For those interested – I’ve been quiet the last couple of days as we’re ploughing ahead getting the new office up and running – got walls up and ceiling almost done – got power and Ethernet in last last night and today we’re off to get wall sockets and worktops. I have to say I have never known joint aches like it!
The dangers of controlling stuff over the Internet became apparent last night as the heating went awol. At some point we realised we were roasting to death at about the same time as the Sky box indicated it has no Ethernet. It turns out that in the process of wiring up Ethernet for the office, we’d uncovered a flaky connection to the main access point which provides all the ESPs with connectivity and feeds the Sky box.
Still – though it took until 1am in the morning to find the problem, at least it’s a clean problem with a clean solution – replacing an RJ45 connector.
Anyway a package turned up for me yesterday which I’ve not had time to do anything with yet – the Omega and a bunch of peripherals – so expect a review coming up soon.
I have a distrust of computer systems (based on working with them for over 30 years) so I built a keep alive into my web based heating system. Basically it’s a retriggerable monostable fed from one of the IO pins. If it doesn’t get a pulse every minute or so (or the power fails), the relay it’s keeping open will close and control will revert to the original thermostat.
It’s a technique used in many control systems where any failure in main system could cause problems and it guards against computer , connectivity (and if the pulse is generated by the same code that does the control) programming problems.
Mark
Hi Pete,
Just a thought on your fault. I’ll bet Mrs. Pete was tut tutting in the background while you tried to track it down! Which begs the question, have you got the design right if a fault like this can bring down your heating system.
I’m just finishing off our Heat Bank and CH controller and originally had all the tank temperatures supplied by an openenergymonitor system that does PV diversion but recently reconnected the tank temperature sensors to the esp8266 system controlling the Heat Bank and CH. In this way I have all of that control isolated in one system which also has in parallel a Honeywell 7 day timer for the CH and the old thermostat control for the Heat bank which can be switched to in the case of failure.
Sure I do some weather compensation, can turn the CH on from outside the house (like from the bus stop in town when we’re been out) but in the worst case scenario I can control the CH with the Honeywell controller and can revert back to the original control system for the Heat Bank.
Maybe your fault the other night is a wake up call for a for a redesign to make each part more autonomous and to build in some fall back systems.
Simon
Looking forward to the Omega review Pete – thanks for taking the time to blog all this new stuff”
Simon does raise an interesting point re “error handling” etc
Do you think some sort of system check regime (watchdog timer?) should be introduced in the node red code for each system that you make?
I’m thinking it would be alongthe lines of “check this is communicating every 6 hours, if not then send error email?”
I suppose you could also extend this to a general “cloud” node red system – that health checks if your whole R.Pi “node red” is powered OK too!
In both cases above, they just send an error email – to indicate to you that there is an error – and if things reset themselves OK (if there has been a power down/up)
I guess inthis case, you could take the view that becuase you’re in the house anyway, you immediately get feedback anyway, so taht should be OK anyway?
Just interested to see what you decide
cheers
Pete