I’m not sure that new mmwave_sensor switch is working as intended, looks like the sleep isn’t blocking the other commands but is executed in parallel. I’ve seen settings change fail a couple of times (with the error saying the sensor isn’t stopped) with that latest code, whereas having the sensorStop / start in each switch was working reliably. Maybe there’s a setting to make sure these are executed sequentially ?
After reading this topic I really have to wonder, hmm, well, basically how practical all this really is just yet.
I mean this technology is fascinating, don’t get me wrong. I wish I had more time to play around with this myself. But when you read about all the false detection problems, the insanely high sensitivity for moving inanimate objects like fans or curtains, the need for aluminum foil shielding contraptions, need to complement with additional PIRs, etc etc. That all sounds so fragile and ready to break as soon as something changes in your rooms layout.
But if putting aside all the technical novelty and tinkering aspects, I really wonder if a simple high sensitivity PIR, like the ultra slight motion types from Panasonic together with a slightly longer timeout wouldn’t be able to achieve exactly the same results for almost all real life scenarios, with much less tinkering and much lower power use (battery powered would work without issues) for all practical purposes. Dunno. I spent a lot of time with the (admittedly technologically much more primitive) RCWL 0516 setups, only to ditch it all in the end in favor of high grade PIRs, because it just wasn’t practical.
I never got there with just a PIR. Combining them is a whole different thread though. I have zero issues with inanimate objects in the combo-build (which probably has old DFRobot yaml). I did end up trying the highest ah EKMC PaPIR and it does indeed work the best of all the EKM’s; but still has nothing on “good mmwave” (aka not RCWL/SeeedStudio) for human micro-kinesis.
Thanks for the feedback. Your right, I think I’ve heard that delay doesn’t stop other tasks. I’ll revert re-add the delay for reliability.
[edit] - kept the switch; just re-added the delay in each action section.
1 Like
Hrmm, I did Google but none of those things 
There are actually clues all over this thread!
More to come…
As for sensitivity. I have changed it, but only have a day of testing and haven’t performed any direct comparisons. Too many other projects!
1 Like
To each their own. I for one have had a DFRobot sensor that I depend on daily for my alarm to detect if I’m in bed. Sitting below my bed facing up. No silly weight sensors needed! I filter out false positives for accuracy levels that are this good: (say today, a rock solid 17 hour chunk that the sensor was off)
1 Like
And if anyone wants to be the Guinea Pig:
60G Millimeter Wave Radar non-contact Human Perception and motion tracking Radar, IR60TR1A
https://www.good-display.com/product/428.html
For anyone interested, from what I can find, the DFRobot mmWave SEN0395 will have more stock coming to a couple places in September. Stock could be gone early from pre-orders.
Bed presence is a good point, I didn’t think of that. Looks like this sensor is ideal for that !
And yeah, to each their own I guess. It’s mostly that I started questioning the practicability of many HA concepts for myself, including fine grained room presence. But if it works for you, great !
1 Like
I’ve got one on the wall facing at the bed and the off is working well, but the on isn’t great. I’ll have to try under the bed (which is much harder to access sadly) to see if it’s better, but I’m not that conviced by it for beds so far.
Even if it was working well it’s still only good for someone / noone, it won’t tell you if two people are there so it’s hard to use for things like alarms if you’re not living alone, the weight sensor as annoying as it is to setup does give you that information.
The mmwave is great for rooms like hallways and bathrooms, where the number of people doesn’t matter imho
I find the mmwave sensor to be more useful in rooms were you are expected to stay idle for long time.
In hallways, kitchens, bathrooms, people are always moving so a PIR and a reasonable timeout (<5 min for me) work great and is cheaper than a mmWave sensor.
I’d say (having used mine for a grand total of 2 days !) that they are more useful in an office or livingroom for example. There you might sit for long periods so PIRs are not satisfactory.
I’ll take it over a PIR anywhere simply because of the very long timeouts they have, at least the ones I’ve tried. The mmwave shuts off real quick, which is nice. I just wish we had some way of making Zigbee devices with ESPHome, I’d bind those sensors directly to the lights, it’d be perfect. Some day maybe
So you folks leave the lights on for minutes even if no one is there?
Old Aqara motion drives me mad with that minute of cool down!
Professional PIRs, like the Panasonic ones, don’t have any timeout at all, they give you raw instant detection data. It’s up to you to process it and add a timeout in the firmware if you need one.
My light timeout, if there even is one, is set between 30 minutes (toilets) and 2 hours (living room) without movement. It’s all LEDs, the power consumption is insignificant, especially compared to heating (we have electric heating). YMMV of course.
Yeah, I heard of the LEDs low consumption, but it just doesn’t feel right to me to have the lights on for so long, for that I rather just use the switches LOL
I was checking yesterday the Panasonic PIR and after a quick look, I saw the cheapest one was around 15€, If you add the dev board, plus whatever you use for the battery plus the time to build it, it’ll be pretty much the same as the ones they sell ready to use, right?
Have a look at the OPs other thread about mmwave, he added a Panasonic PIR to it.
Price wise a DIY build is probably close to an off the shelf unit or even more expensive. But the quality of the PIRs is just a completely different world.
I’ll check it, but having a PIR wired isn’t an option for me unfortunately 
What if I told you this sensor is available in ZigBee format in a round ceiling-mount can? The problem is; China-only and not the usual sites.
Yeah the aqara one ? Everything I’ve tried from that brand has been a huge pain, it doesn’t play nice with proper Zigbee networks so it just ends up being e-waste.
Plus they have a long timeout on it anyway so it doesn’t look that much better than a regular cheap PIR
Leapmmw. They seem to be an ODM so direct sales are challenging. I’d love to try some of their other offerings.