DC to DC Converters / Optimizers

Status
Not open for further replies.
wwhitney said:
...

Also, in the Solaredge architecture, is some of the functionality of a normal string inverter offloaded to the optimizers? I.e. theoretically the inverter itself only has to deal with a more limited range of DC input voltages, so one of its stages might be smaller/less capable than a typical string inverter, due to the DC-DC capabilities of the optimizers.

Cheers, Wayne
Click to expand...
Most simply, MPPT is done at the optimizers and not the inverter.
 
Carultch said:
To follow through with your example, we'll consider modules with an Imp of 5A, and a Vmp of 50V, on an inverter with 400V as the operating input voltage. This would make it a 250W module at this condition of irradiance and temperature. Given 17 single module optimizers in series, that are all uniformly performing at 250W, the total power on this string is 250W * 17 = 4250W. In order to satisfy P=I*V, that means the output current would have to be 10.625A. The input to each optimizer would be 5A and 50V, and the optimizer would transform this to 10.625A and 23.53V.
Click to expand...
This example helped me a lot, thanks! How about the voltage correction factors from Table 690.7(A) (2017 NEC), would you still apply those as well in these optimizer calculations? Just checking if they do still apply.
 
Grouch1980 said:
This example helped me a lot, thanks! How about the voltage correction factors from Table 690.7(A) (2017 NEC), would you still apply those as well in these optimizer calculations? Just checking if they do still apply.
Click to expand...
No. Because the optimizers output voltage doesn't change with temperature.
 
I was thinking on the basic problem of 'how do the optimizers negotiate the voltage distribution (between optimizers)', and realized that the problem is actually quite simple.

The absolute physics requirement is that all optimizers operate at the same output current, because they are connected in series.

So all that is necessary is for the inverter to command the target current value to all of the optimizers. Each optimizer then adjusts its output voltage to whatever gives that target current. No negotiation between optimizers needed.

If the string voltage at the inverter is too low, then the inverter commands a reduced target current. Too high and it commands increased current.

Does this match what real hardware does?

Thanks
Jon
 
Grouch1980 said:
This example helped me a lot, thanks! How about the voltage correction factors from Table 690.7(A) (2017 NEC), would you still apply those as well in these optimizer calculations? Just checking if they do still apply.
Click to expand...

For the input voltage of the optimizer, this does still apply. The optimizer has a maximum input voltage, and the temperature corrected Voc of a single module needs to be less than this. It is rare that this factor governs a design, because rapid shutdown already limits the frigid Voc to 80V. Optimizers are designed with some headroom on the 80V standard.

I have rarely used the voltage correction factors, because it is more accurate to take the temperature coefficient from the module datasheet. Only reason I can see to use the factors from the NEC for this calculation, is for older modules or for custom modules that don't have this figure published on a datasheet,

This voltage never propagates beyond the optimizer, so unlike with normal strings, with SolarEdge optimizer strings, you don't need to accumulate the frigid Voc among the optimizers in series. This answer is Solaredge specific.
 
Grouch1980 said:
How is the output from the optimizers not affected by temperature, but the output from the solar modules is (when there are no optimizers?)
Click to expand...
Because the raw output of the solar cells is determined by physics of sunlight hitting them, which is affected by temperature. And the output of the optimizers is determined by electronic programming, which isn't.

690.7 still technically applies to the conductors between panels and optimizers, but that isn't at all consequential.
 
winnie said:
...

So all that is necessary is for the inverter to command the target current value to all of the optimizers. Each optimizer then adjusts its output voltage to whatever gives that target current. No negotiation between optimizers needed.
...
Click to expand...

I don't think it's that simple because the inverter doesn't know what the target current should be without gathering info from the optimizers. Remember, part of the point of the optimizers is to do MPPT on each panel, and they will not all output the same power as each other.

I could be wrong but my guess is that if an optimizer has more (or less) power available to output, it simply raises (or lowers) its voltage. And probably the inverter can respond by pulling more current, thus lowering the string voltage back to the target. From the inverter's point of view the target voltage is more important for its efficiency than the current is.
 
jaggedben said:
I don't think it's that simple because the inverter doesn't know what the target current should be without gathering info from the optimizers. Remember, part of the point of the optimizers is to do MPPT on each panel, and they will not all output the same power as each other.

I could be wrong but my guess is that if an optimizer has more (or less) power available to output, it simply raises (or lowers) its voltage. And probably the inverter can respond by pulling more current, thus lowering the string voltage back to the target. From the inverter's point of view the target voltage is more important for its efficiency than the current is.
Click to expand...
Central to all this is the fact that a DC to DC converter changes the voltage by trading it for current so that for the optimizer Pin = Pout, and P is determined by insolation on the module. For an unshaded string of identical modules in the same orientation the voltage across the terminals of the optimizers is all the same, but for spotty shade, or different orientations, or different modules in a string (all are OK) the voltage at each output is different but the total string voltage is the the inverter's operating voltage and the current is whatever the highest producing module is producing up to 15A for single modules per optimizer systems. The voltage and current output for an optimizer have a range of operation, so there is a minimum and maximum number of modules you can put in a string, and that depends on the modules and inverter used; there is a table on the optimizer data sheet that provides you the numbers to work that out.
 
Carultch said:
For the input voltage of the optimizer, this does still apply. The optimizer has a maximum input voltage, and the temperature corrected Voc of a single module needs to be less than this. It is rare that this factor governs a design, because rapid shutdown already limits the frigid Voc to 80V. Optimizers are designed with some headroom on the 80V standard.
Click to expand...
jaggedben said:
Because the raw output of the solar cells is determined by physics of sunlight hitting them, which is affected by temperature. And the output of the optimizers is determined by electronic programming, which isn't.
Click to expand...
Thanks guys!
 
Just a bit of a digression on DC to DC converters.
We made quite of those mainly for paper mills. The main supply and dropped from 11kV to 600Vac. It was then rectified to 12 or 24-pulse fixed DC and distributed to DC variable speed drives. These drives could operate from crawl speed to maximum.

On traditional DC drives one of the problems was poor factor. The DC converters removed that at a stroke,
 
Carultch said:
To follow through with your example, we'll consider modules with an Imp of 5A, and a Vmp of 50V, on an inverter with 400V as the operating input voltage. This would make it a 250W module at this condition of irradiance and temperature. Given 17 single module optimizers in series, that are all uniformly performing at 250W, the total power on this string is 250W * 17 = 4250W. In order to satisfy P=I*V, that means the output current would have to be 10.625A. The input to each optimizer would be 5A and 50V, and the optimizer would transform this to 10.625A and 23.53V.
Click to expand...
Using your example again... if each module is operating at 250 watts, meaning each optimizer is operating at 250 watts, and as this would be the ideal condition of max power for each optimizer, 10.625 amps would be the max current delivered to the inverter (if any module produced less wattage due to shading, this would bring the optimizer output current to below 10.625 amps). Would I size the fuse at the inverter, and the source conductors going to it, at 10.625 amps x 1.25 (for continuous current)? so the fuse for this string would be 15 amps, with the wiring sized at #12 AWG. Is this correct?

Also, in a string of optimizers, are they always trying to deliver a constant total voltage? So in this example again, they always try to achieve 400 volts total? or can the total voltage vary?
 
Grouch1980 said:
Using your example again... if each module is operating at 250 watts, meaning each optimizer is operating at 250 watts, and as this would be the ideal condition of max power for each optimizer, 10.625 amps would be the max current delivered to the inverter (if any module produced less wattage due to shading, this would bring the optimizer output current to below 10.625 amps). Would I size the fuse at the inverter, and the source conductors going to it, at 10.625 amps x 1.25 (for continuous current)? so the fuse for this string would be 15 amps, with the wiring sized at #12 AWG. Is this correct?

Also, in a string of optimizers, are they always trying to deliver a constant total voltage? So in this example again, they always try to achieve 400 volts total? or can the total voltage vary?
Click to expand...

SolarEdge has white papers on how to design for the NEC, with their systems in mind, as well as how to relate terms you are familiar with for conventional strings to the equivalent term you'd use with their systems. There is also a white paper that explains why string fusing is not needed for up to 3 strings on the same power stage, which is the case for most SolarEdge systems. The physics behind the need for string fusing on conventional isn't applicable to SolarEdge strings. I expect an update on this will come sometime soon to address what is required for 4 strings, a feature of newer models. At one point, SolarEdge had fusing built-in to their inverters where they anticipated 3 strings, but that has since been removed. The algorithm for fuse sizing is 1.25*Imax, if it is required.

This is a rough guide.
Isc straight = no applicable equivalent.
1.25*Isc gets replaced with Imax.
1.56*Isc gets replaced with 1.25*Imax.
Imp gets replaced with Pstc/Vnom.
Vmp gets replaced with Vnom.
Voc gets replaced with Vmax.
The string voltage of 1 Volt per optimizer is called the shutdown standby voltage. This is what "open circuit voltage" would be, if you measure an open circuit. But it is misleading to call it that, when you usually expect "open circuit voltage" to be a maximum voltage.

Imax = maximum optimizer output current (from optimizer datasheet)
Vnom = inverter's target for string voltage (from inverter datasheet)
Vmax = inverter's maximum input voltage (from inverter datasheet)
Pstc = nameplate power at STC of the string

Second question:
The inverters generally aim for the optimizer strings to to have a voltage of Vnom, which is 400V in this example. This isn't exact, and it depends on the happenstance of how the optimizers each respond to the algorithm. The feedback loop aims to get the voltage as close to the 400V target as it can. In the event that the maximum input current of the inverter would otherwise be exceeded, the inverter raises the target input voltage to compensate. At maximum, the input voltage will be Vmax, which is a rare event.

As an example, given a maximum input current of the inverter of 40A, a Vnom of 400V, and an array power of 17 kW (assuming it is all usable) among all 3 strings, the expected inverter input current is 40A and the expected input voltage is 425V.
 
Last edited:
Thanks! I'll search for those white papers.

Actually what i meant to add, but it was too late so i couldn't edit my comment, do you size the wiring between the largest of 1.25 * Imax OR after applying any adjustment / temperature correction factors... that's how you would design without the optimizers, using regular strings (according to Bussmann's SPD handbook)... I wish the SPD handbook would cover optimizers as well. But it looks like i can just go to these white papers to get my answers.

Carultch said:
This isn't exact, and it depends on the happenstance of how the optimizers each respond to the algorithm. The feedback loop aims to get the voltage as close to the 400V target as it can.
Click to expand...
Understood, thanks.
 
Grouch1980 said:
Do you size the wiring between the largest of 1.25 * Imax OR after applying any adjustment / temperature correction factors?
Click to expand...

I would see that more as an AND, than an OR.

1.25*Imax would apply to terminations (usually the 75C ampacity column)
Imax/(total derate) would apply to wire (usually the 90C ampacity column)

Whichever of the above results in a larger size, is the the calculation that governs.

On top of that, if an OCPD is involved, compliance with article 240 is also required. This usually means both terminations and derated wire ampacity have to "round up" to the size you are using. See 240.4(B) for more detail.
 
Status
Not open for further replies.
Top