Development

Adafruit have a library for the htu21d which works well with Mysensors

Are you using a wifi gateway? Others reported strange values too: http://forum.mysensors.org/topic/2934/wifi-gateway-motion-sensor-value-1107296256-instead-of-tripped/2

@davboecki :+1:

Always hard to say - but seeing issues with relay quite often my advice is to look over the power/ground and radio. The radio is very power sensitive and it can help adding/changing caps GND/VCC on the radio. How you wire the ground from the relay can also make a difference. I have had relays jamming when i ran ground trough the arduino and not straight to source. Also the opposite that it didnt work well if I didnt have ground through the arduino. Range, when you send a on-command there is a ack going back which needs to be ok. Follow your serial logs in node and gateway to find out if the radio traffic is ok - or put a repeater in between. Power - I have had Arduinos (mostly Pro Mini) that was a little to weak to trigger the relay... :( It needs 5v but my pin 5 was only giving 4.7V and some tries resulted in a unsuccessful relay switch even if the logs told me so. To test this you can upload a sketch doing nothing else but setting the pin to high and low with a delay in between.

https://github.com/mysensors/Raspberry

Use: char *hextstring = message.getString(); Even though it's called getString we actually not use the String class (which isn't recommended to use on these small embedded systems)

@Homer I've made a 2 double LED dimmer as well works very good and can easily be extended with more dimable leds. See my project here

Indeed, I use node-red for this. It also has nodes for MQTT but I have not looked into MQTT at all. My setup is a single rf24 based gw which through node-red interfaces with Vera, domoticz, myscontroller and on occasion, mycontroller. It can also push notifications to pushbullet, mail or other endpoints (through add on nodes in node-red). From a controller perspective, node-red uses serial nodes, which in turn become tcp ports through socat that the controllers connect to believing they interface with an ethernet gw.

Found this issue, looks like it NEED TO BE PRIOR to the #include <MySensor.h> #define MY_NODE_ID 7 // id of the node //#define MY_PARENT_NODE_ID 1 #define CHILD_ID 3 // Id of the sensor child #include <SPI.h> #include <MySensor.h>

Well I tested again not defining a radio for a sensor node (relayactuator) and it didn't compile. OK But my issue was with GatewayW5100, if I didn't define the NRF radio, nothing work at all and nodes are not communicating with the GW. In all cases, I just needed to highlight that this may save someone's else time :D

Hmm.. Looks like it could it be that the connection to controller hasn't been initialised when node/gw sends the presentations. Please create an issue on github and create a link to it here.

Do you have something in your eeprom already blocking? Try clearing it using the provided ClearEprom sketch in examples.

@elmezie after a sleep the radio is not powered on again (sleep is for battery power nodes which normally do no expect message to receive) One can wake up the radio again by sending a message e.g. requesting a status or perhaps the current time with gw.requestTime(receiveTime);

@barduino @flopp sorry guys i fixed this last week but forgot to tell anyone. it shou;d be on my github

Yes, it would be possible. But you won't save much juice as keeping radio in listening mode still requires way too much power to be able to run on batteries for any substantial time.

Same problem here, very randomly only blink once when press reset button on node (or upload sketch) And something else... what about : #define MY_RANDOMSEED_PIN 7 // A7 - Pin used for random generation (do not connect anything to this) Must be pin 21 for A7 ?

Got the same problem. Need to change the code to make it works

@rmtucker I also blew the 5V converter on a cheap Chinese UNO clone, by just connecting a power supply. Still wondering why.... The AMS1117 often found on these boards only allow 15V input max. I blew another one by connecting a cheap AC adapter which outputs a higher voltage unloaded.

Some optimization to check: ringerdelay needs to be defined with a type if condition in setup() not necessary initialize buttonpushed and ringSendAt unconditioned normalize logic in if (timenu > buttonpushed && timenu > buttonpushed + ringerdelay) to if (timenu > buttonpushed + ringerdelay), it is the same variable timenu not necessary use millis() directly: if (millis() > buttonpushed + ringerdelay) on relay initialisation the two comments are wrong