I said in another thread that I would put together this post as a chronicle of clean sheet to where we are now, so here it is.

It has been quite a while since my wife and I decided to dip our toes into the world of solar panels and sustainable heating, and boy what a rabbit warren it has been. The reason I've decided to post this new thread is to take our initial install as the starting point and give you an idea of where we've gone from there in case it helps any of you. I'll be bringing in some other topics too where they have a bearing so be prepared for a bit of a "War and Peace" post.

So firstly a bit of background. As you can see from my signature, our home is a bungalow built around the 1920s and since extended by previous owners to give a current floor space of around 105 square metres. It is far enough off the beaten track that whilst it has mains electricity and water, that's about it. No mains gas, no mains sewage and surrounded by a small number of absolutely lovely neighbours and lots of fields and woodland. For us it's idyllic. For someone who likes shops and theatres just round the corner it'd be a nightmare.

When we first moved in the house was heated by a Worcester Bosch oil-fired boiler and there was a 1,000-litre oil tank in the front garden. That needed to be filled twice a year and, with prices as they were then, this meant heating and hot water cost us around £1,300-1,500 per year. Cooking was - and still is - dealt with by means of two 47kg propane bottles which, once again, last us just over a year. Finally, we had a wood-burning stove capable of using the plentiful (once seasoned) supply of firewood from the surrounding area.

My wife and I recognised the oil boiler was coming to the end of its useful life and we didn't want to replace like for like given the patent good sense in trying to look after the planet. We were in a position to be able to spend more than the cost of a replacement boiler and so looked around at what might be a good option. We also thought some solar panels might be a good idea and so researched them too, but weren't sure which we would be likely to do first. In the end, a bit of an expensive hiccup with our boiler convinced us that heating was a higher priority than generating energy and so we concentrated on narrowing down our options; the impending cessation of the RHI scheme also added a degree of impetus to our search. After much eliminating we decided an air source heat pump was the way to go and, once again after discounting several potential suppliers, we found Doré Woodman based in Gatwick. They did a lot more listening than talking and so came up with a plan we felt really did suit our needs with the result we booked an installation ready for December 2021.

Then our local County Council decided to team up with Solar Together to offer a bulk-buy scheme to reduce the cost of the kit and a tendering process to choose an installer - Infinity Renewables in the end - for all the interested people. The cost saving for us meant that effectively we could get a 6kWh battery for free compared with the quotes we'd been getting before so we bit the bullet and said yes to that installation too. As luck would have it, a few customers dropping out meant we were able to be fitted in earlier than expected - December 2021!

So both air source heat pump and solar PV were being fitted at about the same time. I've already mentioned on other threads about our experiences but to summarise I'll simply say that Doré Woodman were outstanding and heartily to be recommended whilst Infinity Renewables were workmanlike; they got the job done but I would neither recommend them nor suggest they be avoided. What it meant, however, was that we ended up with:

• A Mitsubishi Ecodan air source heat pump controlled locally by a Mitsubishi wall unit and controlled remotely by means of the Melcloud app (both web app and phone app).
• A 6.5kWp solar array.
• A 6kWh Growatt battery.
• A Growatt inverter controlled remotely by means of the ShinePhone app or the Growatt web app.

Having already switched to Octopus Energy as our electricity supplier, we inevitably set up an export tariff with them, choosing the Octopus Outgoing Agile tariff. A bit of a runaround with UK Power Networks meant that this wasn't up and running until March 2022 but this still allowed us to be operational for the summer months. This setup, operated largely as delivered, provided us with a significant saving on our operating costs with our heating and hot water bill going down by about a third but then about £500 being paid back to us for electricity exported. The RHI reduced our ASHP installation costs down to about £4,000 net (into the realms of an expensive gas boiler and tank replacement that we'd largely have had to spend anyway) and the £7,500 we spent on the solar PV was easy to see reflected in an increase in the value of the house.

After that first year of bedding in, we started looking at areas we still felt could be improved. The first of those was to look at our tariffs and, whilst we were very happy sticking with the Agile export tariff we decided to switch to an Agile import tariff as well. At the time we were paying about 46p per kWh imported and whilst we were being paid handsomely for exports (at times in the region of 35p/kWh) that was only when the sun was shining. Switching to Agile throughout has meant a reduction to a net average price per imported kWh of around 20p.

All well and good, but my wife and I identified several areas where we could do with using this new kit a bit more intelligently.

1. The ASHP was set up already to use weather compensation but was there anything we could tweak to significantly improve its cost effectiveness?
2. The solar PV solution was already set up so that it would follow a hierarchy of generation and consumption i.e. if it was generating it would fulfill our consumption first, then pile excess into the battery and then, if the battery was full, export the remainder to the grid. Similarly, if it was not generating it would allow us to pull from the battery first and only fall back on the grid after that. However, it was not reacting to times of negative energy prices and it was not husbanding the battery charge to fulfill load at times of highest prices.
3. We were sure there were ways in which the kit could be operated differently during times of sunny weather and times of overcast weather which would maximise the benefits and minimise the costs.
4. We knew we could get all manner of information from Octopus about historic and future prices, about historic imports and exports, about current and upcoming saving sessions, plunge alerts and so on. However, we had no way to make use of this since neither Mitsubishi nor Growatt provided any equivalent to GivEnergy's Octopus integration.
5. We knew it was perfectly possible to retrieve all manner of environmental information from the Internet including weather forecasts and so forth. Once again, could we integrate that information in with the other data to produce something significantly smarter than what we had already?

Contrary to the views of several well respected members on this forum, I was not significantly concerned about the kit communicating with manufacturers' cloud solutions as long as it didn't involve my or my family's personal data being transmitted. As such, I was in no great fight to try to ensure all measured data stayed locally.

Our first step was to create a custom schedule for heating the hot water. My wife reasoned that during the summer it makes little sense to heat the water at night time rates after someone has had a shower just before going to bed (or early in the morning) when you can let the water in the tank stay cold until around 10am and heat it using home grown electricity. I have yet to measure the effect of that but I think it was worthwhile albeit not a huge inroad. It did, however, underline that our thinking was now moving to how best to get the different bits of kit working in unison rather than just how to tweak each separately. As a result, we decided the sensible next step was to get some form of home automation system; a central brain, if you will, that could interact both ways with various different pieces of kit and perform logical changes based on different scenarios. In IT speak, that's IFTTT (If This Then That).

There are several contenders for home automation systems but many need one to buy a hub and many others are rather limited in their support across a range of brands. The one that stood out to us was Home Assistant based on its ability to integrate with a huge range of manufacturers' products. Being open source, it required me to download some stuff and actually install it, and it needed me to have a chat with my work to see if they had an old PC that was being replaced (they had and were kind enough to give it to me instead of scrap it) so that I had something to install it onto. After that it was all pretty straightforward to get up and running. The idea with Home Assistant is that if you have something you want to monitor or automate you tell Home Assistant to install an integration for that device, so I immediately installed the Melcloud integration meaning Home Assistant could then do what I can do with the Melcloud app.

Adding in integrations for Octopus, a couple of weather forecasts and a solar generation forecast (who knew one existed, eh?) meant we were able to get plenty of the kind of data needed to make decisions and we just needed to integrate the solar PV. Unfortunately we found Growatt were making regular changes to the way their systems work which, whether intentionally or not, were frustrating the integration's ability to connect at all. Moreover, Growatt's web and phone apps are fine for a visual representation of what's going on but despite having functionality to change settings manually that functionality doesn't always work - I can't even change the inverter's date and time! No integration, no control, no automation.

At this point I should explain that my and my wife's jobs rely heavily on our computers and our Internet connection. As a result, instead of just using an ISP's router and WiFi and hoping for the best we've got something a bit more robust. We have some structured cabling installed (network sockets in the walls with hidden wires routed back up to a central point) just as you might expect in an office building and that all comes together at a patch panel in a small network cabinet in the loft. Also in that network cabinet is our network switch, meaning we can plug our computers into the wall with a network cable and attach to the local network and the Internet at far better speeds than WiFi would allow. This is all relevant in this posting because of where the inverter is installed - outside.

The inverter itself is a Growatt SPH5000 and the installation manual shows that it comes with several interface connections inside a waterproof cover, allowing the inverter to communicate with other devices - the battery is an obvious example.

The inverter also has a WiFi module connected to the serial port allowing the inverter to communicate via our network back to Growatt's servers. I hoped I might be able to in some way talk with this WiFi interface but all attempts tended to suggest communication is one way - the inverter pushing and pulling, not something else pushing to and pulling from the inverter. This meant my only remaining choice was to talk with the inverter using Modbus. I won't go into detail about what Modbus is since that has been covered by other threads on the forum, but suffice to say it needs a physical connection to the inverter so there's some line over which to talk.

I was fortunate enough that the installers left the 485 interface free. The 485 interface itself is presented as an RJ45 socket so it is possible to plug a standard network cable into it. However, whilst you might well use a network cable, it is most certainly not a network interface and if you were to plug the other end of the network cable into a network switch or router you would kill the port you plugged it into, so DO NOT do that. Instead, think of the RS458 port as a weirdly shaped serial port which, in fact, it is. Given the inverter and battery are installed outside, I couldn't just put a Raspberry Pi or similar next to it and simply wire it into the RS485 socket. Similarly, my home automation box currently sits in a completely different location. Net result is that even with a long cable it would not be practical to just plug one end into the inverter and the other into a computer serial port.

My solution was actually simple even if it sounds complicated. A long network cable with an RJ45 connection on one end was plugged into the RS485 socket on the inverter. That cable was routed in a nice hidden way back into my loft where the other end of the cable was wired into the patch panel. In effect, that meant that the inverter's socket had been "moved" into my loft without moving the inverter itself. From there I plugged another wire into the patch panel and could connect the other end of that wire into either a computer's serial port or, as I did in the end, an RS485 to Ethernet converter. The one I used was a WaveShare box as shown below

The awkward bit was that there are eight wires in a standard network cable and only two wires needed for the Modbus connection. What little documentation was available on the Internet was quite conflicting so there was quite a bit of trial and error to find which two wires were needed and which way round to connect them. For avoidance of doubt, I found the two wires are the middle two (i.e. wires 4 and 5) and wire 4 (normally the blue wire) connects to RS485B terminal and wire 5 (normally blue and white) connects to RS485A terminal. In addition, I needed to spend some quality time in front of the inverter itself pressing buttons to change the RS485 port to VPP mode. With this done, I was able to plug the converter into my network switch, follow the documentation to access its configuration and do the sensible stuff (give it a static network address and change the admin password). I also changed its serial settings to slow down the baud rate to 9600 since I'd read Growatt boxes aren't happy with rates higher than that.

Finally I turned my attention back to Home Assistant and installed the Solax Inverter Modbus integration and configured it to connect to the WaveShare box. Once again, for avoidance of confusion the details are that the inverter type was "growatt" (lower case), the modbus address was 1, the IP address was the one I gave it and the mode was "Modbus TCP"; I wasted a lot of time trying various combinations of wires and addresses whilst choosing "Modbus RTU over TCP" since that's what was suggested in other helpful posts, but "Modbus TCP" did the trick.

That leaves me now with a Home Assistant server that:

• knows what my past, current and future Octopus prices are for import and export,
• knows what the weather is predicted to be,
• knows what the temperature is outside
• has reasonably comprehensive control of my ASHP (knowing room and DHW temperatures and being able to tell it to heat or not) and
• has proper control via local connection of my inverter (knowing how much solar is being produced, battery state of charge and so forth) with the ability to decide where the generated electricity should go and whether or not to discharge the battery to the grid or not.

Phew!

It so happens there are also integrations for other things as well and I use Home Assistant to monitor my network and other fun stuff as well. I am now happy I have a modular and extensible system that can integrate all my smart devices as comprehensively or not as I want - CCTV, propagation lights attached to smart switches, printer and so on. My next job is to actually put in the rules to decide exactly how I want the weather, the time of day and the fluctuating energy prices to influence what my solar PV and ASHP do, but that's another story.