@jkandasa

Brilliant - thank you!!!

@electrik I can try that, but I have 6xpi with small fan and no problems at all. There are capacitors to help with any issues, but I will power down before bed tonight and remove fan connections and disable the python script and see how it goes....As you can see, I really have given it all a lot of thought and effort and that is why this mystery is bugging me!

@tekka Thanks for the reply - there was nothing at all in the mycontroller server log at the time the data simply stopped flowing - all nodes are affecte and all stopped at the same time. So I am guessing it is a GW problem, but I can't monitor that while it is plugged into the pi, can I?

Cheers!

Hello,

I have been having a recurring problem where the GW simply does not seem to send data to the controller.

I have tried 3 different pro mini (3,3V 8MHz) as the GW (this serially connected to the pi with the mysensors 2.3.0 gateway sketch on), replaced all cables, built several 3.3V DC regulator boards (662k and ams1117), replaced the pi3, replaced the radio modules (all have caps fitted), replaced the SD card three times with the last 2 being brand new cards.

The other symptom I get is that the node ID is occasionally corrupted in the controller logs. It just contains junk characters (sometimes quite a few of them!). Again maybe part of the problem or another issue altogether.

Example is last night. At 21.30 GW was sending data from all nodes to controller, at 22:50 I get 1 'junk' node id and then at 23:45 no more data from any nodes for 10 hours. The junk node ID characters appear on the server logs of mycontroller.

It happens randomly, sometimes a reboot of the pi will cure it, other times I need to power off the pi completely and re-start before the data returns.

The only other thing running on the pi is a small python script to control fan speed via gpio pwm pin. It's only a small dc fan to keep the pi cool.

Has anyone any suggestions as to what to try next? It's really getting to me now!

Thanks.

Oh, I see - By 'THIS' I mean this 'build' page and the build example herein.

@mfalkvidd

Certainly - the sensor works by bouncing a IR led light of any particles in the tunnel in the device and measuring the reflected light scattered back to the sensor .
This 'tutorial' does not even have the connections for the led connected(!), so will never work as the manufacturer intended.

Today I got the sensor in the post and looked at the datasheet and application notes - I now have a working sensor. A working sensor requires 6 pins connected (3 won't do) and a resistor and capacitor. When you follow the manufacturers data it works for a general indication of dust/particle level.

I have 2 working sketches both tested with mysensors, a basic one and also one that takes 10 samples over 10 seconds and then averages them into a single value to send to the gateway.

I can post the codes if you like, but the wiring of the application notes must be used and the RC filter must be in the circuit.

@Nca78 That makes perfect sense! - Thanks for the clarification.....Onwards and upwards!

I see from the API that wait() is preferred over delay(). But what about delayMicroseconds? Is it OK to use 'delayMicroseconds' for very short durations or is there a 'waitMicroseconds' function that is supposed to be used?

How was this tested? - The datasheet implies that you need to connect the IR LED and provide it with a pulsed wave. Does it really work like this without the led powered? The datasheet would seem to perclude this from ever working...

@raptorjr said in Communication problem or?:

A 47uF tantalum capacitor worked.

That is good news!

The node started as a prototype with the plan to make it properly when I saw it >was working. But it is more fun to do something new, at least when the node is >working.

I know all about how that one goes, at least it is not just me!