Thoughts on the subject of an accessible open-source DIY solar PV diverter

[Radian]

@billt @NSS I have started this new topic In response to a discussion going off-topic about the prospects for there being a step-by-step guide to constructing a DIY solar PV diverter suitable for almost any skill level, using off-the-shelf modules.

 

In trying to think up a design I concluded that if  there was a plug-in Tuya style SSR with low-latency, zero-crossing switching packaged with a 13A three-pin outlet (yes I know the 13A problem here but hold on...*) then anyone who can put a 13A plug on the immersion lead could do the heavy electrical installation work without calling on a sparks.

 

The rest of the device could almost be achieved with a few more plug-and-play parts and a software download. A Raspberry Pi Zero W with an Audio Bonnet is very close to providing the remaining hardware along with a CT clamp and AC transformer. These sources would be sampled by the audio interface to measure current and voltage at the meter tails. Galvanic isolation is provided by the CT and transformer and WiFi connectivity provides the remote switch. The almost caveat is entered around how to get the signals from the CT and AC transformer into the line level audio inputs without relying on soldering skills and external components. 

 

 

If these obstacles could be overcome, what we would end up with is a two-part unit with remote switching (so the export measuring can be separated from the immersion location) which is at the more premium end of the commercial offerings costing around £500 - for something more in the region of £50. All firmware could be baked into an ISO image and written to the SD card for turn-key operation using the Raspberry Pi foundation's application.

 

All comments welcome.

 

*As for how a full 3kW load could be safely plugged in to a 13A socket, the load switching would deliberately not be continuous - so in the same way we can plug in a 3kW kettle which only boils for a couple of minutes (try repeatedly re-filling and boiling a full kettle and see how hot the plug gets) the load can be de-rated to an arbitrary degree. 2kW being common for domestic appliances and secured with a 10A fuse.

[SteamyTea]

39 minutes ago, Radian said:

A Raspberry Pi Zero W with an Audio Bonnet

My experience with RPi ZW is that some are very reliable, and others are just hopeless, so could it be replaced with a ESP2866/32 or whatever the latest incarnation is, then it is not relying on all the other stuff the operating system is trying to do.

The ones I have already have a analogy to digit converter built in, as well as wireless/bluetooth.

Easier to load an 'image' onto as well, no need to much about with TTL.

Possibly require an easy way to connect it to a network, push button maybe?

 

  

39 minutes ago, Radian said:

The almost caveat is entered around how to get the signals from the CT and AC transformer into the line level audio inputs without relying on soldering skills and external components. 

Sell that bit, but with the option to DIY it as well.

 

 

Edited January 31, 2023 by SteamyTea

[Radian]

1 hour ago, SteamyTea said:

My experience with RPi ZW is that some are very reliable, and others are just hopeless, so could it be replaced with a ESP2866/32 or whatever the latest incarnation is, then it is not relying on all the other stuff the operating system is trying to do.

 

The main reason I suggest using a pi zero is that the firmware can be written to an SD card on a PC using techniques already quite familiar to people. Literally all you need is a card reader/writer which some laptops already have. I think the reliability issues come down to the operating system and how it's set up for use with an SD card. Steps can be taken to significantly improve reliability such as running from RAM and only using the SD as read-only at boot. The nuclear option would be to bare-metal code the required functionality onto the ARM chip i.e. eliminate the OS altogether. Then it becomes a fancier ESP. But still programmed by writing an SD card. Quite a bit of work though.

 

If there was an off-the-shelf ESP32 with a card reader that came with a bootloader for it, that would be good too but I don't know of anyone making that. I'm not planning to make anything available myself. The exercise here is to see if readily available stuff already on shelves can be pulled together in to a fully specified diverter.

[SteamyTea]

3 minutes ago, Radian said:

ESP32 with a card reader

Can get a SPI card reader for the ESP2866, so would assume there is one for a 32.

 

I do like off the shelf stuff, Lego mentality.

[Alan Ambrose]

>>> plug-in Tuya style SSR with low-latency, zero-crossing switching

 

It would be helpful if it switches cycle-by-cycle so you could have all the way from 0 to 100% of a 3kW immersion in, say, 2% steps. Two drives would also be useful for those of us that have 2 immersions in one tank. I think the electrics part (earthing / switching / fallback / wiring / temperature measurement etc) is more tricky than the electronics part - but then I'm an electronics/software guy.

 

Current transformer sits in the meter box? Another on the PV inverter? Need to convert the CT signals to digital to send over wire/wifi?

[SteamyTea]

Just remembered that I have in the past used these header pins that don't require soldering.

 

https://shop.pimoroni.com/products/gpio-hammer-header?variant=35643318026

[Radian]

23 minutes ago, Alan Ambrose said:

It would be helpful if it switches cycle-by-cycle so you could have all the way from 0 to 100% of a 3kW immersion in, say, 2% steps. Two drives would also be useful for those of us that have 2 immersions in one tank. I think the electrics part (earthing / switching / fallback / wiring / temperature measurement etc) is more tricky than the electronics part - but then I'm an electronics/software guy.

 

Yes. The measurement algorithm sums the power (+/-) at least 128 times per half cycle. This updates a Joule counter for every cycle and when an export threshold is reached emits a signal (by UDP or similar low-latency communication) to the switching device(s) to energise. The effect the dump load(s) have is naturally factored into the subsequent power measurements at the meter tails and when the Joules come back in over the import threshold, the load(s) are de-energised. Any number of receivers can play as the only feedback comes through the actual diverted loads imposed on top of the rest of the house.

 

56 minutes ago, Alan Ambrose said:

Current transformer sits in the meter box? Another on the PV inverter? Need to convert the CT signals to digital to send over wire/wifi?

 

The minimum system only needs a CT in the meter box. It's only interested in nett import/export so agnostic about inverters, batteries etc. It's just trying to mop up anything leaving the premises. It's not even necessary to monitor the immersion load with another CT as this can be inferred from the power delta seen at the measuring device. If the immersion stat switches off, it will be evident that there was no response to the dump request and a second tier switching signal (to storage heaters etc. ) can be issued.

 

UDP broadcasts can reach anywhere on the LAN so any number of SSR WiFi sockets could participate. You should be able to see that it really all hinges on finding such a beast. I scratch-built a diverter using a Triac driven by an ESP32 but I'm used to building 240VAC stuff. To remove this hurdle, a commercial SSR with WiFi provided by a Tasmota compatible ESP might be adaptable via Tuya Convert built into the Raspberry Pi. This is what I'm thinking.

[Bramco]

6 hours ago, Radian said:

I have started this new topic In response to a discussion going off-topic about the prospects for there being a step-by-step guide to constructing a DIY solar PV diverter suitable for almost any skill level, using off-the-shelf modules.

 

There already is one - with all the theory -> https://docs.openenergymonitor.org/pv-diversion/introduction/choosing-an-energy-diverter.html

 

I've had the software for the MartinR version running for several years.   You do need to make the SSR board but it's not rocket science.

 

Simon

[MikeSharp01]

Will this help 3.5mm Jack to screw terminals. 

https://www.ebay.co.uk/itm/174582857145?chn=ps&_trkparms=ispr%3D1&amdata=enc%3A1TemzJXDnShykDZcP_jD01A38&norover=1&mkevt=1&mkrid=710-134428-41853-0&mkcid=2&mkscid=101&itemid=174582857145&targetid=1647205088800&device=m&mktype=pla&googleloc=1006886&poi=&campaignid=17206177401&mkgroupid=136851690655&rlsatarget=pla-1647205088800&abcId=9300866&merchantid=420536914&gclid=CjwKCAiAleOeBhBdEiwAfgmXfzHlzp3M4RK9aoX82l1DeVJs1pt-buCVUFugQuRXuyYmR215VNjzhxoCBwAQAvD_BwE

 

 

[Radian]

2 hours ago, Bramco said:

 

There already is one - with all the theory -> https://docs.openenergymonitor.org/pv-diversion/introduction/choosing-an-energy-diverter.html

 

I've had the software for the MartinR version running for several years.   You do need to make the SSR board but it's not rocket science.

 

Simon

 

Yes Simon, I've been posting over on the openenergymonitor.org community for a while but the solutions we've built do require some electronics skills that not everyone has. When I read of people on buildhub piecing together rudimentary diverters using off the shelf WiFi switches with much less efficient control I started thinking about a design along the lines of @SteamyTea's 'Lego mentality' where the bits could just click together.

 

The kind of screw terminal Jack that @MikeSharp01 put up could help with that as at least it could support a couple of resistors without soldering.