Ed Davies

Does the UI let you disable the Spotify plug-in? I'm thinking to get rid of all the warnings for it just to be sure you're only left with the crypto deprecations, which I wouldn't worry about other than to look out for a new version of MusicBox being available. Still, those read errors on some of the music files does point in the direction of the card being flaky.

Or in a rural area, on the end of a long bit of wet string, with just yourself generating quite a few kW. Remember, 253 V is less than 5.5% over 240 V which is what a lot of the distribution network taps are still set to.

Re PV generation across the year, well worth a look at PVGIS: https://re.jrc.ec.europa.eu/pvg_tools/en/tools.html for your proposed house site. On another forum I collated the outputs people were actually getting each month with what PVGIS predicted for their site. Generally speaking PVGIS was pretty reasonable but a bit pessimistic; about a quarter of the time the actual output was below the PVGIS prediction, 3/4 of the time it was above. PVGIS has been tuned since then with different radiation databases added so perhaps it's more accurate now. An important consideration is that usually people optimize the angle of their PV panels to maximize production over the whole year [¹]. This results in the panels being at quite a shallow angle which does clobber winter production (though it increases summer production slightly). Putting the panels at a steeper angle reduces summer production slightly but increases winter production a lot. For off-grid you want to maximized DJF (December, January and February) production. If that's sufficient then you're laughing the rest of the year. E.g., for my plot in the north of Scotland increasing the panel angle from 35° to 60° reduces the production in May from 121 kWh/kW to 108 h but increases the December production from 14.4 to 18.8 h. The gained 4 h of production in December is much more valuable than the lost 13 h in May when you're off-grid. In other words, yes, there's a big difference between summer and winter production but the way people currently have PV set up exaggerates this effect. [¹] or, more realistically, just accept the angle of the roof they happen to have which is typically quite shallow. If you're doing a new build then you don't have to do this - you can pitch your roof appropriately or ground mount your panels.

That's mildly impressive. A decade or so ago they could only get to that level of accuracy with some fairly heavy post-processing once the data was taken back to the office.

The Stefan Boltzman equation is one of my most favourite of the equations. ? The emissivity, e, typically equals about 0.9 to 0.95 for most building materials other than shiny metal. A simpler approximation to @JSHarris's first equation is to just use 10 W/m²·K. I.e., 10 watts for each square metre of warm floor for each °C of temperature difference between the floor surface and the average room temperature.

In theory, LED lights (other than the “filament” type) ought to be a lot less affected by voltage fluctuations. Is that actually true? Houses with battery/inverter systems in them (Powerwall, etc) should also be less affected. I wonder if a saving in cable cost could ever tip the balance in their favour, economically.

Only the supply (to the house) and extract (from the house); not the inlet (from the outside) or exhaust (to the outside) which should be at near-to-outdoor temperatures, of course.

To keep an area at 28 °C, assuming the same room temperature, surface finish and insulation above and below the heating element, will take the same amount of power independent of whether it's 230 V, 240 V, extra-low voltage, water or molten sodium. If the amount of insulation above the element (wire, foil or pipe) is increased then the element will need to be hotter which will increase the losses downwards unless the insulation below is also increased to compensate but, as @JSHarris points out, with sensible amounts of insulation below this loss will be relatively small anyway. It's just possible that a low-voltage element which can be very close to the surface could make significant savings compared with elements buried a bit deeper but only if the insulation below is completely inadequate. I suspect their figures are done on the assumption of no insulation below the elements. If so, that's a cheat IMNSHO.

How about running de-humidifier in the room for a few hours a day? It'll drop the relative humidity and provide some heating, a bit more than the amount of electrical energy it uses as there's also heat released from the water it condenses. Some friends had a rural house in the north of Scotland which relied on convection to get heat from the wood burners downstairs to the upstairs bedrooms. Not surprisingly, the upstairs was pretty chilly and suffered from mould. I suggested a dehumidifier which solved the mould problem and warmed the bedrooms up a bit. Since then they've bought a larger but similar construction house, which at least does have radiators upstairs, and now have 3 scattered around.

kN. SI prefixes for x1000 and below are lower case, symbols for units named after people (Sir Isaac here) are initial capped. I always find these quoted floor loadings surprisingly low. E.g., a rack of bookshelves could easily weigh 3.6 kN per linear metre. Assuming that's spread over a metre each side (2m span) it's still 1.8 kN/m². Or, clicking on the first washing machine on the Argos site, mass is 66.5 kg, width 60cm and depth 52cm so the floor loading is 66.5 * 9.81 / (.6 * .52) = 2.091 kN/m². OK, it'll be spread over a bit more area but there are probably other heavy things near it plus the dynamic loads as it tries to shake itself to bits, as is the wont of such things.

Is it worth trying a fresh memory card?

You could try wget on that URL to see what it gives. On my (Raspian) Pi I get: $ wget -S http://sc5.radiocaroline.net:8010/ --2019-10-13 17:04:50-- http://sc5.radiocaroline.net:8010/ Resolving sc5.radiocaroline.net (sc5.radiocaroline.net)... 78.129.202.200 Connecting to sc5.radiocaroline.net (sc5.radiocaroline.net)|78.129.202.200|:8010... connected. HTTP request sent, awaiting response... HTTP/1.0 200 OK icy-notice1:<BR>This stream requires <a href="http://www.winamp.com">Winamp</a><BR> icy-notice2:SHOUTcast DNAS/win32 v2.5.5.733<BR> Accept-Ranges:none Access-Control-Allow-Origin:* Cache-Control:no-cache,no-store,must-revalidate,max-age=0 Connection:close icy-name:Radio Caroline icy-genre:Classic Rock, Oldies icy-br:52 icy-sr:22050 icy-url:http://www.radiocaroline.co.uk icy-pub:1 content-type:audio/aacp X-Clacks-Overhead:GNU Terry Pratchett Length: unspecified [audio/aacp] Saving to: ‘index.html’ index.html [ <=> ] 282.57K 22.5KB/s ^C Have to control-C to stop the download as it's streaming, presumably. I expect wget is already on the system but if not you can probably install it with: apt-get install wget Alternatively, you could try curl but you'll need to look at the man page to see how to get it to list the response headers.

Probably a bad idea. Again, from the manual:

Is the Pi still on that address? It didn't get reallocated during your network faffing, did it? From the manual:

When you're SSHed into the Pi, can you ping out to the Internet? E.g., $ ping bbc.co.uk

Tanking the walls wouldn't help. A French drain outside likely would depending on the circumstances. I'm thinking of the house in Tongue I rented which had significant damp problems downstairs which I think was at least partly due to it having had a concrete floor put in. The water table was always pretty close to the surface (and sometimes above it) and I strongly suspected that the concrete floor made it worse by not allowing water vapour out.

I'd think that if there's water vapour rising up through the soil below the house then a vapour-open floor will let some of it out reducing the overall vapour pressure under the house and therefore the likely amount of water which would go into the walls. I don't know how much difference that'd make in practice, and it's probably very site and house dependent, but I can understand being wary of retrofitting a relatively vapour-tight floor in a house which didn't originally have that.

Anybody else think mention of the Venturi effect is a complete red herring here? Just in case this isn't obvious to anybody, 22mm won't provide a slower supply of hot water (in litres/minute or whatever) once it's run through to hot but it'll take longer for the water to get hot at the tap when you first turn it on, compared with 15 or 10mm pipe, simply because there's a greater volume of cold water waiting in the pipe which has to be run through and replaced.

Ventilation is good but wouldn't critters getting under the tin and nesting in the membrane, etc, be less good?

I'm with @andyscotland on this; ring finals seem like a fail-dangerous setup. I'm not planning to have any and for most radials I'll go for the minimum plausible current for the breakers, never more than 20 A. There are off-grid reasons for going that way but I think I'd follow that plan even in an on-grid house. I have a notion in the back of my mind that BS1363 sockets are supposed to always be able to supply 13 amps. I.e., you shouldn't put them on circuits with less than a 13A fuse or MCB. E.g., not on a lighting circuit even if you plan to only plug a light into them, hence the use of round-pin plugs for remotely switched corded lamps. Obviously there's diversity allowed so three sockets together don't have to be able to supply 39A off a 32A ring final. Or am I imagining that and it would be OK to have BS1363 sockets on a 6 or 10A radial?

Common sense says that's a sensible solution but thinking, instead, about regulations: why wouldn't these have to be in a metal enclosure? I.e., what's “similar switchgear” to a consumer unit and what isn't?

That'll likely not be a globally routeable address but rather one in one of the local address blocks [¹]: 10.nn.nn.nn or, more likely, 192.168.nn.nn, probably 192.168.0.nn if you've got default router settings. Almost everybody's home LANs (Wi-Fi and wired) use addresses in these blocks; there are many millions of routers all with the address 192.168.0.1. There's no way to directly use that address to access the device from outside your LAN. If all the devices in people's home had unique globally routeable IPv4 addresses then we wouldn't be running out of IPv4 addresses now, it'd have happened a decade or more ago. Instead such communication has to be either via a server (run by SG, presumably) which does have a globally routeable IP address or it could be sent to the globally routeable IP address of your router if that was configured to forward packets (with some particular destination port number) to your SG controller. There are two complications with configuring your router to forward packets to your SG controller: firstly the way it's set up varies from router to router and holding user's hands to get it configured properly would be a support nightmare for SG. It's possible that their controller could talk to your router and get the configuration set up automatically (using UPnP [²]) but that's not always going to work. E.g., I've got UPnP disabled on my router - can't remember if that was the default or I turned it off myself but whatever… The second complication with using the router to forward packets to your controller is that the external IP address of your router could change. Some ISPs assign a static IP address to each user, others just assign one from a pool of numbers when a user connects so any time you turn your router off and on again you'll likely get a different IP address. Again, a support nightmare for SG if they tried to rely on that. [¹] https://en.wikipedia.org/wiki/IP_address#Private_addresses [²] https://en.wikipedia.org/wiki/Universal_Plug_and_Play#NAT_traversal

Just re-read your post and noticed this bit. So it's not a matter of not being able to talk to your controller remotely, it's a matter of being able to access it but not command it. Speculating wildly, perhaps SG have realised that their protocol to whatever server is used to find the controller from the app and get the status is secure enough for those purposes (for privacy) but not secure enough to safely allow remote control. If so, 1 point to them for thinking it through sensibly.

To be able to access a device on your own LAN from outside on the Internet requires a couple of things: 1) Your router needs to be set up to allow packets addressed to the device to pass through from the Internet. This is doable but router specific and a right PITA. 2) Your app needs to know the global address of your router or device. Both of these are tricky. Normally routers are set up to only allow in packets from the Internet which are replies to recent outgoing packets or are specially configured both for security reasons and for address-space reasons. In an act of bone-headed idiocy [¹] the original internet protocol (IPv4) was designed with only 32-bit addresses allowing roughly 4 billion devices to be addressed. Since most people use many more than one device needing addressing and there are quite a few people on the planet we've basically run out of addresses. A typical domestic DSL line only gets one IPv4 address to be shared by all the devices in the home using NAT. So your SageGlass device won't have its own globally routeable IPv4 address. As well as the addressing problem there's also security. Given the dire state of security of most computers and related devices it's best if most of the world's hackers don't have access unless it's specifically required. There are two possible solutions. One is the current version of the internet protocol (IPv6) which allows many more devices to be addressed (many billions of billions of billions, literally). Unfortunately, as a matter of general tardiness and bloody mindedness the computer industry has been very slow to take up the protocol. The standards were mostly stabilised in the late 1990s but it's still not in common use 20 years later, which is a bit pathetic really. What most IoT devices do is call back to a central server somewhere which the app can also call in to to establish communication. Presumably this is the approach SG will use for their next version. It usually works OK but it has some critical flaws: 1) You're dependent on your internet connection working to do simple things, often even if you're at home on the same LAN as the device being controlled. 2) You're dependent on them continuing to run the server. Possibly OK if you have some sort of subscription to motivate them but very dubious if they're just promising to do so indefinitely out of the proceeds of the original sale. 3) There may be serious privacy concerns depending on the function being controlled, the software quality and ethics of the company concerned and so on. [¹] Arguably based on racist assumptions, or maybe just a lack of confidence in the likely spread of internet usage.

It's worth pointing out that Glyphosate doesn't stop stuff growing permanently. It stops the stuff that's there permanently (i.e., kills it) but is specifically designed to degrade in the soil so allowing replacement plants, of the same or different types, to grow in place.