WiFiThing

A breath of fresh air, or dead in the water? Let’s take a look.

But firstly an update – as of today (Feb 1, 2016) the Kickstarter campaign has been cancelled. Now see my comments below..

WiFithing overall: So WiFithing comprises a master board in a nice white box and up to 8 slave boards (in nice white boxes), handling for up to 4 groups of FS20 radiator valves and 10 Orvibo Smart sockets – and flying in the face of many new developments, uses radio – but not WIFI to connect devices.

Kickstarter: The project is on Kickstarter and has 3 days to go to achieve the goal of £15,000. They are currently sitting at £11.8k which is a worry as most successful projects tend to go way beyond the original goals so close to the end date and we’ve already seen a couple of projects in here fail at this late date – but stranger things have happened.

The Orvibo saga: So first things first and a little diversion. I opened up the box and inside was a a WiFithing master, a WiFithing slave and the Orvibo Smart Socket along with some instructions. Even from China the Orvibo units are £17 each (Amazon seem to have stopped stocking them and don’t know when they will be back in) and so not exactly the cheapest solution for mains socket control – but at least they had a proper British plug integrated.  This is the mains socket that WiFithing have chosen to support so it is worth a further look.

Well, I promised a diversion…. I investigated at some of the Apps for the Orvibo device and “still not very reliable” was the first comment I noted (referring to the app). One guy was pleasantly surprised as the app worked “80% of the time”. Nice looking, too expensive and by the look of it not too good on the App side.  I’m looking for MQTT-supported similar devices and of course, Espressif-based devices are coming online now. Espressif themselves during the summer trip to Boston I took with them, showed me  a unit selling in China (sadly not outside) for just a few dollars and the Sonoff board would probably nicely fit into a case similar to this – that device has already been hacked by at least two of us to handle MQTT and is nicely priced as some of you know.

Most of the reviews for the Orvibo-compatible apps achieved poor scores on Android (a shame as it’s a lovely looking device). One app looked promising with a high score… Orvibo AllOne Wifi Alternative.  Well, I got no-where with that as, on installation it spotted my two NOWTV boxes and simply would NOT let me go any further until I set those up.  I scrapped that and instead installed the WIWO app and plugged the Orvibo into the wall. The App found the device immediately and I made a note of the UID.

The App turned the Orvibo on and off. Success. I moved the socket a few feet to a place where I could plug a lamp in – and the App promptly crashed. However thanks to THIS article https://nathan.chantrell.net/20160101/orvibo-s20-wifi-mains-socket-with-node-red I was able to get the unit to work under MQTT – on and off – first time, no problem.

BUT – wait for it…. turn the unit on – disconnect the power, reconnect the power – and …. it does not remember the state it was in.

That for me would be a money-back-requiring deal-breaker for that particular unit.  Imagine using this in an environment with faltering power (and there are MANY such environments, particularly in rural areas ) – this socket would be a non-starter as are many of the cheap and cheerful B&Q mains controller sockets.  It is worth noting that in the modified SONOFF controller software, I (and presumably the other guy who’s done the hack) ensure the power-up output remains as it was before the power cut or brownout!

Of course, one could easily have Node-Red refresh the state of the socket or sockets every minute and that ultimately is what I did. Works a treat.

Anyway, I digress. The Orvibo is just part of the picture. The Kickstarter project for WiFithing makes a big deal of not using WIFI “As you have probably heard by now, IOT devices tend to be very low bandwidth and bursty in their communications. WIFI overhead is pretty painful”.

I’m not sure I follow the argument against using WIFI – I just finished setting up 40 IOT devices and having them talk to Node-Red as some of you know. The overhead in terms of use of the router appeared insignificant and the cost of each unit almost irrelevant – so I must be missing something in the above. They also refer to WIFI only solutions as being heavy on battery – but I know people in here who have successfully put ESP devices to sleep, occasionally waking them up to report back via WIFI. 

The WiFithing product does not use mesh networking and so each WIFITHING slave is paired up with a master. 

Again as many of you know, I’ve been down that route with the very DIRE NRF24L01 – in fact even with a mesh network I spent months on what ultimately was a waste of time so I am very wary of claims about such radio systems – to be fair they are using 800Mhz and not 2.4Ghz. Even at 800Mhz I guess no-one who tests these lives in houses with thick stone walls.

The package I received comprised a WIFITHING master, an Orvibo socket, and a small unit containing a battery connector and a board by panstamp.com So the Panstamp devices uses the CC1101 low power radio. and depending on the model, an MSP430MCU or an Atmega 328p. I’m not sure about the box, very nice and small but there was only room for the board and 2 AAA cells inside. Not entirely sure what you’re supposed to DO with it given no room to fit anything inside.

The WiFithing master itself, again a very pretty box but inside was the requirement for a round battery – and I started the review on Sunday when the shops are closed. As no such battery was supplied that brought that to a halt. The master unit will apparently cost $57 which puts it on a par with a Raspberry Pi 2 with WIFI. Ok, this is a finished product which claims you can have control “code-free” whereas Pi solutions tend to involve at least flashing a chip…In the campaign they compare their board to a Launchpad MSP430FD4969 + CC3100 booster + Anaren CC1101 board  – and that’s fine – if I were about to design a controller, those three would  NOT be on my list of candidate boards.

Radiator control: The FS20 radiator valves mentioned seemed to me fine but a little expensive – including postage they seem to work out at £45 each. A handy solution complete with batteries – they have to make more sense than the really STUPID thermostatic valves that most people find their radiators attached to but again I’d rather have a WIFI-controlled stat I could hack or that already used MQTT. Personally I’ve been thinking of hacking one of these… and sticking an ESP to control it – but if anyone has a better idea that is no more expensive – do let me know.

Open Source: The WiFithing project is open-source but the code is not yet on Github. Also the radio certification is not yet complete. See the Kickstarter page for more info.

I noted that they intend to have the product manufactured in the UK. Will that jack the price up? In an ideal world I’d like to see everything manufactured in the UK but I’ve seen enough conversations in this forum to know that many folk head off to China for best prices.

If you want to use the web app they charge £1 (£1.50) a month for unlimited devices per single house/site.  I think I’d rather pay once and be done with it – others may disagree. Many of the features are “goals” which will be progressed following a successful launch. With only 3 days to go and a £3k+ shortfall on already modest goal I would not hold your breath but certainly worth taking a look if you want simple radiator and remote control with uber-security and no code.