On 22/11/14 20:08, Jeremy Nicoll - news posts wrote:
>> Here's a very nice graph of the same:
>>
http://i62.tinypic.com/105wbwx.jpg
>
> What does the dial saying 1.32 kWh represent?
1.32 kWh in the last hour. If you click on the "Daily", "Weekly" etc
button it will change accordingly.
>
> Is the power used by the switches themselves tiny?
Good question - I will be able to measure that when I have some more,
hopefully next week. Not sure how sensitive they are - might not be as
accurate as a KillOWatt type device.
> If youhave a house power-cut, do all the devices come back to life when teh
> power comes in, and talk to each other again. (I realise you may not know
> this yet!)
Yes. I have power cycled the controller.
The one thing this one is too dumb to do is "catch up" on scheduled
operations that happened when it was off - a bit like cron won't run
stuff if you turn the computer off but anacron makes an effort.
>
>> This device is a bitch to pair though - took me about 10 goes. Trick is
>> to power it from USB while pairing so it does not go to sleep every 2
>> seconds before it finishes the pairing (it actually looks like 4
>> devices, not one, so you get the master device, the PIR sensor but end
>> up missing the rest).
>
> Once a device is paired, is it paired for good? Or will it forget its
> pairing at some point?
It shouldn't do.
In fact, bringing devices over, I had to do the "unpair" thing first to
make them forget about the Vera.
This seems reliable even with mains devices unplugged and battery
devices with the batteries removed - I guess that magic numbers are
stored in flash or EEROM.
>
>> This one shows the gui logic for a simple scheduler for the heater switch:
>>
>>
http://i60.tinypic.com/2lmb11w.jpg
>> You have to pay 2x the price for the next model that lets you write
>> actual LUA code (we'll you can hack mine but results are not guaranteed).
>
> When these scripts run, is there any log file produced that shows what they
> did? Maybe the gui logic is impossible to make a mistake in, but I'd also
> wonder if one was writing in LUA (never used that...) whether it can
> generate disgnostics, log files etc...
>
The graphical scripts are hard to debug. I forced mine to run 2 mins
into the future until I got the basic logic right - then I changed the
times to the correct ones.
Not sure what LUA is like on the HC2.
>
>> However it does sport a REST/JSON HTTP based API that's really simple to
>> talk to. eg:
>>
>> curl --request GET \
>> "
http://USERNAME:PASS...@fibaro-hclite.example.com/api/devices" \
>> | python -mjson.tool | less
>
> So... do these device have ip addresses just on your LAN, or are they
> addressable by anyone in the outside world? Would you have used that curl
> command in that form (with the .com address presumably having whatever value
> DNS provides) or a local address?
>
The controller only has an IP address. You can NAT it, firewall it or
expose it on a public IP as you wish - treat it like any other service.
In about a week I should have some idea of the overall reliability.
Due to ongoing lack of proper CH, I am using these to control my oil
filled electric rads. Also the energy monitor is useful as it will help
me size the CH for next year. After that, the switches will be deployed
on a few desk lamps as burglar deterrents.
I might have a little ZWave in the CH - like the radiator valves. That's
why I'm interested. In the medium to long term I'd liek to move to an
open source project to drive it with maybe a Raspberry Pi but frankly
those projects are not there yet. I think from what I've seen, ZWave
*could* be a reasonably solid technology, for 98% reliability.
Given it can be queried (ie comms are 2 way) there's no reason, with
proper logic and a correct scheduler algorithm (that knows the final
state at any given time without relying on passing through the switching
times) it ought to be able to check and force the devices to get the
correct state. I think the Fibaro is too thick to do that, but that's
not a fault of ZWave itself.