How exactly does grid-tied hybrid inverter detect loss of grid?

Thanks, read that but I still feel I'm missing something... in that description it says..

Disconnect from AC input is a bit more complicated. If grid drops out, there is a momentary overload on inverter as inverter tries to power the collapsed grid, up to possibly the surge current limit of inverter, before it releases AC input pass-through relay...

but that doesn't apply if inverters are outputting to grid...

Toughest thing for inverter to do is detecting when AC input goes open circuit, like when you open AC input breaker. Zero AC input current is a legit condition and voltage on AC input does no good since inverter is feeding through pass-through relay to AC input.

Most inverters normally use the up/down pulses from phase detector, to determine when to release pass-through relay. When a legit AC input is present there should be a repetitive average of up and down phase correction pulses. Some inverters, when AC input is opened, just wander off in frequency until they hit max inverter run AC frequency limit before disconnecting pass-through relay...

But if there are 2 or more inverters, won't they keep each other in sync - or at least for quite a while?

Recognition of loss of AC input can take from a couple of seconds to a minute before it releases pass-through relay...

But for UK ENA Type-tested inverters, disconnection has to take place within 500ms, and I'm guessing the same for other jurisdictions?

Thanks, read that but I still feel I'm missing something... in that description it says..

but that doesn't apply if inverters are outputting to grid...

Yes it does. It periodically pushes a phase wiggle against grid to test that grid is still present.

But if there are 2 or more inverters, won't they keep each other in sync - or at least for quite a while?

When hybrid grid interactive inverter is connected to AC input source it is slave to AC input freq/phase/voltage. Once pass-through relay releases the inverter goes back to being its own master.

If you have inverters running in parallel, or series connected 240/120vac for split phase, there is usually an interconnect cable to keep units in sync when there is no AC input.

In addition to sync'g the two hybrid units, there is usually other interconnect control lines, like if one unit shuts down the other unit also shuts down, or if one is charging battery, the other is also charging battery and they both switch charging states (bulk, absorb, float) at the same time based on first unit jumping states. Again, if AC is present, cross sync'g control is overridden and they obey the phase of their respective AC input as slaves to AC input.

Inverter-generators that can run in parallel require AC output be connected first before one generator is started first and it becomes master for phasing. Second inverter-generator, seeing AC voltage present on its input when started, becomes slave to first generator started.

But for UK ENA Type-tested inverters, disconnection has to take place within 500ms, and I'm guessing the same for other jurisdictions?

Only way to detect an open circuit disconnect on AC input is by inverter pushing against phase lock to AC input. Pass-through relay allows inverter output to be present on open circuit AC input so voltage detection is of no use. There will be a current deviation push back when a legitimate grid is present. Grid is an 'immovable' object when inverter tests for grid presence by attempting to wiggle it phase lock to AC input.

Most cheaper, non-UL compliant inverter just bias their phase locking circuitry so if AC input goes open circuit there is no phase locking reference and the inverter frequency will wander higher in frequency until it hits maximum allowed frequency spread limit of inverter before it releases pass-through relay. Some count how many zero crossings occur in AC cycle without any phase correction generated. If too long a period without phase corrections then it assumes AC input has been lost and releases pass-through relay.

I mean it makes sense to me that once the grid drops, it becomes this vast bottomless pit and your inverter cannot supply enough energy to support it, so it goes into overload and turns off. All in the blink of an instant.

While that is the most likely situation There could be instances of grid loss where the only loads are near your home and within the capacity of your system. Thus a danger to someone not prepared to find a voltage on the grid. I find RCinFla explanation more comprehensive since the anti-islanding protection is not reliant on simply overload.

How it accomplishes this is in the Voodoo electronics land to me though.

Yes it does. It periodically pushes a phase wiggle against grid to test that grid is still present.

I think many here don’t understand what this means or misunderstand it. I will try to explain further what you are saying, hopefully I don’t muddy it more:

1) the “grid” is huge. If you want to speed the grid up (increase the frequency) you must supply enough power to the grid to measurably increase the speed of every single motor on the grid. Not just your neighbors, but everyone in your county, state and neighboring states. In the US there are several large regions that form their own “grid” I forget exactly how many.

2) Their is no way your inverter can do anything to impact grid frequency. A large 1,200 Megawatt nuclear unit cannot impact grid frequency on its own in normal circumstances except a very fast transient if it trips for example.

3) when he says “wiggle” what he means (I believe) is that, for example, the inverter is programmed to attempt to increase frequency periodically. If it is actually attached to the grid, the frequency will not budge but the inverter power output will increase sharply. If frequency does increase, it knows it’s no longer connected to a grid.

4) I’m sure the electronics to accomplish this as fast as it is required are complex.