DIY solar immersion diverter and hybrid inverter?

Does anyone have any experience with DIY diverters and hybrid inverters?

I’m using an SSR to make use of an “energy bucket” design, with a Solis 5G hybrid. When the diverter is off, the inverter CT clamp (Eastron communicating over modbus) mostly reads zero as you’d expect, but when the SSR is operating even with fairly rapid (eg 5Hz) switching, the CT clamp is all over the place reading import and export of large values (which I guess is correct!). The smart meter (Aclara) is behaving as I’d expect with a bucket system.

Is the wide variation in the CT readings likely to cause any trouble for the inverter in the long term? I was concerned that when it reads import it’ll try to draw power from the batteries for part of a second or similar.

I don’t have the option (easily) of excluding the immersion from the CT reading. I can change the frequency of the SSR switching easily (right now I’m using a fixed frequency for testing).

Thanks for any advice and thoughts!

Welcome, Tom, to the OEM forum.

We’ve certainly had one or two battery systems with diverters - which prompted a ‘Learn’ (now part of Docs) page about their operation.

I take it you’re referring to Robin Emley’s description of the way his Mk2 PV Router works - which is in fact very much the same as Martin Roberts’s.

I’m quite sure the inverter won’t know anything about the C.T. readings - what you really want to know is, is the inverter likely to not work properly, or is it or the batteries likely to be damaged, by rapidly switching from generating to consuming?

I think the fundamental question you need to answer is, what are you trying to achieve with the inverter and the diverter? Which has priority? What control do you have over the inverter/battery charger?

What I’d really question is, why are you switching at 5 Hz? It clearly needs to handle the induction hob sort of switching speeds, but to the best of my knowledge, that’s a fair bit slower than 5 Hz. Have you considered a much, much lower frequency - minutes rather than fractions of a second?

Is this because of where the immersion heater is in relation to the rest of the installation? In that case, how are you measuring and how are you controlling it?

Thanks for your help - I had a look through the docs, which have been very helpful on the bucket concept - but in terms of hybrid inverters could only find reference in the “multiple diverters” section, which is effectively what this is I guess - the hybrid inverter is diverting power to (and from) the battery in response to what it sees on its CT meter, which is why I’m having these issues.

Yes - I have been hoping to control the SSR (zero cross type) in the way described to minimise export (at the moment we’re not being paid at all, and in the future it’ll be less than we pay to heat our water), plus possibly help with battery management by slowing down charging a bit.

The CT readings I’m referring to are the inverter’s clamp (sorry, there’s loads on our board!) - it has an Eastron SDM120 connected over modbus which it uses to try to keep grid input/output as close to zero as possible. Generally it’s very good at this and it tends to read 0, occasionally + or - 40W. I experimented this morning with the kettle and when turning it on or off the inverter doesn’t ramp up and down instantly, but within a few seconds (4-5?) it’s back to reading zero after a brief pull or push to/from the grid.

Minimal control over the inverter - that’s the real issue - so when the inverter “sees” the demand from the diverter there’s a risk of it starting to drain the battery if, for example, there’s 1kW spare power but it ever ‘sees’ the immersion operating at 3kW (eg if I were switching over minutes rather than seconds or fractions thereof)

I’ve been trying a variety of frequencies whilst testing - I figured keep it stable for the moment to see how the inverter behaves before I complicate things by adjusting on the fly. Initially I was using phase control with a RobotDyn 24A dimmer - but even with 12A passing through it the temperatures were way too high for my liking so I decided to try a ZC SSR which seems to operate much cooler (but restricts me to burst mode). As mentioned above, if I cycled over minutes the inverter would ramp up its output to power the immersion heater and start draining the battery - so ideally I’m trying to find a way to “fool” not just the meter but the inverter too.

Yes - the immersion heater wiring comes off the main consumer unit and disappears immediately up into the ceiling. The inverter’s CT (the one I’m concerned about) sits on the meter tail so it’s seeing the whole house input/output. You can in theory move it onto the load side but I gather it makes it much worse at balancing (because it’s no longer simply targetting zero through the grid connection but trying to make its output equal to the sensed load)

I’ve been experimenting with various different ways round this this morning and haven’t found a satisfactory solution - I think I may have to go back to the phase control plan and work out how to upgrade the cooling of the Robotdyn module (unless there are better triacs I could solder on there!)

I think you don’t understand triacs! They all drop pretty much the same voltage, so irrespective, they’ll all need to dissipate roughly the same power. It’s how you remove the heat resulting from that lost power that governs how hot it gets. Your zero-crossing SSR should be dissipating roughly the same power for the same current, because it’s still a triac, so it’s probably got better cooling.

And then you’d recharge it again until it was full and you start to export, when it would power the dump load again. I take it you don’t want this to happen - or should it be a question for the battery supplier: what will this do to the battery, in terms of reliability and life?

I was also thinking of flicker and what this will do to you and your neighbours.

Meaning that at present, the inverter is seeing the diverter as part of the house load, and even if you wanted to change that around (so that the battery is part of the house load) you can’t?

Can you get at the meter tail where it enters the C.U, to put the inverter’s c.t. there, and rewire the immersion backwards through the c.t, so it doesn’t see the dump load at all?
Can you get an identical c.t. to the inverter’s and put it on the immersion feed, wired in series/parallel with the grid one, to subtract the immersion current? (see the N.America page for adding the outputs of two c.t’s.)

Possibly - it was a suggestion from another forum which I think might have been referring to the individual triac perhaps being cheap and not up to scratch! Though I’ve seen a few datasheets with slightly different voltage drops quoted. Certainly the panel-mount SSR I’m using at the moment is far easier to cool as I have it firmly bolted with thermal paste to the inside of an aluminium heatsink enclosure.

Yes - trying not to cycle energy through the batteries too much, partly because of the losses but mostly because of the potential effect on battery life. I’ve been experimenting with different ways of using the ZC SSR today and have arrived at a relatively flicker-free solution, but I think in the longer run some kind of phase control is probably going to be preferable for my use-case. If there was an equivalent to the Robotdyn dimmer in a panel-mount enclosure I’d be sorted!

Unfortunately the meter tail goes into a Henley block to which the inverter also connects, so attaching the CT anywhere else would mean the inverter wouldn’t be able to “see” its own output. The two-CT idea is definitely interesting, though it would make one of my use-cases (slowing battery charging) impossible as if the inverter can’t see the immersion heater load at all then when the batteries aren’t full it will continue charging them, and the immersion would be powered from the grid.

Having played with things today, the ramp-up-and-down of the inverter is I think slow enough that my current method works - though makes the inverter report look ugly as the grid CT is oscillating about zero rather than being mostly sat there. I’ll have a better look to see if I can improve the robotdyn dimmer cooling - or I even wondered about switching for an instant-on panel mount SSR and just using the ZC detection from the robotdyn device (or a smaller one).

What does the battery charger/inverter tell you? Is there an output you can use for controlling your diverter, because you haven’t mentioned what you have that’s doing the “energy bucket” bit. If it’s essentially Robin’s design, i.e. something you have all the information about, it’s simply a matter of swapping the triac driver and you can have a heatsink of whatever size you need. Unfortunately, you’ll probably need significant filtering to get within acceptable limits for the harmonics you generate. (Though I wouldn’t like to say if you’re obliged to filter as it’s not something you will be selling, it would be kind to yours and your neighbours appliances if you did.)