mmWave Wars: one sensor (module) to rule them all

I also want to power my sensor from smoke detectors but failed.
Could you show more details about how to do it?
I use a DC6-24V to 5V converter but it seems not working.

@cyben76 :Hi,Abdullah, can you please share link for english version again? Your first one on Wetransfer expired…
would love to have an english app…it’s still not available on their official site, as it seems.

There’s a link a few lines into this page if that is what you are looking for

Sadly that link guides only to chinese version of android app. Ok, one “solution” is to use windows app on BT enabled laptop, sincer Windows app is in english. But app would be easier…

Anyone who have setup LD1115H sensor has experience any false triggers??

Yes. Many. I stopped using it.

oh!? That is a shocker. I do have a LD2410 and that has been solid.

Yes, the LD2410 was better than the LD115H for me as well.

Distance settings should be a mandatory feature.

Do you have the STL for that small enclosure? Thank You

Update: my LD2420B arrived and after tinkering a bit with it I can share my opinion.

As it was previously said, it’s Bluetooth range sucks. Prepare to have it really close to a Bluetooth proxy in order to get a stable connection (I have it at 40 cm distance. I tried to place it at 3.2 m and while it connected, it lost the connection after a minute).

The test I performed was to turn a light on and off when I enter the room and when I exit respectively. I couldn’t use the motion or the occupancy sensor as they were fluctuating too often to be stable. What I used is the static energy sensor (it is disabled by default in home assistant) where I found that if you stay still it has a value about 10-20. When moving, this value increases up to a 100. Therefore, I built an automation that has the static energy as trigger and turns the light on when the value is above 90 for more than a second and turns the light off when the value is below 5 for 10 seconds.

This way, I don’t have false positives and since the light on trigger has such a high value, my cat is triggering the sensor but not the automation. Also, with my cat staying still in the room, I have seen that it goes down to value 2-3 so the light will turn off.

The most important thing is the placement. I have it a bit higher than the eyes level when sitting so it won’t “see” the floor where is more likely for pets to walk.

It takes some trial & error but eventually it can be used.

The biggest set backs was the Bluetooth range and the fact that requires very thoughtful placement. If you are willing to solder it on an ESP32 you will solve the BT range issue however it might still limit you a bit on the placement as it would depend where you have a plug.

Overall, for the 3 euros I paid it is ok :slight_smile:

Update 2: i realized that using the static energy was not so stable and resulted to false positivies turning the light off even when some one is in the room. A better way to do it is to use the static target distance which will tell you how far away from the sensor you are sitting. Stay at the door of the room and check the distance and then create an automation to turn the light off when the distance is higher than that.

3 Likes

This has been a fascinating journey and I am quite happy with what I was able to do. Thank you for your research and here is my implementation of the LD2410 BLE sensor.

Project:

  • WeMo D1 ESP development board for Bluetooth Proxy
  • LD2410 BLE from AliExpress
  • 110/210v to 5v converter from AliExpress

Problem:

  • PIR is not good enough for bathroom usage. Many times people are not moving enough to make the PIR detect motion.
  • Wanted to determine if #1 or #2 was potentially being conducted and in turn if the later turn on the exhaust fan if not already on
  • Once occupancy by the MMW was clear; run fan for an additional 5 minutes while turning off the lights immediately

Worked out quite well in my bathrooms. Pictures showing the device and placement. Works flawlessly. :slight_smile:

4 Likes

How far is the BT proxy from the sensor? Do you face any disconnects?

Each sensor is about 15-20’ away from a Bluetooth proxy. I haven’t seen any issues yet. I am now having BT disconnects due to the ESPHome BT Proxy. Trying to troubleshoot and solve like others now. Boo!

It was stable for a week or so, but eventually my BT adapter itself times out… :frowning:
Gave up with BT now, probably will solder it to the Wemos D1 mini inside my IKEA VINDRIKTNING.

I gave up on the LD2410B. I was trying for days do avoid false positives but I the lights kept turning off when I was staying still. I will throw them to the trash

1 Like

My experience has been that connecting these to Home Assistant via bluetooth will not give stable or reliable operation. Bluetooth is very useful for calibrating the output via the Android app, but any other use seems to just be a headache. Even using the company’s Android app on a Samsung tablet, the connection to the device isn’t stable for more than a few minutes - not even in the same room, less than a meter away.

That said, these are utterly fantastic when properly connected via a separate microcontroller. I have several LD2410s working very reliably for weeks now (with device uptime in the hundreds of hours) in order to detect presence. One is using an ESP32 that runs some other devices, and the others are connected via ESP-01s (ESP8266). Despite being connected via UART, bluetooth still works, so you can connect via the Android app to calibrate the sensitivity curves. This is a much better experience than using the Windows program or incrementally changing values in YAML before recompiling and uploading an ESPhome binary.

I used a USB-C breakout with power running to the LD2410 and to a 3.3v buck, which then sent power to the ESP-01. The LD2410 is then connected via UART to the ESP-01. Including a cheap plastic case sourced from Amazon and a 3 meter USB-C cable, the total BOM is under $9.

Thanks for your comment. My problem was not the disconnections via Bluetooth but rather the false negatives. I ended up being in the dark a bit too often :).

I was assuming that the sensor can detect the slightest movement such as breathing but based on my tests, if I would stay still (obviously breathing) the sensor will not detect me.

It’s static distance would turn to 0 cm and the static energy was 3. It’s the same value as if the room is not occupied. I tried for days to build automations to counteract this behavior either ending up increasing the sensitivity to high that it detected presence from other rooms, or i would reduce the sensitivity and delay turning the light off for a few minutes hoping that it will detect another relatively large movement. What is the difference then from a PIR sensor if I keep delaying the automation action?

Maybe my problem was the correct calibration I don’t know, but overall I didn’t find it reliable.

The Android app is practically required to calibrate it - it gives you a realtime visualization of the distance and energy level so that you can tune your thresholds at every gate (each gate being 0.2 or 0.75m depending on the setting). I’ve been using one in my bedroom and it detects me while I’m sleeping. It even detects my cat sleeping in the room. Unless I’m holding my breath, it’s very hard to get a false negative.

It’s really a completely different experience once you calibrate it inside the app (which gives far more detail than the HA integration).

If you used default module configuration it will be like this. You have to change it’s sensitivity and threshold for each of the range distance “sectors” to adjust it for your particular room. I have one in toilet as well for 2 months and finally after many years of PIR sensor whole family is happy as the lights never goes off when you sit longer. Only if you try to be still and hold your breath for more then 10 seconds eventually lights goes off and immediately when start breathing it goes on.
So from my prospective this sensor is super perfect.

2 Likes

I have 3 of them. The only one I have managed to set up properly never missus a beat. The other 2 are 90% reliable but just need another tweak or 2 before they are perfect. Once the sensitivity is set they are great little sensors.