SolarEdge operational question

[inetdog]

Thank you (assuming that was directed to ButchDeal who was shouting at me).

Don't assume.
Butch saild "OUTPUT" a couple of times for emphasis, Not necessarily shouting.
"OUTPUT OUTPUT OUTPUT!" is not just emphasis.

But much less civilized behavior has gone unremarked on the Form in the past, so let's just drop it now.

[solarix]

Ah yes, I had a long conversation about this with the SolarEdge people when they first came out.
How does the control algorithm for the optimizers work when they have no communication link to the inverter other than the power flowing in the string? The bottom line was that this was their core IP and did not want to talk about it.... I did get out of them that the control loop for the optimizers runs much faster than the control loop for the inverter. The inverter just adjusts its input impedance to try and keep the input voltage at roughly 350Vdc. The optimizers then dither their output voltage around (I'm thinking) to search for the right output current that is best for the whole string. Because it is doing this much faster than the inverter does its adjustments, the optimizers can easily track any change in the string current. A very clever and marvelous system design.
On the other hand, they are also modulating the string current with digital info in order to talk between the inverter and the optimizers - although at a really slow rate (maybe once every 10sec). The inverter interrogates the optimizers one by one so that it can accumulate all the individual module data. Unfortunately, they do it randomly and if you have a problem with one of your optimizers it can take like half an hour before all the optimizers report in. If (I mean when) one of them has failed, you can hook up a laptop to the inverter and follow the optimizers as they report in, but it is always a process of elimination, waiting (and waiting) for them to all randomly report in to find the one that doesn't. When installing SolarEdge systems, always make a map of the optimizers with their serial number I.D. so that you can know what is where later on when you start having failures....

[jflorey2]

How does the control algorithm for the optimizers work when they have no communication link to the inverter other than the power flowing in the string? The bottom line was that this was their core IP and did not want to talk about it.... I did get out of them that the control loop for the optimizers runs much faster than the control loop for the inverter. The inverter just adjusts its input impedance to try and keep the input voltage at roughly 350Vdc. The optimizers then dither their output voltage around (I'm thinking) to search for the right output current that is best for the whole string. Because it is doing this much faster than the inverter does its adjustments, the optimizers can easily track any change in the string current. A very clever and marvelous system design.

So this is the opposite of what other people are saying. It is how I first assumed the system worked, but I keep running into the same conceptual problems. How do the individual optimizers know what the string voltage is, so they can make good decisions on what output voltage to target?

The inverter can certainly adjust its input impedance to provide some kind of a signal. But if it does that, then the optimizers have to track that as well; if they don't match the input impedance you lose power. Plus, each optimizer is seeing the impedance of the inverter plus the impedance of every other optimizer in the string, and the other optimizers are changing their impedance as they dither their own conversion ratio. So it's hard to see how the optimizer would see that impedance change and turn it into useful information (like "go up in voltage" or whatever.)

If what you say about the PLC is true (random transmit time from the optimizers) then it seems like it would be very difficult to use that for system voltage control on larger systems.

[solarix]

The optimizers have no idea what the string voltage is and yet somehow they adjust their output voltages to stay in the neighborhood of 350V all together. I surmise that if shade hits one optimizer's panel and its power goes down, then it drops its output voltage to keep its output current the same. This will drop the string voltage a little which after a while the inverter will raise its input impedance in order to get the string voltage back to nominal. This is somehow detected by the optimizers and they all adjust their outputs to run at a lower current which lifts their output voltages a little and gets the string back to 350V.
The interrogation of the optimizers is definitely way too slow to use in the control loop and is just used for housekeeping tasks like monitoring, initialization and re-pairing.

[jflorey2]

The optimizers have no idea what the string voltage is and yet somehow they adjust their output voltages to stay in the neighborhood of 350V all together. I surmise that if shade hits one optimizer's panel and its power goes down, then it drops its output voltage to keep its output current the same. This will drop the string voltage a little which after a while the inverter will raise its input impedance in order to get the string voltage back to nominal. This is somehow detected by the optimizers . . . .

That's the problem. When that one optimizer drops its output voltage, then its impedance goes up. (R=V/I.) So all the other optimizers see the impedance go up - even without the inverter doing anything. In fact, impedance will change throughout the day as all the optimizers change impedance as the sun gets more or less intense. So it can't use an observation of impedance to set output voltage, or just the normal changes in insolation would make the string voltage vary all over the place.

[foo1bar]

This is somehow detected by the optimizers and they all adjust their outputs to run at a lower current which lifts their output voltages a little and gets the string back to 350V.

My impression was that the inverter sends to the optimizers a message "I need you at 1% higher current" so it would get lower voltage - and then they'd respond appropriately.
Just because the reporting mechanism is slow doesn't mean that there can't be a faster communication channel from inverter to optimizers (possibly with limited messages so not much bandwidth needed)
I was envisioning a digital message - but it's quite possible that it's more like you're describing and is done via a more analog means.

[solarix]

My impression was that the inverter sends to the optimizers a message "I need you at 1% higher current" so it would get lower voltage - and then they'd respond appropriately.
Just because the reporting mechanism is slow doesn't mean that there can't be a faster communication channel from inverter to optimizers (possibly with limited messages so not much bandwidth needed)
I was envisioning a digital message - but it's quite possible that it's more like you're describing and is done via a more analog means.

No, the SolarEdge engineers told me that the optimizers are operating autonomously and they do not communicate with the inverter in order to control the string voltage.

[ButchDeal]

No, the SolarEdge engineers told me that the optimizers are operating autonomously and they do not communicate with the inverter in order to control the string voltage.

that does not mean that the inverter isn't broadcasting information though, that the optimizers can independently work on.

[jflorey2]

that does not mean that the inverter isn't broadcasting information though, that the optimizers can independently work on.

That's communication.

It would be interesting to set one of these systems up and then sniff it to see what (if any) communication is happening between inverter and optimizers. So 8 is the minimum string size, huh?

[foo1bar]

It would be interesting to set one of these systems up and then sniff it to see what (if any) communication is happening between inverter and optimizers. So 8 is the minimum string size, huh?

You'd need to have some idea of what they're doing to communicate.
Otherwise it'll take a long long time to figure out what is communication and what is noise.and/or optimizers reporting in.

[ButchDeal]

That's communication.

It would be interesting to set one of these systems up and then sniff it to see what (if any) communication is happening between inverter and optimizers. So 8 is the minimum string size, huh?

He stated that "optimizers are operating autonomously and they do not communicate with the inverter"
the fact that the optimizers are not communicating with the inverter does NOT preclude the inverter form communication with the optimizers.

It would be difficult to sniff the communications as you have no idea what form it is taking on the DC leg.

[jflorey2]

You'd need to have some idea of what they're doing to communicate.
Otherwise it'll take a long long time to figure out what is communication and what is noise.and/or optimizers reporting in.

Well, there aren't many forms it can take, since it's one closed DC circuit. It can modulate current or voltage from DC into the ghz. And it's quite likely that if you start the system up, then make a rapid switch from the inverters to a 350VDC source, you'd get only the inverter signaling on the (now fake) network.

[foo1bar]

Well, there aren't many forms it can take, since it's one closed DC circuit. It can modulate current or voltage from DC into the ghz. And it's quite likely that if you start the system up, then make a rapid switch from the inverters to a 350VDC source, you'd get only the inverter signaling on the (now fake) network.

If there's nothing where the inverter expects communication back and you've got a really good frequency analyzer and you're in a low-noise environment with a good power supply that might be enough to start to figure out whether it's broadcasting something when you raise/lower the voltage.


Somehow it tells the optimizers "you're connected, you can go to operational voltage instead of 1V"
So I expect it's doing some broadcasting.
But there are a lot of options on what frequency it could use and what encoding method it could use.

[ButchDeal]

Somehow it tells the optimizers "you're connected, you can go to operational voltage instead of 1V"
So I expect it's doing some broadcasting.
But there are a lot of options on what frequency it could use and what encoding method it could use.

also it tells the optimizers very quickly the opposite, (rapid) shut down now

lots and lots of options on the frequency....

[bcroe]

No, the SolarEdge engineers told me that the optimizers are operating autonomously and they do not
communicate with the inverter in order to control the string voltage.

Perhaps each optimizer just keep pushing up its output voltage until until enough power is drawn to put
its panel at MPP. The inverter could just keep taking more energy from the optimizer string until the
total voltage drops to around 350V. The potential for instability is high.

I have thought about building up some modules to do this for some panels in some strings here, so that a
whole string wouldn't drop out when shade shifts over a few panels. Bruce Roe