←back to thread

280 points RyanShook | 2 comments | | HN request time: 0.401s | source
Show context
zahirbmirza ◴[] No.45144523[source]
Connected thermostats are great in theory! But they should not have to rely on a cloud connection. A local network with the option of internet connectivity would be awesome; but, it seems, no company is going to become uber successful if there isn't the option of forced upgrades and cloud subscriptions. Look at Ring...
replies(8): >>45144548 #>>45144575 #>>45144606 #>>45144681 #>>45144736 #>>45144819 #>>45145083 #>>45148418 #
stavros ◴[] No.45144681[source]
I bought a $20 Zigbee thermostat from AliExpress and it has been fantastic. It turns on when it's cold, and off when it's hot. Anything else, I can do with software, because it's just Zigbee.
replies(1): >>45144953 #
gerdesj ◴[] No.45144953[source]
I go for Zwave by choice but Zigbee comes a close second. It does share 2.4GHz with wifi but its many tiny bands fit within the "edges" of the wifi bands. If you stick to 1,6,11 for wifi, Zigbee will co-exist very happily. Even if you don't, it will still work fine - the messages are tiny.

Both Zwave and Zigbee build mesh networks with multiple routes. Wifi devices ... don't. Wifi is fine for IoT but it isn't optimised for it. My fridge/freezer uses wifi as does my oven and microwave. It doesn't matter if they lose comms sometimes and there is no choice anyway.

My light switches are Zwave. Thanks to way modern UK wiring is done, most of my switches end up with an extra conductor and so are permanently powered and act as hubs for the battery powered window sensors and the like.

My cameras are all PoE ethernet, including the door bell. All Reolink.

I have two UPSs with at least 30 mins run time. I could easily put in a genny or a battery or even use my car (EV) but its not important enough (yet). So far everything will work without the internet.

I have deployed two VLANS for IoT - THINGS, and SEWER for the really worrying gear on it!

Home Assistant runs the show.

replies(1): >>45146169 #
1. kevin_thibedeau ◴[] No.45146169[source]
> If you stick to 1,6,11 for wifi, Zigbee will co-exist very happily.

I worked for a company that converted a legacy wire protocol with no QoS guarantees to be used over a proprietary modification of Zigbee. One of the managers complained that their volume control would randomly climb to the max loudness. The protocol used press/release packets for button presses and if the volume-up release packet was lost due to interference, you got a runaway increase in volume from the system assuming it was still held down. This usually happened when the channel assignment was in a band used for active wifi.

replies(1): >>45162710 #
2. gerdesj ◴[] No.45162710[source]
That is pretty daft. Surely, you should always send the intention rather than the mechanism when a dodgy medium is involved!

If the medium was ATM or hard wired ethernet then sure why not send button presses. Those are reliable media.

The obvious fix would be transmit "vol+1/Pressed" on button press and "vol+1/Release" on button release. On receipt of v+1 do just that and no more. Note a /Release to colour a widget correctly, perhaps. Holding down V+ would transmit multiple v+1 or use a wheel as an old school Walkman did to send actual values.

Nothing new is old or something 8)