'Home battery' sanity check?

[mjc506]

Apologies if this is the wrong subforum, it seemed the 'most' appropriate.

This seems a bit too simple, and I'm wondering what I'm missing:

Prius or similar inverter, plus control board(? with modded software). Use two IGBTs (one of the three half-H-bridges) from the MG1 side as mains (230V, 50Hz) live in, and MG2 as 'mains' out to the CU, and then DC to/from battery of course... This would allow charging a largish battery from the mains at cheap times on a TOU tariff (or from a 3phase wind turbine etc, or even using the 'spare' half-H-bridges to take other DC (solar) or AC (single phase gens) inputs. Would have to 'choose' one input at a time.) and power a house from the MG2 side of the inverter, plus keep everything working through power cuts... Could even use each of the three half-H-bridges to power groups of circuits (ie, house on one phase, EV charging from another phase, workshop on the third, for example)

Voltage on neutral would have to be monitored, and the inverter shut down if it differs from earth voltage (to protect against against a broken neutral upstream), and a local earth would be required (or just turn everything off if there's a power cut).

There's round-trip battery and inverter losses, but a TOU tariff is plenty sufficient to be ahead financially. Add in local renewables charging and that improved further. I guess a set of contactors could switch between mains and inverter, but that's additional complexity and the switch may be too slow during a power cut, and there would have to be some way of synchronising mains and inverter phase and frequency. Doable, but the efficiency gain may not make it worth it.

Ignoring the legalities... what've I missed?

[NiHaoMike]

If you don't need whole house backup, it's far less invasive to make it a zero export inverter. That is, it connects to the breaker panel via a dedicated circuit and sensors are used to limit the current it sources so that it does not export to the grid. In doing so, it can simply plug into that dedicated circuit and be considered an appliance instead of building wiring, which gets around many of the code issues.

One big catch is that you'll have to learn how to design and build your own inductors for the output filter since such large inductors are not readily available off the shelf.

[mjc506]

It'd be a shame to lose the backup, and I'm not worried about rules and regs particularly. Let's keep it hypothetical

True, inductors will have to be manufactured as well. They'll be fairly large!

[LRBen]

You'll have to ensure that on a power cut it stops exporting energy to the grid. Otherwise it will put grid engineers in danger who are expecting to work on unpowered lines.
The Tesla power walls have some fancy extra box for exporting energy I think and one of the main features of it is to stop grid export as soon as it senses a power cut.

[mjc506]

Definitely. Initially at least, I wouldn't be exporting at all, but a serious consideration if/when that's implemented.

A 'simpler' method would be a battery charger, with the/an inverter powered from the DC, (so impossible to export to grid) but it seems neater to use an all-in-one unit like the Toyota inverters.

[NiHaoMike]

It'd be a shame to lose the backup, and I'm not worried about rules and regs particularly. Let's keep it hypothetical

True, inductors will have to be manufactured as well. They'll be fairly large!

In my design, I'm using the third phase as a dedicated UPS output. That allows for a zero switchover time.
viewtopic.php?f=14&t=419

You'll have to ensure that on a power cut it stops exporting energy to the grid. Otherwise it will put grid engineers in danger who are expecting to work on unpowered lines.
The Tesla power walls have some fancy extra box for exporting energy I think and one of the main features of it is to stop grid export as soon as it senses a power cut.

All you need for that is a contactor connected to a circuit that checks to make sure the voltage and frequency are in range. You'll also have a feedback signal from the contactor (auxiliary contacts) to signal the inverter to switch from current control mode to voltage control mode. That's how such a simple voltage and frequency check can be so reliable - a varying current source connected to a varying load quickly becomes unstable. An export blocking circuit on top of that makes it even more reliable - it simply throttles back the inverter if it tried to push current back to the grid.

It could be possible to implement zero export control in a STM32 if you're using hardwired sensors, but I decided to go with a FPGA since I plan to use PLC connected sensors. The main challenge is making sure the zero export condition remains met at all times, which puts the response time needed on the order of a few hundred microseconds. Decoding the PLC signal fast enough would be challenging to do with a microcontroller - the FPGA's inherent parallelism makes it easier to meet the strict timing requirement. (That's probably why I haven't seen any zero export inverters with wireless or PLC current sensors.)

A 'simpler' method would be a battery charger, with the/an inverter powered from the DC, (so impossible to export to grid) but it seems neater to use an all-in-one unit like the Toyota inverters.

You could have the inverter operating full time with either some unused inverter phases or an external bridge rectifier as the rectifier. Much easier from a control perspective and allows for zero switchover time. That's how many datacenter UPS units work.

[mjc506]

Ah, thats neat using three phases of an MG both both mains in and out. You use an isolated neutral then - I'm trying to work out in my head how to tie neutral and earth and still get 230VAC RMS on live without popping fuses (230VAC RMS = 650V peak to peak, higher than the voltage booster output I think, and I don't fancy a battery at that voltage... so I'd need to use two phases with live and neutral across them? More reasonable battery voltage, but I've got to be more careful with the neutral-ground tie) I think the inverter case is connected to 12VDC ground (I'd use the 12VDC to charge a small 12V battery to run the control and inverter) but the HV battery and MG outputs are all floating.

Yes, I'd be aiming for an 'online' UPS - no switching time, and fewer contactors. I can't see a massive difficulty ensuring no export - just make sure it can only 'regen'. That said, probably worth making sure there's sufficient hardware there to support safe exporting for if/when that's desired.

[slow67]

230VAC RMS = 650V peak to peak

How do you figure that? 240 x 1.414 = 339V peak to peak

[mjc506]

...duh, you're of course entirely correct, not sure how I managed that...