Hardware

@LastSamurai said: Perhaps this helps. As far as I know and have read 1Mhz doesn't really change that much because you sleep most of the time anyways. Try to disable the BOD, that uses most of the power. I don't think the idea behind changing the clock to 1Mhz is to save energy, but to allow the mcu to run at lower voltages (allowing it to keep running when the batteries are already considered dead for other uses). The Atmega328 can run down to 1.8V but only if your clock speed is 4Mhz or lower. We change it to 1Mhz because its easier to do that in software, rather than solder a new crystal.

@m26872 said: I've low power hacked APM and bare bone 328p. As I think I've linked from here, at 1MHz (internal) 9600 baud was the maximum and each baud rate required their own bootloader version. To do it easy I just used precompiled Optiboot for this. Thanks you, thats a very interesting read! I will try some of that myself.

Depending on your sensor setup they should be "sleeping" to conserve battery (if using battery). The sensors will only send messages when ordered to do so in code. The controller might send out ack signals depending on your setup but it is mostly in receive mode. On a side note if you are worried about RF interference it goes down to what frequency and what channel you are on as well as your settings for power. Also any other emitting device in your house could interfere. If you are worried about health issues you shouldn't be. You get orders of magnitude more EMF radiation from the sun than you could get from these devices.

If you are going to let the node go into sleep between sensor readings (f.ex. Wakeup every 5 minutes or so) you don't need the troubles of switching to 1Mhz. When you sleep the atmega, you also turn off the oscillator for the MCU. So it doesn't use any power. In fact one could argue that it will result in a longer wake time if you are running at 1Mhz instead of 8Mhz, due to the slower execution speed. And so it could end up using the same amount of power, as when running on 8mhz. This is however untested/unverified, and just a theory that I have.. :)

@mvader - If you can get some cheap Android phones/tablets, you can use the VoxWav app on it and mount this devices in each room. The VoxWav app act as a wireless microphone and remote for VoxCommando http://voxcommando.com/mediawiki/index.php?title=VoxWav here is a nice demo video: www.youtube.com/watch?v=0RCZ7aGGaro I use this app also on a android smartwatch to control things which you can see here: www.youtube.com/watch?v=TMPstHDvnM8 I hope this give you an idea ;-)

I'm thinking about a similar project. One thing I'm struggling with is how to detect the current relay status. You will not notice when another push button is used, so your MySensors device in your controller will only switch between on and off without any meaning to the physical relay on/off. Maybe it is possible to use a mosfet with the 24v relay connected as gate to sense a switch-event triggered by any push button to update the status of your light. Just an idea at the moment, maybe someone with more knowledge on electronics can form a curcuit out of that.

Microview? https://www.sparkfun.com/products/12923

Ok, the heat issue was new to me regarding the ssr's. I already have one SSR-40DA and its quite big as it is without a heatsink. I think I wil go with the standar relays instead. But What kind of fuses should I get to make it as sefe as possible. A Thermal fuse Do I need some more fuse? if so a ebay/aliexpress link would be great.

@Cliff-Karlsson : maybe it would need to make a childboard and connect it to MYSXconnector 1.5 but sorry, I am not planning to make a HLK childboard for the moment.

Problem solved. For those interested. The issue was the 3.3V Step up board. When the radio turns on there is a large power draw which causes the voltage on the 3.3V rail to drop. This caused the temperature reading fluctuations I was seeing. I added a large cap between the Vout / GND of the step up. Problem solved.

Hi, or use a small bulb instead neon...should also work. But a cap in parallel is always a good idea for this.

@NotYetRated Here's a thread with some great options: http://forum.mysensors.org/topic/1789/external-connector-options. Cheers Al

yeah those seem pretty pricey.. pro mini is pretty small.. so are the sensebender boards.. for much cheaper

@Dwalt said: I have not had problems with enclosures per se, but I have seen temperature differentials due to locations exposed to direct sunlight and up against thermal masses such as a external walls or internal walls with enclosed heating ducts or hot water piping. I have four sensebenders in vented enclosures mounted on interior walls and have no problem with their accuracy. They are also loosely mounted inside their enclosures which may help them avoiding picking up heat from the walls. Overall, they seem much more consistent and accurate than the several DHT or Dallas sensors I have in my system. Thanks for the reply, perhaps my issue stems from my enclosures being mostly sealed. No venting what so ever. I did a quick test of a few of my sensors, opening the boxes and letting them sit in open air and the differentials came down substantially. Will look in to adding some venting!

Finally began tinkering with the boards I ordered a couple of months ago. Built 2, using regulated 5V. Both tell me they cannot init radio. I have yet to pull out the multimeter and begin diagnosing, but wanted to post up and see if anyone else has experienced this? I have not actually come across this error before with MySensors. I did solder pin 1 and 2 on the two pads, attached a 5v to 3.3 volt step down, cap in the slot direct above the radio. Blanking on what could be wrong, other than something witht he baords I received.

Check the bmp180. Its a presure sensor, but it includes a fairly accurate temperature sensor and is rated down to 1.8v.

On my 5V minis, the 16mhz crystal is much larger (long silver oval component stamped 16.00) than on 3.3v. At least on mine.