MortarThePoint

I'll have to remind myself which I decided against individual pipes per appliance. On thing I don't like about it is the shear number of pipes I have to pass through a cavity wall which currently doesn't have holes in it for the pipes. If the local manifold is accessible, you don't end up with many more connections than without a local manifold. Consider for example a 3 appliance bathroom (shower, loo, basin) The connections are as follows Local manifold: at main manifold, 4no. at local manifold, 3no. at appliances --> 8 TOTAL No local manifold: 3no. at main manifold, 3no. at appliances --> 6 TOTAL It feels unlikely that I would need to isolate one appliance whilst keeping the rest of the room operational. On a statistical basis: You are almost 3 times as likely to have an issue with a long inaccessible pipe crossing the house than if you just have the 1 such pipe. Chances very low still though.

I was hoping that most of the stuff that would need access on the cylinder would fit on the front 120 degrees of the cylinder and have reasonable access. The plan was for the tank to go in before the wall is put up, but that tempts the installer to put something somewhere difficult. I am planning to fit a computer fan (120mm) to blow air out of the cupboard. They push about 90m3/h = 0.025m3/s and the heat capacity of air is around 1kJ/m3K. That means a 120mm computer fan can remove 0.025m3/s * 1kJ/m3K * 10K = 0.25kW with the cupboard 10C above the neighbouring room's temperature. The cylinders loss figure is about a third of that, so I should be OK (hopefully). There will be loss from the UFH too, but I could always add another fan. Cold manifold is one feed per room that uses water (all 15mm), garage and outside tap. DHW manifold is 10mm to 4 basins (3 bathrooms, 1 loo), 15mm to 3 showers* and 1 bath, 15mm to Kitchen, 15mm to Utility, 2 spare. Feels like adding a spare or two to the cold. Note: two of the showers are about 6m run from the manifold, the third is about a 15m run and (*) actually shared with a bath which will mostly be used more as the children grow up so waiting for hot isn't too bad. I have run the numbers on this before and they looked OK (10 seconds per 15m run at 10l/min shower, pure hot). I keep hearing people say I should have a Water Softener, but have never liked how the water 'feels' or tastes with them. I'm near Cambridge, so water is hard but not mega. I expect shower TMVs are the main risks with this so keeping them accessible (one through a hatch in the wall) and the other two are on shower bar so readily replaceable. We've had the same kettle with this water for ages, but do use a Brita filter (which gets replaced very infrequently). Network rack is going under the stairs with the other GF UFH manifold (toasty, so another venting computer fan or two is likely). I'm far from Passive Haus standards so have to have a boiler/ASHP and so makes sense to use it to heat the water. I do wonder if I might regret not having a secondary tank coil, but don't currently have plans for anything that could use it. Possible future PV could be hooked up to this tank's immersion heater I presume. Without the experience, I have to think it through ahead of time rather than rely on instinct. Many a time have I stood on site staring at something wondering what 'the right way to do it' is. Equally, sometimes you just have to get it done and move on.

Sorry, cross post. Is it not relevant to Stainless Steel though?

It's a stainless steel cylinder so is that only relevant to copper cylinders? It's not going to be the official Mitsubishi one as there have been issues with supply. Here's a thread on the cylinder: But this is the datasheet:

@Nickfromwales I guess it's the fact it's sacrificial and so wears out.

PDGF link doesn't work so here it is as an image:

I've just ordered my ASHP & Cylinder package and am progressing my planning out of the Plant Room / Cupboard. The house design leaves a small room 1m x 2.6m for the following: DHW cylinder (ASHP): has a reasonably small footprint 575mm circle), though will no doubt have pipes all over the outside. Plan to leave extra 200mm all round. It is very tall, ~1800mm. DHW and cold water manifolds ASHP controls: Mitsubishi Ecodan. hopefully doesn't take up much space UFH primary pump and filters: I am going with a single large pump pressure driving a 3 manifold system. If I get issues, I can later add pump(s) to manifold(s), but no more space needed in Plant Room. UFH manifold: 4 port, 3 used. Plus controller. Electrics CU: 3-phase connection. This will be many way as I am having lots of circuits. I will also have a feed to the garage, which may be in a separate CU box. I guess it could end up as actually being 3 CUs, one for each phase where the phases are slit House / ASHP / Garage. Lighting control: a metal cabinet housing relays and control circuitry for the lights Battery bank: Potential to add this later. Server rack style. May choose to have it outside the house (e.g. garage or meter kiosk) Forgotten something? Rough layouts below. Walls & Floor Finish: Before the ASHP and DHW cylinder work can start, I need to decide what I am doing about the walls and the floor. They are currently bare blockwork and screed respectively. For the walls I see three options: As a minimum, I would seal the blockwork with Passive Purple for airtightness Wet plaster the walls and paint Simple tile As for the floor, I see four options: Leave as bare screed Seal with a paint (like garage floor paint) Simple tile Tank and create some form of sump around the DHW cylinder. Would need a drain provision though What have others done with their plant rooms?

But for it being super weird, there is a lot of sense to having the manifold on the ceiling shooting out its however many pipes horizontally off to where ever they need to go. Not very accessible (would need steps) but avoids the need to send each pipe up about 1000mm and then through a right angle into the ceiling void of the room next door. Feels too weird though. I could just imaging the look on a professional's face if they saw it.

Some useful dimensions for 3&e cable: https://www.expertelectrical.co.uk/6243y-cable#product-specs-dimensions

Some prices for reference, all per 100m drum from TLC (though Prysmian is around for similar): T&E : 1.0mm2 £33.50, 1.5mm2 £43, 2.5mm2 £64, 4.0mm2 £108, 6.0mm2 £162, 10.0mm2 £275, 16.0mm2 £428 3&E : 1.0mm2 £44.50, 1.5mm2 £61

You're a a bigger scale than my humble domestic needs. 6A at 48V is nearly 300W and that may be more than the entire house's lights. At 10m would be 20m of current path s twice as bad as you suggest. I'd hope that I would only have maximum 10no. 10W lights on at any one time so that's 100W total. If they were all at 10m on the same 1.0mm2 spur that would be (100W / 48V) * (2*10m) * 44mV/A/m = 1.8V --> 4% which is too high. I'm preferring the star network approach in which you have only single light or light groups on each spur. If a group they all get controlled together. I don't think any group would be more than 4 lights and so unlikely to exceed 40W. 1.0mm2 cable at 10m would suffer 0.7V --> 1.5% under that scenario which is OK, not great. I'd prefer to use 1.5mm2 cable as it isn't much more expensive (+37% at TLC). Interesting, I guess there is no equivalent to an RCD or RCBO as your toast too quickly for it to trip.

It would be foolish anyway to work on an automated light setup only taking note of the wall switch and ignoring the fact that the automation could have turned the light on. The number bayonet is a much better solution that a threaded bulb for this sort of reason. The threaded section of such a bulb is Neutral.

I'm concerned that the Mains Smart has Neutral side switching. Subject to confirmation, it looks to be allowed by BS7671 but still feels funny. It feels less safe than Line side switching as the light fitting will still have Line voltage present when changing a bulb with the light off, rather than in the normal situation when there would be Neutral present which is close to Earth potential.

Well here is a wiring strategy that falls back to a very standard approach by only rewiring back near the Consumer Unit. Nothing needs doing at the lights or the switches. The only thing that is weird about the "Mains Normal" end result is that it is a star circuit rather than a radial. That's because each light (or group) and switch is on a spur off a central set of bus bars. The Sense Input in the mains "Mains Smart" case is a bit tricky but could be done in a couple of easy ways: Relay switching e.g. 5V logic using a mains coil Mains optoisolator (e.g. Aliexpress, though that uses about 0.4W for the opto which is disappointing)

If going the more standard route, I had planned to do 3core&e between light and switch. You can now get smart switches that don't require Neutral (e.g. Sonoff). I guess they periodically charge a capacitor via the light at too low a current to light the bulb. May not work with all bulb types, I don't know.

For clarity and to stay on thread topic, all of the before diagram could be 48VDC (or 24VDC), as indeed could the after diagram. [Note: with 1.0mm2 T&E BEFORE: (2*10m)*44mV/m/A*(10W/48V) = 183mV --> 0.4%, AFTER: double that so 367mV --> 0.8% if still at 48VDC or 77mV --> 0.03% if mains].

This is perhaps a clearer drawing of that rewire star circuit (before and after): red and different blue for clarity, earth omitted but would be wired. The main downside of this type of star network is that there are two round trips for the current path to the CU and so the voltage drop is larger than if just one current path to CU and one to switch. Probably weighing up 2x10m vs 10m + 4m and probably comparable to a radial circuit that has longer current paths to the light as have gone via other lights rather than direct.

It is a lot of wires to pull, but in my case that doesn't take long as I have an open ceiling void atm. I count the following to the light in the before case: T&E from radial IN T&E from radial OUT (not for all) T&E to switch single&E to CTRL And at the switch: T&E to light (only E connected) bell to CTRL The bold is what a normal wiring would have. Your setup is need having only one cable to the light and only one cable to the switch. I get that you could revert your setup to something a bit more normal, but that would still be quite an unusual setup. As I understand it, my original BEFORE/AFTER drawing (3 posts back) is your layout with the addition of the light to switch cable. That drawing could be improved by pre-connecting the light to switch cable at the light end (but note, the blue wire would need a brown sleeve in the switch). This newer layout would allow it to revert to a totally standard radial setup with some wires left behind. The main issue I see (other than the amount of wires) is the bell wire shares the same containment as the T&E between light and switch. Sparkies would normally twitch at that (low voltage and mains sharing the same containment) but there are two things in its favour: All the bell wire (and the single& earth) would use permitted (aka 'safe') zones as if it was mains The CTRL would be in an earthed metal cabinet with appropriate labelling. If signals need to exit this metal cabinet for attachment to something (e.g. Raspberry Pi / PC) then an optically isolated USB connection could be used. If you have used T&E between your switch and the cabinet, then you could revert to a system without relays: At cabinet, connect one switch wire to Line At cabinet, connect other switch wire to SWL of light This would create an usual star circuit, but be easy for a Sparkie to understand. Here is a rewired example with just two lights, but more would just be similar: AFTER: red and different blue for clarity, earth omitted but would be wired.

In fact you could use Single&Earth (example, though more expensive than t&e so defeats the point slightly) for the automation wiring just to the SWL & Earth terminals of the lights and the rest of the light wiring is as per a normal radial. CTRL and 'Normal Radial' would have to share their Neutral so as to work with an RCD or MCBO which you would definitely want. Could even completely wire the light ready for the change and only need the switch rewiring. CAREFUL: couldn't be done in the no switch scenario as then there would be a hidden dangerous wire (though should only be Neutral as you wouldn't power the radial's Line). All of this will kill your cat!

If adopting any weird and wonderful wiring solutions, I think it would be wise to make provision for it to be changed to a more normal setup if required (e.g. selling the house). In the centralised approach, that would be a matter of running t&e or 3core&e from light positions to switch positions. In the future a Sparkie could come along and connect that cable to the light and the switch and then you'd have a more normal setup. Still not a conventional radial, but I think it would still be classed as a radial, just with one light on each spur and a mega junction box. The extra cost of running t&e or 3core&e from light positions to switch positions would suck but think of it as insurance. It could be done with cheap 1.0mm2 t&e so in my GF case cost 84m*£0.33/m = £28 of cable. If you allowed a Sparkie 30minutes for each one (generous) then it would cost about £20 each to rewire. Would be much harder is you omitted switches entirely. Then I guess you could still route the 'not connected' t&e cable to the appropriate location (in oval conduit with slack) for a backbox etc to be cut in later and hope that nobody damages it before it's needed as there wouldn't be a safe zone to remind you the wire is there. In the BEFORE case you could connect the Earth to the light and switch and so avoid the cable wandering off. For reference, I saw MK Logic Plus switch (K4871WHI) has the following Terminal Capacity: 4 x 1mm², 4 x 1.5mm², 3 x 2.5mm², 2 x 4mm², 1 x 6mm²

Of course that centralised approach to automation isn't limited to low voltage DC lighting. I have just done a radial lighting layout for the house GF and it uses 79m of 1.5mm2 t&e and 84m of 1.5mm2 3core&e (both could be thinner). I think with a centralised approach at the CMU that could all be replaced by 173m of 0.5mm2 t&e. If the centralisation was better (e.g. at hall), then I think that could be reduced further. The 84m of 3core&e is for switches and assumes 4m for each switch. If I didn't want to use wireless switches, I could use low voltage/current bell wire costing about 5p per metre or 50p per switch. 79m of 1.5mm2 t&e and 84m of 1.5mm2 3core&e 79m*£0.43/m + 84m*£0.61/m = £34 + £51 = £85 173m of 0.5mm2 t&e and 126m of bell wire 173m*£0.20/m + 126m*£0.05/m = £35 + £6 = £41 [£81 is using 1.5mm2 t&e for light] _________________ Cables: https://www.tlc-direct.co.uk/Main_Index/Cable_Index/Twin_and_Earth/index.html https://www.tlc-direct.co.uk/Main_Index/Cable_Index/Three_Core_Grey/index.html https://cpc.farnell.com/pro-power/bellwirecca/bell-wire-cca-conductor-100m/dp/CB14615 [though would prefer multi strand] Potentially dubious PCBS: 16way WIFI relay board £12 https://www.aliexpress.com/item/1005003297328162.html 16way USB relay board £14 https://www.aliexpress.com/item/1005001666658945.html 32way ESP32 relay board £110 https://www.aliexpress.com/item/1005004281943758.html [NOTE: choose correct version if doing own firmware]

Interestingly, if you adopted a completely centralised approach where all lights are directly powered by a centralised controller, you could probably wire it in 0.5mm2 T&E (which is apparently going to be available) if using 48VDC. Extending Table F6 to 0.5mm2, that should be 88mV/A/m. A 10W 48VDC light at 33m (66m round trip) would then suffer a voltage drop of 88mV/A/m * (10W / 48V) * 66m = 1.2V or 2.5%. I would expect that with a good choice of central control location, pretty much all lights could be <10m so suffer less than 0.367V --> 0.8% voltage drop. Another plus for 48VDC over 24VDC which would suffer 4x the percentage voltage drop. You'd have lots of long cables to run, but have only one central brain to which all switches communicated. A 10m 0.5mm2 cable would probably cost about £2.

Oh, I'd incorporate a thermal shutoff and PTC fuse. But as I say I haven't done enough thinking yet

Not so sure what you mean about thermal safety, but some thoughts (48VDC): Wiring: Similar current limits will apply, though perhaps a factor of Sqrt(2) will come into play (DC vs AC), so de-rate On Site Guide figure by say 33%. I expect if limited to 5A / 7A, you'd be OK in all installation methods with 1.0mm2 / 1.5mm2 and have up to 240W / 336W of light at your disposal with out any diversity, probably double with diversity. 240W is something like 30no. 8W bulbs or 75m of LED strip. DC:DC converters: A DC:DC converter could run with an efficiency >95%. If 90% efficient and a 9W light, you need to dissipate 1W of heat. Radiative heat loss alone would get rid of 1W from around 5cm x 5cm at 75C (22C ambient), but that wouldn't be the only heat loss method. [Note: 75C is hotter than the PVC allows] Convection if open will probably dominate and conductive if a spot will help. I haven't really run any proper calculations on this, but it feels manageable. Intelligence: Average power would be in the uW range when light not on. When light on, it would be dominated by the relay (if not MOSFET controlled) and likely 0.1W. Local Switch: like a Bluetooth remote so power in uW range. The low voltage itself goes a long way. What are you thinking of for the intrinsically safe? To have it as a product you'd have to do EMC testing etc.

I've just found this thread looking for one I thought I had started on 48V lighting but can't find. I have a couple of questions (I'm learning): Why 24VDC rather than 48VDC? Where do you get your bulbs? I have seen lots of 24VDC strips, but not bulbs Bulb fittings, I would like to use standard bayonet or threaded bulb fittings I am an electronics engineer and would like to go all the way and make my own custom driver/control PCBs and wireless switches too, perhaps even lights. Bluetooth or Zigbee rather than WiFi. This means I could have no wire running from the light to the switch (good for a retrofit) and with a Li-SOCl2 coin cell I can have 15 years of battery life. All lights can then be locally controlled by a wall switch, centrally controlled by home automation as well as phone/tablet controlled. If you are worried about cost, a light switch, back box, oval conduit and 4m of 1.5mm 3-core probably costs about £10 depending on tastes. A wireless switch and controller probably about the same off-the-shelf. I had thought to go with 48VDC to minimise voltage drop and because it seems to be the most common voltage used by server style battery banks (based on a very quick look). I see the main benefits of a SELV system (24VDC or 48VDC) as: I can play with it outside of Part-P and other regulations [TBC and for how long] A single high quality regulated supply in place of lots of distributed LED drivers each of which is lower efficiency, gets hot and has to be found space for. I can make my own lights. I have the technical capability to do this for mains too, but that would be much more of a regulatory headache. Designing my own 48VDC (or 24VDC) light bulb electronics would be quick, the hard bit would be the fitting. Automation opportunities. Plenty of this for mains powered lights too. You wouldn't start here. If you were designing a lighting approach from scratch (no prior restrictions) I think this is where you would end up. Safety (ish)