Inter coupling of microinverters

The present invention relates to DC-to-AC convertors for electric power systems. Lower cost, high-power, efficient, DC-to-AC convertors are of interest for solar energy economics. In the prior art, DC-AC inverters are the second highest cost item next to the photovoltaic panels. For high efficiency and low heat dissipation, commutation of DC to produce AC preferably uses solid state switches that are either fully on or fully off, and do not dwell more than a microsecond or so in an intermediate state. Therefore it is more complicated to produce a sine wave that takes on all values between the negative peak and the positive peak. On the other hand, producing a square wave which switches between the positive peak and the negative peak produces a form of AC that is not suitable for all loads.

Various manufacturers provide prior art DC-AC convertors that fall into one of a few broad classes and operating modes. The class of “modified sine wave” converters maintains both the same rms and the same peak voltage as a sine wave, while still employing only on-off commutation. This is done by switching the voltage between the desired positive peak, zero and the negative peak, spending 50% of the repetition period at zero, therefore achieving both the same peak and the same rms values as a true sine wave, and being compatible with a greater variety of loads.

Still, there are loads that do not tolerate the modified sine wave; for example appliances that present inductive loads, such as induction motors, some cellphone and laptop battery chargers, fluorescent lamps and tumble dryers, and any device with an internal power supply that uses capacitive reactance as a lossless voltage-dropping means, can malfunction on modified sine waveforms. Moreover, there is a potential problem with radio and TV interference due to the high level of harmonics of the modified square wave converter. Such a waveform is therefore not a candidate for coupling solar-generated power into the utility network or into house wiring.

“True sine wave” is another class of prior art DC-AC converter, and is required for coupling power into the grid or into premises wiring.

Another categorization of convertor relates to whether they are designed to power loads directly, or whether they are designed to feed and sell power back into the electricity grid. A load inverter that can power loads directly is said to operate in standalone mode, and is also called a “standalone inverter”, while a grid-tie inverter is said to operate in grid-interactive mode and is also called a “grid-interactive inverter”.

For safety and other reasons, the latter have to meet different specifications than the former, especially under fault conditions. In particular, a load inverter should be a constant voltage source, while a grid-tie inverter does not have a constant voltage output but must adapt to the voltage of the grid, and is rather a controlled current source. Moreover, a load inverter is always used with energy storage such as a rechargeable battery, and should maintain efficiency at both light and heavy loads and have low, no-load power consumption, so that the battery is not discharged while the inverter is idling at night. Grid-tie inverters however do not have the same a requirement for no-load power consumption, as they do not operate at night.

Many prior art inverters used low-frequency transformers in the synthesis of sine waveforms, but the large amount of copper and iron required for low-frequency transformers adds significant cost and weight.

Transformerless inverters are known in the prior art, particularly for utility-interactive inverters, which use high-frequency switching or pulse width modulation to approximate a sine wave. However, a disadvantage that arises in certain of these these converter concepts is the imposition of the high-frequency switching waveform on the solar array, which can capacitively couple through the glass cover upon touching it, potentially causing RF burn to personnel or damage to the solar panel, as well as causing the solar array to radiate substantial radio interference. Thus a design is required that can create a more benign common-mode voltage fluctuation on the solar array DC conductors.

One known method of making grid-tie inverters to convert DC power from a solar array to AC power than can be back fed into the grid is to employ multiple microinverters connected to the grid in parallel. This arrangement has been pioneered by, for example Enphase Inc.

When microinverters are attached to each solar panel, the advantage is the elimination of DC wiring, for which the National Electrical Code has specified new, unusual and onerous regulations.

However, multiple microinverters are more costly than a single large inverter. There is therefore the desire to reduce the cost of multiple microinverters to equal or better the cost of a single large inverter while retaining the benefit of eliminating DC wiring inside the premises.

Microinverter apparatus and arrangements for combining multiple microinverters are described that allow the use of some common components among the multiple microinverters for reduced cost, as well as achieving a benign common-mode voltage fluctuation at solar array terminals.

Each microinverter apparatus comprises input terminals for a floating DC input from one or more strings of photovoltaic cells, a pair of AC output terminals, a low pass LC filter with its output connected to a first of the AC output terminals and an H-bridge of switching transistors for selecting either the positive or the negative polarity of the DC source to be connected to either the low pass filter's input or to a second of the AC output terminals, and a controller for controlling the switching of multiple microinverters using any switching approximation using two or more signal levels such as delta or delta sigma modulation or pulse width modulation to provide an approximation to a desired waveform. The H-bridge alternates between connecting the positive and the negative of the associated solar array to the second of the AC output terminals only at the low, sign-changing rate of the desired AC output, thus avoiding imposing a high common-mode frequency on the array.

The microinverter outputs may be combined in series to increase the total output voltage or in parallel to increase the total output current, or in a combination of parallel and series connections. In all cases, the controller may optimize the switching of each microinverter so that the combined voltage or current waveform or both approximates a desired sine wave with as little as possible unwanted ripple due to residual unfiltered switching components.

A common mode filter is employed to slow the edges of the low-frequency common-mode switching signal imposed on the array to reduce the export of radio frequency interference. It is disclosed that a common, multifilar common-mode mode choke can be used with inventive methods of combining the microinverter outputs, thus reducing total component count.

FIG. 1 shows the basic arrangement of a microinverter

FIG. 2 shows two microinverters connected in series to a single phase 120 volt rms grid.

FIG. 3 shows two microinverters connected in series to a split-phase 240 volt rms grid.

FIG. 4 shows antiphase inverters connected in series to a 240 v split-phase grid

FIG. 5 shows the common mode waveforms of the arrangement of FIG. 2 .

FIG. 6 shows a common mode hash filter

FIG. 7 shows the transient response of a common mode hash filter

FIG. 8 shows two antiphase microinverters sharing a common-mode choke

FIG. 9 shows another arrangement of microinverters sharing a common-mode choke.

FIG. 10 shows a series connection of four microinverters sharing two common-mode chokes

FIG. 11 shows the common mode waveforms of FIG. 10

FIG. 12 shows a another series connection of four microinverters

FIG. 13 shows a parallel connection of four microinverters

FIG. 14 shows an interleaving arrangement for panels within a solar array connected to respective microinverters A, B, C and D

FIG. 1 shows the basic circuit arrangement of a microinverter. This arrangement of an inverter was disclosed in U.S. Pat. No. 9,735,703 to current Inventor, which is a continuation of U.S. Pat. No. 8,937,822. These patents are hereby incorporated by reference herein in their entirety.

The circuit of FIG. 1 shows a DC input connected through a common mode filter comprised of C 2 ,C 3 ,C 4 and L 2 to an H-bridge of switching transistors Q 1 ,Q 2 ,Q 3 ,Q 4 . One of the H-bridge output lines is connected to a first AC output terminal labeled N while the other H-bridge output line is connected through the low-pass filter comprised of L 1 and C 1 to the second AC output terminal labeled L. If used alone as a grid-tie inverter connected to a single phase of the grid, the AC output terminals L and N would signify the Live and Neutral lines of the single phase grid. In that case, for a 120 volt RMS grid, the DC input voltage would have to be at least 120√=169.7 volts. When L 1 is small, it was discussed in the above-incorporated patent and applications how achieving the desired power flow into the grid requires precise control of the AC output voltage and phase generated by the H-bridge.

Description of the operation of FIG. 1 is summarized below.

When the desired output is a positive-going half cycle of the 120 volt RMS grid, transistor Q 2 is turned on to connect the negative polarity of the 170 volt DC supply to terminal N, while transistors Q 3 and Q 4 connect filter inductor L 1 alternately to the +ve and −ve polarity of the supply with a duty factor that, after filtering with L 1 ,C 1 , provides the correct mean positive voltage corresponding to some point of the positive half cycle of the desired sine waveform. The duty factor variation can be in the form of multiples of a constant time period at + or − according to a delta-sigma bit sequence, or can be in the form of variable time periods according to a pulse width modulation approximation.

On the other hand, when the desired output is the negative-going half cycle of the 120 volt RMS grid, transistor Q 1 is turned on to connect the positive polarity of the 170 volt DC supply to terminal N, while transistors Q 3 and Q 4 connect filter inductor L 1 alternately to the −ve and +ve polarity of the supply with a duty factor that provides the correct negative mean voltage point on the negative half cycle after filtering.

Since the +ve and −ve of the supply are thus alternately connected to the neutral terminal N (which is ultimately grounded and thus at zero volts), the common mode signal before common-mode filtering is simply a low-frequency square wave.

An advantage of the arrangement of FIG. 1 is thus that the common mode signal created on the solar array feeding the DC input is relatively benign, consisting of a square wave at the line frequency (e.g. 60 HZ) with the sharp edges removed by the common mode filter.

The simplified common mode filter shown in FIG. 1 incorporates no damping; therefore the response to a common mode square wave would exhibit very large ringing of long duration. To avoid this ringing, a practical common mode filter must incorporate damping as was shown in the above-incorporated references. The design of common mode filter from the references is shown in FIG. 6 , and its transient response is shown in FIG. 7 .

In FIG. 6 , the common mode damping is provided by the 47 ohm resistor connected from the center of tap of the two 0.1 uF capacitors to neutral and ultimately ground. Simulation showed that, while the filtering of the common mode signal results from the 1 mH common mode inductance and the two 0.01 uF capacitors, the damping is only effective if the capacitors coupling to the damping resistor are an order of magnitude greater. This can couple some 60 Hz and harmonic power in the several hundred milliwatt region into the damping resistor as waste energy, which however is reduced to about 183 mW by paralleling the 47 ohm resistor with a 5.6 mH choke, its value being chosen to optimize the transient response as shown in FIG. 7 . Thus, filtering the common mode signal inevitably requires a damped filter, and the filter damping can cause energy wastage unless care is taken to minimize it.

The common mode filter design of FIG. 6 is for a particular inverter power level of 60 amps times 120 volts RMS, i.e. 7.2 kilowatts, and values can be impedance-scaled for other voltages and currents. For example, if a microinverter is designed to output 60 volts and 30 amps RMS, the impedance scaling is unity and the same filter can be used.

FIG. 2 shows the connection of two 60 volt RMS inverters in series to produce 120 volts RMS for coupling to a single-phase grid. To produce 60 volts rms out, the DC input to each microinverter must in the 85 volt region at a minimum.

The series coupling of FIG. 2 couples the L terminal of the lower microinverter to the N terminal of the upper microinverter. If the lower inverter produces the voltage output labeled AC OUTPUT A and the upper inverter produces the voltage output labeled AC OUTPUT B, then the total output across the grid L and N lines will be the sum of AC OUTPUT A and AC OUTPUT B voltages. Thus the two 60 volt RMS microinverters in series generate the required 120 volts RMS.

It is of interest to examine the common mode waveforms of the arrangement of FIG. 2 . The common mode waveforms before common-mode filtering appear at circuit nodes ( 110 , 111 , 112 and 113 ). FIG. 5 shows these waveforms.

Since the lower inverter has its N output terminal connected to the grid neutral, its common mode waveforms are exactly the same as for FIG. 1 , except with half the voltage swing due to the 85 volt supply instead of 170 volts. The negative node ( 110 ) alternates between 0 volts (neutral) and −85 volts (when the plus is grounded), while the positive node ( 111 ) follows it with a constant offset of 85 volts. The offset is constrained to be constant due to capacitor C 2 of FIG. 1 , which is large. C 2 must be large, as described in the incorporated '822 patent, as the DC current taken by an inverter is a raised cosine waveform of twice the line frequency and having twice the peak AC output current. Since the array can only supply a constant current, C 2 must act as a reservoir to smooth out the current demand to the array and is typically a plurality of large aluminum electrolytic capacitors in parallel.

FIG. 5 shows that the common-mode signal on the upper inverter's array terminals ( 112 and 113 ) looks completely different. This is because the upper inverter's N terminal is not connected to the grid neutral, but rather to the 60 volt RMS output from the lower inverter's L terminal. Thus the common mode signals at nodes 112 and 113 comprise an 85 volt peak to peak 60 Hz square wave added to a 85 volt peak sine wave, and their fundamental components are in phase, which increases the amplitude to 255 volts peak-to-peak. Now referring to the common mode filter of FIG. 6 , if the center tap of the two 0.01 uF capacitors and the 47 ohm damping resistor were taken to ground/neutral, the 60 Hz component being dissipated in the 47 ohm resistor would be substantially higher; on the other hand, if taken to the N terminal of the upper inverter which is connected to the L output terminal of the lower inverter, the common mode signal would contain all residual, unfiltered high frequency switching components of the lower inverter. Thus there is dilemma, indicated by the ? mark in FIG. 2 , as to where the common-mode filter return should be connected. It may be that the center tap of the upper common-mode filter's 0.01 uF capacitors should be returned to grid neutral to ensure that the inverter hash from the lower inverter is attenuated, while the damping resistor is returned to the N terminal of the upper inverter to reduce 60 Hz dissipation in the damping resistor; however, the upper and lower microinverters would then be slightly different. This may not be an issue, but some other configurations that preserve symmetry will be examined.

Another configuration is shown in FIG. 3 . In this case two microinverters 101 -A and 101 -B of the FIG. 1 type each generate 120 volts RMS, in phase, and so may be connected in series to generate 240 volts RMS, with the L terminal, now labeled 2, of the lower inverter ( 101 -A) being connected to the N terminal, now labeled 1, of the upper inverter ( 101 -B). Moreover, the 240 volt output is connected to a split-phase grid having an L 1 terminal and an L 2 terminal, both carrying 120 volts RMS, but in antiphase, so that the difference between them is 240 volts RMS.

The common mode waveforms at nodes 110 , 111 , 112 , and 113 are now indicated in the same FIG. 3 . The waveforms can be explained by realizing that they are as per FIG. 5 , except that the grid neutral terminal of FIG. 2 has been replaced by a grid L 1 terminal carrying an antiphase 60 Hz sine wave of 120 v RMS. The waveform on L 1 thus subtracts from the waveforms shown in FIG. 5 , removing the sine wave from nodes 112 and 113 of FIG. 5 and leaving only the square wave common mode signal, while the common mode signals on nodes 110 and 111 of FIG. 3 are the sum of a square wave and an antiphase 60 Hz sine wave. The latter is very desirable, as the 60 Hz component has been substantially reduced, actually by the factor 4π−1, or 0.27. This gives lower dissipation in the common-mode filter damping resistor. Likewise, the dissipation in the upper inverter's common mode filter (relative to the total output power to the grid) is reduced as compared to the waveform of FIG. 5 .

The difference between the common-mode waveforms of the upper and lower inverters of FIG. 3 arises because the lower inverter ( 101 -A) alternately connects the +ve and −ve of the solar array to a grid line, while the upper inverter ( 101 -B) alternately connects its solar array +ve and −ve to terminal 2 of the lower inverter (which is virtual neutral at a mean of zero volts), and not a grid line. This asymmetry is removed in the arrangement of FIG. 4 .

In FIG. 4 , two microinverters ( 101 -A and 101 -B) are connected in series by connecting their terminals ( 1 ) together while connecting their terminals ( 2 ) respectively to the L 1 and L 2 grid lines. Since the combined voltage is now the difference between their respective output voltages V 2 and V 1 , they must be operated in antiphase to obtain constructive voltage addition. It is known to do this in the prior art of inverters such as the Xantrex SW4048, which, however, do not produce common mode signals on the DC input lines as they use isolation transformers to obtain the 60 Hz AC outputs, and are intended to operate from a common DC source.

It may be seen in FIG. 4 that this arrangement cancels the 60 Hz sine wave component from the common-mode of both DC inputs, leaving the familiar square wave common-mode signal, and both microinverters are operating identically. This is because each inverter alternately connects the +ve and −ve polarities of its associated DC source to the same common point of the junction of the inverters' output terminals ( 1 ), which should be a virtual neutral/ground and therefore at nominally zero voltage. It's voltage is actually minus half the sum of the inverters' respective output voltages V 1 and V 2 , which, since V 1 and V 2 are in antiphase, should cancel to zero. It is not exactly zero however, but carries residual unfiltered components of the switching waveforms, abbreviated to “ripple”. The ripple thus gets imposed on the terminals of arrays 100 A and 100 B and may cause radio interference by being radiated from the arrays unless adequately attenuated by filters 101 -A and 101 -B. On the other hand, the 60 Hz component of the common mode signal is now out of phase on array 100 A compared to array 100 B. If the arrays are physically close, or even better, if individual panels or even individual solar cells are interleaved, this results in a substantial reduction of any radiated 60 Hz hum component that could couple to inadequately screened audio devices on the premises. For example, FIG. 14 shows one possible interleaving arrangement for panels or sub-arrays A, B, C and D of FIG. 10 . The 60 HZ common mode signals on sub-arrays A and D are in antiphase as are the common mode signals of sub-arrays B and C. Therefore the interleaving of sub-arrays A and D horizontally and of B and C horizontally combined with the vertical interleaving of A/D rows with C/B rows has the effect that adjacent sub-arrays have anti-phase 60 Hz common-mode components. Capacitive coupling of 60 Hz hum to nearby systems is thus substantially reduced.

Before discussing other arrangements, it may be pointed out that the similarity of the common mode waveforms, albeit in antiphase, suggests the possibility of a joint common-mode filter. FIG. 8 shows how a quadrifilar common mode choke can provide the filtering inductance for both microinverters.

Joint common-mode choke ( 200 ) now has a quadrifilar winding. The four wires carry the DC+ and DC− from both arrays ( 100 -A and 100 -B) to their respective microinverters. The insulation between wires of the quadrifilar winding must be sufficient to support the DC voltage difference between them. Because the common mode signals for the two inverters are out of phase, the inverters are connected to their respective wires at the opposite ends of the winding of choke ( 200 ). It may be shown that the common mode inductance required can be halved by this arrangement, that is the number of turns can be reduced by the factor 0.7 compared with using separate common-mode chokes for inverters 101 -A and 101 -B. Since large toroidal cores are one of the larger, heavier and more expensive components, all metrics are improved by the use of a joint common-mode choke.

In FIG. 8 , the 0.01 uF common mode filter capacitors are returned to neutral while the 47 ohm damping resistors are returned to terminals ( 1 ) of the microinverters, mainly to show this variation rather than for any confirmed advantage. There is a reason behind returning the filter capacitors to neutral rather than ground, however, and that is to close the filtering current loop within the inverter so that high frequency ripple currents are not flowing through ground all the way to the service entrance (which is the only place neutral may be connected to ground) and then back on neutral. Thus it is avoided to export common-mode ripple on the power and neutral lines; this is valuable in reducing common-mode noise, with regard to non-interference with the power-line communications scheme for Smart Appliances disclosed in U.S. Pat. No. 9,614,588 by current inventor, and U.S. Pat. No. 9,785,213, also to current inventor, both of which are hereby incorporated by reference herein in their entirety.

One remaining issue with FIG. 8 is that, because the ripple on inverter pins 1 is transferred in phase to all common mode signals, common mode choke 200 as connected in FIG. 8 does not provide any filtering of this ripple, leading to the consideration of yet another arrangement.

FIG. 9 shows yet another arrangement for connecting two microinverters in series to a 240-volt split-phase grid. In FIG. 9 , output terminals 1 and 2 of both inverters have been interchanged compared to FIG. 4 . The result of switching terminals 1 and 2 is that the common mode signals have become of the form of those for inverter 101 -A of FIG. 3 , namely a 60 Hz sine wave subtracted from a 60 Hz square wave. Thus all common mode signals now exhibit a reduction of their 60 Hz components by the factor 0.27. That there is still some residual 60 Hz component has the benefit that ground leaks on any DC input line may be detected by using a 2-pole AC GFI breaker on the L 1 ,L 2 outputs. The L 1 , L 2 currents are normally equal and opposite and thus do not trip the GFI breaker. If any DC line has a leak to ground however, the specific current phase that it has will be in phase with one of L 1 and L 2 , thus adding to that current, and out of phase with the other of L 1 and L 2 , thus subtracting from that current. The resulting imbalance current is the 60 Hz component of the ground leak, and an imbalance current of only 6 mA is normally sufficient to trip a GFI breaker. Since the 2 pole GFI breaker will open both hot legs, all connections to ground, neutral and grid hotlegs are removed and the whole solar system is then floating, cutting any possible return path for ground leak current and reducing it to zero ground leak current to zero. The utility of a common mode signal in detecting ground leaks was already disclosed in the above-incorporated '822 patent. To preserve the ability to employ this ground leak trip feature, no permanent electrical connection to ground, neutral or live should be made by any component in the inverter system, hence the component labeled “output relay” in FIG. 6 . Arrangements should be made to open this relay upon detecting a ground fault, such as by using a signal from the above mentioned GFI breaker, which could for example be used to trip off the inverter, thus opening the output relay(s).

It was pointed out above that the output filter denoted by L 1 ,C 1 in FIG. 1 may not succeed to eliminate all high frequency switching components from the AC output of an inverter. This is especially true when the switching frequency is not very much higher than the low-pass cut off frequency of filter L 1 ,C 1 . When both microinverters of FIG. 8 or FIG. 9 have residual ripple, it is desirable that the dominant spectral components of the ripple should be in-phase on terminals 1 of the FIG. 8 case or terminals 2 of the FIG. 9 case, such that there is no ripple difference voltage between corresponding pins of the two inverters; thus when those pins are joined, no ripple current flows and thus no ripple current flows to the L 1 and L 2 utility grid connections. Thus we desire the dominant components of ripple to be in-phase even though the inverters are generating out-of-phase 60 Hz voltages. This may be approximately achieved by coordinating the timing of the switching within each inverter through use of a common switching controller. Use of a common switching controller also reduces cost compared to uncoordinated microinverters.

FIG. 10 shows the series connection of four microinverters to a 240 v split-phase grid. Each inverter produces 60 volts RMS AC, for which the associated DC supply voltages from Sub-Arrays A, B, C and D must be at least 85 volts The two outer inverters labeled H-bridge A and H-bridge D are connected as in FIG. 9 , namely with the output of the slowest switching pair of transistors connected to L 1 or L 2 . The common mode waveforms on sub-arrays A and D are thus 85 volt peak to peak square waves with a 120 volt RMS sine wave subtracted therefrom. These waveforms are shown in FIG. 11 , from which it may be seen that the common mode waveforms on sub-arrays A and D are merely antiphase versions of the same waveforms. A joint common-mode choke CML 1 may thus be used for sub-arrays A and D to connect to their respective H-bridges. Due to the waveforms being in antiphase, sub-array A connects to two wires at the opposite end of the winding of CML 1 compared to sub-array D. FIG. 10 omits details of the connections of common-mode filter capacitors and damping components for clarity, as its purpose is only to show the use of joint common-mode chokes CML 1 and CML 2 .

The inner two inverters comprising H-bridges B and C powered respectively by DC from sub-arrays B and C are substantially identical to FIG. 8 except for operating with half the DC voltage input and half the AC voltage output. The common mode waveforms are thus 85 volt peak-to-peak square waves as at nodes 111 and 112 of FIG. 8 , scaled down by a factor of 2:1. Inverters B and C may thus also use a joint common mode choke CML 2 .

The circuit of FIG. 10 may also be re-arranged so that the inner two inverters are as per FIG. 9 instead of as per FIG. 8 , thus producing the common mode waveforms of FIG. 9 . If the output pins of all four inverters are swapped, the circuit of FIG. 12 is produced. Since the inner two inverters H-bridges B and C are now connected as in FIG. 9 , the common mode waveforms on their respective sub-arrays B and C are as per FIG. 9 , namely an 85 volt peak to peak square wave from which is subtracted an 85-volt peak sine wave.

The common mode waveforms for the outer inverters A and D are the result of adding an 85 volt peak sine wave to an 85 volt peak to peak square wave, resulting in the waveforms 112 and 113 of FIG. 5 , but of half the amplitude. Due to the reduced amplitude, the concern for power dissipation in the damping resistors is eliminated.

Inverters A and D of FIG. 12 may also use a joint common mode choke CML 1 to connect to their respective sub-arrays A and D. FIG. 12 shows the filter capacitors of the common mode filter (the 0.01 uF capacitors of FIG. 6 ) being returned to utility grid neutral, as this is expected to be the cleanest available line. The damping resistors may be returned to different points however, to reduce dissipation. For example, the damping resistors for inverters B and C can be returned to the 0-volt mean point, while the damping resistors for inverter A may be returned to the junction of C 1 and C 2 , while the damping resistor for inverter D is returned to the junction of C 3 and C 4 . Since the common mode waveforms in the case of 4 microinverters are of reduced amplitude, and thus the concern for damping resistor dissipation is reduced, the damping resistors may be returned to a consistent place for all four microinverters, namely to the junction of C 1 and C 2 for inverters A and B and to the junction of C 3 and C 4 for inverters C and D. This allows the construction of identical modules for all four microinverters.

As mentioned previously, when two microinverters such as A and B are joined at the junction of C 1 and C 2 , the residual unfiltered ripple that passes through their respective filters L 2 ,C 2 and L 2 ,C 2 should ideally be in phase at least for its dominant spectral component. Thus no ripple current at that frequency is created. When now the two inverter pairs, each with their dominant ripple component canceled, are joined at the junction of L 2 and L 3 , the second most dominant ripple component from each inverter pair should be in-phase at that point before they are joined, thus avoiding creating ripple current also at the second most dominant ripple spectral component. This can be approximately arranged by using a common switching controller to coordinate the switching of all H-bridges to minimize total ripple currents or voltages exported to the grid that are not at the desired line frequency. In fact, using N inverters in series (or parallel) theoretically allows N−1 unwanted ripple components to be canceled.

Ripple cancellation of parallel- or series-connected microinverters may be achieved by the following consideration: When the clock frequency phase of a switching inverter is shifted by Phi, Fourier components at N times the clock frequency are shifted by N.Phi. Thus a number M of inverters connected in series or parallel and having their clock phase relatively shifted by 0, Phi, 2Phi, 3Phi . . . (M−1)Phi will have ripple frequency components at N times the clock frequency shifted by 0, NPhi, 2NPhi, 3NPhi - - - (M−1)NPhi. If Phi is chosen to be 2π/M, the sum of such ripple components is substantially canceled for N=1, 2 . . . (M−1), thus canceling the first M−1 harmonics of the switching frequency.

Microinverters may also be combined in parallel, as has been pioneered by Enphase Inc. However, according to the current invention, physically combining microinverters in groups of two, four or more can produce not only cost reductions due to sharing components such as common mode chokes, switching controllers and mechanical housings, but also produce technical benefits such as ripple reduction through using the common controller to coordinate switching to that end.

In the case of paralleling microinverters, output currents and ripple currents add; therefore it is desired that the dominant components of the ripple currents cancel as far as possible even though the 60 Hz current components shall add in phase. This is again facilitated by coordinating the switching by means of a common controller.

FIG. 13 shows four microinverters ( 200 -A, 200 -B, 200 -C, 200 -D) broadly conforming to the basic scheme of FIG. 1 , connected in parallel to feed power to a 240 volt split-phase grid connection though 2-pole GFI breaker ( 500 ). Each microinverter now has to produce the full 240 volts RMS, which requires a DC input from their respective photovoltaic sub-arrays ( 300 -A, 300 -B, 300 -C, 300 -D) of at least 340 volts. By co-housing the microinverters, they can share common components and in particular the octofilar common mode choke ( 400 ).

All of the microinverters of FIG. 13 operate in phase at 60 Hz, and so create identical common-mode waveforms. As there is no connection to ground or anything else (apart from capacitance of the solar arrays to ground), Kirchoff's law requires that the sum of the currents in the L 1 ,N and L 2 paths through 2-pole GFI breaker ( 500 ) is zero; this is exactly the condition for not tripping the breaker. On the other hand, if an accidental path from any array terminal to ground were created, the current imbalance in GFO breaker ( 500 ) would cause it to trip, interrupting the ground fault leakage current. Moreover, disconnection of a grid-tie inverter from the grid will cause it to stop inverting, thus open the above mentioned output relays and eliminating any AC or DC voltages on the array relative top ground.

Any or all of microinverters 200 -A to 200 -D of FIG. 13 may be any of the inverters 3, 4, 8, 9 10 or 12. For example, if inverters 200 A- 200 D each comprise a pair of series connected inverters as in FIG. 4 , the eight inverters may jointly use a 16-filar common-mode choke. Moreover, using a common switching controller for all eight inverters allows seven ripple components to be canceled by coordinating their switching.

In the prior art, microinverters were integrated with respective solar panels, thereby achieving the benefit of eliminating DC wiring, but hindering the use of common components and thus increasing cost. The inventive microinverter arrangements described above may achieve minimum system cost when all coupled microinverters are co-housed to allow use of common components. DC wiring is nevertheless still substantially eliminated if the co-housed microinverters are located sufficiently close to the array such that no DC wiring is more than 10 feet away from the array. The 10 foot distance is defined in the National Electrical Code as the distance, beyond which the above mentioned new, unusual and onerous safety requirements for DC wiring apply, and which it is desired to avoid. A person skilled in the art may devise many configurations of coupled microinverters using the teachings disclosed herein, all of which are considered to fall within the scope and spirit of the invention as described by the attached claims.