@terence-faul said in Interupt not waking sensor:

when the Interupt is called does that invoke the loop from the beginning?```

No, it resumes. But in your case, as sleep is your last function in Loop(), loop is relaunched.

I just quick looked at your code (no time).
Have you tried to remove others sensors code, and run a reedswitch only code.
It should be the first step in your debugging. Better do step by step to check what's the culprit.
If your reedswitch code is working ok, if there is any interference with other sensors/libraries, the logic order/algo etc.

Are you using sleep() for saving energy (so a batt powered device)?? then imho I would have choosen a better sensor. I don't get why people still uses these dht..
Eg. cheap si7021, Bosch BMEs : better power consumption, voltage range.

DHT libs has lot of delays, yuk!
eg. minimum 10ms just for a read start signal (more for the complete reading), imho it needs refactoring to be used in non blocking way.

So, you can:

• disable your DHT code and try reedswitch code only.
• if it works. then you have a problem with 1) your code logics 2) external lib (but imho it's a bad sensor choice)
• Check 1) by maybe changing a bit your code logics(order of functions calls like testing your reedswith right after your sleep call).
• if it doesn't fix, rewrite sensor lib in non blocking way to be used with your sketch
• if you don't feel enough good for rewriting lib, then change sensor brand/model, and you may fix this problem

As a rule of thumb, libs are not all well optimized even if their examples works, but sometimes it needs tinkering so it can work in larger projects like a mysensors multisensors sketch.
So before using a lib, I often check&fix if there are any delays/loops/blocking stuff and others details which could bring some problems and slow down my dev.

Edit: ahah I didn't notice, you beat me @mfalkvidd

@alowhum
you're confusing things a bit.
to do this, microbit board use an additional mcu with a special fw. Which means,

• option 1) the 2nd mcu includes jlink reprogramming tool and targets nrf swd pins, in that case, no bootloader needed in the nrf
• option 2) the 2nd mcu reprograms the nrf by serial, then there is a bootloader in both mcus.

but the two ic, when blank, doesn't know how to do usb, usb-cdc, serial etc, even if one has builtin support, it needs a first programming tool. they only know special reprogramming protocols..nothing new, like I said above, all mcus work the same, so read&experiment to better get the picture and it can be useful to know, in case you want to teach your customers/users, or they ask you, how to change/update for another bootloader, erase the mcu etc. what would you say, it is too advanced?? not true! you don't know what you don't know.. preconceived ideas on programming tools (limiting yourself and future answers to your users too)

I think this is off topic, and maybe polluting Sancho's project. Sorry Sancho.

you can get compiled fw from arduino, then upload it if a dedicatedf usb/dfu bootloader is there.
But, you still need a jlink when mcu is blank, bootloader issue, to erase it, or when you want to update bootloader for new feature.
Whereas for an avr8 you would need avrspi, for nrf5, samd and lot of ARM mcus (if not all) you need an swd programmer. I think it's been mentioned multiple times on forum. Better source some hw and try, to better get processes.

Still I think it's good habit to have minimal tools, and know how to upload to the regular way mcu, else when it fails you're stuck without tool (eg. I already saw issue with a samd usb bootloader and it needed reprog the bootloader because no more usb boot..). I'm not saying your customers should become tech guys, but there is a minimum to learn, especially so many points in diy HA to pay attention.
Same to you, if no more boot, then how would you know if it's hw or bootloader, without being able to restore bootloader?
This also applies to avr8, like arduino nano/mini.., how do you upload Mysbootloader/dualoptiboot for FOTA, or update the crappy bootloader and its wdt from cheap clones??).

you can copy compiled fw directly to usb. But from arduino, I think it's swd/jlink only, if i remember.
with this dongle, I don't think you'll get a huge range, less range as regular 20db PA boosted nrf24, it might be directive too (orientation may matter).
Cool for dev and experiments, but for production? depends how such tiny thing and antenna is plugged and hidden.. I would prefer a lot an external antenna.

@mfalkvidd
well, I think it should also apply to other rf24 etc too. not the csma itself as the topic suggest, sure, but better knowing or feature for adapting to environment.

I coded something for my gw, when creating a new network, :

• it scan/sample a range of freq/channels for their energy levels. Make a list of best low energy channels
• scan selected channels for existing networks. Sort your list of favorites by number of existing networks, less the better of course. Store it for future freq hopping, antijam etc
• Pick best channel and create new networkd id with customized settings (csma threshold etc) based on datas you got

To run automagically, it needs commissionning nodes and some changes (not realease-able yet..), as first time they are powered they don't know the channel, network...
Process could be done manually of course (running a dedicated sketch, then set defines etc, just less friendly still simple process)

about this topic, I agree that maybe define description is missing at some place, and why not changing the default value to something more reasonable, might do the job

@mfalkvidd said in CSMA / LBT for RFM69 new driver:

Is the default limit if -95 good, or should we change it?
Would it be appropriate to dynamically change the limit?
any other ideas?

I don't remember exactly why it's -95.. I think first times it was on -90. Even if we refactored the rfm69 code a lot, we kept some settings from lowpowerlab lib (-90) for compatibility with already existing environments, and because it worked during tests, it can depend on some variables

That said, for others knowledge, as you already know, to be on safe side you can change this threshold with this define:

#define MY_RFM69_CSMA_LIMIT_DBM (-95)

I think it's a good practice, when building a new diy network, no matter the radio you're using, to analyze/learn your environment else there might be some surprise

If your environment is quiet, fixed csma might be enough once you get your threshold right.
Dynamic, yes why not. can get tricky depending on the usecase in the network, and I think it's more than a dynamic csma that need to be implemented, to improve packet delivery reliability and security (at network creation, and for longterm) like some commercial stack do too.
eg. neighboorhood and overlapping networks, overlapping frequencies when using different protocols in same bands (like mysensors nrf24, zigbee, ble, thread, wifi.., hopefully subghz is often less crowded), jamming etc.

Hi.

Do you use MySensors master branch??
If so, try to use the development branch, it should work.

Side note:

• you don't need to add spi.h
• I'm not sure if you can use esp as a node in MySensors. As nodes, if I'm not wrong, people use esp as gateways+sensors connected (when you need), but not as nodes.
  I never tried esp as mysensors nodes; in that case I usually use direct mqtt or http api. I get it why using them as simple mcus for nodes, but I prefer better lowpower mcus for nodes.. so I can't tell you more on this point.
• for a simple esp32 gw sketch you can take a look in your Arduino IDE/Examples/MySensors, and of course in MyConfig.h for others defines.

I hope this helps.

@alowhum
It might be just a feeling because I still see many people on forums ordering zwave because they get some troubles and don't like range/reliability of 2.4ghz zigbee..
Do you think zigbee will beat wired HA ? I'm kidding of course

subghz is still very efficient, and reliable (better penetration etc)! vs a 2.4ghz crowded indoor band. BLE5 etc are nice sure, then it depends on your environment, rf performance of your device, because BLE5 doesn't mean bigger range if antenna is too small and inefficient, and not everyone live in aprtment too, so the subghz solutions.. Each solutions have their set of pros&cons.

An ultimate "thingie" is very subjective, as some people could say they don't agree with you about a rpi0 or better rpi3, because for them the ultimate ha, controller or whatever, should run virtualized with others services on a low tdp computer machine, nuc-like for example.
Some could also say they need to be able to connect to multiple brand of devices, protocols, or wired KNX etc.. when you build a fresh new house for example, it would be pity to use radio.

Another point you forgot in your list, the storage.. sdcard or hdd/ssd for an ultimate controller device? not same end price but not same reliability.

I agree I like a lot Snips too. But Snips needs one rpi3 as server at least.
Rpi0 isnt enough powerful for that, it can be used only as a snips satellite.
If rpi0 and cheap are the keywords, then it's limited to simple voice commands without AI (nlu etc).. not what I would call ultimate controller
And regarding the microphone, the simple rpi 2mics hat is cheap&nice for starting but it has no true DSP for audio treatments, so, not as good as the more expensive alternatives like the respeaker core2 .

About user friendly, and all the opensource stuff vs commercial solutions, I don't know anyone, who wouldn't be a geek or computer guy, to invest lot of time, writing notes etc to learn how to run and handle a Linux system.. and I'm not talking about all the security stuff like what's ssh, a fail2ban etc.. I know a lot of smart people in their work, some have little spare time, or not, but all look at me with wide opened eyes when I use command line, and they say they don't want to learn that for running their house, and there'll always be a moment when you'll need to use cli..

Very subjective

Hi,

weird that you got problems with nextion.
I think you have something wrong in the gui creator tool, sdcard, or code, else it should work.
I have two here, and no problem. I'm not using them for HA, not a fan of "diy" displays, with small dpi, for home..But I always thought they are noob friendly (serial, easy to use api and gui creator tool, saving some coding time), so I'm sure you'll get it working.

Hello,

never tried this way, but I don't see why it wouldn't work. it's software spi..
Maybe show us your sketch, and add this define, so you can check in logs which pins are really used by the radio driver

#define MY_DEBUG_VERBOSE_RF24

imho, hardware spi is always preferable, so I would have fixed it the other way, like cutting traces and soldering thin wires to get it working again