No neutral in wye system

[electrofelon]

It's a question with several answers.

1) Some inverter topologies use the neutral as a current carrying conductor. If the 3ph inverter is made up internally of 3 power units connected in a WYE configuration for example, then any imbalance in energy production between the 3 will require the use of a neutral. It seems like a lot of transformerless inverters use the neutral as a current carrying conductor.

I dont understand that though: If the 3 power units were connected to a common DC buss, you would think they would - or could at least easily be made to - have equal power processing thus be a balanced system. Now if the inverter had 3 MPPTs, than it is very understandable that each one would no process the same amount of power, however I have never seen an inverter with 3 mppt's.

 

[GoldDigger]

I dont understand that though: If the 3 power units were connected to a common DC buss, you would think they would - or could at least easily be made to - have equal power processing thus be a balanced system. Now if the inverter had 3 MPPTs, than it is very understandable that each one would no process the same amount of power, however I have never seen an inverter with 3 mppt's.

Well, it's sort of a chicken and egg situation. To make sure that the three inverter sections with the same DC supply produce the same ouput power is not terribly hard, although there would be a margin of error. Differences in component tolerances would require an actual comparison of output rather than just an intention to make the three units match.
But with a grid interactive inverter if the three grid wye voltages are not perfectly matched then matching power output would not result in matching currents.
And if you just match current, the easiest way to do that would be to sense and minimize current in the neutral. But if there is no neutral, you cannot do that.
Finally, if each inverter section is trying to match the grid's line to neutral voltage for that phase you surely need to have access to the neutral for voltage measurement even if you are not intending to run current through it.

If the three inverter sections each go line to line then the grid reference can be taken from the three lines and no neutral is needed.
Among other things the manufacturer will have to take into account during the design process that line to line inverter sections will have to operate at a higher voltage than line to neutral inverter sections, requiring a higher DC input voltage, which will either put a minimum on string voltage or require an MPPT circuit that can either boost or buck the DC voltage.

 

[ggunn]

Well, it's sort of a chicken and egg situation. To make sure that the three inverter sections with the same DC supply produce the same ouput power is not terribly hard, although there would be a margin of error. Differences in component tolerances would require an actual comparison of output rather than just an intention to make the three units match.
But with a grid interactive inverter if the three grid wye voltages are not perfectly matched then matching power output would not result in matching currents.
And if you just match current, the easiest way to do that would be to sense and minimize current in the neutral. But if there is no neutral, you cannot do that.
Finally, if each inverter section is trying to match the grid's line to neutral voltage for that phase you surely need to have access to the neutral for voltage measurement even if you are not intending to run current through it.

If the three inverter sections each go line to line then the grid reference can be taken from the three lines and no neutral is needed.
Among other things the manufacturer will have to take into account during the design process that line to line inverter sections will have to operate at a higher voltage than line to neutral inverter sections, requiring a higher DC input voltage, which will either put a minimum on string voltage or require an MPPT circuit that can either boost or buck the DC voltage.

It really boils down to RTFM, doesn't it?

 

[jaggedben]

Well, it's sort of a chicken and egg situation. To make sure that the three inverter sections with the same DC supply produce the same ouput power is not terribly hard, although there would be a margin of error. Differences in component tolerances would require an actual comparison of output rather than just an intention to make the three units match.
But with a grid interactive inverter if the three grid wye voltages are not perfectly matched then matching power output would not result in matching currents.
And if you just match current, the easiest way to do that would be to sense and minimize current in the neutral. But if there is no neutral, you cannot do that.
Finally, if each inverter section is trying to match the grid's line to neutral voltage for that phase you surely need to have access to the neutral for voltage measurement even if you are not intending to run current through it.

If the three inverter sections each go line to line then the grid reference can be taken from the three lines and no neutral is needed.
Among other things the manufacturer will have to take into account during the design process that line to line inverter sections will have to operate at a higher voltage than line to neutral inverter sections, requiring a higher DC input voltage, which will either put a minimum on string voltage or require an MPPT circuit that can either boost or buck the DC voltage.

Isn't source voltage another consideration? i.e. there's an optimum DC voltage for a given grid voltage (and number of phases?), and this is especially constrained when your inverter doesn't have a transformer. It seems to me that manufacturers are taking their technology they developed for 240 single phase tweaking it for 277V L-N in 480/277 wyes. Because presumably that's easier than stretching their technology to deliver at 480V L-L. I'd guess you could have the same inverter components that can deliver 240V single phase, 208V L-L 3-phase, or 277V L-N 3 phase, depending how their arranged internally. Simplifies your manufacturing operation. I believe every inverter I ever saw that said it could be configured for delta without neutral was for 280V, not 480V. I don't know if a 600V vs. 1000V DC limit also makes a difference here, but maybe. (If that's raised to 2000V maybe things change?)

 

[pv_n00b]

Isn't source voltage another consideration? i.e. there's an optimum DC voltage for a given grid voltage (and number of phases?), and this is especially constrained when your inverter doesn't have a transformer. It seems to me that manufacturers are taking their technology they developed for 240 single phase tweaking it for 277V L-N in 480/277 wyes. Because presumably that's easier than stretching their technology to deliver at 480V L-L. I'd guess you could have the same inverter components that can deliver 240V single phase, 208V L-L 3-phase, or 277V L-N 3 phase, depending how their arranged internally. Simplifies your manufacturing operation. I believe every inverter I ever saw that said it could be configured for delta without neutral was for 280V, not 480V. I don't know if a 600V vs. 1000V DC limit also makes a difference here, but maybe. (If that's raised to 2000V maybe things change?)

There is a minimum DC voltage needed to generate a given P-P AC waveform. This was not much of a problem in transformer based inverters, they all put out a voltage in the 200 to 260Vac range and used a transformer to provide the correct output voltage. In transformerless inverters you either have to have a high enough DC input voltage from the array to create the AC voltage you need or you need a DC boost circuit and then you take an efficiency hit so no one wants to use them if they don't have to. Bucking the DC is relatively efficient, boosting is not.

The peak voltage of your standard sinusoidal AC wave form is the sqrt(2)*Vrms. So for a 480V system you need more than 679Vdc. This is why the market for 480V 3ph transformerless inverters did not really kick off until it was easy to install at 1000Vdc. But 240V and 277V transformerless string inverters were fine with a 600V array.

It's actually more complex than this but this is as deep as I want to go here. There is a lot of information on the internet about how inverters work if anyone wants to really dig into it.