Micro inverters at the edge of array or deep inside the array

[agarg]

I have an array of 22 panels. A micro inverter other than enphase typically supports 2 or 4 panels. I am thinking of placing micro inverters on the edge of the array and use long PV cable to connect the micro to the panels deep inside the array. The reason, I am thinking like this is if I need to service any of the micro-inverters, I can just disconnect and pop in a new and power that back up again by restoring grid back to the array. The micro have many many electronic parts that may fail where as panels have just a few diodes. So, somehow I expect micro to fail more often than panel.

There is another part of my mind says using shorter cable is not only efficient, its also cleaner and so what if the micros end up being where hands wont reach. And, assuming the micro failure being rare, I can also get the panels lifted out and replaced back after the defective one has been replaced. This is more so as many people say that we never need to change out these micro and they are more likely to go obsolete than fail.

What are the thoughts of you learned folks on this forum.

Much appreciate.

Anil

PS: Illustration of the array attached.

 

[400bird]

Do you remember the Walmart solar fires with Tesla or was it Solar City solar?
Those were caused by the MC4 connectors. If you run extensions to put all your micro inverters up top each extension adds more connectors and more points of failure.

How long do you expect the inverters to last? Are you planning on one fail per year? Would you fix this yourself or need to hire out the solar contractor to do the work?
Pulling a panel isn't hard, but that's my perspective and I installed my system and went through the permitting process.

 

[agarg]

Oh. That is interesting. MC4 connectors starting fire. It does sound logical and perhaps that is exactly the reason arc fault has been added in the NEC. And, yes, I hadn't thought but more MC4 connectors means more potential locations for spark after they have been given repetitive hot and cold treatment. So that means find location that needs least number of mc4 cables between inverter and panel.

 

[zanydroid]

I've thought about some creative ways of using 2- and 4- port microinverters, in similar vein to what you're thinking about. Though I haven't thought of this one before. As long as the cables are within the same array (modulo the array boundary definition) you're probably OK using the Rapid Shutdown exemption granted to microinverters in the code. Usually I end up convincing myself to just do things the orthodox way.

It's possible that some topologies using 4-port microinverters will anyway need extension cords for some of their ports. In the install I'm planning for my house I expect to need them on some oddly configured parts of my house. But I still plan to put the microinverters in a centrally located place, with maybe a little bit of attention paid to servicing.

Microinverters *should* fail independently and at random, so you could even accept some lost microinverters as the cost of doing business. (If there's a systemic error with one production batch, that may not be true. Good luck asking your distributor to give you microinverters from a randomized assortment of batches)

In your picture, you may be able to achieve what you want (easy access to microinverters) for large portions of the large array, with only some on the inside. For the small 4 panel arrays, you just need to install them the standard way for a 4-port microinverter, and they're a pretty good match for microinverters anyway (EG small arrays on a unique roof slope). So your question is moot.

Keep in mind the current at peak output of 72cell solar panels is pretty high. Assuming Vmpp is 40V, you're looking at 10A. What's the voltage drop at the lengths you are thinking of? The total length of wire in this string might be longer than a single DC string of the same current, which translates to more losses. I haven't done the math to see how important this feels.

Now, for the large array, you could instead do that as a DC string with RSD devices (and maybe optimizers if you feel spicy). They are on the same plane, so microinverters provide less gain. And it may anyway be easier to integrate with your storage.

I believe you are presuming that a solar contractor doing work on your house will pass the labor savings onto you. I dunno if that's a valid assumption. Especially if it becomes really confusing which microinverter corresponds to which panel.

 

[agarg]

I've thought about some creative ways of using 2- and 4- port microinverters, in similar vein to what you're thinking about. Though I haven't thought of this one before. As long as the cables are within the same array (modulo the array boundary definition) you're probably OK using the Rapid Shutdown exemption granted to microinverters in the code. Usually I end up convincing myself to just do things the orthodox way.

It's possible that some topologies using 4-port microinverters will anyway need extension cords for some of their ports. In the install I'm planning for my house I expect to need them on some oddly configured parts of my house. But I still plan to put the microinverters in a centrally located place, with maybe a little bit of attention paid to servicing.

Microinverters *should* fail independently and at random, so you could even accept some lost microinverters as the cost of doing business. (If there's a systemic error with one production batch, that may not be true. Good luck asking your distributor to give you microinverters from a randomized assortment of batches)

In your picture, you may be able to achieve what you want (easy access to microinverters) for large portions of the large array, with only some on the inside. For the small 4 panel arrays, you just need to install them the standard way for a 4-port microinverter, and they're a pretty good match for microinverters anyway (EG small arrays on a unique roof slope). So your question is moot.

Keep in mind the current at peak output of 72cell solar panels is pretty high. Assuming Vmpp is 40V, you're looking at 10A. What's the voltage drop at the lengths you are thinking of? The total length of wire in this string might be longer than a single DC string of the same current, which translates to more losses. I haven't done the math to see how important this feels.

Now, for the large array, you could instead do that as a DC string with RSD devices (and maybe optimizers if you feel spicy). They are on the same plane, so microinverters provide less gain. And it may anyway be easier to integrate with your storage.

I believe you are presuming that a solar contractor doing work on your house will pass the labor savings onto you. I dunno if that's a valid assumption. Especially if it becomes really confusing which microinverter corresponds to which panel.

When I assume low failure rate and deal with small failures as cost of doing business, it gives me lot of freedom to place micros deep beyond the edge to optimize the dc cables already built in. Most micros are about 10"x10" and they come with a few inches of pig tail. Using trigonometry, I have concluded that I won't likely need any DC extension cables for the panels, save one. And, I can optimize the AC trunks too but there the voltage is 6 times higher at 240V and so the losses are an order of magnitude less. On the DC side, assuming a 5 ft cable and 12 awg, my calculations tell me a voltage drop of about 0.25v and not meaningful.

I can easily keep a map to pin point where the micro is and I wasn't too concerned with saving contractor money and in many cases, I may chose to do it myself. In most trades, there are fewer people chasing break-fix. Businesses survive better by chasing new deals!!

 

[zanydroid]

OK. I think conventional sales pitch for microinverters is that they either last long enough, or provide redundancy.

With these multiport micros you do have a larger failure domain.

The 2- and esp 4-port micros are quite competitive with string inverters, and should handily beat solar edge, cost wise. What is the reason to go with micros instead of strings? Micros still increase the complexity of designing and operating storage.

 

[agarg]

With these multiport micros you do have a larger failure domain.

Good point. However, it starts to lean towards the touted advantages of string inverters (fewer things to fail).

The 2- and esp 4-port micros are quite competitive with string inverters, and should handily beat solar edge, cost wise. What is the reason to go with micros instead of strings? Micros still increase the complexity of designing and operating storage.

I may be wrong, but I think that 2x and 4x are essentially 2 or 4 MTTP with one string inverter. Less electronics so less costs.

I picked micros because, inherently, it offers resilience. It is not one go all go. Most people when they talk about DC voltage being risky don't realize that AC causes heart fibrillation in an instant and hence more risky IMO. Besides when people compare 240V AC to DC, they should compare 370V of DC to 240V of AC. The peak voltage of AC 240V is actually close to 370V. So I don't buy that DC on the roof is risky (that is just ingenious marketing). My reason for micro is resilience and in choosing 2x or 4x, I am actually doing hybrid string. Let me know if my facts are misplaced.

 

[zanydroid]

IIRC HoyMiles 4x is cost-competitive with smaller SMA with RSD factored in, but you get way more MPPTs with it. Some microinverters have 1:1 MPPT to input port. The HM1200/1500/etc 4 port series is 1:2 MPPT to input port, but IMO you are unlikely to have a problem since installing it in a code compliant way strongly forces all panels to be in the same aarray.

I think DC coming out of solar panels isn't terribly bad under 300V. However DC is subject to more stringent regulation by NEC, for instance rapid shutdown and putting the DC conductors in metallic conduit throughout the circuit (MC is OK once it gets into the roof). While AC can be done with regular NEC wiring methods. DC also has unique calculations for conductor sizing and when fusing is needed, though TBH if you don't understand that you can just outsource it on Greenlancer or whatever, and implement the plan they provide (I expect they're also using design tools that will add the fuses as appropriate). 250V MPPT also is a little small for larger arrays, I think it's limited to 4s2p with 72cell panels (~400W PTC) if you want to avoid fusing.

A difference between 4x micro of 1500W-2000W (Hoymiles supports that range) and an equivalent string inverter is in how the production is controlled. In both cases, the output is controlled by an inverter or charge controller. However, suppose you need new functionality, EG to charge a battery.

With microinverters, you can still AC couple into a battery, however when operating off-grid it requires the AC interface to the battery to communicate with the microinverter so that the microinverter power output is kept in balance with consumption + charging. If it is not, the microinverters will eventually trigger anti-islanding. The standard protocol for this (frequency/volt) has its limitations, the proprietary protocols are better but require you to stay within one brand, or two brands specifically partnering with each other.

With string inverter, you can replace the inverter with a solar charge controller or AIO going into a battery without going back on the roof, and there is a pretty large variety of hardware to choose from. The coordination is much easier in this case: you can't trigger anti-islanding because the battery is full (b/c it doesn't make sense, this is a function to protect against pushing into AC grid, and that is not involved). And the control schemes for determining if battery is full, etc. have been used for longer in the community with inter-brand interoperability (and somewhat lower consequence if they're implemented wrong).

All that said, I decided to go with microinverters for my DIY install later this year, since they require less expertise to get right for grid tie.

 

[agarg]

I think DC coming out of solar panels isn't terribly bad under 300V.

I started my engg career with HV DC and actually 240V AC is 370V at its peak. So conductors rated for 600V AC are actually good for 1000V DC. The DC does not have a reversing nature and so it become dangerous on making or breaking the circuit. Breakers rated for 600V AC will at best be OK for 200V DC, IMO. Way early in my career, as an apprentice, I have accidentally welded my screwdriver to breakers! Haha! We were fearless back then! And dumb as a rock!!

I have an approved permit for this:

My current plan

I have all the SMA and BYD equipment acquired and tested. I am now in the process of completing my BOM for PV+Micro side of the sketch. Everything after the combiner box is already in place and some parts are inspected and interim approved by the city.

I am heavily leaning towards the Hoymiles and will decide in the next day or so. Actually AP System is also pretty cost competitive with the Hoymiles. Check out their DS3. It has two discreet MPPT, two PV inputs, and at 880va and sub 250 dollars is very cost competitive with the HMS-2000 at 1900va. I just think that Hoymiles is innovating faster and is more DIY friendly.

The SMA ABU acts as an in between the load and the grid and in case of grid failure, creates a micro grid that enphase has now copied from SMA who did this several year ago.

I am amazed at the amount I am learning everyday from you guys and in sparring I discover my blind spots too.

I am now deciding to tuck the micros deep inside the array and hope that their 25 years of warranty would never get a chance to be cashed and will deal with failures if and when they arise. Hoymiles is troublesome in this respect as it will kill 4 panels if one were to fail (it is a string inverter with 4 MTTP but tucked under the panels).

 

[Mike 134]

If you tossed all the "innards" of microinverters on a table alongside the "innards" of an auto's electronics, you'd be hard pressed to say which is which both are just printed circuit boards with various electronics soldered on.

That said the auto's electronics can be subject to -40F if you live in Canada to 120 on a summer's day in Arizona. Much the same environment as your roof. But add one more for the auto's, they are subject to vibrations and shocks daily that your microinverters never will experience. I wouldn't worry about mounting them deep.

 

[agarg]

If you tossed all the "innards" of microinverters on a table alongside the "innards" of an auto's electronics, you'd be hard pressed to say which is which both are just printed circuit boards with various electronics soldered on.

This is a very compelling argument in favor of MLPE. Thank you.

 

[zanydroid]

HoyMiles has a pretty well documented Modbus // TCP modbus API that you get access to if you get their DTU-S device. It has features like monitoring, throttling microinverter output, … I believe there’s some open source wrapper libraries for this now and probably home assistant integration.

The HMS is sub-1G which may or may not be better than the 2.4G radio band HM micros.

I don’t know if AP systems uses wireless or wired PLC communications between their monitoring device and the microinverters. I’m not sure if there is enough forum experience with how flexible or reliable the advanced use cases are, it depends on both the raw capabilities and the reliability as you stack the whole system together.

 

[agarg]

Hoymiles is also now OEM to few others. Recently, Aptos Solar started selling relabeled Hoymiles in US. Not sure if the data units are same/compatible. I have a set of friends/family who are not too excited about Hoymiles. The HA integration is awesome and could be a game changer for gaining followers.

 

[zanydroid]

I haven't used the HA integration yet. Haven't wanted to mess with the DTU that I got with my turn-key installation since it's tied to their monitoring. When I get a new DTU for my self-installed system I'll give it a shot.

It's pretty standard actually to have HA integration with solar stuff that has an OK API, there's a pretty active community.

 

[svetz]

... Breakers rated for 600V AC will at best be OK for 200V DC...

No.
You can't use AC breakers for DC applications. With alternating current the voltage swings up and down crossing zero volts at the frequency. AC breakers only trip when the voltage is near zero when the current is exceeded to avoid arc quenching. With DC there is no voltage swing and the breaker must handle the arc. DC breakers are also often marked with a +/- and must be installed with the correct polarity.

 

[Bluedog225]

Please keep us posted. I’d love a diy non-enphase solution.

I have often wondered why there are not more diy rooftop inverters (non-micro). Seems a reasonable middle ground.

I don’t understand the resistance to 3M butt splices instead of problematic mc4 connectors. They are not that hard to implement or replace.

 

[Bluedog225]

@ncsolarelectric

 

[Bkord]

No.
You can't use AC breakers for DC applications. With alternating current the voltage swings up and down crossing zero volts at the frequency. AC breakers only trip when the voltage is near zero when the current is exceeded to avoid arc quenching. With DC there is no voltage swing and the breaker must handle the arc. DC breakers are also often marked with a +/- and must be installed with the correct polarity.

Some breakers are rated for ac and dc. I often use QO breakers for 48v dc telco equipment. The same breakers are in my house main panel for ac.

 

[svetz]

Some breakers are rated for ac and dc. I often use QO breakers for 48v dc telco equipment. The same breakers are in my house main panel for ac.

Arc is a function of voltage and medium, he was talking 200VDC. I don't doubt you can find breakers rated for either, but it would be dangerous to assume as the OP did:

... Breakers rated for 600V AC will at best be OK for 200V DC...

 

[kundip]

I have an array of 22 panels. A micro inverter other than enphase typically supports 2 or 4 panels. I am thinking of placing micro inverters on the edge of the array and use long PV cable to connect the micro to the panels deep inside the array. The reason, I am thinking like this is if I need to service any of the micro-inverters, I can just disconnect and pop in a new and power that back up again by restoring grid back to the array. The micro have many many electronic parts that may fail where as panels have just a few diodes. So, somehow I expect micro to fail more often than panel.

There is another part of my mind says using shorter cable is not only efficient, its also cleaner and so what if the micros end up being where hands wont reach. And, assuming the micro failure being rare, I can also get the panels lifted out and replaced back after the defective one has been replaced. This is more so as many people say that we never need to change out these micro and they are more likely to go obsolete than fail.

What are the thoughts of you learned folks on this forum.

Much appreciate.

Anil

PS: Illustration of the array attached.

You can put microinverters anywhere.
I have mine at ground level where I can easily replace, work on them or check them.

I have an array of 22 panels. A micro inverter other than enphase typically supports 2 or 4 panels. I am thinking of placing micro inverters on the edge of the array and use long PV cable to connect the micro to the panels deep inside the array. The reason, I am thinking like this is if I need to service any of the micro-inverters, I can just disconnect and pop in a new and power that back up again by restoring grid back to the array. The micro have many many electronic parts that may fail where as panels have just a few diodes. So, somehow I expect micro to fail more often than panel.

There is another part of my mind says using shorter cable is not only efficient, its also cleaner and so what if the micros end up being where hands wont reach. And, assuming the micro failure being rare, I can also get the panels lifted out and replaced back after the defective one has been replaced. This is more so as many people say that we never need to change out these micro and they are more likely to go obsolete than fail.

What are the thoughts of you learned folks on this forum.

Much appreciate.

Anil

PS: Illustration of the array attached.

I put all my microinverters at ground level. Easily accessible.
Originally I laid them out but now I just hang them up on a rod.
None get hot.