Gridtied AC coupled batt inverter/charger, and contingency operations

[Hedges]

I have a NEM agreement for 3kW peak

That agreement probably allows you to export up to 4kW (allows 10% or 1kW increase, whichever is more)
More hours at peak wattage, they didn't restrict. So a second system (or added PV string) of different orientation could slip under the limit.

The biggest loss of efficiency I’m seeing with DC-coupled is how poor it will be in handling shade. MPPT-per-panel is much less practical with DC strings versus microinverters...

DC coupled MPPT should work fine with shading in the case of a single series string of panels. If multiple strings with greatly different shading, then it suffers. If similar shading so strings remain matched within 10% maybe 20%, would be OK.

I’ve already got a 4kW grid-tied system under NEM1 that covers my annual consumption (at least before they start reducing credit by shifting peak TOU hours this summer).

So maybe you've already taken advantage of the extra 1kW

 

[fafrd]

That agreement probably allows you to export up to 4kW (allows 10% or 1kW increase, whichever is more)
More hours at peak wattage, they didn't restrict. So a second system (or added PV string) of different orientation could slip under the limit.

The actual agreement states that any modifications will be reviewed and approved by the utility, so yes, I have that 1kW of headroom, but that’s where the ‘asking forgiveness versus asking permission’ comes into it. I don’t want to risk being in bad standing versus my existing NEM1 agreement (so I’ll offset self-consumption all I want, but I don’t want to risk exporting of 3kW).

DC coupled MPPT should work fine with shading in the case of a single series string of panels. If multiple strings with greatly different shading, then it suffers. If similar shading so strings remain matched within 10% maybe 20%, would be OK.

The issue gets more complicated once you throw half-cut panels into the equation.

A half-cut panel will put out ~Imp/2 @ Vmp when only half of the panel is shaded (and without the use of any bypass diodes).

On a string, that would drop the entire string to half-current and just over half-power, so the entire half-cut panel gets cut out of the string by the MPPT SCC.

That allows the remaining unshaded panels in the string to output full power, but wastes half-a-panel’s worth of power which would be available in a parallel string.

I have a half-array-wide shadow from a tree working it’s way from one side of the array to the other over the course of the first three hours of every morning, and the top-halves of half of my half-cut panels will be unshaded for a good hour near peak radiance (10:30am to 11:30am).

So yeah, finding a series solution which will work as effectively as a full-parallel string (or strings) for this specific shading issue is a challenge.

So maybe you've already taken advantage of the extra 1kW

No, I haven’t, but I know my utilities understanding of what I need to do to take advantage of that additional 1kW and I don’t want to get into an argument with them. Also, I’m planning for a new 4.5kW array which should be outputting out over 3kW peak... (so 1kW doesn’t get me where I’m trying to go).

 

[Hedges]

The issue gets more complicated once you throw half-cut panels into the equation.

A half-cut panel will put out ~Imp/2 @ Vmp when only half of the panel is shaded (and without the use of any bypass diodes).

On a string, that would drop the entire string to half-current and just over half-power, so the entire half-cut panel gets cut out of the string by the MPPT SCC.

That allows the remaining unshaded panels in the string to output full power, but wastes half-a-panel’s worth of power which would be available in a parallel string.

I have a half-array-wide shadow from a tree working it’s way from one side of the array to the other over the course of the first three hours of every morning, and the top-halves of half of my half-cut panels will be unshaded for a good hour near peak radiance (10:30am to 11:30am).

So yeah, finding a series solution which will work as effectively as a full-parallel string (or strings) for this specific shading issue is a challenge.

A dumb MPPT starting at Voc and moving downward may stop at Imp/2 due to one (or more) such panels partially shaded. That misses half the power from all the other panels.
A different MPPT would explore lower voltages periodically, discovering a higher peak at Imp but lower voltage. That uses bypass diode to skip over a few half-shaded panels to get full power from the unshaded panels.

Optimizer doing 50% buck should capture max available power in this case.

As your tree's shadow moves, how much of the array has panels capable of delivering Imp/2?
My guess is most will get both halves shade, so nothing more to gain from them.
I think the panels with 2p of half-cut cells would only really shine in a situation where a structure shades an entire row of the similarly - commercial array where next row of panels shades edge of this one. or Los Gatos where ugly panels are required to be hidden behind parapets.

 

[fafrd]

A dumb MPPT starting at Voc and moving downward may stop at Imp/2 due to one (or more) such panels partially shaded. That misses half the power from all the other panels.

Yes, that would be a pretty dumb M

A different MPPT would explore lower voltages periodically, discovering a higher peak at Imp but lower voltage. That uses bypass diode to skip over a few half-shaded panels to get full power from the unshaded panels.

if you’re talking about a series string,it actually has to drop voltage by ~Vmp/3 + Vdiode for each bypass diode it activates.

If there are 2 panels in the string that are putting out less than ~2/3Imp @ Vmp that can put out full Imp with only a single bypass diode activated, string voltage will need to drop by ~2/3 Vmp + 2Vdiode.

Unfortunately, with half-cut panels, activating a bypass diode throws away potential energy if the shade in on only 1/2 of thr

Optimizer doing 50% buck should capture max available power in this case.

I believe they way they classify it, doubling current would be called 100% buck, not 50% buck. The name doesn’t really matter as the result, and we both agree that an optimizer capable of buck-converting up to Imp/2 @ Vmp to Imp @Vmp/2 would be able to capture all of the available energy from a string of half-cut panels with one or more half-shaded panels.

Unfortunately, Tigo’s optimizers are limited to a maximum of 33% buck, meaning that they can convert Imp/2 @ Vmp to no more than 2/3Imp @ 3/4Vmp (meaning all unshaded panels in the string will be operating at something over only 2/3 of available power (voltage on those panels had to increase to over Vmp to reduce current to 2/3Imp, so a full 1/3 of available power is mot

As your tree's shadow moves, how much of the array has panels capable of delivering Imp/2?

I’ve been taking notes every morning (should probably just set up a time-lapse camera).

There is a good hour late in the morning when 4 to 5 panels out of 10 are half-shaded

My guess is most will get both halves shade, so nothing more to gain from them.

Yes, I’ve identified 4 panels that are fully-shaded and become unshaded vertically one column at a time, so those 4 work well in a 2P2S array. It’s the others that are more problematic because they unshaded vertically and all together (the top of the tree).

I think the panels with 2p of half-cut cells would only really shine in a situation where a structure shades an entire row of the similarly - commercial array where next row of panels shades edge of this one. or Los Gatos where ugly panels are required to be hidden behind parapets.

Exactly (which is unfortunately exactly the situation I find myself in for an important hour+ every late morning).

 

[GXMnow]

One of my early ideas before I went with the Schneider was to use a microcontroller, like an Arduino to monitor the output current of the Enphase Microinverters. I know my house base load is just under 1,000 watts. So any time the solar was making less than say 1,200 watts, it would do nothing and let the solar feed the house, and maybe export a little. Above 1,200 watts, have it turn on a solid state relay that powers up a 600 watt charger on the battery bank. If solar goes over 2,000 watts, turn on a second 600 watt charger. And if solar tops 3,000 watts, turn on a third charger. Build in a bit of hysteresis, so as the production falls, if solar goes below 2,800 watts, one shuts off etc. I already have one 600 watt (58.8 volt 10 amp) charger that I have used to charge my battery bank. It is a proper CC CV so it will taper off current when the batter bank becomes full. When you get to the peak rate time, have all 3 chargers shut off, and let the system export. This time of year, my 16 300 watt panels are still cranking out 1,400 watts, so it runs my house and still exports 500 watts or so as I go into the peak rate 4 pm to 9 pm block. And this is also when I trigger the Schneider to go into "Grid Sell". In your case, this is when you turn on the grid tie battery inverters.

 

[fafrd]

One of my early ideas before I went with the Schneider was to use a microcontroller, like an Arduino to monitor the output current of the Enphase Microinverters. I know my house base load is just under 1,000 watts. So any time the solar was making less than say 1,200 watts, it would do nothing and let the solar feed the house, and maybe export a little. Above 1,200 watts, have it turn on a solid state relay that powers up a 600 watt charger on the battery bank. If solar goes over 2,000 watts, turn on a second 600 watt charger. And if solar tops 3,000 watts, turn on a third charger. Build in a bit of hysteresis, so as the production falls, if solar goes below 2,800 watts, one shuts off etc. I already have one 600 watt (58.8 volt 10 amp) charger that I have used to charge my battery bank. It is a proper CC CV so it will taper off current when the batter bank becomes full. When you get to the peak rate time, have all 3 chargers shut off, and let the system export. This time of year, my 16 300 watt panels are still cranking out 1,400 watts, so it runs my house and still exports 500 watts or so as I go into the peak rate 4 pm to 9 pm block. And this is also when I trigger the Schneider to go into "Grid Sell". In your case, this is when you turn on the grid tie battery inverters.

Funny, I was thinking about exactly that today (a bank of LiFePO4 chargers connected to switches controlled to maintain drain to chargers larger than export power. I may need to get you to send me your design (never worked with an Arduino before).

I’ve seen his simple the clamp sensors to control my GTIL inverters are and believe it should be pretty straightforward to control a variable-power charger in a similar manner.

A staircase of smaller chargers would be the next best thing. On the one hand, getting chargers that are more than 80% efficient is a challenge. On the other hand, the GTIL inverters I’m using are only ~75% efficient at 25V, so powering the self-consumption by Microinverter-produced AC will be much more efficient that drawing out of the battery or the DC-coupled PV to generate AC.

I’ll be generating something like 3kW peak and self consuming ~350W during the day on average (primarily fridges)

So I could use a one or two panels to generate 450-900W of AC (actually peaking at 337.5 to 675W based on my existing 4kW AC-coupled array.

Whenever there is any export on either leg, I can turn on a 10A 8S LiFePO4 charger which charges at ~250W and consumes ~312W @ 120AC.

So with a maximum of 2 chargers per leg, I should be able to assure that there is never any export, And if I use 600W chargers that would allow even more AC power to be cintrolled for zero-export.

Of course, that still leaves the issue of shutting down the Microinverters once the battery is full, but that seems doable with a switch to turn off the grid signal to the Microinverter-based array.

I’d sure prefer some company putting together a more integrated / elegant solution rather than having to throw together this frankenstinian-solution myself, but desperate times call for desperate measures...

 

[Hedges]

I’d sure prefer some company putting together a more integrated / elegant solution rather than having to throw together this frankenstinian-solution myself, but desperate times call for desperate measures...

There are, I think this accomplishes the goal (suck up juice to achieve zero export, or charge battery off-peak and discharge on-peak

Sunny Boy Storage 3.8-US / 5.0-US / 6.0-US | SMA America

Discover the flexible Home Energy Solution: The SMA Sunny Boy Storage 3.8-US / 5.0-US / 6.0-US battery inverter! ► Find out more about its benefits!

www.sma-america.com

The inverter itself is not unreasonably priced ($2500 for 6kW). Compatible batteries is what might cost more than you want to spend.
If a supported BMS was available, then DIY would be an option.

https://files.sma.de/downloads/SBS-Batteries-TI-en-17.pdf?_ga=2.30590399.1447024346.1614721797-621369231.1589916091

For instance, Battery Box H, 10 kWh for $6500

Search results for: 'byd battery box h10.0 complete 10.24kwh system sku part number battery box h10 0'

www.thepowerstore.com

AxiStorage 7.5 kW for 3800 GBP

AXITEC AXIstorage LI 9 SH 7.5KWH Li-Ion Solar Battery Storage

The AXITEC AXIstorage LI 9 SH 7.5KWH Li-Ion Solar Battery Storage is manufactured in Germany and perfectly combines performance, safety and longevity. As an ideal supplement to the solar modules, AXITEC has promoted the development of energy storage systems.

www.zerohomebills.com

 

[fafrd]

There are, I think this accomplishes the goal (suck up juice to achieve zero export, or charge battery off-peak and discharge on-peak

Sunny Boy Storage 3.8-US / 5.0-US / 6.0-US | SMA America

Discover the flexible Home Energy Solution: The SMA Sunny Boy Storage 3.8-US / 5.0-US / 6.0-US battery inverter! ► Find out more about its benefits!

www.sma-america.com

The inverter itself is not unreasonably priced ($2500 for 6kW). Compatible batteries is what might cost more than you want to spend.
If a supported BMS was available, then DIY would be an option.

Yeah, that’s the gotcha (at least today). And $2500 will purchase an awful lot of PV kW in 2023...

https://files.sma.de/downloads/SBS-Batteries-TI-en-17.pdf?_ga=2.30590399.1447024346.1614721797-621369231.1589916091

For instance, Battery Box H, 10 kWh for $6500

Search results for: 'byd battery box h10.0 complete 10.24kwh system sku part number battery box h10 0'

www.thepowerstore.com

AxiStorage 7.5 kW for 3800 GBP

AXITEC AXIstorage LI 9 SH 7.5KWH Li-Ion Solar Battery Storage

The AXITEC AXIstorage LI 9 SH 7.5KWH Li-Ion Solar Battery Storage is manufactured in Germany and perfectly combines performance, safety and longevity. As an ideal supplement to the solar modules, AXITEC has promoted the development of energy storage systems.

www.zerohomebills.com

Understand that my constraints are different than many, but I’m looking to throw together a system that generates enough power to get ~5-10k miles of EV driving per year, meaning 1500-3000kWh/year valued at $300-600 off-peak.

More than 5 years to break-even all-in starts to seriously eat away at my motivation.

So any $5000 boxes are pretty much of a buzz-kill.

So having 3-5 strings with 3-5 budget MPPTs charging a battery and self-consuming that battery energy through self-consumption using my 2 $275 GTIL inverters (especially once peak hours kick-in) fits my budget much better than an expensive all-in-one.

I’ve got some shading issues which will cost me some lost energy, but another string of 2-3 panels and another MPPT goes a long way to compensating for that at a cost of under $700.

So I’m seeking an alternative that won’t cost much more than that.

There are new budget optimizers coming out of China: https://m.alibaba.com/amp/product/1600186277853.html

So perhaps one of these will support 100% buck.

With an optimizer supporting 100% buck compensation I’d be able to convert any half-shaded half-cut panels from Imp/2 @ Vmp to Isc @ Vmp/2 and get pretty much the same power out of my DC-coupled array as I’d get from a full-parallel string from a few lower-current series strings...

Though the microinverter + smart zeto-export charger is another appealing solution (at the right cost).

 

[Ampster]

But it is annoying to me that there is not yet a solution to allow AC-coupled battery charging designed for zero-export (full consumption of excess generation).

Maybe I missed something in earlier comments. The Enphase Envoy can be configured for zero export. If you put a charger on a timer then the zero export of the Envoy would power the charger to the extent there is sufficient solar generation. The only thing this solution can't do that a hybrid inverter can do is only charge the batteries from solar. In this scenerio the charger will run and draw from the grid if there is not enough solar.

 

[Hedges]

Instead of zero export, could the Envoy be programmed for 3kW export?

Then fafrd could add a second grid-tie system with different panel orientation from the first and square-up power production at 3kW for an extended period of the day. That would give more for charging electric car at night.

 

[Ampster]

Instead of zero export, could the Envoy be programmed for 3kW export?

That is a good question and I don't know the answer. In my situation I AC coupled the micros which I added subsequent to my PTO. My existing GT inverter which was approved in my PTO was also AC coupled and the hybrid inverter acted as the gatekeeper and only allowed the approve kW to go to the grid.

I am sure my production looks strange if PG&E were to look at it on a day that I had small loads. It would ramp up much quicker and ramp down much later than a typical 3.8 kW inverter with a flat top for much of the day. I would be producing a lot more kWhrs than my approved inverter would be able to produce but in my case there are loads that mask that during most of the day. Some people on other forums have remarked that I may be in violation of my PTO. Their theory is that my generation is flowing through my unapproved hybrid inverter. It is a grey area and I am already on NEM 2.0 so I am taking the risk. I understand how someone would make a different decision. As we know the electrons do not carry a fingerprint as to which inverter they originate from.

 

[Hedges]

As we know the electrons do not carry a fingerprint as to which inverter they originate from.

I see you haven't heard of the Chinese and their quantum-entangled communications.

I would be inclined to use the NEM approved GT inverter but over-panel with morning and afternoon oriented PV strings. That should peak at 3kW early and remain there most of the day.

 

[Ampster]

I would be inclined to use the NEM approved GT inverter but over-panel with morning and afternoon oriented PV strings.

Actually my NEM approved inverter is over panelled with a DC to AC ratio of 1.50 to 1 which would also mask my additional non NEM approved panel production. I agree it is a grey area and I am sure the lawyers at PG&E implied more when they used the term "system" in my NEM agreement.

I should note that having the system constructed to code and approved by the local building officials is an important assumption. When asked, my local officials said, there was no requirement to get any equipment "behind the meter" approved by PG&E as long as it did not export to the grid. My viewpoint is that my hybrid inverter us acting as a switch and the electrons are merely passing through that switch.

 

[GXMnow]

I am in a very similar situation. My NEM 2.0 agreement is only based on the 16 Enphase iQ7 microinverters. I can export 900 KWHs a month and a peak rate of 16 amps or 3,840 watts. Nowhere in the agreement does it mention at what time of day. My export would hit about 3,000 watts when no one was home at solar noon. Now the Schneider XW-Pro is charging my batteries at solar noon, so my export then never exceeds 1,500 watts, and the highest I see when the battery bank is topped off is about 2,000 watts. So I am well under what I am allowed to export. But where I used to have to buy 800 to 1,200 watts for the 5 hours from 4 pm to 9 pm, I now export or consume less than 500 watts. Of course, once I am stuck running the A/C I will be back to consuming. If I ever get questioned.... I just added a battery backup since we are being told we will have "Public Safety Power Shutoffs". It is all still connected back to my main panel using the same 20 amp breaker. And the Schneider is fully safety approved with a nice UL logo. If I do add more solar panels, that might cause an issue. But if they only charge the battery bank, does my utility have any say in that? That is one reason for having my additional panels DC coupled into the battery bank.

 

[Hedges]

You have a device that can use a clamp-on current transformer to implement net-zero(grid)

If you routed the wires from NEM approved GTinverter through the same current transformers it would implement net-zero(grid - GTinverter) or net-zero(grid + GTinverter) depending on direction of wire.

I always say, hardware is cheaper to implement than software.

With greater difficulty I could implement net-zero(grid + 3kW). Or more precisely net-zero(grid + 3000W/240V) without correcting for actual grid voltage.
What I would do is use a resistor 4800 ohm, 3W or larger to establish a 25 mA current from each 120V leg. Put 1000 turns of wire through the current transformer so it measures 25A. That offset would equal 3kW at 240V.

 

[Hedges]

My NEM 2.0 agreement is only based on the 16 Enphase iQ7 microinverters.

Enphase advertises their inverters based on how many watts of PV can be connected more than how many watts of AC they can produce.
How much over-paneling do yours have? How many hours clipped at max output?
If you added "Y" connectors and paralleled another panel of similar voltage, oriented toward different time of day, you would get more Wh.

 

[GXMnow]

Enphase advertises their inverters based on how many watts of PV can be connected more than how many watts of AC they can produce.
How much over-paneling do yours have? How many hours clipped at max output?
If you added "Y" connectors and paralleled another panel of similar voltage, oriented toward different time of day, you would get more Wh.

My current panels are rated at 300 watts even. The iQ7 inverters are rated for a constant 240 watts out. So I am over paneled by just 25% now. On my best cool sunny days, I have seen all 16 inverters clipped out for a bit over 2 hours. Most of the year, they barely reach clip, so adding more panel to the input could spread that out, especially if a parallel a good evening aimed panel with the ones that get shadowed at 4 PM. If I added the 4 panels on my garage roof, and had them aimed a bit west. I might be able to pull another 2 to 4 KWH's a day. Those 4 inverters would likely climb faster, even in the morning, clip out longer, and hold power later in the day. Under the Rapid Shut Down rules, it "might" still be legal. This would require about 50 feet of cables from the new panels to where they would parallel with the existing panel iQ7s. Those wires could potentially be at VOC voltage of one panel. That is just 40 volts on my 60 cell panels. Is that acceptable? I could also add 2 more on the end of the lower roof array. They would be shadowed in the morning, get full sun along with the other panels for about 4 hours, so the inverter will clip, but then keep making power as the ones on the other end of the array go into shadow from the remaining palm trees. The ones on the flat garage roof will be much easier to turn to face west. I'll have to watch the sun travel angle in summer when I need the extra power to run the air conditioning. If I can keep up 800 watts or so to 7 pm it would help a ton. At this time of year, my array is dropping to zero by 5:45 pm, but add in "Daylight Savings Time" and the longer summer days, with the panels aiming that way, I think 7 pm is doable. My current arrays are actually 25 degrees to the east. I can probably hit about 30 degrees west with a row of 4 panels on the garage. My flat roof area is about 18 feet square. I want to be 55 degrees off axis of the roof, so the right triangle ends up about 7.5 feet by 10.7 feet. I just sketched that out. It seems like a pretty extreme angle. But when I think about it, that would be facing right into the evening sun. How far up to I want to tilt it? The sun will start to drop behind trees, so I will get the most power if I aim them still above the trees. Also the flatter they are, the earlier in the day they will start to produce as well. I may only tilt them up about the same 20ish degrees of my current panels. I'll have to plug that into one of the insolation calculators and see what I get. This may be the cheapest way to gain that extra power in the summer.

I just put the numbers into the "solarelectricityhandbook" calculator, and having the array turned from my current 25 degrees East to 30 degrees West will make nearly identical total power per panel in a day. It just shifts the peak generation about 3.66 hours later in the day. This looks like it could clip out the inverters from before 11 am to after 5 pm. And depending on the tree shadows, might still have some power out to almost 8 pm in June. But I may have to cut down one more palm tree.

 

[fafrd]

Instead of zero export, could the Envoy be programmed for 3kW export?

Then fafrd could add a second grid-tie system with different panel orientation from the first and square-up power production at 3kW for an extended period of the day. That would give more for charging electric car at night.

Thinking over what we were discussing, it’s really just setting up an AC battery charger as a dump load.

A single Microinverter (yes, that’s what NEP is calling their new 500W Microinverter) fed by a 450W panel will peak out at ~335W of production on my roof.

I consume an average of 350W all day long (primarily 5 fridges) so in general, I’ll be using all of that generated AC power before it reaches the grid.

With a clamp meter, I can detect when consumption drops to under 100W and use that to turn on a switch powering my 10A 8S LiFePO4 charger which consumes 350W. Essentially, whenever some combination of the fridges is consuming the power being generated, the AC charger will be off, but whenever the combined consumption of the fridges drops under the power being generated, the AC charger kicks in to take up the slack before it exports. Because the fridges all already duty-cycle, I wouldn’t even need to worry about a timer (thrashing should be low).

And battery fully-charged can be used to disconnect the Microinverter until after the sun goes down.

So it is possible, even tempting, but it’s complicated and the benefit is marginal.

I’m coming to the conclusion that running more wires to have more MPPTs driving more strings in parallel is the best bang-for-the-buck as far as dealing with my shade issues.

MPPT SCC and Watts of panel production are the least expensive elements in a system and DC-coupling makes production throttling so much easier than AC, as well as providing a bright line between production/storage and generation/consumption.

 

[opticalcarrier]

can someone escribe this NEM/net/zero/export thing? is it simply trying to not export power that for whatever reason you would end up not getting paid for?

 

[Hedges]

can someone escribe this net/zero/export thing? is it simply trying to not export power that for whatever reason you would end up not getting paid for?

Basically.

Biggest reason would be if utility forbids export (for instance in Hawaii there was so much PV installed the grid couldn't handle more without becoming unstable.)

Next would be if costs to do net metering make it uneconomical, like utility rates would be raised higher and your PV production is less than consumption at peak times, resulting in your bill going up not down. Or just high monthly fees.

The original grid-tie PV just shoved everything it could into the wires and usually back-fed the grid. A zero-export setup would let power be drawn from the grid but would never backfeed.

And then there are battery systems to store whatever would have gone out to grid, later try to supply from battery rather than drawing from grid. Seems to me that needs at least a relay to disconnect GT PV if necessary to prevent any backfeed when battery full.