I changed the radio to be powered by the external power supply (it can power both 3.3V and 5V) - and it's much more reliable until eventually it stops - I think the power supply struggles supplying both the servo and the radio.

I'm going to try hooking up another Nano tonight to purely provide the 3.3V to the radio, see if that improves things.

But I was wondering what the best power supplies people use to power this stuff? I do seem to have a lot of issues with reliable power. Also, I still would like to implement an ACK feature - I could get the sensor to send a state update on MQTT when the state changes I suppose, but would be nice if the ACK's were built in.

Hi Foto,

Thanks for the response, but I already had that code as I forked your code. I don't think the message is getting to the client full stop, the lights on the node don't flash when the node doesn't respond. It's almost as though I need the gateway to insist on getting an ack from the node, and it it doesn't get one, keep sending it till it does.

The actual radio issue though strikes me as a power issue but, both the gateway and the node use 100uf capacitors which should provide enough reserve power to ensure the radios are fed well?

After testing. It seems like it's happening more often than not. I might have to send the same mqtt instruction 4 times before the sensor moves the servo.

I unfortunately can't debug the gateway as I have to compile it with debug off for the mqtt client gateway to fit on my uno.

Hi Folks,

I came across an issue where sometimes a MQTT message sent to an actuator using the MQTT Client Gateway fails (http://forum.mysensors.org/topic/524/mqtt-client-gateway/32) which means that OpenHAB think's the actuator is in one state, when in actual fact it's in another. In my case it's a Servo connected to a Thermostat to control my heating - so OpenHAB sends a message to turn it off, but it didn't get through, and it doesn't seem like it's retried - so the heating stays on - obviously this is a problem!

Is there a way to resend the data if it didn't go through correctly?

I'm using the MQTT Client Gateway with OpenHab and Mosquotto Sub (when testing) as the controller

I think I might have fixed this now, I think the capacitor may have been connected a bit loose.

However something I'm slightly concerned by, and I've noticed this with 5v and 3.3v - is occassionally the message won't come through. Now that's not a massive problem when you're dealing with temp sensors, but when this is a message that controls heating and could mean that it's not switched off as the message didn't come through, it's a bit more worrying. Is there a retry method of some kind?

Hi Folks,

Tried with a different radio, on 3.3v - same result - I see it appear in my MQTT broker speaking to the network - announcing it's presence - but sending it messages results in nothing happening.

Hi Folks,

I've had some troubles with the radio and ACKs not being sent properly - as a background.

Anyway today I set up an acuator - a Servo which controls my heating. It all works fine (although the Servo is a bit unpredicable when starting up, even with an external power supply and a capacitor - but I digress) - but I just suddenly had a realisation that I hooked up the radio to 5V not 3.3V - so I went to fix my mistake - however when running off 3.3V, I found that the radio would send it's hello messages, I saw them on MQTT - but it wouldn't do anything when sending it a MQTT message - switching back to 5V and it all works fine.

I'm using a capacitor, a 100uF one.

My questions really are, is this safe leaving it on 5V? Is the worse that could happen that the radio just stops working (ie is it a fire hazard?) - Also, is it just possible that my batch just prefers 5V? It just seems odd! The radio is the only thing running off the arduino nano power supply - the servo is on it's own.

Any ideas?

Right, I've fixed this, temporarily anyway.

So I hooked up a spare uno and hooked it's 3v up to the gateway radio, making sure to link all the grounds up. Still not working. So I hooked up an uno to my nano sensors radio and then it all worked perfectly. Now obviously this isn't a permanent solution as I don't want to have to use two power sources for all my sensors, but I have capacitors on order so I'll try with them next week to see if that fixes the problem. But it's definitely a power problem...

Sorry, what did you do for everything to work fine?