diy solar

diy solar

Help with question about wire derating

<I put a 48VDC fan in the attic and have a couple 12V panels in series feeding that, so afternoon sun vents the attic. Also used to power fans on my older passively cooled inverters.>

What did that rig run you? Rather than burning up one of my rack positions, I think I’d rather just accept the 25% efficiency loss and power with AC (or pick up one if those fan/vents with an integrated panel).

About $50 at Weird Stuff Warehouse (RIP).
Some Pabst fans for $12 at HalTed Specialties (RIP).

The trick is finding one that works with slowly waking up PV. Some will shutdown until the next power cycle.
I hung the 48V fan which had been from top of a rack in a gable vent, put Pabst box fans lying on top of SWR2500U heatsinks (but they didn't like rain exposure.) Couple spare panels in the yard, and I get variable cooling in proportion to sunshine. No timer or thermostat required.
 
<I put a 48VDC fan in the attic and have a couple 12V panels in series feeding that, so afternoon sun vents the attic. Also used to power fans on my older passively cooled inverters.>



About $50 at Weird Stuff Warehouse (RIP).
Some Pabst fans for $12 at HalTed Specialties (RIP).

The trick is finding one that works with slowly waking up PV. Some will shutdown until the next power cycle.
I hung the 48V fan which had been from top of a rack in a gable vent, put Pabst box fans lying on top of SWR2500U heatsinks (but they didn't like rain exposure.) Couple spare panels in the yard, and I get variable cooling in proportion to sunshine. No timer or thermostat required.
Cool.

If I end up needing to cool my attic, I will hit you up for advice...
 
Ok, now you’ve got me interested-enough in an AC-coupled solution that I think I should make sure the wiring I am installing will support AC as well as DC-coupling.

I did a little Google research I found a few Enphase / Envoy threads here but nothing definitive.

As far as wiring (the subject of this thread), I think it’s pretty straightforward:

If I’ve got 4500W connected through 240V Microinverters, that would mean 18.75A, no problem on 8/2 wiring even with all deratings applied.

If I’ve got 4500W connected through 120V Microinverters, that would mean 37.5A, taking me beyond the capability of 8AWG and meaning I’d need at least a single 6AWG run to handle that current.

Is it allowed to run a single 120V AC feed on multiple 8AWG wires? Can the microinverters be divided up into seperate 120V AC strings?

I’ve got an idea of what a ‘perfect’ AC-coupled system would look like from my perspective, but I’ll start a new thread for that discussion...

Of course you can have a 120V 8 awg circuit, with suitable breaker (like 50A). You could have two runs of 8 awg, but still only 50A breaker.
You should be able to allocate 120V inverters between neutral and two hot legs with a 2-pole 50A breaker. Unless they are required to be on a smaller breaker, due to what their wires can handle if shorted.

I would rather find a string inverter that supports zero export. I think the SMA ecosystem does, but I'm not clear on how to use the add-on boxes. They don't seem to document theory of operation and system configurations like they used to. Probably a number of vendors do now, with markets like Hawaii blocking export (maybe California's Rule 21 and Hawaii's similar rule would open that back up. But frequency/watts doesn't seem good to me for a distributed grid having localized excess. Volts/Watts should cover that.)
 



But I still can't find details on the hardware and how to hook it up.
 
Of course you can have a 120V 8 awg circuit, with suitable breaker (like 50A). You could have two runs of 8 awg, but still only 50A breaker.
You should be able to allocate 120V inverters between neutral and two hot legs with a 2-pole 50A breaker. Unless they are required to be on a smaller breaker, due to what their wires can handle if shorted.
Yeah, if the 120V inverters are used to generate separate L1 and L2 legs, turn it’d only by a max of 18.85A/leg.

And also, I don’t recall having to do the A*125%*125% derating for my AC array (but may have forgotten) - are AC and DC -coupled PV systems all derated the same way?
I would rather find a string inverter that supports zero export. I think the SMA ecosystem does, but I'm not clear on how to use the add-on boxes. They don't seem to document theory of operation and system configurations like they used to. Probably a number of vendors do now, with markets like Hawaii blocking export (maybe California's Rule 21 and Hawaii's similar rule would open that back up. But frequency/watts doesn't seem good to me for a distributed grid having localized excess. Volts/Watts should cover that.)

The frequency/Watts control is attractive but I’ve heard it’s still being debugged. The most appealing thing about a Microinverter-based system is no-cost emergency disconnect (built-in).

But in my case, I’d need to assure that the communication from wherever the controller is to the new microinverters doesn’t interfere with my existing NEP Microinverter-based system (so an frequency shift on either L1 or L2 needs to either be isolated to the new grid or extremely tight tolerance).

Don’t quite understand what you mean by ‘Volts/Watts should cover that’?
 



But I still can't find details on the hardware and how to hook it up.
My biggest issue with many of these systems are that they are closed and only work with specific (generally very expensive) batteries (is: Schneider Conext).

Do any of these solutions (Sunny, Enphase) support DIY LiFePO4 batteries yet?
 
And also, I don’t recall having to do the A*125%*125% derating for my AC array (but may have forgotten) - are AC and DC -coupled PV systems all derated the same way?
For AC, we just do continuous amps x 125%. That's 25% headroom to avoid nuisance trips.
PV gets an extra 25% because panels can receive more than one full sun, due to light reflected off clouds and glowing through clouds.

The frequency/Watts control is attractive but I’ve heard it’s still being debugged. The most appealing thing about a Microinverter-based system is no-cost emergency disconnect (built-in).

SMA Sunny Island has been doing it for many years and it works fine. Some other vendors had issues along the way, and one member's experience using Enphase with Schneider is detailed on a thread. He's also actively bumping parameters around to manage battery charging and export.

But in my case, I’d need to assure that the communication from wherever the controller is to the new microinverters doesn’t interfere with my existing NEP Microinverter-based system (so an frequency shift on either L1 or L2 needs to either be isolated to the new grid or extremely tight tolerance).

I think Envoy uses power line modem (high frequencies superimposed) for control.
Nobody can do frequency shift when tied to the grid.
My Sunny Boys use frequency shift while off-grid. Because they would deliver 100% at 60 Hz, before reconnecting to grid Sunny Island raises frequency beyond limits causing Sunny Boy to drop off while it connects; during that time it supplies the load. When Sunny Boy return they can deliver 100% because it feeds the grid.

Don’t quite understand what you mean by ‘Volts/Watts should cover that’?

Rule 21 has one implementation that requires GT inverter to ride through frequency and voltage excursion for 299 seconds.
Alternate implementation is to reduce power with increasing frequency (watts/frequency) or with increasing voltage (watts/voltage)
Watts/voltage seems like the more obvious approach, because if grid voltage starts to rise, simply delivering less power would avoid pushing it too high. That should manage neighborhoods at the end of a long wire without affecting others in the middle of the city.

It appears the current model Sunny Boy can have frequency/watts enabled. That lets them work with Sunny Island for grid-backup (previously only said to be compatible for off-grid setups.)

There is now a supported data link to the new Sunny Boy as well. That may be how zero-export can be implemented. Link shown in the data sheet I linked here.


Also shown in this data sheet

 
My biggest issue with many of these systems are that they are closed and only work with specific (generally very expensive) batteries (is: Schneider Conext).

Do any of these solutions (Sunny, Enphase) support DIY LiFePO4 batteries yet?
Good article - thanks.

‘Over-sizing of inverters is becoming more commonly accepted; to achieve a subsidy the CEC allows a ratio of 133 per cent over-sizing. This is not at all the best ratio. The economic sweet spot, in-terms of return on investment (ROI), is somewhere between 150 per cent and 200 per cent over-sizing.
But, for now, the CEC (in its infinite wisdom) has decided that inverters must be oversized by no more than 133 per cent if a solar system owner wants to receive STCs.

Yeah, exactly - PG&E allowed me to put in a 4kW array with 3kWs-worth of Microinverters.

A central grid-tie inverter could be attractive to me if it supports DIY battery but it’s also got to be reasonably-priced.

It sounds like all of this is still on a state of flux, but when my cheapo Sunpower GTIL inverters poop out on me, try his is likely to be the way I go.

Don’t care too much about Microinverter-based versus central-inverter based so for me right now, the main thing to focus in is that my string wiring will support all of these options.

5 1S2P strings of 450W panels brought from roof to basement in 8AWG wires should allow me a lot of flexibility - it’l like having 5 quad-cut 900W super-panels wired all the way to the basement.

If I need higher voltage I can go as high as 5S2P (or even 10S1P) just rewiring the PV connections on the roof and sending the string(s) down through one or two of my 8AWG home runs.

For where things are now, wasting ~$100 on 8AWG home run wiring I may not end up using seems like a worthwhile (very modest) investment.
 
For AC, we just do continuous amps x 125%. That's 25% headroom to avoid nuisance trips.
PV gets an extra 25% because panels can receive more than one full sun, due to light reflected off clouds and glowing through clouds.
Yeah, that’s right (now I remember) - 125% derating for AC (and thanks for the explanation about what’s different with DC).

So 8AWG could carry 4500W of 120VAC (but not with much of any room for temperature derating). My AC gone runs through the same attic in Romex and I don’t recall whether I had to oversize for temperature derating or not. Of course, for an AC-coupled Microinverter solution, it would ideally be split into the two phases, so only 18.75A per 8AWG wire (plenty of headroom for a hot att
SMA Sunny Island has been doing it for many years and it works fine. Some other vendors had issues along the way, and one member's experience using Enphase with Schneider is detailed on a thread. He's also actively bumping parameters around to manage battery charging and export.

It’s supposed to be an open standard. I talked to Victron about using it with APS microinverters (they made an announcement together) but they basically advised against it... Outsude if walled-gardens, it sounds like it’s still gel
I think Envoy uses power line modem (high frequencies superimposed) for control.

PLM is how my NEP monitor communicates with the microinverters. If these communication islands are not isolated, you need to worry about ontetfetence
Nobody can do frequency shift when tied to the grid.

Yes, that’s right (duh!)
My Sunny Boys use frequency shift while off-grid. Because they would deliver 100% at 60 Hz, before reconnecting to grid Sunny Island raises frequency beyond limits causing Sunny Boy to drop off while it connects; during that time it supplies the load. When Sunny Boy return they can deliver 100% because it feeds the grid.
Yeah, that’s a classic backup solution. So the idea they have is to reserve frequency shift for off-grid backup and use PLM for ‘smart’ features like zero export while grid-tied?
Rule 21 has one implementation that requires GT inverter to ride through frequency and voltage excursion for 299 seconds.
Alternate implementation is to reduce power with increasing frequency (watts/frequency) or with increasing voltage (watts/voltage)
Watts/voltage seems like the more obvious approach, because if grid voltage starts to rise, simply delivering less power would avoid pushing it too high. That should manage neighborhoods at the end of a long wire without affecting others in the middle of the city.
Didn’t know about voltage-raising control - is that part of the Rule 21 standard that all vendors must suppprt? And is that also only when off-gr
It appears the current model Sunny Boy can have frequency/watts enabled. That lets them work with Sunny Island for grid-backup (previously only said to be compatible for off-grid setups.)

Not understanding - so you need Sunny-Island to isolate the load from the grid for this all to work (remember, I’m interested in a full-parallel solution with no/minimal rewiring needed...).
There is now a supported data link to the new Sunny Boy as well. That may be how zero-export can be implemented. Link shown in the data sheet I linked here.


Also shown in this data sheet

I’ll need to digest those datasheets, but my main concerns remain:

-get away from walled gardens and assure support for DIY LiFePO4 batteries

-zero-export support net of existing grid-tied array production (and without interfering with that existing AC-coupled array)

Whether Microinverter or central inverter-based, if there was a battery-tied inverter-charger that also communicated either to microinverters (AC-based) or had multiple MPPTs (DC-based) to control battery charge/discharge based on zero-export and array output (AC or DC) based on DIY battery being fully-charged and more power available than current loads (zero export), that would be the product I want.

I think it’s still too early and all this stuff is in a state of flux and development, but that’s one of the reason I’ve planned for my ‘real’ build to be in 2023 (but hopefully supported by the roof-to-basement wiring I’m putting in now...).
 
For AC, we just do continuous amps x 125%. That's 25% headroom to avoid nuisance trips.
PV gets an extra 25% because panels can receive more than one full sun, due to light reflected off clouds and glowing through clouds.



SMA Sunny Island has been doing it for many years and it works fine. Some other vendors had issues along the way, and one member's experience using Enphase with Schneider is detailed on a thread. He's also actively bumping parameters around to manage battery charging and export.



I think Envoy uses power line modem (high frequencies superimposed) for control.
Nobody can do frequency shift when tied to the grid.
My Sunny Boys use frequency shift while off-grid. Because they would deliver 100% at 60 Hz, before reconnecting to grid Sunny Island raises frequency beyond limits causing Sunny Boy to drop off while it connects; during that time it supplies the load. When Sunny Boy return they can deliver 100% because it feeds the grid.



Rule 21 has one implementation that requires GT inverter to ride through frequency and voltage excursion for 299 seconds.
Alternate implementation is to reduce power with increasing frequency (watts/frequency) or with increasing voltage (watts/voltage)
Watts/voltage seems like the more obvious approach, because if grid voltage starts to rise, simply delivering less power would avoid pushing it too high. That should manage neighborhoods at the end of a long wire without affecting others in the middle of the city.

It appears the current model Sunny Boy can have frequency/watts enabled. That lets them work with Sunny Island for grid-backup (previously only said to be compatible for off-grid setups.)

There is now a supported data link to the new Sunny Boy as well. That may be how zero-export can be implemented. Link shown in the data sheet I linked here.


Also shown in this data sheet

The Sunny Boy Storage looks very interesting. My two issues are the required battery voltage and it’s only able to backup critical loads (which means rewiring).

The GTIL inverters I have will not run during power outages. I’ve got a 3kW PSW inverter that I can use to create a local grid during power outages (after manually disconnecting the main breaker) and then if I need additional power, the 2 GTIL inverters can have their sensors repositioned to add a further 1kW + 1kW.
 
Rule 21 is required in California, but several choices of which feature to implement.

Watt/frequency happens to be what SMA has been using for years for "island" off-grid, or for "grid backup" when temporarily disconnected. Exact behavior and parameter limits may vary.

For years (like a decade and a half) most if not all Sunny Island had a mode of adjusting output power in response to frequency. Some models were flagged as "off grid" only, and some "grid backup" but not "off grid"?? Never saw the explanation for that one.

Latest models -40 and -41 Sunny Boy lacked the RS-485 bus previously used as part of "grid backup" to tell Sunny Boy if Sunny Island was taking on the UL-1741 function and if off-grid. For older models, I've watched Sunny Boy change state, but I never managed to trigger on the data bus and capture a waveform of that message. But I was able to see communication by a different device "Sunny Data Control". Normally RS-485 is single-master, but somehow they have multi-master working.

The -40 and -41 do have Speedwire. It looked like European model Sunny Island had that bus and used it for communication with Sunny Boy. Bus adapter is available for US model, but not for that purpose (probably not in the firmware). Looks to me like the external gateway and data manager now take RS-485 from Sunny Island and translate to talk to Speedwire and make things work. (easier to implement a hardware solution than a software/firmware one.)


You can use DIY lithium batteries with Sunny Island; it requires a compatible BMS. REC makes one (about $500).
I'm not clear on good ways to use Sunny Island to shift time of consumption/production with grid tie. It provides off-grid or grid-backup. Their newer product Sunny Boy Storage (400V lithium) is all about shifting time of use and backup requires external hardware.
 
The Sunny Boy Storage looks very interesting. My two issues are the required battery voltage and it’s only able to backup critical loads (which means rewiring).

The GTIL inverters I have will not run during power outages. I’ve got a 3kW PSW inverter that I can use to create a local grid during power outages (after manually disconnecting the main breaker) and then if I need additional power, the 2 GTIL inverters can have their sensors repositioned to add a further 1kW + 1kW.

It has an ability to run loads while grid is down, but not recharge from PV/Sunny Boy. Kind of like the new Sunny Boy have "secure power" to do PV direct to AC without battery.

Primary function of Sunny Boy Storage is shifting time of use, for instance to implement zero export with storage.

To do full backup, it requires "ABU", Automatic Backup Unit. That is a transfer switch and 120/240V transformer.
Last I read, only a single Sunny Boy Storage, 6kW and anemic surge capability, could be used. No paralleling (but of course you could have two separate backup systems.) Perhaps that has changed by now.

LG's RESU10H was one battery for it a year ago. Supports 3 batteries, so that would have been 30kWh storage. I think 3x 20kW or maybe 30xW batteries are available. But not RESU10H - recalled last December for fires (not LiFePO4)

Sunny Island is and was the heavy duty off-grid solution. Latest data sheet only shows one for 120V or two for 120/240V. But it also supports 2s2p and 3-phase. The European model will probably reach US eventually. Don't know what features it might offer.
 
It has an ability to run loads while grid is down, but not recharge from PV/Sunny Boy. Kind of like the new Sunny Boy have "secure power" to do PV direct to AC without battery.

Primary function of Sunny Boy Storage is shifting time of use, for instance to implement zero export with storage.

To do full backup, it requires "ABU", Automatic Backup Unit. That is a transfer switch and 120/240V transformer.
Last I read, only a single Sunny Boy Storage, 6kW and anemic surge capability, could be used. No paralleling (but of course you could have two separate backup systems.) Perhaps that has changed by now.

LG's RESU10H was one battery for it a year ago. Supports 3 batteries, so that would have been 30kWh storage. I think 3x 20kW or maybe 30xW batteries are available. But not RESU10H - recalled last December for fires (not LiFePO4)

Sunny Island is and was the heavy duty off-grid solution. Latest data sheet only shows one for 120V or two for 120/240V. But it also supports 2s2p and 3-phase. The European model will probably reach US eventually. Don't know what features it might offer.
Cool. From the other thread, think I understand your setup now. I’m hoping to avoid any rewiring (and we’ll see if I get away with it). Avoiding rewiring and any additional failure point is a higher priority for me than having fridges automatically stay on when the grid fails and there is no one home.

I also suspect a top-cabin solution like that cost you a pretty penny. Because it’s a ‘practice/learner’ build,I’m cheaping-out and going with mainly cheapo Chinese components:

30A HQST 12/24V 780W MPPT $35
3kW WZRELB PSW Inverter $200
(2) 1kW Sunpower GTIL inverters $275 ea.

So all-in, I’ve spent less than $800 for charger & inverter components that should generate well more than the 1200kWh PG&E’s TOU changes will cost me (tied to the 3 380W panels I’ll be feeding the MPPT with).

A couple years from now I’ll buy better components and sill have a much better idea of what I need.

You’ve helped to see that the mainstream inverter companies are moving much faster than in the past as far as solar+storage for zero-export - I was starting to think they only the Chinese knew how to drive any real innovation in this industry anymore...
 
Back to the original subject of this thread, I have some more questions about NEC and wiring guidelines I would appreciate feedback on:

My MC4 connectors and MC4 Y connectors are rated for 50A @ 125C (yes, C, not F).

So as long as those are protected by a 40A fuse, they will be safe (125% x 40A = 50A).

8AWG PV wire is rated for 80A in air, so derated to 87% for 45C, it is rated for up to 69.6A and needs to be protected by a fuse no larger than 55.68A (so more than protected by a 40A fuse).

So the only restriction on 2P versus 3P on 8AWG PV wire comes down to whether that 40A fuse suffers from nuisance trips or not.

NEC says to use Iscmax x 125%, so in my case 23A x 125% = 28.75A for 2P (so fine protected by a 40A fuse which is almost 40% bigger), while 34.5A x 125% = 43.125A for 3P (which obviously cannot be protected by a 40A fuse).

On the other hand, the MPPT controller I’m using is limited to 780W, so at Vmp, it will never draw more than 780W / 50V = 15.6A. 125% if that realistic max current is 19.5A and that can be protected by a fuse which is at least 125% of that, or 25A (and certainly by a 40A fuse).

And in practice, my existing array never goes beyond 75% of rating, which translates to 27A at Vmp for 1350W of panels (3x450W). 125% of that realistic current in the case of an SCC failure/short is 33.75A and a 40A fuse is only 18.5% beyond that realistic short-circuit current, but in this case I’d want a trip, so less margin is better protection.

So realistically, the only way I’ll ever have more than 40A of current is if there has been a failure in the SCC (short) AND there is exceptional irradiance driving current beyond Iscmax.

So even though the NEC says I should have wiring and fuses able to handle 125% of 43.125A (meaning 54A), there will never be an unnecessary nuisance trip if that 3P string is protected by a 40A fuse.

So I’m questioning the meaning of the NEC guidelines in the case of over paneling a string (which is apparently becoming increasingly common).

Any perspective appreciated - I’m basically back to considering 3 3P strings wired through 2 6/3 MC conduits...
 
Last edited:
If SCC limits input current to a particular maximum, then I'd consider 1.25 times that to be minimum fuse size.
If PV array voltage was (or could be) lower, then SCC would draw more current to get enough watts to deliver its maximum output current. Output current is what it regulates.

An SCC may try pulling array lower than Vmp, therefore going higher than Imp, as it looks for maximum power point (of which there can be multiple in the event of series/parallel arrays. But probably it would not if it is already clipping at maximum output.

Sizing OCP for Isc (times 1.25 for extra illumination) and multiplying that by 1.25 for margin is the most certain way to ensure OCP doesn't trip.
Sizing the wire for that is necessary to avoid wire overheating in the event of a short; SCC doesn't come into play.
 
If SCC limits input current to a particular maximum, then I'd consider 1.25 times that to be minimum fuse size.
If PV array voltage was (or could be) lower, then SCC would draw more current to get enough watts to deliver its maximum output current. Output current is what it regulates.
Yes, the key is to be certain that the SCC cannot lower voltage to a level where current can exceed the 80% of the level you fused for. For example, my SCC has a minimum of Vbattery+2V so I never need to worry about voltage dropping to ~1/3 Vmp_panel (~13.7V) when charging a 24V battery with my 41Vmp panels...

But 2/3 of Vmp_panel = 27.3V which means my SCC can drop to that voltage if Vbattery is 25.3V or lower (and it is not saturated). So I need to assure that my fusing can handle any overradience at that voltage (either fused for 125% Iscmax or assuring SCC will have saturated by 125% Iscmax @ 27.3V, in which case I can fuse for W / 27.3V (max current before saturation @ 27.3V) x 125%
An SCC may try pulling array lower than Vmp, therefore going higher than Imp, as it looks for maximum power point (of which there can be multiple in the event of series/parallel arrays. But probably it would not if it is already clipping at maximum output.
Yes, that is the ley
Sizing OCP for Isc (times 1.25 for extra illumination) and multiplying that by 1.25 for margin is the most certain way to ensure OCP doesn't trip.
Most certain and easiest, yes. But places unnecessary constraints in wire size and fusing in the case of overpanelling (especially with 1S parallel arrays).

Sizing the wire for that is necessary to avoid wire overheating in the event of a short; SCC doesn't come into play.

Sizing wire for 125% of 125% of maximum current is unnecessary if that extra current resulting from extra radiance has nowhere to go.

Imagine the home run wires being left open - current is zero so no need to fuse or size wire for any specific current.

Now if those wires can be shorted by whatever they are connected to, you need a fuse to protect the wires - the fuse must be no more than 80% of wire rating (detested as appropriate).

If a short occurs and that fuse doesn’t trip, it’s not a problem. And if a short occurs and that fuse does blow, it’s not a nuisance trip (because a short has occurred). Extra radiance has nothing to do with it.

The SCC does have something to do with it because it’s the SCC which determines whether a trip was a nuisance trip or not. SCC working as expected and extra radiance causing a trip = a nuisance trip.

SCC shorting and fuse being blown with or without extra radiance and that is not a nuisance trip (because there is a serious issue).
 
Rule 21 is required in California, but several choices of which feature to implement.

Watt/frequency happens to be what SMA has been using for years for "island" off-grid, or for "grid backup" when temporarily disconnected. Exact behavior and parameter limits may vary.

For years (like a decade and a half) most if not all Sunny Island had a mode of adjusting output power in response to frequency. Some models were flagged as "off grid" only, and some "grid backup" but not "off grid"?? Never saw the explanation for that one.

Latest models -40 and -41 Sunny Boy lacked the RS-485 bus previously used as part of "grid backup" to tell Sunny Boy if Sunny Island was taking on the UL-1741 function and if off-grid. For older models, I've watched Sunny Boy change state, but I never managed to trigger on the data bus and capture a waveform of that message. But I was able to see communication by a different device "Sunny Data Control". Normally RS-485 is single-master, but somehow they have multi-master working.

The -40 and -41 do have Speedwire. It looked like European model Sunny Island had that bus and used it for communication with Sunny Boy. Bus adapter is available for US model, but not for that purpose (probably not in the firmware). Looks to me like the external gateway and data manager now take RS-485 from Sunny Island and translate to talk to Speedwire and make things work. (easier to implement a hardware solution than a software/firmware one.)


You can use DIY lithium batteries with Sunny Island; it requires a compatible BMS. REC makes one (about $500).
I'm not clear on good ways to use Sunny Island to shift time of consumption/production with grid tie. It provides off-grid or grid-backup. Their newer product Sunny Boy Storage (400V lithium) is all about shifting time of use and backup requires external hardware.
Interesting but still pretty much of a ‘walled garden’ if it only functions with a specific BMS (especially for $500).

A Microinverter-based system combined with a smart and efficient AC battery charger / inverter which tracks PV generation and self-consumption and charges battery to absorb any excess generation as well as discharge battery to cover any self-consumption would be attractive to me, especially if it is smart and fully-programmable as far as controlling times and actions based on battery SOC.

I basically get most of that with my GTIL / MPPT rig:

MPPT will throttle PV output once battery is fully-charged.

GTILs can cover up to ~2kW of self consumption on a schedule controlled by 2 cheapo 1kW (8.3A) lamp timers.

GTILs can be programmed for minimum battery start voltage following LVD - it’s not as precise as SOC, but provides another way to delay start of supplying self-consumption in the morning as the battery begins to charge following being drained overnight,

Biggest gripe I have is the 75% efficiency when operating @ 24V but either higher-efficiency products will materialize or I’ll move up to 48V...
 
It has an ability to run loads while grid is down, but not recharge from PV/Sunny Boy. Kind of like the new Sunny Boy have "secure power" to do PV direct to AC without battery.

Primary function of Sunny Boy Storage is shifting time of use, for instance to implement zero export with storage.

To do full backup, it requires "ABU", Automatic Backup Unit. That is a transfer switch and 120/240V transformer.
Last I read, only a single Sunny Boy Storage, 6kW and anemic surge capability, could be used. No paralleling (but of course you could have two separate backup systems.) Perhaps that has changed by now.

LG's RESU10H was one battery for it a year ago. Supports 3 batteries, so that would have been 30kWh storage. I think 3x 20kW or maybe 30xW batteries are available. But not RESU10H - recalled last December for fires (not LiFePO4)

Sunny Island is and was the heavy duty off-grid solution. Latest data sheet only shows one for 120V or two for 120/240V. But it also supports 2s2p and 3-phase. The European model will probably reach US eventually. Don't know what features it might offer.
My biggest problem is that it looks like it requires rewiring. The most beautiful thing about the Sunpower GTILs is that they operate in parallel through standard outlets (1 per leg).

So they can be removed anytime - you have not changed anything to make your normal grid hook-up different...
 
MPPT SCC have various max Voc, typically 50V, 100V, 150V, 200V, 250V, 600V
At least in the case of Midnight Classic, 200V and 250V cost more per watt than does 150V.

How many panels in series depends on the panel specs. Nominally 12V, 24V, 36V, sometimes a bit higher.
I had to hunt back to find this earlier post (for context).

Largely inspired by this exchange with you, I’ve analyzed my shading and how MPPTs work with series strings and I’ve discovered that series strings will, in fact, work at least as well if not better for my extreme shading issues.

I’ve got three panels which will be clear all day except for 1 hour in the late morning when a moving flagpole-like shadow will impact the lower half of 1 panel (they are split-cell, so 2 half-panels in parallel.

3P would have dropped to 83% production during that hour, but 3S will drop the voltage to 77% (2 bypass diodes activated) to 88% (single bypass diode activated) and will be roughly the same efficiency.

I’ve got another 4 panels that will have their bottom haves largely shaded while their upper haves are clear, and a properly-designed 2P2S array (functioning like 4P2S because of the split cells) will work as well or better than a 4P array at half the current).

And the final 3 panels are interesting because they are on the boundaries of the moving shadow, so a 3S string allows one panel to always be clear while two panels on the opposite boundaries are partially shaded and a 3S array will perform much better than a 3P array because it will allow bypass diodes to be activated to get partial output from the partially-shaded half-cells (while a 3P array would have merely bypassed all partially-shaded half-are arrays).

So pretty much triggered by your comments on how much happier you are with how your series strings performed, I’ve changed course and will now be going with 2 40A 200V MPPTs for 2 3S1P strings and one 60A 200V MPPT for a single 2S2P string.

I add isn’t correctly understand how bypass diodes work until I made this effort, so thanks for your persistance in suggesting I might be making a mistake thinking they a full-parallel array would be superior for my situation...
 
Largely inspired by this exchange with you, I’ve analyzed my shading and how MPPTs work with series strings and I’ve discovered that series strings will, in fact, work at least as well if not better for my extreme shading issues.

...

I add isn’t correctly understand how bypass diodes work until I made this effort, so thanks for your persistance in suggesting I might be making a mistake thinking they a full-parallel array would be superior for my situation...

Key assumption is that bypass diodes work, don't fail if continuously carrying Imp. Some brands/models can't handle that.



 
Back
Top