Something that I see rarely discussed when it comes to solar power and inverters is how a large part of the expense of an inverter stems from its internal temporary energy storage eelements (could be capacitors, could be inductors).
Assuming illumination is constant in timescales of minutes (so thousands of electrical AC cycles), one can observe the following conundrum: a solar panel produces DC electrical power, while most consumer devices assume AC power input. So when designing (or choosing) an inverter one can make 2 choices in theory.
If one draws on average half the current a solar panel can supply, the other half of the energy will simply recombine in the solar cell, heating it up.
I will assume a resistive load (as a well designed product should not reflect energy back in the grid).
An AC grid voltage is sinusoidal, and for a resistive load the current is also sinusoidal (the relation of course being U=RI)
The power is P=UI=RI^2. so the power is also oscillating sinusoidally, centered about the average power.
This seems to forces us to choose between the following 2 options:
1) have the inverter use all the incoming power (requiring electrical energy storage within the inverter: energy storage capacitors or inductors, which is expensive) or
2) have the inverter draw half the solar panel power (the other half is wasted as heat in the ), or have 2 solar panels (half the energy again wasted) to reach the same output AC power as 1); the inverter can theoretically convert the DC power to AC this way
But in reality there is a third way, if you don't mind having 2 outlets where the AC voltages come out in quadrature. If two AC outlets deliver the same average power with the voltages in quadrature, then the powers will be in counterphase, so no energy storage capacitors or inductors are needed (cheaper inverter per watt), and no energy is wasted (full utilization of the solar panels). The only downside is that you can't deliver the total power to one and the same device. But consider some high-way stop convenience store in the desert with multiple refrigerators, say an even number of off-grid refrigerators, then who cares if half of the refrigerators use AC power in counterphase with the other refridgerators?
A similar reasoning could be used for houses in a street, where the even houses and the odd houses are powered in counterphase (voltages in quadrature), which would substantially decrease the costs for inverters.
The hilarious thing is that so many modern appliances are moving towards variable frequency drives and pile on even more waveform synthesis. Would be nice if they could just accept direct DC input.
> Would be nice if they could just accept direct DC input.
The thing is, lots of appliances do accept DC input (most electronics and probably all inverter-based stuff like washing machines and aircos). If only there was a standard for DC they could adhere to. Then they could add yet another adhesive to proudly proclaim the feat.
>> If only there was a standard for DC they could adhere to.
There are. 12v automotive (13.9v) is one. Then comes 24v, a standard in aviation. There was once a push for 48v in cars so that air conditioners and braking systems could be made all-electric, but it never became widespread.
In any case, neither system appears to have defined a plug standard, which IMO is the real issue as far as supporting end user applications, and where USB and the 12V cigarette lighter plug have been such winners. It's only ever going to be useful for installed applications (RV fridges and the like) if it's something you have to wire in.
You still wouldn’t want to, especially if your house is fairly large. DC transmission line losses are actually pretty big over distances you might think are short.
I'm not sure that's correct. My understanding is that for a given voltage the transmission losses are basically the same (actually possibly slightly better for DC since they don't suffer from the skin effect). The problem comes when you try to get the same power down a low voltage DC circuit which naturally requires high currents.
My understanding is AC is safer at higher voltages than DC - and if you have a choice between 12v DC and 120v AC, the latter needs radically smaller cables.
I haven't seen any truly high quality data on voltage safety, due to the obvious ethical issues with performing properly controlled experiments. It's possible the reason you see 120v in every home but 120v DC almost nowhere is inertia rather than safety.
So, there is a lot of bad, anecdotal information out there about which is worse, AC or DC. The reality is that it _depends on the situation_.
In the case of shock and electrocution DC is _less dangerous_ than low frequency AC (sub ~1kHz). The "let go" currents for DC are several times higher than that of low freq AC, meaning it requires a higher DC voltage to prevent someone from being able to let go. The same is true for the currents where danger of injury and death can occur. DC is still safer than low freq AC.
This has been scientifically tested numerous times in both ethical and non-ethical ways. Here is a paper that shows actual numbers for "let go" currents and dangerous currents vs frequency (from DC up to 10kHz): http://www.wright.edu/~guy.vandegrift/wikifiles/Electric%20s...
In particular look at Fig 3 on page 3 of the above PDF. (One really interesting thing to note in this paper is that women have lower "let go" and dangerous current levels!)
However! There is another factor here where AC can be safer than DC. Fire safety! It is much more difficult to prevent DC from arcing and potentially caused fires than AC. This is because the zero crossover of AC which you mentioned generally causes any arcs to quickly extinguish at lower voltages. DC doesn't cross zero volts and will produce far more arcing at the same voltage.
This is why if you look at the ratings for switches, relays, plugs, etc the DC rating is always much lower than the AC rating.
Why woud AC be safer than DC? AC has a higher peak to peak voltage (factor sqrt(2)), so if it's breakdown voltage related DC would be safer. but in cases dielectric breakdown happens for both AC and DC, I can imagine AC being safer since it regularly crosses zero current, potentially allowing the breakdown plasma to extinguish...
the 12V DC vs 120V AC is an apples oranges comparison, with 12V AC and 120V DC it's the other way around. both would be false comparisons.
why would voltage safety tests be unethical? why would one actually test flesh, instead of the theoretical safety models?
The reason we have AC everywhere is simple, historically it was easier to step up and down with transformers (which don't work with DC), but nowadays DC-DC converters are a solved problem.
I don't know if it's only anecdotal or if there is some actual study to verify it, but the typical reasoning on why AC is "safer" than DC is that AC has a "zero-crossing" point, whereas DC (obviously) does not. Why is this considered "safer" (again, possibly anecdotal)?
Because if you accidentally contact DC at a high enough voltage to shock you, your muscles contract - and stay contracted. AC, on the other hand - at least at the relatively low frequencies typically used (50/60 Hz) - crosses a "zero point" where the voltage is "zero" - and lets your muscles relax - briefly - long enough to be able to move away from the current (or in worst case - ungrip your hands).
Again, I don't know if any study has been done on this potential "mythological" reasoning (I would be surprised if there hasn't) - but that's usually the reasoning given.
> why would voltage safety tests be unethical? why would one actually test flesh, instead of the theoretical safety models?
Well, Edison did it that way because it made for better PR for people to watch criminals or elephants being killed by deadly ac, than to have them read papers on theoretical safety models.
Stepping AC up/down is relatively easy, all you need is a transformer. I'd say the invention of the transistor / IC made it possible for DC. But it took a while to perfect those designs - wall warts changed from transformers in the 1990s?
hmm, mostly mass manufacture, but also very low ON resistance mosfets, very emcheapened "microcontrollers" (yesteryear's microprocessors) with more performance, dedicated SMPS chips, and of course ... china
As others have already noted, cheaper and high-power low-resistance MOSFETs are likely the reason - and that did mostly happen in the 1990s.
Something to look at from that period are hobby-grade RC (radio control) cars. Most used NiCad battery packs, and had some extreme amounts of power behind the motors (which were all brushed DC - BLDC was in the future). These motors pulled a lot of amperage (550 and 750 can styles), which the battery packs could deliver, however, there weren't motor controllers small enough to control that much power.
So instead - pretty much up until the 1990s at some point - hobby RC cars used a "resistor speed controller" - something like this one (also known as a "mechanical speed controller":
Basically it was a multi-tapped high-power resistor (or multiple smaller value high-power resistors) that was tapped in a "variable rheostat" manner with a switch operated by a servo. You would usually have three speeds - high (direct to battery), medium, and low; the resistor would "bleed off" excess current as heat (boy, did they get hot!). Yes, it was inefficient, but it was small, robust, and simple to repair or replace.
Of course, there usually wasn't a "reverse gear" (though I am sure someone hacked something together back then). Most of the time, this wasn't a real issue in the hobby - you spent most of your time going forward.
Such controllers actually have a long history - the earliest electric cars used a similar system (just much larger resistors - usually open coil):
In both cases, switching was done either mechanically, or using large relays or contactors. While it is very inefficient, it is also fairly robust if designed right. Which is why it is still used in a lot of automobiles (though this is rapidly changing with newer models using electronic PWM control) - where?
The AC/heater blower motor! On many cars, there's a "resistor pack" that plugs into the control switch/knob for setting the speed of the blower, and it looks virtually the same as ever - here's one for an older vehicle:
About the only difference is the addition of a heat sink. Newer models from even more recent vehicles don't look much different, and they all work on the same principle. They are usually installed in the blower duct work, so that the air rushing by keeps them cool. Unfortunately, if they are designed improperly, or they don't get enough air cooling (or the fan motor dies) - they can heat up extremely hot and melt or catch the car (ductwork - which is usually plastic) on fire! This is especially true if the fan is on "medium" or "low" speeds and the motor seizes (maximum current draw); high speed wouldn't be a problem because the load would short things out and hopefully a fuse would blow (though - not always - sometimes the "fuse" is the wire itself!). This would cause the resistors to get extremely hot - glowing red even - and can cause a fire. I'm certain more than one automotive fire has started this way.
Today, though, thanks to low cost and highly efficient mosfets - and BLDC motors - more and more cars are implementing true variable speed blowers, and using more efficient motors as well. This comes at a cost of more complexity and (depending on how it's implemented) more difficult to repair/replace control and motor systems, but they tend to be safer, and more efficient (this isn't really an issue with ICE vehicles, but very important on electrics for obvious reasons).
strange that you would call a pair of wires carrying DC a "transmission line".
what makes you think identical lengths of identical cable with the same resistance R powering identical loads R_L will dissipate more heat when carrying DC than AC?
the total resistance of the pair of wires R and the load R_L form voltage divider
in the DC case: P_cable=RI^2
in the AC case: P_cable=R*(I_RMS)^2
they should dissipate the same heat, you may want to brush up:
>For alternating electric current, RMS is equal to the value of the direct current that would produce the same average power dissipation in a resistive load.
and
>Because of their usefulness in carrying out power calculations, listed voltages for power outlets (e.g., 120 V in the USA, or 230 V in Europe) are almost always quoted in RMS values, and not peak values.
The only association with higher resistance losses would be when using low voltages but high currents... and even then the resistance losses would be equally high with low voltage high current AC since the fraction of energy dissipated in the cable versus the load is the same in both cases I^2 R / R_L
correct, but I don't know how large these are typically, an anecdote: I visited a friend who had a lamp socket (with shade) above his desk, and he was complaining about the fluorescent bulb he put in, in comparison with the resistive filament bulbs he usually put in: it was ticking even though the switch in the wall was turned off. At first I suspected a bad contact, but that didn't really make sense. Then I realized what was probably happening: capacitive leakage, so I ask if the socket has 2 switches? sure enough. the distance between the switches was long enough so that the parallel wires embedded in the wall effectively formed a long capacitor, so in both of the off states AC current was flowing through this unintentional capacitor. AC will also have inductive losses yes.
It's ac that has transmission line losses, not dc. DC just has resistive losses, which are the same for ac (rms) and dc, but bigger for low-voltage systems. DC is safer, in most ways, at the same voltage as ac (rms).
So the issue isn't that running your house on dc is less efficient; it's that running your house on 12 volts is less efficient.
Vacuums and hair dryers invariably use universal motors that run the same on ac or dc. AC motors aren't that common in household appliances because their speed is set by the powerline frequency, which is slow. Typically you find them in microwave oven fans.
They're more common in industrial equipment, but a lot of them there are being converted to run off VFDs, which of course internally run on dc.
But currently every appliance has it's own inverter, so running a fixed voltage add through the house would do little good since every DC appliance wants a different voltage.
All those modern lightweight wall warts are switching DCDC converters anyway—they just have a rectifier and capacitor on the front end to get an approximate DC waveform to start from. That's also why modern power supplies are almost always universal for 240@50Hz or 120@60Hz. The frequency is irrelevant and the DCDC can adapt to a huge range of input voltages because the ratio isn't locked in by the wiring in a transformer.
Modern power electronics are not just more flexible, smaller, and lighter, they're more efficient as well.
this is certainly true, but in the mean time there are still a lot of appliances in current operation, just rolled off the assembly line, and still coming off assembly lines that do not accept DC input.
a lot of the material effort that was put in to them would be wasted if we replace them before they break down
and if we keep using them, and keep opting for the single AC outlet inverter, we are wasting energy storage elements at volume of 0.01 Joule per Watt (think of the total PV energy production at least in residential solar [I would be uncomfortably surprised if commercial solar parks don't use this already, but then again that would require the grids to be split in an in-phase grid and out-of-phase grid, which would also be new knowledge for me...])
this also affects adoption, as less expensive storage-less inverters would decrease the time until the solar set-up pays itself back!
Large-scale power generation, including solar parks, is invariably three-phase, which has a more constant total absolute voltage than the two-phase system you're suggesting.
... except I am suggesting a 4 phase system: 2 outlets with each the usual 2 phase AC!
consider 4 phases in 90 degree turns (I wish HN had a button to render LaTeX or so when a reader wants it on demand):
V_0 = V/2 sin(w t + 0 pi / 2)
I_0 = I sin(w t + 0 pi / 2)
V_1 = V/2 sin(w t + 1 pi / 2)
I_1 = I sin(w t + 1 pi / 2)
V_2 = V/2 sin(w t + 2 pi / 2)
I_2 = I sin(w t + 2 pi / 2)
V_3 = V/2 sin(w t + 3 pi / 2)
I_3 = I sin(w t + 3 pi / 2)
Now outlet "in" (in-phase to avoid clashing with "I") is across terminal 0 and 2, and outlet "qu" (quadrature) is across terminals 1 and 3, so the current and and voltages are:
Vin = V/2 sin(w t + 0 pi / 2) - V/2 sin(w t + 2 pi / 2)
= V/2 (sin(w t) - ( - sin(w t) ) )
= V/2 (2 sin( w t) ) = V sin(w t)
Iin = I sin(w t)
Vqu = V/2 sin(w t + 1 pi / 2) - V/2 sin(w t + 3 pi / 2)
= V/2 (sin(w t + 1 pi / 2) - ( - sin(w t + 1 pi / 2) ) )
= V/2 (2 sin( w t + 1 pi / 2) ) = V sin(w t + 1 pi / 2)
= V cos(w t)
Iqu = I sin(w t + 1 pi / 2) = I cos(w t)
and the powers flowing through the outlets are:
Pin = Vin Iin = V I sin(w t) sin(w t) = V I sin^2(w t)
Pqu = Vqu Iqu = V I cos(w t) cos(w t) = V I cos^2(w t)
so the total power is
Ptot = V I (sin^2(w t) + cos^2(w t)) = V I
constant total power out, just like constant DC power into the inverter so no need for storage capacitors / inductors.
QED
It makes sense for solar parks to use 3 - phase because the back-bone power distribution of the grid is 3 phase.
Do you know if the inverters from solar park panels to 3-phase use storage capacitors to convert to 3 phase?
Do you know if any residential inverters are commercially available that split into 4 phase (2 AC outlets) like I describe, without unnecessary storage capacitors / inductors?
>China now leads in total solar energy capacity followed by Europe with 114 GW.
>Remarkably, 64 percent of solar systems in the EU are installed on rooftops, 26 percent of them residential, 18 percent commercial and 20 percent industrial.
so residential solar in Europe is 26% of 114GW = 29.64 GW
so a line frequency of 50Hz in europe implies single AC outlet inverters need an energy store of 0.01 Joule per Watt (in the US only 0.008333... J per W, because they have 60Hz)
This amounts to 296.4 MJ (mega joule) of capacitor / inductor energy storage, elements whose raw materials must be sourced, must be built, must be bought by the consumer aand which are potential points of failure (what is not present can not break down). That could have been avoided. Which we still can avoid for future inverters, without replacing all our consumer electronics with DC consumer electronics.
constant total power out, just like constant DC power into the inverter so no need for storage capacitors / inductors.
Whoa, you're right! Assuming a perfect power factor and perfect balancing on the loads, of course. Your four-phase scheme is very ingenious! I'm sorry I didn't appreciate this at first.
But here's a thing I'm not understanding: yes, the powers sum to a constant. But the way an inverter produces real sinewave power (as opposed to modified square wave) is to PWM an H-bridge to ramp the voltage up and down, filtering it with an inductor (and usually a capacitor, and maybe more than one of each, but those are inessential). An inductor's time integral of voltage, and thus its average voltage drop (disregarding losses to winding resistance, hysteresis, eddy currents, etc.), must be zero to keep its current finite. So the output side of the inductor has the same average voltage as its input side. So, for example, a 25% PWM duty cycle produces 25% of the full-scale output voltage.
However, in your four-phase scheme, when one of the phases (let's say in) is at ±25% output voltage and thus 6.25% peak output power, the other (qu) is at 93.75% output power and thus ±96.8% peak output voltage. This implies that, during that part of the wave, the active in MOSFET needs to be on 25% of the time, while the active qu MOSFET needs to be on 96.8% of the time. That means that between 18.7% and 25% of the time, both MOSFETs are on, so the two phases are actually shorted to each other (though not on the load side of the inductors). Is that okay? I guess it means that the current generated by the solar cells is being shared between the two phases during that time.
Still, it makes me worry that an imbalance of loads or power factors between the four phases could produce some kind of hazardous condition.
You say that the four-phase system doesn't need any energy-storage elements. To a first approximation you're right, again assuming well-balanced, power-factor-corrected loads on the phases: there's no need to store energy harvested close to the zero-crossing for, assuming 50 Hz, the average 5 milliseconds until it can be released close to the peak. 5 milliseconds is a long time, so these storage or filtering elements need to be quite large, 5 millijoules per watt (I think you dropped a factor of ½ in your calculation there, presumably calculating for a full half-cycle instead of a quarter cycle). By contrast, if you're PWMing a sine wave with a PWM frequency of 100 kHz --- a reasonable thing to do with modern IGBTs or power MOSFETs --- the filtering elements only need to store the energy for a maximum of 10 microseconds, and less if the PWM duty cycle is above the minimum. 10 microseconds is 10 microjoules per watt, 500 times smaller. So you only need 0.2% of the energy storage elements you need for what you're describing as typical current systems.
(All that happens if your power factors or loads are imbalanced is that the inverter isn't drawing a consistent amount of current, so potentially the panels become less efficient.)
A thing I'm not sure about is the junction capacitance of photovoltaic cells. A photovoltaic cell is a diode, reverse-biased in normal use, and reverse-biased diodes have a junction capacitance; we'd expect their enormous junctions to have significantly larger junction capacitance than the pF-scale capacitances we see in small-signal diodes. How large is the ½CV² = 50 nJ/microfarad (at V=0.316 V) energy-storage capacity of the PV panel itself? In particular, is it much larger or much smaller than the 5 mJ/W = 5 ms number you'd need for a single-phase inverter to not be wasteful?
As for your questions about the current designs of power-generation inverters and residential inverters, no, I don't know about them. Typically, in both the US and here in Argentina, one side of a normal power outlet is "neutral" (see http://amasci.com/amateur/whygnd.html for the reasons around this) and violating that expectation might cause some problems --- notably, electric shocks from the outside of Edison-screw-type lightbulb sockets.
However, I don't think you need to violate that expectation; you've described a four-phase system, but I think you get the same advantages with a two-phase system, with the phases in quadrature just as you proposed, but with the two phases sharing a neutral wire. The voltage from neutral ("ground") to the in terminal would be V sin(ωt), much as before, while the qu terminal would be at V sin(ωt + ½π), which is to say, V cos(ωt) --- both relative to the neutral wire. In effect you need only a single H-bridge with four MOSFETs (or IGBTs) and two inductors to produce the two voltages, controlled with a scheme slightly different from the usual H-bridge scheme, because it makes sense to have one side of the H-bridge turned on (in, say) while the other side is turned off.
Option 4: output square wave power instead of sinusoidal and hope you don’t destroy the downstream equipment.
In practice, isn’t most of that storage used as part of a boost converter to get a higher output voltage than you input? You either need to switch capacitors between series and parallel or feed a step-up transformer with low-voltage, high-current A/C, which ends up being enough copper wire to get heavy and expensive.
transforming to a higher or lower voltage can be done with small capacitors / inductors (just switch at higher frequency, for each doubling of frequency you can halve the capacitance and thus cost of capacitors / inductors)
but given a desired AC output wattage implies a specific output power, cycled at twice the line frequency, so with only 1 outlet you need to either throw away half the solar power energy, or have storage capacitors or inductors to store energy for the timescale of the line frequency (which is much much longer than the timescale for using capacitors or inductors for merely stepping up / down the voltage.
for example an ideal inverter drawing 10 kW DC and delivering 10 kW AC at 50Hz line frequency will by definition require an energy store of 10 kW / 2 50Hz = 100 J,
square waves would also work, but would still require 2 outlets, and would require the downstream equipment to tolerate it.
You’re assuming there’s a common ground between the two sides of the inverter. If it’s allowed to float, whichever terminal is supposed to be lower at the moment can be connected to the DC ground while the other is connected to the stepped-up voltage.
This is dual to using a full-wave bridge rectifier to get DC from AC, where a half-wave rectifier is simpler but needs energy storage to ride through the negative half of the cycle.
yes, now you display you understand what it would address (a good design that delivers clean sinusoidal wave forms for all combinations of output power -which sum to smaller than or equal to the total supported power, a square region of power combinations- would be non-trivial but is theoretically absolutely feasible without energy storage elements [other than small ones for stepping up or down, control etc])
it would be nice if a flexible (supporting all operating points in the power square) 1 DC-in 2 AC-out quadrature voltage storage-less inverter had an open-source design.
[One can not make a single outlet sinusoidal without equivalent storage capacitors / inductors, its simple mathematics, constant DC average power in - clean AC power out = power stored and released or simply wasted in sinusoidal oscillating fashion (regardless of implementation).
to avoid storing energy, while requiring clean sinusoidal output, 2 AC outlets is the lowest number of output outlets that admit an exact solution since the sum of 2 power sinusoids in counterphase result in a constant output power.]
the power operating point square is delimmited by 0 and 1/2 total power for each AC outlet.
Why? The use case of plugging a solar panel directly into a device is pretty niche, IMO. It's much more common and practical to put solar panels on your house so they can power everything, but you can still draw from the grid at night.
I agree that it would be niche today, but for devices that don't switch location in a house, one time wiring for DC "straight from the panels" so to speak makes sense for those devices that can draw a lot of power: it decreases the energy storage requirement on the inverter
Redesigning them so that in addition to AC, the devices have terminals for DC would probably not add that much to the cost. And off-grid may be niche, but partially because it's inconvenient to connect some of the larger loads that may require AC to your solar panels: if your going to buy energy buffering AC inverters for the house anyway, and don't understand or want to think about the price gradient for the energy buffer, people will obviously choose the setup you describe.
Totally off the wall fourth way (which probably has a ton of reasons it would not work): have shutters in front of each panel, and as power the load draws fluctuates, block and unblock panels so that a given panel only generates power when that power can be fully used.
If the panels are small enough, you could get a pretty good approximation to a sinusoidal output from the array of panels.
The shuttering system doesn't actually have to be shutters. Anything that can block panels with the right timing would do. You could probably do something with rotating discs with holes or slots in them, where the phase between adjacent discs can be adjusted to control have often the holes or slots align to let light through.
How is what you’re proposing different than split-phase, which is already how houses are wired, and how inverters connect? This is why inverters generate 240VAC and why your electrical load panel had alternating circuits across the split phase.
Just provide 3-phase power. The load is nearly constant, and lots of equipment already handles it fine, e.g. motors much prefer 3 phases even at low power, due to this constant power/symmetry.
sure 3-phase would work towards the future, and saves a conductor too, I'm trying to save all consumer equipment (since they universally use 2-phase AC) AND the energy buffer capacitors / inductors in the inverter.
Maybe the real thing here is to get more inductive load equipment shipped with 3-phase as the default. Dryers and stoves are, but they're mostly resistive, and it's for load reasons— 1500W just isn't enough for those applications.
But thinking of air conditioners, fridges, washing machines, central vacs, etc; those are all typically shipped today with a two-phase plug but could use 3-phase. OTOH, if they're not actually driving the motor with the 3-phase and are all just rectifying the power and generating their own waveform with a VFD then there's no point; they should have an option to accept DC.
a single AC outlet still pulses power sinusoidally, could you provide a diagram? the 240V split phase is just with a center tap to ground, so each phase is 120V with respect to ground... I don't see the relevance here (unless I misunderstand you)...
The GP is saying (IIUC) that you should have 2 outlets and opposite phases. But that’s exactly what split-phase looks like in house wiring, right? (Whether that’s how the inverter functions or not).
with 2 outlets he means 4 conductor terminals, and with "opposite phases" he means voltages in quadrature (so that power is in opposite phase, since out of phase voltages means in phase power)?
Assuming illumination is constant in timescales of minutes (so thousands of electrical AC cycles), one can observe the following conundrum: a solar panel produces DC electrical power, while most consumer devices assume AC power input. So when designing (or choosing) an inverter one can make 2 choices in theory.
If one draws on average half the current a solar panel can supply, the other half of the energy will simply recombine in the solar cell, heating it up.
I will assume a resistive load (as a well designed product should not reflect energy back in the grid).
An AC grid voltage is sinusoidal, and for a resistive load the current is also sinusoidal (the relation of course being U=RI)
The power is P=UI=RI^2. so the power is also oscillating sinusoidally, centered about the average power.
This seems to forces us to choose between the following 2 options:
1) have the inverter use all the incoming power (requiring electrical energy storage within the inverter: energy storage capacitors or inductors, which is expensive) or
2) have the inverter draw half the solar panel power (the other half is wasted as heat in the ), or have 2 solar panels (half the energy again wasted) to reach the same output AC power as 1); the inverter can theoretically convert the DC power to AC this way
But in reality there is a third way, if you don't mind having 2 outlets where the AC voltages come out in quadrature. If two AC outlets deliver the same average power with the voltages in quadrature, then the powers will be in counterphase, so no energy storage capacitors or inductors are needed (cheaper inverter per watt), and no energy is wasted (full utilization of the solar panels). The only downside is that you can't deliver the total power to one and the same device. But consider some high-way stop convenience store in the desert with multiple refrigerators, say an even number of off-grid refrigerators, then who cares if half of the refrigerators use AC power in counterphase with the other refridgerators?
A similar reasoning could be used for houses in a street, where the even houses and the odd houses are powered in counterphase (voltages in quadrature), which would substantially decrease the costs for inverters.