Single Phase Inverters on 208 3 Phase

[jaggedben]

Perhaps I am confused on a couple of things. It appears to me that if I have X current and Y voltage coming from a group of PV panels to get 7.6 kW out of the inverter on 240 and connect the same configuration to 208, I just LOST 1 kW on each converter. Am I confused on that part?

First, as I said already, you wouldn't (necessarily) connect the same group of panels. You would connect fewer panels to the inverter with the lower power rating.

Second, more importantly, the energy you get ultimately depends on the sunshine. Whenever sunshine doesn't reach the inverter power limit, you are losing nothing. See again the thread I posted about clipping. If you connect 1.2kW of PV panels to 1kW of inverter you will typically lose only about 1% of annual energy production to clipping (not 20%).

On the transformer aspect, would a typical 7.6 kW single phase inverter provide one sine wave per 1/60th of a second, or two offset by 120 degrees to match the two phases of the two legs? I would think the same energy loss in combining two phases into one for consumption ((2) 120 legs = 208) would produce further losses as this process is attempted to be reversed, especially if the inverter is only sending one signal per 1/60th of a second.

A typical 240 or 208V grid-tied solar inverter provides a single line-line sine wave, and the transformer it's connected to provides the line-to-neutral voltage and current. Pretty much the same as the utility does on the primary of the transformer.

 

[bellington]

Yes. The inverter can take your X amps @ Y volts DC and either convert it to Z amps @ 240V, or Z * (240/208) amps @ 208V. Same power production, to first order.

[Although it is theoretically possible that the maximum DC voltage allowed with the inverter set to 208V AC output would be lower than the maximum DC voltage allowed when it is set to 240V AC output; not sure if any commercially available products have that limitation. That could require a different stringing arrangement.]


A 208V inverter for use on a 208Y/120V doesn't need to act as two 120V inverters, it can just act a 208V 2-wire source. You seem to have some misconceptions about 3 phase power.

Cheers, Wayne

The inverter specifications recognize the loss and states the output in apparent power will only be 6.6 kW on 208. Your statement is asking the inverter to make up the difference in power loss by increasing the current. It can't.

For a single phase inverter to provide a matching output to two legs of a 3 phase transformer, it must provide two separate sine waves every 1/60th of a second that are offset by 120 degrees to the same side as the two 120 legs. I don't think that is a misconception, that's the reality of the situation. Going backwards through the transformer will produce less than the original 120 volts for each leg.

 

[winnie]

I'm not seeing what you mean by the above. Agreed that a current-limited inverter will have lower capacity at 208V AC output than at 240V AC output. But I see no capacity loss differential between 1 such inverter vs 3 such inverters connected to a 208Y/120V system. What am I missing?

Cheers, Wayne

Note: not relevant to the OP, because the OP will have multiple inverters in a 3 phase set feeding loads in a 3 phase set.

I might have had it backwards; there might actually be an _increase_ in capacity. I was simply noting that the phase angle of the L-N loads is not aligned with the L-L source, and whenever you have different phase angles you can get differences in apparent VA of the various systems. I'm more familiar with this going in the other direction, where you have (say) a pair of 20A 120V L-N inverters only able to supply 4160VA of L-L load, but if you have three 20A 120V L-N inverters you can supply 7200VA of three phase load.

-Jon

 

[wwhitney]

The inverter specifications recognize the loss and states the output in apparent power will only be 6.6 kW on 208. Your statement is asking the inverter to make up the difference in power loss by increasing the current. It can't.

The inverter is current limited on its AC output, so yes my previous statement is subject to the restriction that Z * (240/208) amps is still less than the current limit.

Basically each inverter is reduced from a 7.6 kW inverter to a 6.6 kW inverter. As such, you will need more inverters, with fewer panels per inverter, to get comparable performance. But that is the only effect. Once you do that, a given amount of sun on a given configuration of panels will produce the same power output (to first order) for both the 208V and 240V inverter options.

For a single phase inverter to provide a matching output to two legs of a 3 phase transformer, it must provide two separate sine waves every 1/60th of a second that are offset by 120 degrees to the same side as the two 120 legs. I don't think that is a misconception, that's the reality of the situation. Going backwards through the transformer will produce less than the original 120 volts for each leg.

You seem to be thinking that the transformer behaves differently for forward power flow and backwards power flow. It does not for grid-following inverters. The grid connection on the primary side sets up the voltage and phase relationships, and the inverter just syncs to it.

Think of it this way: you have a 480D : 208Y/120V transformer. You put a 50A 208V resistive load on two legs of the secondary. What are the primary currents from that load, and how much real power does that represent? The real power will be 208V * 50A = 10.4 kW, even though the currents on the primary side will be out of phase with their respective voltages. The same thing happens with a grid-following inverter on the secondary that produces 50A at 208V, just with minus signs.

Cheers, Wayne

 

[winnie]

For a single phase inverter to provide a matching output to two legs of a 3 phase transformer, it must provide two separate sine waves every 1/60th of a second that are offset by 120 degrees to the same side as the two 120 legs. I don't think that is a misconception, that's the reality of the situation. Going backwards through the transformer will produce less than the original 120 volts for each leg.

But this is not the situation that you have. You will not have a single inverter trying to interface with the 3 phase system. (And yes, if you had a _single_ inverter there would be an issue, which Wayne and I are discussing.)

You have multiple inverters (ideally a balanced 3 phase set) each producing a single phase output that is phase matched to the particular pair of phases it is connected to. The net result is that your PV system is operating as a 3 phase system, even though individual components are single phase components.

This is no different than the wye secondary of a 480/277V transformer supplying delta connected loads.

-Jon

 

[wwhitney]

I'm more familiar with this going in the other direction, where you have (say) a pair of 20A 120V L-N inverters only able to supply 4160VA of L-L load, but if you have three 20A 120V L-N inverters you can supply 7200VA of three phase load.

The above is somewhat of an accounting slight of hand. In that both systems can supply a maximum single L-L load of 4160VA. The second system is better at supplying multiple L-L loads; it can supply (3) 2400VA L-L loads in a balanced arrangement.

But I think the only impact of this phenomenon for the current discussion is that for conductors and transformers whose size is controlled by the inverter output current, it will be more efficient to have a multiple of 3 single-phase L-L inverters. There will no first order effect on power transmission efficiency when everything is grid-tied and will balance out on the grid even if unbalanced on the premises wiring.

Cheers, Wayne

 

[jaggedben]

For a single phase inverter to provide a matching output to two legs of a 3 phase transformer, it must provide two separate sine waves every 1/60th of a second that are offset by 120 degrees to the same side as the two 120 legs. I don't think that is a misconception, that's the reality of the situation. Going backwards through the transformer will produce less than the original 120 volts for each leg.

I think you are likely overthinking this, at least from the solar side of things. You have a primary source that is powering the transformer from the grid side and provides the waveforms. The primary source is basically filling in all the blanks. A single phase 208 inverter contributes current to the two 120V legs it's connected to, but that current leaves the neutral point to other places. If you balance the outputs of multiple inverters around the three phases then it ultimately finds its way to other solar inverters. The transformer balances things out regardless.

I'm not saying there are no power factor implications, but they would ultimately be addressed by your primary source, the 480/277 'grid'. (Not that the latest smart inverter features shouldn't be enabled to help. I'd be pretty confident your supplier supports that. In a microgrid configuration that's something to be aware of.)

One thing that may help to understand is that the grid tied inverters act (essentially) as current sources, not voltage sources. This means, for one thing, that your solar inverter power factor isn't low when measured on its own circuit.

 

[bellington]

But this is not the situation that you have. You will not have a single inverter trying to interface with the 3 phase system. (And yes, if you had a _single_ inverter there would be an issue, which Wayne and I are discussing.)

You have multiple inverters (ideally a balanced 3 phase set) each producing a single phase output that is phase matched to the particular pair of phases it is connected to. The net result is that your PV system is operating as a 3 phase system, even though individual components are single phase components.

This is no different than the wye secondary of a 480/277V transformer supplying delta connected loads.

-Jon

The aspect of "a single phase output that is phase matched to the particular pair of phases..." is my concern. If one inverter is connected to phases A-B, the single phase output will either match phase A, match phase B, match an average between the two, or simulate both phase A and phase B, creating a sine wave for phase A and another for phase B 120 degrees later. If it could do the latter, then it would almost become one leg of a three phase inverter. I am almost certain it does not do the latter.

 

[wwhitney]

The aspect of "a single phase output that is phase matched to the particular pair of phases..." is my concern. If one inverter is connected to phases A-B, the single phase output will either match phase A, match phase B, match an average between the two, or simulate both phase A and phase B, creating a sine wave for phase A and another for phase B 120 degrees later. If it could do the latter, then it would almost become one leg of a three phase inverter. I am almost certain it does not do the latter.

The above would only make sense if you had a 3-wire inverter putting out power using L1, L2, and N. Then in the last sentence you'd be generating power on two of 3 phases of a 3 phase system, not just one phase.

But my understanding is that single phase grid following PV inverters are generally just 2-wire sources. They sync to the L-L voltage waveform and push out current in phase with that. If they even have a N connection, it is just used for qualification that nothing funny is going on on the grid side that would preclude them from pushing out power. I.e. they might check that the L-N voltage is close enough to 120V, without regard to the L1-N and L2-N phase difference. But they generally don't use N in the power output circuit..

Basically you seem to be concerned about a non-issue for grid-following inverters. All the 3-phase complexity is handled by the voltage system that the inverter(s) are syncing to. All the inverters need to do is push out current in phase with the 2 L-L wires they are connected to.

Now grid-forming inverters, which can set up a microgrid when disconnected from the utility grid, that's a different story. There the only way multiple 2-wire inverters can jointly form a 3 or 4-wire microgrid (be it 120/240V single phase, 3 phase delta, or 3 phase wye) is for the inverters to communicate with each other and jointly act as a 3 or 4-wire inverter.

Cheers, Wayne

 

[synchro]

The aspect of "a single phase output that is phase matched to the particular pair of phases..." is my concern. If one inverter is connected to phases A-B, the single phase output will either match phase A, match phase B, match an average between the two, or simulate both phase A and phase B, creating a sine wave for phase A and another for phase B 120 degrees later. If it could do the latter, then it would almost become one leg of a three phase inverter. I am almost certain it does not do the latter.

I think a source of confusion is the common use of the terms "phase A", "phase B", "phase C" when referring to a one of the ungrounded conductors in a 3-phase system, which can be misleading. A phase shift only has meaning on a relative basis between two voltage or current waveforms (which are often sinusoidal in AC power systems). And so the A-N voltage waveform and the B-N voltage waveform would normally be 120 degrees apart in a 3-phase system, or 180 degrees in a common 240/120V single (aka split-phase) system. If a 2-wire single-phase inverter is connected across lines A and B, then it would output a current that is in-phase (i.e., at 0 degrees) with the voltage across A and B (unless it is programmed for and capable of power factor correction with a non-zero angle). This is true regardless of the phase shift that is between the A-N voltage and the B-N voltage.

If there is only one single-phase inverter and it is connected across lines A and B, its current will be at 30 degrees from the A-N voltage and 30 degrees from the B-N voltage. But if we connected another inverter across B and C that is outputting the same amount of current, then the sum of the two inverter currents provided to B will be in-phase with the B-N voltage because one current leads by 30° and one lags by 30°. And if we add a third inverter across A and C with the same output current level, then all of the line currents contributed by the inverters will be in-phase with their L-N voltages. This is one of the reasons for balancing the inverters across A, B, and C. What I just described is identical to what happens when you put balanced L-L loads on a 3-phase system.

I see Wayne has just addressed these same issues in his post, and has provided more specific info relevant to PV system implementation.

 

[bellington]

All the inverters need to do is push out current in phase with the 2 L-L wires they are connected to.


Cheers, Wayne

Each pair of 2 L-L wires are phases A-B, B-C, or C-A. Can a single phase inverter output 2 phases to match any of those pair of phases?

 

[winnie]

Echoing @synchro and @wwhitney phase difference is _meaningless_ if you are talking about a 2 wire connection.

The connection A-B has a _single_ sine wave, with a single frequency, magnitude and phase angle. The inverter matches this sine wave.

The fact that A-N and B-N have _different_ phase angles is not relevant. All that matters is that the inverter match the A-B sine wave.

When you have 3 inverters, connected A-B, B-C, and C-A, then each will see a _single_ sine wave, and the combination of all 3 will be a properly balanced 3 phase set.

-Jon

 

[wwhitney]

Each pair of 2 L-L wires are phases A-B, B-C, or C-A. Can a single phase inverter output 2 phases to match any of those pair of phases?

The short answer is that there are no "2 phases" for it to match. Let me expand on synchro's and winnie's comments:

If you have one wire, which we might call A, then you can measure the current on that wire. If we want to measure voltage, that's defined between two different points in a circuit, which could be two wires, say the pair A-N. [And if the wires are carrying current, and aren't superconductors, then the voltage we measure between those two wires will depend on where we measure along the length of each wire, due to the resistive voltage drop that occurs on the wire.]

But talking about phase only makes sense when we have two different waveforms. So for two voltage waveforms, we would need 4 wires, or 3 wires with one wire used twice. So we can talk about the phase difference between the A-N voltage and the B-C voltage. Or between the A-N voltage and the B-N voltage. [Or the A-N voltage and the A-N voltage, but that phase difference will always be 0, so we can skip the case of 2 wires with each wire used twice.]

So if you're talking about a single wire, say A, it's best to call it a leg or line; we can call it a phase as slang to refer to the A-N voltage's phase in comparison with some other voltage's phase, but that's the only sense in which it's a phase, and that's rarely what we really mean.

The upshot is if you have only two wires, there's only one voltage waveform, and there's no sense of voltage phase in that part of the circuit on its own. But if you put a load or source across those two wires, then you can also measure the current through that load or source. That can give you a current waveform, and we can talk about the phase relationship between the current and voltage waveforms. This is where the notion of "power factor" comes in. For a resistive load, the current waveform will have no phase difference (it will be in phase) with the voltage waveform.

Likewise, a 2-wire single phase inverter in the usual case will just push out current in phase with the single voltage waveform it sees.

Cheers, Wayne

 

[bellington]

Echoing @synchro and @wwhitney phase difference is _meaningless_ if you are talking about a 2 wire connection.

The connection A-B has a _single_ sine wave, with a single frequency, magnitude and phase angle. The inverter matches this sine wave.

The fact that A-N and B-N have _different_ phase angles is not relevant. All that matters is that the inverter match the A-B sine wave.

When you have 3 inverters, connected A-B, B-C, and C-A, then each will see a _single_ sine wave, and the combination of all 3 will be a properly balanced 3 phase set.

-Jon

If connection A-B in a 3-phase wye system were actually a single sine wave (while several measuring devices would indeed average the two into one, there are two distinct waves), then A-N would be 120V, B-N would be 120V, and A-B would be 240V. But, since A-B is a combination of 120V A-N phase A and 120V B-N phase B, then the combination of those separate phases combined is the result of the addition of those two waves separated by 120 degrees and equal 208 volts, not 240. The 32 volts is lost in the combination of the two waves, with sections of A sine wave rising correspond to B sine wave falling and thus cancel each other.

 

[jaggedben]

...Can a single phase inverter output 2 phases ...

No. Because it's a single phase inverter. ;-)

Also, it doesn't need to. Any more than a 208 load needs to draw from 2 phases.

 

[jaggedben]

If connection A-B in a 3-phase wye system were actually a single sine wave...

It is.

The sum sin(x)+sin(×+120deg) is also a sin wave.

 

[wwhitney]

If connection A-B in a 3-phase wye system were actually a single sine wave (while several measuring devices would indeed average the two into one, there are two distinct waves), then A-N would be 120V, B-N would be 120V, and A-B would be 240V.

No. The voltage A-B is always a single sine wave (for AC systems). Its definition depends only on A and B; N is not relevant and there may be no N. You just measure the voltage between the two points as time varies, graph voltage versus time, and you get a sinewave.

Now if you happen to have 3 wires A, B, and N, then you can take 3 different voltage measurements: A-N, N-B, and B-A. So you get 3 different sine waves, and you can ask how they are related, both in magnitude (RMS amplitude) and phase. [The choice of letter N here is suggestive of the application we have, but at this point is it arbitrary, not necessarily the neutral point of some polyphase system.]

Kirchoff's voltage law tells you that the sum of these 3 waveforms has to be zero, i.e. the sum of the voltages is zero at any point in time. So that means there's really only one degree of freedom unspecified in the above (if you fix the A-N and N-B magnitudes), e.g. the phase angle between A-N and N-B.

So take the case that A-N and B-N both have magnitude 120V. Then if the phase angle is 0, B-A has magnitude 240V (but as a signed quantity is the negative of A-B). This is your usual 120/240V single phase system. If that phase angle is 60 degrees, then B-A has magnitude 208V; this is your usual 208Y/120V system. In theory any other phase angle is possible, with corresponding B-A voltage magnitudes between 0V and 240V.

Anyway, I think the relevant point for your initial concern is that the 32V difference between the two cases in the previous paragraph is not a "loss." Certainly not in the sense of power.

Cheers, Wayne

 

[jaggedben]

...

Now if you happen to have 3 wires A, B, and N, then you can take 3 different voltage measurements: A-N, N-B, and B-A. So you get 3 different sine waves, and you can ask how they are related, both in magnitude (RMS amplitude) and phase. ...

And for the record this is what the inverter almost certainly does, to determine what kind of source it's connected to and therefore the appropriate voltage window to operate in.
If A-N and B-N are 180degs apart (or inverted, if you prefer), the inverter will only output if A-B is within the voltage window for 240V nominal. If they are 120deg apart, it will operate if A-B is appropriate for 208 nominal. If the phase shift isn't tolerably close to one of those, it won't operate.

(Some inverters are able to operate without the neutral connection, but in my experience they require you to alter the default settings to do so. i.e. tell it what the grid voltage is when it can't tell on its own.)

 

[wwhitney]

But, since A-B is a combination of 120V A-N phase A and 120V B-N phase B, then the combination of those separate phases combined is the result of the addition of those two waves separated by 120 degrees and equal 208 volts, not 240. The 32 volts is lost in the combination of the two waves, with sections of A sine wave rising correspond to B sine wave falling and thus cancel each other.

To put things a different way, the partial cancellation you correctly describe still gives a sine wave as the result.

And so if you measure a 208V sinewave between A-B, you (and electrical theory) have no way of knowing if there happens to be an N elsewhere in the circuit for which A-N and B-N are both 120V and 120 degrees apart. It could be that A and B come from a 2-wire single phase engine generator that has been set to put out 208V.

Cheers, Wayne

 

[bellington]

To put things a different way, the partial cancellation you correctly describe still gives a sine wave as the result.

And so if you measure a 208V sinewave between A-B, you (and electrical theory) have no way of knowing if there happens to be an N elsewhere in the circuit for which A-N and B-N are both 120V and 120 degrees apart. It could be that A and B come from a 2-wire single phase engine generator that has been set to put out 208V.

Cheers, Wayne

Agreed from the load side. You can always measure the results of the two waves and they can be easily obscured with many measuring devices. An oven doesn't care if it is being fed two phases offset by 120 degrees; it only sees the results of the addition of the two sine waves and the loss of 32 volts. RMS sees the result of those two waves and the loss of 32 volts. Apparently, a single phase inverter also sees the resultant and outputs the sum of the two phases. Going back into leg A and B of 208/120Y transformer, does it sync to phase A, B, or 60 degrees between the two? I want to make sure I can get power back to the battery in another building.