PVOutput shows unusual dip to zero power

[ButchDeal]

Soooo, although the optimizers are reporting randomly to the inverter, the individual reports are NOT available at the inverter.

the last individual optimizer reports are available on the (solarEdg) inverter

The inverter then reports via a PLC-ish protocol carried in TCP/IP packets to the Solar Edge/Enphase mothership.

No The inverter and envoy reports via some means, possibly ethernet since it has an ethernet port on it to the internet cloud servers.

What Sensij and I were saying is that they use PLC for enphase and PLC similar for SolarEdge to communicate between the optimizer/micros and inverter/envoy, then they use ethernet (or possibly PLC depending on individual installation). But all of it has a TCP/IP protocol that has collision avoidance and wait times to rebroadcast.

If that is the case, the only thing I have to 'unlearn' is that you can't just ask the inverter for the last known panel reports. Gotcha. (unless I am still confused).

you are still confused. The inverter has the last known optimizer data.
here is an example of the (remote view) of the local display from one, showing a single optimizer data last reported : Layout.jpg

Been involved with the TCP/IP suite of protocols and IPv6 since the '80s. At one point in my life, was the NIST NVLAP accreditation agent for the GOSIP protocols, including ISDN, that were in the same layers of the network protocol model. In fact, I won the organizational pool on the exact month that TCP/IP would kill off the entire GOSIP program...

congratulations. Of course you know that IPv6 RFC was only submitted in December of 1998, though right?

https://www.ietf.org/rfc/rfc2460.txt

[jpoet]

On the inverter, both the Error and Warning logs are empty.

[AzRoute66]

Soooo, although the optimizers are reporting randomly to the inverter, the individual reports are NOT available at the inverter. The inverter then reports via a PLC-ish protocol carried in TCP/IP packets to the Solar Edge/Enphase mothership. If that is the case, the only thing I have to 'unlearn' is that you can't just ask the inverter for the last known panel reports. Gotcha. (unless I am still confused).

Been involved with the TCP/IP suite of protocols and IPv6 since the '80s. At one point in my life, was the NIST NVLAP accreditation agent for the GOSIP protocols, including ISDN, that were in the same layers of the network protocol model. In fact, I won the organizational pool on the exact month that TCP/IP would kill off the entire GOSIP program...

[ButchDeal]

Twice over the last few weeks, I have seen the PVOutput data show a brief drop to zero power. The latest one just happened 45 minutes ago (2017-08-28 11:15am): https://pvoutput.org/intraday.jsp?id=58790&sid=52694 . I don't remember a cloud going by at that time, and even if there was one, the power level should not drop to zero. When I look at the SolarEdge dashboard, it shows a corresponding dip, but there it does not go all the way to zero, and bottoms out at 3kW.

Is this just a data glitch, or is this something I should be concerned about?

TIA

I am with sensij on this. you should look through the solaredge monitoring for issue. The voltage information looks odd as well.
Get the installer to give you full access on solarEdge monitoring portal and configure your account as owner.
Then you can go to the web site and look at all the optimizers, DC voltages etc and see what is going on

https://monitoring.solaredge.com/solaredge-web/p/home

[GRickard]

If you get your inverter manual out, you will find the process to get into the error log. It sounds like you may be getting an error and shutting down. If so, it will take 5 minutes to come back on line.

Greg

[sensij]

Can I get the one minute explanation of why I shouldn't expect to be able to cover up a panel and see it immediately on my _____. I don't think I need to be spoon-fed, just a quick pointer or two to put me in the fray. I don't mind 'grinding' out the details, but I do try to avoid it...

I'm not as familiar with Enphase, so I'll stick to SolarEdge.

Each panel has an "optimizer" that does DC-DC conversion to develop at steady 350 V DC for the string into the inverter. The optimizers report the instantaneous output of the panel, but do so on irregular intervals. The communication uses high frequency signaling on top of the DC production, a form of powerline communication. When two optimizers talk at the same time, the message is lost, so they have randomizing element in their broadcast timer that helps minimize those collisions. The end result is that a single panel only updates its output over an unpredictable interval, usually between 5 and 20 min, depending somewhat on the number of panels in the system.

The inverter itself displays instantaneous power for the whole system on its screen, which has an update rate of something like 10 Hz to my eye. The inverter only stores and broadcasts power to the monitoring portal in 5 min intervals though. The portal displays those 5 min intervals as a 15 min average on most of the screens, but more detailed reports can be generated to see the 5 min data. I think the 5 min readings are instantaneous, not average, but figuring that out when overlaid with the calibration error of the inverter is hard, and I gave up quick once I saw what a mess it would be to determine. I didn't integrate a revenue grade meter (I use one standalone), and the SolarEdge software treats that a bit differently, so what I'm writing may not be entirely applicable to that configuration.

When you cover up a panel, you'll immediately see on the inverter the impact on overall power generated. If you are trying to trace that particular panel, you need to leave it covered long enough to be sure that you catch it during one of its successful reports.

PVOutput uses an API to poll the monitoring portal and pull out the 5 min data (among other things). It is something of an acquired taste, but is much richer in the information it can convey than the SolarEdge portal is, especially when you use its ability to make comparisons across time or systems.

I think the 5 min time periods are attempts to limit the amount of bandwidth required for data transmission, and limit the amount of data that SolarEdge need to manage.

[ButchDeal]

Can I get the one minute explanation of why I shouldn't expect to be able to cover up a panel and see it immediately on my _____. I don't think I need to be spoon-fed, just a quick pointer or two to put me in the fray. I don't mind 'grinding' out the details, but I do try to avoid it...

Quick spoon feed answer is that "Real time" has to be defined. Nothing digital is ever really real time. The term real time means that it is fast enough to be of use for the application.
PVOutput is just a third party presentation system that presents data from another device like solarEdge or enphase. So it only adds more delay.

You can't cover up something on the roof and see it immediately because there is a delay in the hardware as well as the protocol. With both SolarEdge and enphase the module level devices wait a random amount of time to transmit to avoid collisions on the transmission network. Even so there can be collisions so they try again.

To really understand it you would have to understand basic internet protocol or TCP/IP which they based their communications on.

So basically you shouldn't expect it because you have been told that there is a delay based on the communication protocol that they use.

[AzRoute66]

Not to hijack anything, but I'm a little confused. Remember, I know nothing about inverters and even less about grid tied inverters. But this is not the first time I have heard a whiff of 'some lag and/or averaging' with the monitoring of these optimized systems. I just looked at Enphase and Solar Edge web pages for their monitoring and both claim real time feedback as one of their 'features'. So, from what I see we have Enphase 'Enlighten', Enphase 'Envoy', and Solar Edge 'Whatever'. And apparently all of these feed (are polled by?) something called PVOutput. I went to PVOutput briefly and it appeared so irrelevant that I can only assume you need to create an account to get a feel for what it does for you.

Can I get the one minute explanation of why I shouldn't expect to be able to cover up a panel and see it immediately on my _____. I don't think I need to be spoon-fed, just a quick pointer or two to put me in the fray. I don't mind 'grinding' out the details, but I do try to avoid it...

[sensij]

I'd be concerned about it. PVOutput shows 5 min nresolution, SolarEdge portal by default shows 15 min, so the data point is probably getting averaged out a bit there.

If you have enough access, you can see if an error is being logged by the inverter at that time.

My 2 year old system started showing similar dips a couple of times a week at the beginning of the month. (See 8/17 for an example of a long one, some have been shorter). Error codes indicated L-N voltage errors. Last week, the tenant reported "flickering lights", and after some quick troubleshooting at the service panel, I had SDG&E come out and look. Turns out, the feed conductors are failing (house was built in 1962), and a new service drop from the pole needs to be run. Still trying to get the work scheduled with them, but hoping to get it taken care of this week.

Again, I'd say step 1 is to find out the error code... there are other more benign reasons it could be getting knocked offline.

[jpoet]

Twice over the last few weeks, I have seen the PVOutput data show a brief drop to zero power. The latest one just happened 45 minutes ago (2017-08-28 11:15am): https://pvoutput.org/intraday.jsp?id=58790&sid=52694 . I don't remember a cloud going by at that time, and even if there was one, the power level should not drop to zero. When I look at the SolarEdge dashboard, it shows a corresponding dip, but there it does not go all the way to zero, and bottoms out at 3kW.

Is this just a data glitch, or is this something I should be concerned about?

TIA