5 minutes ago, IdlePhaedrus said:

You don't have to answer, I can research it myself when I have some time.

I expect the answer will be "it's complicated"

37 minutes ago, IdlePhaedrus said:

Thanks.  May I ask what the mechanism is that a grid tied inverter decides to turn off when the grid drops? 

You don't have to answer, I can research it myself when I have some time.

My guess is that it is shades of this:

The grid represents an infinite bus. in other words something so vastly huge and immovable that the the little 'ole inverter couldn't cause the system to alter the presented frequency or voltage or phase shift. Then there will be a feedback algorithm that tries to do exactly this (with some portion of every cycle say), and if it can actually do it, it knows the anchor of grid isn't there so it switches off.

Least ways that's how I'd do it.

Edited January 9, 20206 yr by phil.g00

12 minutes ago, phil.g00 said:

Least ways that how I'd do it.

Thanks.  So yes, it appears to be "complicated". 

It also suggests that the decision by the inverter to shut down might be down to the version of firmware installed and how that version measures the various parameters such as current, voltage, frequency and phase shift.

Bugs come and go in software / firmware so in this scenario if the inverter doesn't have a second set of immutable parameters stored separately from the general running firmware a particular firmware version could conceivably render the original certification useless.

Definitely an interesting conundrum. 

 

10 minutes ago, IdlePhaedrus said:

Bugs come and go in software / firmware so in this scenario if the inverter doesn't have a second set of immutable parameters stored separately from the general running firmware a particular firmware version could conceivably render the original certification useless.

Exactly.

16 hours ago, phil.g00 said:

The grid represents an infinite bus. in other words something so vastly huge and immovable that the the little 'ole inverter couldn't cause the system to alter the presented frequency or voltage or phase shift. Then there will be a feedback algorithm that tries to do exactly this (with some portion of every cycle say), and if it can actually do it, it knows the anchor of grid isn't there so it switches off.

There's actually some industry standard ways to do it, but not all grid codes require the same level of "paranoia" if you will. NRS097 requires some basics (voltage within the right range, frequency in the right range), plus one active measure (you get to pick one). Active measures include (I am sure there are more...):

1. The inverter usually has a phase locked loop locked onto the grid frequency. You can introduce a slight error into this so that the frequency would increase up once the lock with the grid is lost. The frequency then drifts out of range the moment you lose the grid and you disconnect as you hit 53Hz (as you normally would).

2. You actively try to change the power factor by shifting the voltage and current a few degrees left and right.

Within the literature, which I admittedly understand only partly, there is something called an NDZ (non-detection zone), which happens when you somehow manage to have conditions just right so that the detection fails. @IdlePhaedrus's inverter could possibly be non-faulty and he observed the one-in-a-million NDZs... or... and this I actually think is more likely... he has an inverter with a large NDZ which was not picked up in testing.

As another analogy, this is now happening with crash testing in cars. There is a specific set of tests that are run. It is standardised and manufacturers know what they are, so they specifically optimise for them. A car can pass with flying colours, and then you may discover later that it has weak points where you didn't look. That's why crash tests added an "small offset frontal" test. Cars were getting good ratings... as long as large enough part of the frontal face is involved in said crash 🙂

 

This topic quickly went from registering to a discussion about something completely off this topic.

I believe the following

1. If your system is less than 20KW and off grid you should not have to register

2. You should be able to produce a COC for the installation to proof that it is safe if asked by the authorities and your insurance

3. If your system is grid tied and you can feed back it should be registered in order for the workers to be safe.

4. I believe the building regulation should be changed that all new building should have at least 3kw of solar installed when a new house is build.

5. i believe generators must be banned from residential areas  

Edited January 10, 20206 yr by Krokkedil

5 hours ago, plonkster said:

Within the literature, which I admittedly understand only partly, there is something called an NDZ (non-detection zone), which happens when you somehow manage to have conditions just right so that the detection fails. @IdlePhaedrus's inverter could possibly be non-faulty and he observed the one-in-a-million NDZs... or... and this I actually think is more likely... he has an inverter with a large NDZ which was not picked up in testing.

Thanks @plonkster.  That is really interesting and helpful. 

The inverter is running the same version of the firmware as I was running when the problem occurred.  I also quickly tested on the G59 grid standard again before changing to the NRS standard on the same station with the well point pump running by simply turning off the grid supply.  It still failed to turn off.  After the change to the NRS standard it worked as expected.

The chap from Ginlong in the UK had this to say on the matter:

Quote

The UK standard has a whole bunch of frequency shifting functions that might have caused this type of behaviour.

PS, I have no idea what a frequency shifting function is.

Edited January 10, 20206 yr by IdlePhaedrus
bit of extra info

1 hour ago, Krokkedil said:

This topic quickly went from registering to a discussion about something completely off this topic

Apologies @Krokkedil.  Threads on this forum tend to do that.  Expect it is down to the broad nature of the subject.  Will bow out of this thread now. 

Have an awesome weekend.

6 hours ago, IdlePhaedrus said:

.  Threads on this forum tend to do that.  Expect it is down to the broad nature of the subject.  Will bow out of this thread now. 

You're welcome to bring it back to the original topic if you feel something more needs to be said about it.

I'm not sure about this Level 6.  Does it mean up to 10 hour blackouts/day?   How will cell-towers/repeaters handle this?  I understood that MTN  has back-up for 4 hours. What happens then?   Do these towers shut down and then come back when ESKOM returns?

It sounds like a major catastrophe to me.