Latest Threads 
Page Index Toggle Pages: 1
How a grid tie inverter detects grid down, anti islanding (Read 17935 times)
YaBB Newbies

Posts: 5
How a grid tie inverter detects grid down, anti islanding
May 20th, 2013 at 10:17pm
It may not be apparent to most people how a GTI detects that the grid is down. There may be a number of methods to do that and a number of circuits that would be a dependable way to do it.

Anti-islanding is mostly a safety feature for power company line workers that may not follow proper insulation procedures or short circuit the powered off lines to ground like they should as instructed by company rules and training. It also prevents damage to the GTI itself or other equipment in your home since a GTI is normally not output voltage regulated.

Normally when a entire city's grid goes down, your house wiring becomes a zero ohm hard dead short, in principal. There is no way your little grid tie can push power into that.

You have to take into consideration that the GTI is generating a AC waveform in sync with the grid and needs to constantly detect the grid frequency in order to sync. This is most likely the main method used for detecting that the grid is gone.

Syncing a GTI to the grid is not easy to do with a CPU, if you wanted to build one from scratch, once you get into the design work you would start to understand how difficult it is. For example, you trigger a pulse on the DC side of your output transformer and that comes back into your AC line sensor within a few microseconds so timing is important.

However, since you are in control of the output "pulse", you can now distinguish between the grid and your "pulses" so in the design stage with testing you are going to be able to come up with a pretty dependable way to detect if the grid is down.

Combine that with voltage detection and it's going to be pretty hard to make it fail.

The testing method suggested by this company called metlabs uses reactive loads either capacitive or inductive, you can see the diagram there as an example of situations a home might have for loads.


UL 1741 suggests that anti-islanding happen within 2 seconds. How they came up with that number is anyone's guess, maybe they figure a linesman could withstand a 2 second jolt or maybe it takes more than 2 seconds for a line truck to reach the location of the power outage and get all set up, you never know. I would call it pulling a number out of a hat. But OK, it's reasonable and gives engineers somewhere to start.

2 seconds is a long time in grid AC 60 Hz cycles, 120 cycles to be exact. And that's a lot of CPU cycles so there's no problem detecting it in software for even the lowest cost CPUs. Meaning that you don't need some fancy expensive high Ghz CPU in order to do anti-islanding properly.

Most modern GTIs with CPUs probably cutout within a few AC cycles, even the cheap ones.

The main concern for possibly tricking a GTI into continuing to generate power after the grid goes down would be a reactive load such as a motor which would somehow magically and PERFECTLY generate a proper CONSTANT feedback signal that would trick the GTI AC detection circuit and CPU programming into thinking the grid was still connected.

This is probably a pretty hard balance to achieve in the real world, a situation would have to be just right and is a extremely rare occurrence.

You would need a house to become disconnected from the grid, wires fell off the main pole and that wonderful perfectly matched motor would have to be on and working just at the right time with no other motors or loads on at the same time. One other small motor on at the same time and the whole thing would fall out of sync.

If the physical load on the motor varied even a little, or if another electrical load came on or turned off, it would all fall out of sync (resonance) and the GTI would drop off.

In the real world, the power (watts) requirements of the reactive load would have to be just right and match the GTIs output power perfectly, all other loads would also have to somehow match the situation. Any variance and it drops out of sync.

Depending on where the power outage occurs is also a factor in balancing things out, so a lot of things have to be just right in order for anti-islanding to fail.

For instance, if the line pole transformer is still in circuit, now you have to overcome the resistance of that (push enough power/watts to overcome it). Have you ever seen the step down side coil inside a 40 KV to 240V transformer? It's a pretty short piece of aluminum and would look like a hard direct short to any small 1000W or less GTI.

You are on the low voltage end of the pole transformer. You need to push a lot of amps out to create just a small amount of amps on the 40,000V side of the transformer. So it's easy to short out the 40,000V side with even small wires or a small load if something was producing power.

Then think about all the loads down the line, the house next door, the next transformer down is also pulling down on the whole circuit with the dead loads in other houses on that transformer.

The grid is really an infinite load unless you are pushing megawatts into it.

Meaning when the grid goes down it looks like a zero ohm resistor you can try to throw 10,000 watts at and have it not budge a bit.

I did a google search after hearing that there might be testing results available showing that some modern day small GTIs can fail anti-islanding due to some kind of reactive load such as a refrigerator compressor motor being in circuit when the power fails. But I could find no such test results.

I can see how in a lab experiment you could create the exact perfect conditions to trick one particular detection circuit you happen to be testing at that time into thinking the grid is up, but in the real world that prefect situation would be a extremely rare occurrence.

NEC code seems to require a outside disconnect for solar installations, so there is a redundant safety system in place there for linesman that might follow the rules and bother to go house to house and pull the disconnect, or at least turn off the main breakers before working on a close by line.

But if you have ever watched linemen work, you will see that they work with the power energized all the time, they have the proper insulated equipment and training for that and it saves the company money (time) so you can bet they do it whenever they can. When direct contact is necessary, you will always see them using jumper cables to short all wires they are working on to ground. Standard procedure.

A lot of people think that the cheap small ebay type grid tie inverters are not safe because they are not UL "approved". But UL is not the end all of safety, a lot of UL approved stuff has started house fires so there is no guarantee that a UL approved device won't cause you problems.

UL just lays out suggestions for something to be safe, more like guidelines.

The small ebay grid tie inverters do meet UL 1741 - notice the word "MEET", which is the requirement in a lot of places, words like "comply"  also come to mind, you see that a lot. They cut out very quickly when the grid is gone.

The power company is worried about clean AC, and the small grid tie inverters meet that also, less than 5% total harmonic distortion is the spec and really, you have to be pretty clean if you want high efficiency.

So read that again in case the power company starts making it hard for you to go solar. No modern grid tie produces dirty power because it would be inefficient! All of them have anti-islanding !

And they are like any other appliance, they plug in, have a fuse (even on the DC side), are in a metal box.

Do you need a permit to plug in a big power sucking appliance? Why do they want to prevent you from using a small plug in grid tie then?

I would like to hear what other knowledgeable people here think and if anyone has links to actual real world tests where anti-islanding failed or reports where linesmen were injured by small GTIs.

Back to top
Page Index Toggle Pages: 1