Grid tied solar system.

[posplayr]

An MPPT algorithm adjusts the PWM to the panels to obtain maximum power. Modifying the algorithm to push away from maximum power in an attempt to regulate output voltage seems theoretically possible to me, although some care would need to be made to make it responsive enough to handle a sharp drop in load.

http://www.instructables.com/id/ARDU...and-Algorithm/

Well yes of course but it has little to do with MPPT. The GTI is basically a programmable DC to AC convertor that have internal control loops based on sensed current and voltage (on both ends). In a standard GTI mode, it cannot regulate output voltage, as it would end up fighting the grid, so it has to float above the grid voltage. The AC power that the GTI pushed out is (with the operating efficiency) the same as the power coming in on the DC lines from the PV. For constant PV power, if line voltage goes up , then current must go down. This is closed loop control but only internal to the GTI; it does not control total power or output voltage. So it can only affect the AC amps for a given PV power in.

And as you note, some new software could be used to control based on new objectives. In the simplest view it is possible to sense the AC line voltage and control the PV power holding the AC voltage within spec and assuming constant loads the current will find it's own level. This is basically what a UPS with double conversion would do. None of this is impossible, just impractical, because the primary design features of the GTI inverter is that it changes from GTI mode to OFF mode off when the line voltage does away. I have little doubt that to get whatever approvals are required, that the GTI manufacturers had to demonstrate this type of fail safe to meet ENC! So there is only GTI and OFF, nothing else. It is purposely designed to have not only nothing else, but failsafe to NOTHING else. So to come in the fact and say, OK we are now going to design a fail safe GTI that has a software controlled UPS mode? You are now going to get into safety interlocks for mutually exclusive operation and the GTI can not accomplish that backfeeding a breaker at the bottom of a panel.

So while the GTI could perform the function with added UPS functionality, there is no way to implement it easily and keep the GTI failsafe "OFF mode". So for the most part until something radically changes in the nature of what a GTI is, any islanding of GTI will probably be by controls outside of GTI itself. Those two modes of control would both achieve line voltage regulation of the island. Load control (e.g. water heater load dump) of source control (e.g. shorting panels) or a combination of both are much more feasible without trying to change the GTI with it's fail safe. Load control would likely not have much impact on the GTI, while I could envision interoperability issues depending upon how much or fast you want to chop the PV power. I guess you could also do load control at the GTI input, but that that might be more difficult than panel control.

You would basically add the output voltage controller to this diagram along with a whole house transfer switch.

http://www.solarpaneltalk.com/showthread.php?22001-Grid-tied-solar-system&p=163439&viewfull=1#post163439

[sensij]

It would be worth reading up on how Solaredge implements their zero feed in algorithm. It is done by monitoring current and adjust away from maximum power point on the panels (this is not chopping pv power up, the difference matters).

With a small dump load to act as the grid, and appropriate monitoring equipment, it might be possible to regulate an island this way.

[posplayr]

It would be worth reading up on how Solaredge implements their zero feed in algorithm. It is done by monitoring current and adjust away from maximum power point on the panels (this is not chopping pv power up, the difference matters).

With a small dump load to act as the grid, and appropriate monitoring equipment, it might be possible to regulate an island this way.

This is more than just a different algorithm, they have added a feature to limit power to a maximum as measured from a separate power meter. It would be an easy matter to hack a RS485 message from a micro controller that is measuring output AC line voltage and commanding a power limit to achieve that output power regulation. Using the UPS's you now have an supply side control GTI Island. You could couple this with load dump as well.

http://www.solaredge.com/files/pdfs/products/se_feed_in_limitation_flyer.pdf

[posplayr]

It would be worth reading up on how Solaredge implements their zero feed in algorithm. It is done by monitoring current and adjust away from maximum power point on the panels (this is not chopping pv power up, the difference matters).

With a small dump load to act as the grid, and appropriate monitoring equipment, it might be possible to regulate an island this way.

This is more than just a different algorithm, they have added a feature to limit power to a maximum as measured from a separate power meter. It would be an easy matter to hack a RS485 message from a micro controller that is measuring output AC line voltage and commanding a power limit to achieve that output voltage regulation. Using the UPS's you now have an supply side control GTI Island. You could couple this with load dump as well.

http://www.solaredge.com/files/pdfs/products/se_feed_in_limitation_flyer.pdf

[posplayr]

Posting is Fed up.

Basically what SolarEdge have done is added a programmable power limit to the GTI standard mode. Basically it will start to cut back if:

if(Pmeas-P_set>0)
DECREASE POWER
else
INCREASE POWER

It will limit cycle at P_set.

If you dynamically change Pmeas and pack it into a RS485 message

VAC_setpoint=120
Read(VACmeas)

Verr = ( VACmeas - VACsetpoint) // voltage control error
VTerr = dt*Verr+VTerr // Integral error
Pmeas = Kp*Verr + Ki* VTerr // PI control

send_485(Pmeas)

[Naptown]

Did not read everything
However a makeshift maguivered system will not pass code without a NRTL blessing.
Even if you use all NRTL listed equipment.
They are funny like that.

[posplayr]

Did not read everything
However a makeshift maguivered system will not pass code without a NRTL blessing.
Even if you use all NRTL listed equipment.
They are funny like that.

Of course, that is what I have been saying is the biggest challenge. Although given the Solar edge feature, I can't help but think they (SE) are not easing into something like this. The GTI safety disconnect is not defeated, and a whole hose transfer would have to be implemented. As far as certification the rest should be pretty easy. You just have a RS485 device that plugs into the wall. That can't be that hard to certify.

[Naptown]

Of course, that is what I have been saying is the biggest challenge. Although given the Solar edge feature, I can't help but think they (SE) are not easing into something like this. The GTI safety disconnect is not defeated, and a whole hose transfer would have to be implemented. As far as certification the rest should be pretty easy. You just have a RS485 device that plugs into the wall. That can't be that hard to certify.

I believe the entire system has to be approved for the intended use.
I can take UL listed products and use them in a way not consistent with approved use and fail.
An inspector may not pick up on this but in the event of a fire you can be sure the insurance company's engineer will

[posplayr]

I believe the entire system has to be approved for the intended use.
I can take UL listed products and use them in a way not consistent with approved use and fail.
An inspector may not pick up on this but in the event of a fire you can be sure the insurance company's engineer will

And yes, I agree. However, I would assume that approval would be for representative equipment and not specific combinations. The lynch pin is if Solaredge say that it is an approved use, which they would probably be more likely to have a proprietary system for doing this to sell systems and there is no reason that any GTI manufacturer not do to their own style.

I'm not really suggesting anybody go out and build one of these as a one off it is probably cost prohibitive to get it approved, and the respective manufactures probably already have plans for something similar already; just wait on it. The main reason they might not be motivated is the need for a whole house transfer and the typical expense to do that, what is their angle for trying to make money off of it. Maybe in the end it is just the flexibility that they provide to the consumer by adding value to the GTI that does it.

Bottom line this is not a technical problem, it is more market demand driven.

[sensij]

Posting is Fed up.

Basically what SolarEdge have done is added a programmable power limit to the GTI standard mode. Basically it will start to cut back if:

if(Pmeas-P_set>0)
DECREASE POWER
else
INCREASE POWER

It will limit cycle at P_set.

If you dynamically change Pmeas and pack it into a RS485 message

VAC_setpoint=120
Read(VACmeas)

Verr = ( VACmeas - VACsetpoint) // voltage control error
VTerr = dt*Verr+VTerr // Integral error
Pmeas = Kp*Verr + Ki* VTerr // PI control

send_485(Pmeas)

This has some merit, but I would be worried about the relative communication cycle time of the RS485 and the response time required to prevent an overvoltage condition if a big load drops out. SolarEdge claims fast response time though, at least when using the embedded communication, so maybe that fear is misplaced.

Designing a circuit with a dump load behind a thermally protected MOV might buy enough time for the digital control loop to take action. With that, you could install the meter to watch the power going to the dump load and use the RS485 as SolarEdge intends, taking action to reduce the power generated whenever power at the meter is detected (See example two, 0 W feed-in, in this white paper). The non-linear nature of the MOV might screw that up though... the idea of faking the Pmeas with an independent control loop on voltage might be better.

[posplayr]

This has some merit, but I would be worried about the relative communication cycle time of the RS485 and the response time required to prevent an overvoltage condition if a big load drops out. SolarEdge claims fast response time though, at least when using the embedded communication, so maybe that fear is misplaced.

Designing a circuit with a dump load behind a thermally protected MOV might buy enough time for the digital control loop to take action. With that, you could install the meter to watch the power going to the dump load and use the RS485 as SolarEdge intends, taking action to reduce the power generated whenever power at the meter is detected (See example two, 0 W feed-in, in this white paper). The non-linear nature of the MOV might screw that up though... the idea of faking the Pmeas with an independent control loop on voltage might be better.

Dynamic response is going to be an issue in any event. I would assume any response the solar edge has to faked power measurement is going to to be relatively slow. Unless Solar Edge actual design for voltage control with performance consistent with load variations, I would not rely be too much on it's dynamic response. On the other hand if there is a load cutout, the load dump to a water heater should be very quick and would keep from over voltaging. Same thing for increasing load. So in effect you can look as the load dump as a high bandwidth voltage controller taking up the slop left over from the pseudo power control loop of the "feed-in limitation" control.

RS485 throughputs should easily be able to hit 20-30Hz if not faster if you get the baud rate up to 512K or so. Closed loop bandwidths through the SolarEdge would then probably be limited by about 1/10 of that 2-3Hz. Not really high frequency. With direct control of load dump you should be able to get closer to 20-30 hz response as you only have to flip a switch based on a direct VAC measurement but even 1/2 cycle of 60 hz is 8msec. I can see needing some type of grid capacity to absorb load turn off transients especially(when voltage will rise) that you are just not going to get the dynamic response from the SolarEdge loop.

[bcroe]

This has some merit, but I would be worried about the relative communication cycle time of the RS485 and the response time required to prevent an overvoltage condition if a big load drops out. SolarEdge claims fast response time though, at least when using the embedded communication, so maybe that fear is misplaced.

Designing a circuit with a dump load behind a thermally protected MOV might buy enough time for the digital control loop to take action. With that, you could install the meter to watch the power going to the dump load and use the RS485 as SolarEdge intends, taking action to reduce the power generated whenever power at the meter is detected (See example two, 0 W feed-in, in this white paper). The non-linear nature of the MOV might screw that up though... the idea of faking the Pmeas with an independent control loop on voltage might be better.

I would ask if either of you have actually built some power converters? I think
the whole idea is just this side of impossible, and all this discussion is entirely
retorical.

If I wanted to generate some island AC direct from panels, I might get a 48VDC
input sine inverter. Arrange my panels to deliver Vmp at a lower point in the
inverter range, hopefully Voc would also be in range, or a bit more (series)
circuitry would be needed to limit V applied to the inverter. With a big DC cap
across the inverter input, it ought to run until the load exceeds what the panels
can deliver at that moment, and it will shut down. Guess it will oscillate at that
point, what did you expect?

Using higher voltage panels, use an MPPT controller designed for a 48V battery.
In other words, build a classic panels-MPPT-battery-inverter system but replace
the battery with a fat cap. Bruce Roe

[sensij]

Using higher voltage panels, use an MPPT controller designed for a 48V battery.
In other words, build a classic panels-MPPT-battery-inverter system but replace
the battery with a fat cap. Bruce Roe

Agreed.. the most practical way to power an island would be to just bypass the grid-tie inverter completely and go through redundant off-grid equipment as you suggest, be it with batteries or a fat cap (with a hard interlock disconnecting the grid when making the transfer).

The rest of the discussion is somewhat rhetorical, but since SolarEdge has built in the ability to zero-backfeed with the grid-tie inverter, tapping into that logic in a way that would power an island is (to me) an interesting thought experiment. I don't need to build power conversion equipment, SolarEdge has already done that. The problem is figuring out the most responsive way to use the meter feedback control loop to regulate the PV power.

[sensij]

And since I can't edit... using hybrid equipment would be even more practical than the grid-tie / off grid system in parallel, but since the discussion is mostly around how to island with a grid-tie inverter, the "just use hybrid" solution is not interesting.

[SunEagle]

And since I can't edit... using hybrid equipment would be even more practical than the grid-tie / off grid system in parallel, but since the discussion is mostly around how to island with a grid-tie inverter, the "just use hybrid" solution is not interesting.

I wish I knew why I can edit on my work computer but not my home.

Both running windows 7 and both using Foxfire browser. I also log off from SPT on both computers and while I start or reboot my home laptop every day I do the same for my work Laptop about 1 a week.

There has to be something that the SPT Forum Software is either not recognizing on my home computer or is somehow stopping some functionality on it.

I can't even post with spaces between paragraphs on the home one.