Internet of things is a concept I really buy. Home automation in the form of intelligent lighting, heat control, devices which you can ‘train’ to learn from your behaviour (yes, I’m mainly targeting my coffee-machine here).
So, for a start I want to replace my AC switches and heater thermostats in my house. And I want to use the .net Micro Framework to do it. Thing is, I have a goal that the electronisc of each unit, switch or thermostat, shall only cost $20 tops to replace. This on grounds of volume: I have approx 25 switches and 7 thermostats, equals 32 times $20 = $640. It all adds up.
So I’m beginning to think of how to go about it. The requirements are 1) it must be .net Micro Framework and 2) it must be a wireless solution.
I’m thinking RF and a relay for the switches, RF and small servo-motor for the thermostats. That’s max $10 in parts. That leaves $10 for a .net-enabled micro controller which needs only one SPI and two-three GPIOs and Analog Inputs. Possible? I’m reading a page on mountaineer.org’s site which states that “We have ported NETMF to the STM32F1”. Might be a way to go?
7 thermostats? Does every room in your house have a separate one?
If you can figure out how to do this for $20 per you’ll be a rich man I’d recommend a NETMF central computer that coordinates the other modules based on a simpler MCU. Others will be quick to point out also that by doing this you’re basically voiding your insurance if you have a fire w/o having the proper certifications on your hardware.
The STM32F103RET6 that Oberon ported NETMF to initially runs about $8 per. Add another $1 or $2 for the necessary crystal(s) and decoupling capacitors. Include $1 per board (let’s very generously assume you can fit it on a 5cm by 5cm board from DFRobot or Seeed). That puts your MCU and associated paraphernalia at roughly $10.
If we assume that nrf24l01+ radios would suffice, then they can be had for ~$1.50 per.
If this is a “normal” residential HVAC setup, you’ll need a 24VAC power supply at each location. I’ve used the 78SR from Murata, it’s awesome, but you can’t afford them at ~$9 per. Let’s be generous and say you can put together some sort of horribly inefficient linear power supply for $5 (which you certainly can’t).
If this were a “normal” residential HVAC setup, then you’d need 3 small signal relays to run the heating/cooling system (call for heat, call for cold, call for fan). They run about ~$1 per.
I doubt you could actually achieve that in practice, however. This number also doesn’t include any hardware, buttons, case, or other bits. Further, as Ian said, you likely do NOT want homegrown stuff attached to your heating/cooling system when the insurance adjuster shows up.
As for your AC switches, I assume you mean wall switches. Again, this is an area where that little UL logo is very important. Your power supply will be more complex, the board will be MUCH more important to do correctly (mains voltage maims/kills/burns down the neighborhood), and when things go wrong, you’ll have a real live disaster on your hands.
I agree with Ian. You need netmf for central unit and major units but small swiches connect to central unit and not Internet. This means a $1 micro is good enough.
insurance is not a major problem per se, here in Denmark we’re allowed a deal of freedom in messing with the electrical systems if - conditioned - a trained electrician deems it correctly installed. Fewer lives are lost than you’d think, that’s typically when no-one cared to chase that ‘approved’ signature down… Changing existing AC wall switches doesn’t even require someone skilled looking into it, though one should lest insurance is voided.
$20 was the limit for the electronics only - buttons and mountings and such are not the most expensive things in the world and I can machine the casings quite easily. So thanks godefroi for your estimate (and for teaching me how to solder small! - Soldering headers onto the GHI Cerb40 module - YouTube) that tells me it’s possible. But practical? I’m not sure I could bear moving to a different MCU platform for the units, not when development is so easy with these .net-enabled chips…
I’m currently planning something similar.
I want to use one or two G120HDR as intelegence and use some Atmel Tiny µC’s in thermostats and propably Switches.
As communication Backbone I was thinking about RS485. Easy to use and robust on a single twisted pair wire.
Or are there any other options which support wire length >20m?
btw. I have 11 Thermostats in my house.
Abou costs:
A small tiny costs less then 2€, Atmel Studio for coding is free (but you need to code in C or C++, not a problem for me, and also only verry simple programs for switches and thermos). Then abou 20€ for a programmer.
An empty 2 layer circuit board costs me about 6€ at a size of 50x50mm if I take 15 pices.
Plus a couple of other parts (resistors, …): might be under 20€ per thermo.
As a housing I use my old ones, where I need to replace one per year in average anyway.
None that I know of, cheaper than what you describe. I do envy you; wish I could cut up my walls and install wires, the project would be much easier.
[quote=“Reinhard Ostermeier”]Abou costs:
A small tiny costs less then 2€, Atmel Studio for coding is free (but you need to code in C or C++, not a problem for me, and also only very simple programs for switches and thermos). Then abou 20€ for a programmer.
An empty 2 layer circuit board costs me about 6€ at a size of 50x50mm if I take 15 pices.
Plus a couple of other parts (resistors, …): might be under 20€ per thermo.
As a housing I use my old ones, where I need to replace one per year in average anyway.
[/quote]
Yea, looking at the Atmel’s, too. By all means let me know how you fare, and what choices you end up with, am much interested herein.
Yea, looking at the Atmel’s, too. By all means let me know how you fare, and what choices you end up with, am much interested herein.
[/quote]
Fortuenatelly I have tubes (ductwork) for all cables in my house.
I want to make some 1st circuit drawings over the next couple of weeks. Fortuenatelly I have a college at work who is very familiar with such things.
Why do not you start using X10 standard ? Controlers and swithcs are already existing, and everything can be controled by a NETMF Device, see GHI Electronics – Where Hardware Meets Software
Maybe because x10 is a really poor implementation of automation? You never get confirmation that an issued command is acted on, that’s pretty hard to maintain knowledge of state in a controlling application.
I’d recommend a NETMF central computer that coordinates the other modules based on a simpler MCU. Others will be quick to point out also that by doing this you’re basically voiding your insurance if you have a fire w/o having the proper certifications on your hardware.
