to have ~6.5V as output (input for rcwl). |
|
Thanks for the code.
Quick question: How often do you read the sensor ?
(I want to know if this will work if the processor is doing many other
things apart from reading the analog input.)
Regards,
Rajaraman
India
…On Fri, Aug 7, 2020 at 4:57 AM BernardRat ***@***.***> wrote:
Hello,
I changed my code to compute the derivative of the analog motion output
(dy/dt) which so far seems to have increased sensitivity with less false
alarms. Here is the updated code:
int readAnalogRCWL(){
long sum=0;
int i, t;
for(i=0; i<2; i++){
t=analogRead(motionAnalogPin);
delay(100);
t=abs(t-analogRead(motionAnalogPin));
sum+=t;
}
return sum/i;
}
The above function returns the time rate of change of the analog output
(dy/dt). In my application if the function returns a value greater than 70,
then true motion has been detected. You can use the serial plot in your
application to help determine what values work best. Have fun.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#2 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABXAAH6KCX65D7UTV64TE6DR7M35TANCNFSM4DHRNINA>
.
|
|
@nextgenman I would like to have 1 or 2 sensors for my garden. Are you selling and shipping your designed board? I would like to place it in a 3d printed housing. Is your board working with ESPeasy and can I power it with LiPo battery? |
|
any experiences with xyc-wb-dc sensors? I just replaced a rcwl-0516 showing false triggers even with vcc-filters. I could even connect the xyc-wb-dc to the wemos without filters and the result was surprisingly good, not showing any spurious peaks during whole night recording. The price of these sensors is very cheap as well. |
can you also test outside? |
seems to work outside too but didnt test influence of moving trees etc. |
Did you do anything special? I am using them in combination with a Sonoff module flashed with Tasmota, and some Wemos Modules also flashed with tasmota and I am operating them at 3.3v from the esp board. It seems to give a lot of false triggers, even moving it ~30cm away from the ESP still gives the triggers. Added a 100uF cap over the vdd/gnd on the sensor, but no joy so far.. |
Sorry to hear. The difference might be that I am operating the sensor from the 5V pin on the Wemos with 10 Ohms resistor in series and the 100 uF on the sensor side. The whole setup is powered using the usb plug on the Wemos. Software is ESPeasy. Still no problems here. |
|
Ok, will try the 10 ohm resistors series first. Than revert to 5v. |
|
no, the 10 Ohms are in the VCC lead between Wemos and the sensor, forming a RC-filter with the cap at the sensor side. |
|
So even with the 10R in between (i.e. the pi filter) I still see enormous amount of false triggers. |
|
Hi, hope it work for you |
|
Thanks will try that. Would really be nice to have a solution to put it all in the same enclosure... |
|
I use ceramic cap. Based on your environment, maybe you have to increase or decrease the capacitance. If you a have noisy power supply, take a look at the link below: |
|
Hi, I also had troubles with RCWL-0516’s false triggering, and tried a pi-filter (1 mF-10ohm-1mF), sole 1 mF capacitor across Vin and ground of RCWL-0516, 10k pull-down resistor between the output of RCWL and ground, and local DC power feed as close as possible to the RCWL, and different combinations of these, with no success. However, when directly sourcing the RCWLs with 12VDC from my home automation center’s good quality PC-power supply through few meters of unshielded KLM-cable or about ten meters of CAT6-cable, without any of the mentioned filters or local power supplies, there has been no false triggering anymore. Just rock solid detection, with all three RCWLs which I installed few days ago. Just wanted to share this, since I really wasted my time with all my unsuccessful experiments! |
|
Hi, just wanted to share my experience also. |
and others
I used this sensor with a esp8266 and i have a lot of false triggers. Seems to be strongly affect by the wireless network , in general , by strong electromagnetic fields . I tried to place him at a considerable distance from esp8266 and I used a lot of decoupling capacitors. Sure that the false switching operations were far fewer .
I would like to know if anyone met this problem and how to solve it .
If I mount a metalic shield on the back of the sensor, may this affect its detection properties ? Thanks .
The text was updated successfully, but these errors were encountered: