7 ms·

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 tempo

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.

dmurray 7y ago
Or a single refrigerator with two cooling elements.
- DoctorOetker 7y ago
  correct, but if we are redesigning all consumer devices, we might as well redesign them for solar DC power
  - dahfizz 7y ago
    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.
    - DoctorOetker 7y ago
      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.
kd5bjo 7y ago
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.
- DoctorOetker 7y ago
  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.
  - kd5bjo 7y ago
    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.
    - DoctorOetker 7y ago
      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.
jtc331 7y ago
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.
- DoctorOetker 7y ago
  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)...
  - jtc331 7y ago
    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).
    - DoctorOetker 7y ago
      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)?
- namibj 7y ago
  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.
  - DoctorOetker 7y ago
    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.
jcims 7y ago
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.
- DoctorOetker 7y ago
  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!
  - kragen 7y ago
    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.
    - DoctorOetker 7y ago
      ... 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? according to: https://www.energycentral.com/c/cp/solar-photovoltaic-pv-power-europe-report-2018 https://www.energycentral.com/c/cp/solar-photovoltaic-pv-pow... >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.
tzs 7y ago
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.

dmurray 7y ago

Or a single refrigerator with two cooling elements.

DoctorOetker 7y ago
correct, but if we are redesigning all consumer devices, we might as well redesign them for solar DC power
- dahfizz 7y ago
  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.
  - DoctorOetker 7y ago
    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.

kd5bjo 7y ago

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.

DoctorOetker 7y ago
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.
- kd5bjo 7y ago
  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.
  - DoctorOetker 7y ago
    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.

jtc331 7y ago

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.

DoctorOetker 7y ago
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)...
- jtc331 7y ago
  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).
  - DoctorOetker 7y ago
    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)?
namibj 7y ago
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.
- DoctorOetker 7y ago
  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.

jcims 7y ago

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.

DoctorOetker 7y ago
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!
- kragen 7y ago
  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.
  - DoctorOetker 7y ago
    ... 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? according to: https://www.energycentral.com/c/cp/solar-photovoltaic-pv-power-europe-report-2018 https://www.energycentral.com/c/cp/solar-photovoltaic-pv-pow... >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.

tzs 7y ago

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.