TerryE

Looking at your elevation, you are going to have major issues with solar gain for 3 seasons a year. Better to design your thermal solution now. Lots of discussions on the forum of the issues and approaches.

You will have to do the UFH calcs before going this route, IMO. It's a brilliant solution if you have a passive class house, but might not work for a typical minimum compliance thermal design. There are very few days a year when we need much more than 1 kW average to heat our reasonably large 4 bedroom house, so using a 3kW heater to heat the slab only needs to be on for 8 hrs a day. However, a more typical grand design with acres of glass at a U value of roughly 1 will require maybe 4 times that, so a large ASHP or gas boiler is pretty essential. You will need a buffer tank or TS simply because you can't pump the output of a typical boiler directly into the slab, and as you say you need to have a reasonable cycle time. I would have thought that a buffer tank + SunAmps would be a better fit in your case.

We had a fan assisted lower vent on our old larder. We found this quite useful because we could purge the larder. You could something like this one a simple timer so you could do this say 3-6am in the summer.

I am also not sure where the sweet spot for this is. If you want to stay native RPi then the RPi0 seems a better fit as an I/O processor. This NanoPi NEO4 is similar form-factor but in most ways outperforms the RPi3B (2Gb RAM, eMMC, USB3, Proper thermal management / heatsink) at RPI3B prices.

Pete, looking very good Just like the plans you showed us. Hope that the airtightness goes well. PS. one reason that I am posting is that Jan tracks my posts and she will be interested in this as well

Cutting off an old olive on a pipe tail is a total PITA, IMO. It can be done with care, but it just so easy to introduce longitudinal scratches on the pipe tail, and if the previous olives were brass then your pipe tail will almost certainly be slightly crimped. If you aren't going to reuse the existing tails and olives then my suggestion is to cut the pipe back and put in new tails with some Yorkshire straits or whatever. I normally use end-feed, but soldering 28mm in place is one occasion where I prefer the safety and tell of Yorkshire. @Nickfromwales, please correct me if I am talking out of my a*se.

Since we use SunAmps to provide DHW, we don't have any control on the HW setpoint other than the master TRV in the DHW manifold. As I commented above, we also keep or temp at around 23°C. At the moment, I set a configuration parameter in the the Database to change this, but I also have a browser dashboard that we can use to monitor the system and change it -- not that I want anyone to do so trivially, because it takes a few days for the house to settle down even if you want to change this by a degree. So functionally we do pretty much the same thing. My implementation is in software, but the advantage of having kids in the IT business is that this isn't an issue for us.

@newhome, My situation is somewhat different I've got a couple of sons and a son-in-law who understand programming and the last also does this stuff just to the chagrin of my daughter -- "I've just ended up marrying a copy of my dad!" (And she's done a bit of programming in her time but has made a point of selective amnesia here ?) But coming back to Jeremy's point, I think that there is a flaw in this argument: a passive house like yours or mine is a completely different beast than a conventional one. The time constants and rates of heat loss are just in a different league. Unless the control system is well-matched to the system it is controlling, the net result won't be stable control.

Like Jeremy, I have a lot of instrumentation one my system, but this is largely to enables me to characterise how my house as a system behaves. the thermometers are daisy chained on a One-wire bus and cost pin money. All of the smarts are done in one RPi. Most people find using software-based systems daunting, but I've used dozens of languages over 50 years of programming so this is second nature to me. And the nice thing about software base control is that you don't have to start buying extra control bits and rewiring stuff. In Node RED, you mostly drag and drop inside a browser window. I've pretty much got to the point where I can write all this up properly for others. Most aren't going to be interested in the journey to where I've ended up where we are, but more just as simple an explanation as possible of the end point and some understandable supporting data on how the system works.

@Ed Davies, sorry -- I've just cross posted. As I said I do collect and log the daily local met office forecast, but we have found so far that we don't need to use it. If I do see evidence of our simple algo not tracking well enough then I will add a feed forward term. I used to do this sort of control stuff on some of my work projects, so find it interesting but I am also a bit time constrained at the moment. It's the sort of refinement that I can make any time of the next few months when I have the free time.

@dpmiller, I will expand Jeremy's comment: If we have a Δt of 8°C and don't compensate for it, the time constant of the house will mean that the house will drop about a ¼°C as a result. There is also a decrement delay through the walls; I saw this in my modelling scenarios: if you and a step function in the external temperature, it takes a couple of days for this to make its way through the stone skin and walls. If the slab temperature drops (or rises) from the computed set point then the daily algo tops up or trims the daily heating amount. At the moment, I have this step capped at 1hr, so it would take 3 days to step from 5 hrs heating to 8 hrs or v.v. but I find that this is quite responsive enough. The system also takes in the weather forecast from the met office each night and I could plug this is as a feed-forward term. Next month, I am adding a thermometer in the air-gap between the outer stone skin and the timber-frame and might use that instead if it is a cleaner estimator. Using this sort of data as a feed-forward term could drop the variability even more but this gets me into further characterisation, Kalman filtering etc. and I have other priorities at the moment. @joe90, this is a case of personal preference, and one that can only be developed from experience. Jan and I started out thinking along the same lines as you, but we've come to really like the stable temperature. OK, we end up working in T shirts is we are doing something a bit physical, but that's what we do in the summer anyway: you just have to get used to the fact that the house internally is like the summer all year round -- and we have. You say that there is nothing worse than a cold bathroom in the morning, but that's my point we don't have any cold rooms any time. The house was at 22.6°C last night at midnight and at 23.0°C this morning when we got up. After my old draughty and hard to heat farmhouse, this is just heaven. Yes, we could have the whole house a couple of °C cooler and save maybe £50 p.a. on our Willis heating costs but our personal preference is that we'd rather pay the extra £50 a year and be able to wander around bare footed and in T shirts.

@MikeSharp01, Mike I find myself wanting to raise a mild dissent with your post. For those not familiar with what this is all about and have some mathematical toleration, try the Youtube Computerphile videos by Mike Pound on this general topic. Amazing stuff. However, I see this is more of a classical control systems issue rather than a deep learning one. Deep learning requires huge amounts of training data across a catholic set of scenarios, and this just the case for the thermal control of a passive class house. In another topic @Ed Davies has given a working definition of time constant and in our case at a ΔT of 15°C our house loses about ½-1°C/day with the heating off -- let's round this at the worst end, and this gives a time constant of ~15 days. IMO, even if it was a third of this, it would still make no sense to me to try to control it with a bang-bang system that can react in seconds. Yes, with some careful crafting you might be able to get this to work, but my thought is really why on earth should I bother? At the moment, my Node RED system runs a set of calcs at 17:00 each day to work out how much heat is needed for the next 24 hrs and then sets up a simple two period ON/OFF execution plan to add this heat. (though it is going to have to get about another 3 or 4°C colder for the second period to be used). I do have a roomstat now -- or at least a DS18B20 sitting inside one of the internal partition walls against the backside of the plasterboard and measuring the average room temperature, which shows a ripple of about ½°C over the day. My day-to-day 17:00 slab temperature does wonder slightly -- I have been refining the algorithm and the 1-σ is currently about 0.3°C; with this stable performance and the cost of the entire H/W control system a few hundred quid, I am not sure how deep leaning will help me refine this. The system can also be trivially adapted for ASHP heating the slab directly, so future-proofing for this isn't an issue either.

@lizzie we spend 3-4 months a year on one of the Greek islands (when we're not busy building houses that is), so we also keep our house warm -- typically a base temperature of around 22°C. This is about 1°C warmer than when we first moved in last Dec, but we find that we are comfortable with this. Even using our little Willis (and mostly E7), the cost of the extra heating for this °C increase is around 15p /day for 6 moths a year say or £27, so it's in the noise and this little hike means that Jan can wander around bare-footed and in a tea shirt all year round. (Whilst this £27 might seem a relatively high increment, our overall energy costs are so low that we are still debating whether we can make the numbers work to justify the capital and installation costs of installing an ASHP.) The issue comfort variations is an issue -- so like you when sitting quietly in the living room in the evening we find that we prefer the room to be a degree or two warmer, but at the moment our answer is to put on a thin jumper in the evenings. We do have a level of natural zoning in that our 1st floor is about 1°C cooler than the ground floor (we have no heating at all upstairs), but this suits us. @joe90 In my case it's not just Jan that I have to train; it's myself. If I am going to the shed to pick up a tool, then I know that I should close the back door behind me as I leave the house and reopen it 1 minute later when I return. Leaving it ajar for a minute is just sloppy, but I already use the utility as an air lock so I probably lose less than a 1p's worth of energy -- life is just too short; I find it easy just to accept that the overall temperature in the utility/ toilet will drop by a degree or so and recover over then next 15-30 mins. IMO, there is nothing wrong with having a room stat so long as you are aware of the potential limitations resulting from placement, and put it somewhere that represents a sensible aggregate environment -- such as an inner corridor, and avoid putting it in a room with an external door or a room where you will be generating excess heat such as by cooking. There is also a fundamental difference between my and Jeremy's approaches for slab measurement: he used spot probes and I use the averaged temperature of the UFH returns which effectively provides an aggregate temperature averaged across the whole slab. If you zoom into my graph above then you will see that there is 2 min sampling quantisation noise, but apart from that the reading is stable to better than 0.1°C

Out rear utility room is basically a corridor with the toilet on one side. with the door to our central hall closed it is still like a fish tank with one side missing: open the back door and there can be a 10-15°C temperature drop across the doorway. The hooter air will spill out of the top of the door and the external cold air flood into the room at ground level. The mixed air in this fairly small volume will still be a degree or more colder as a result. @lizzie, your approach doesn't really work for our house: it is basically like a Thermos flask: all of the insulation is on the outside so if you don't have open doors, then the temperature is amazingly uniform both spacially and in time: If I turn the heating off entirely at the moment then the house as a whole cools by ½-1°C per day.

I don't really use an environment thermostat as an active component in my CHS. The following graph might help to explain why. I've got temperature sensors all over my system, and plot shows the two sets of readings: one set for the Ground-floor toilet which also houses my SunAmps, manifolds and Willis; and one set for the average return flow temperature for the slab. These are for two days: the 4th and the 6th Nov. So some explanation: The reason for the big humps (peeking at 7am and 7pm) are that my CH calcs put in a primary heating pulse in during the E7 off-peak period (00:00 - 07:00 for us), and if needed an extra top-up in the late afternoon. The outside temperature leading up to the 4th was really cold here and the top-up was needed, but it had warmed up by the 6th and so it wasn't. The small saw tooth is because I use the on-the-hour temperature as an accurate slab temperature. I hae found by trial that if I run the UFH pump, then the return thermometers stabilise after 6 mins or so, so once an hour (at 54mins to the hour), I turn on the UFH pump for 6 mins and then use the on-the-hour reading as reference. This 10% duty cycle is enough to get a good slab reading and reflow any local solar gain around the slab. The toilet thermometer reads slightly higher than the other rooms because there is a little residual heat loss from the SunAmps, etc. The ripple over the day here is about 1¼°C where it is nearer ½°C elsewhere. When we are not in the house, this reading is pretty stable (as on the 4th when we were visiting Bristol). But when we are in the house then I see big occasional drops, and the drop is aligned to one of us leaving the back door open to go out to hand out washing or to get something from my shed. My point here is that I have real problems using local air temperatures as an estimator for overall house temperature. These are just far too responsive compared to the actual overall house temperature. If you put the thermo in the hall then just opening an external door dumps enough heat to drop the measured temperature a degree or more. Any cooking does the same in the kitchen as does 3 or more people just sitting in a room. Using a single wall mounted thermo as environment estimator is just too noisy for me to use sensibly: I don't want my UFH system turning on just because I've opened the back or the front door for a minute. One option would to add a low -pass filter (putting my sensor in a box, or behind the plasterboard), but at the moment I don't bother and do everything based on the slab profile. As a footnote, this is more of a geeky debate, hence my posting this here rather than in one of the heating forums. However, if the mods feel it fits better elsewhere, then please move it.

I keep humming and haring about doing something like this but to be honest, my Pis typically run at a few % utilisation and well within their memory constraints. There are lots of reviews of these RPi super-clones, but when you them look at the forums, stability and project / community support is always an issue. I find myself saying -- let's wait another 3 months to see if something better comes along. Certainly, it's a lot easier if you only want headless server support and for most people DietPi makes configuring such devices very easy. My main two RPis use SSDs for primary storage and eMMC seems to be an excellent way of avoiding this as an issue. The SBCs that I've seriously considered are the Odroid C2 or the Asus Tinker S (must be the S version to have on-board eMMc). There are better performing SBCs out there and being reviewed, but buying them is another issue.

Barney, You seem to imply that a SA option is the only option for electric only. I don't understand this logic at all and it doesn't make much sense to me. The slab is a bloody great and very cheap thermal battery. So why heat it via the SunAmps? This seems an odd decision to me. OK, if you've got lot's of PV, then there might be a couple of months a year when you have excess over your DHW needs that you could use the SAs to top up the slab during E7 peak periods, but the payback in extra instrumentation and control costs is going to make this a marginal design choice. Far better to have a low temperature ASHP loop direct (say at 30°C via a PHE) into the slab where you can get a solid CoP, and where you are operating the ASHP is a gentle regime. IMO, the 50-60°C PCM isn't matched to ASHP operation as you will been to run it flat out and at a crap CoP. 18kWh just isn't enough to heat a reasonably large passive house as the temperatures fall into single digits or lower. My SunAmps aren't even used in the UFH circuit. Ed, A relay might be able to switch low voltage AC or DC at 30A, but definitely not at 230V. You need special relays to do this 230V switching; they are called contactors not relays (even though they are a type of relay). This issue is the make / break arcing will knacker the contacts of a normal relay and so they rapidly burn out. You need a special type designed to take this type of contact -- hence the name. Designing these circuits is not for the notice because of the back EMFs that can generated during make / break. These also typically need 24V coil voltage to operate them so you end up cascading them with smaller relays, SSRs, or a Darlington driver. At Jeremy's suggestion, I just use 230V 20A SSRs which are driven by 5V TTL logic. A lot easier. These SSRs delay the break timing up to ½ a cycle to position it at a voltage zero-crossing, so there is no instantaneous power at the make / break. The only downside is that they are not 100% efficient, and so they give off 30W or more each when operating so you need to be careful about thermal design / management of any enclosure. PS. I find that the ½ hourly usage data from my supplier is enough to track the power demands.

Also remember that with most passive houses - at least out MBC ones that use cellulosic filler - that the decrement delay is so long that spot lows are not really relevant; it is the daily average that's important. Last winter / spring was colder than average, and I don't think that we had a daily average below -5°C. My house is larger than Jeremy's and a marginally lower spec, but even so based on actuals I am still at 1.85kW at this temperature, so only having one Willis isn't a real issue for us.

You do need to be very careful when ordering timber. The quoted sizes are nominal, and can be less in practice. I ended up oversizing and using my planer/thicknesser to size all of mine. Linings and architrave full to the floor. The tiler / fitter can multitool off any excess of need for fitting.

@Gav_P, yes, at the moment we only use the Willis. We have prepared for adding an ASHP, but it will only be value for money if I self-fit a budget one. As Jeremy says I am looking at a budget of around £2K, with a payback of ~3 years. But that assumes that I will be doing my own maintenance, and not paying a third party.

I have mine mounted vertically across my UFH manifolds. By the time that we are down to a sustained external temp of 0°C we need about 8hrs a day to keep our largish 4 bedroom house toastie. (Each deg C drop adds about 40 mins on-time to keep the house in balance). I will be fitting an ASHP next summer because the numbers work and it will be good for peak summer cooling, but the Willis will remain an essential backup component. Why? Because the ASHP won't be a mission critical component in our heating system, so I can take a far more relaxed attitude to maintenance / spairs, etc. At the moment, my Node RED system does a daily calc using the weather forecast, slab temperature profile, etc. to work out the heating profile for the next day - typically a solid block overnight during cheap rate and and extra top-up late afternoon if the weather gets cold. No other feedback. Dear simple. Good enough to keep the house temp better than ±½°C which is good enough for me.

As far as the external power bus goes, my current thinking is to use something like 20awg multistrand twin. Tuofeng do a decent cable that gets 4½+ ? reviews. This has a resistance of 1 ohm / 30m, which gives an acceptable diversity for the size of my garden. This will get enough power to any of my external projects, including monitoring my supply meter.

5V is a bit marginal unless you go for heavy duty cabling. 60W of devices will require 12A and remember that at a PD of 5V, the diversity will become a big issue. IMO, better to use 12 or 24V and per-device Buck step-down DC-DC convertors (which cost pin money) to drop the voltage to 5 or 3.3V as required.

I think that we've split the discussion in to to separate usecases: The ability to collect periodic aggregate power usage - say every half hour -- mainly so you can automatically monitor and plan your ToD usage and overall power billing. The ability to collect fine grain usage so you can get a perspective on individual device usage. Jeremy's solution is very elegant but is also invasive in that it involves intervention in the primary power circuit, and this is something that I would prefer to avoid which is why I tend towards pulse counting as this still gives me 0.001kWh granularity. In my view, the whole issue of designing power constrained devices is interesting. A typical gaming PC will draw perhaps 300W, a modern laptop with SSD perhaps 20-50W depending on the processor, a RPi about 10W. A pair of NiH battery might has ~ 5000mAh capacity and if you want the batteries to last 3 months, say then the average draw of your device must be around 1.5mA. Whilst ultra-low power devices do exist, many active components have non-trivial draws. For example the PCF8593 has a typical draw of 300mA; the LM393 200mA; the light sensor diode 100mA, so the only way that you can run an device for months on a battery is to do a repeat of <sleep a long time at 10s of μA> then <wake up for a short time and do some measurements>. So IMO, battery powered devices are limited to a small number of periodic monitoring applications. But it is pretty straightforward designing IoT devices with a ~1W power draw. This is why I am currently thinking of stringing a 12V supply around my house exterior and garden to power my devices as needed.

ED, TerryE is also one the core developers of the NodeMCU Lua firmware for the ESP8266 / ESP32, so I am also a very useful resource to know when it comes to programming ESPs and this is also why I prefer to use them over Atmel or RPI0 devices for IoT applications Deep sleep on both the ESP8266 and the ESP32 powers down everything apart from the RTC and a small RAM scratch pad in it. Wake from Deep Sleep is effectively a cold boot, except for any context saved in the RTC RAM and in the Flash SPIFFS. The power draw here is ~20μA. I have yet to check whether pulses are at 0.001 or 0.01 kWh but if they are at the former then a power draw of 30kW through the meter (which is entirely possible at peak) will result in 30,000 pulses per hour, or one every 120mSec or so, and a more typical average busy draw of 3kW will still be waking the ESP every sec or so I need to benchmark how quickly the ESP comes out of deep sleep, but this will be tight and still have high peak draws if it powers the Wifi modem. Perhaps a better alternative here is the light sleep option which draws about 1mA and this keeps the CPU registers and RAM powered but shuts everything else down. GPIO events will trigger wakeup from from light sleep and wakeup here takes under 5 mSec. The big advantage of using a separate chip to do the counting is that you only need to wake up from time to time -- say once every 30mins, interrogate the counter and go to sleep again. @MikeSharp01 Looking at the DS1371 datasheet, it doesn't look as if it can be programmed in the mode that I want. It seems to be driven off the mast Xtal and the counter provides a watchdog function, viz it triggers an event after a preprogrammed number of ticks. Of course there is always the fallback option of using an ATTiny85 to do the event counting, which is even cheaper to implement and will bring the overall power into the battery domain, but this is going to involve getting into Arduino development -- ho hum.