Converting grid tied to DC

[TorC]

OK, I think my system is relatively close to what you foresee ending up with, so I'll give a brief description of layout and let you ponder it with the rest:

8 strings of 3 panels going to four Outback FM80s in a 3S2P configuration. The FM80s put out ~48VDC to a DC load center for a LFP battery bank. Also connected to this DC load center are Outback Radian inverters which draw from the battery bank to power my loads. A generator or grid can be used as backup power if the batteries run flat. The power company should have nothing to say or do about this thus far.

---

As a separate feature, it sounds like you have your system selling power to the utility under unfavourable, and therefore unprofitable terms. The Outback Radian inverters are capable of communicating with the Outback SCCs to know how much power can be collected from the panels and sell it to the grid when your batteries are full. These inverters (or another hybrid inverter) can therefore replace your micro-inverters to sell power back to the grid. I don't know if changing the setup in this way will require approval from the power company.

Good luck, and come back when you have had a chance to rest and come up with more questions.

 

[zanydroid]

Are you planning to do DC to code? EG comply with rapid shutdown, etc.

That is more complicated than microinverters, which are exempt / get these capabilities built in, along with eliminating the need to understand how to plan out the config of the strings.

To properly do this incrementally and to code you will need to understand what wiring methods are allowed for different parts of the array. For instance, your microinverter strings do not need wiring contained in metal conduit / MC (wiring methods for AC are much relaxed vs DC). Although some installers do that anyway.

Most likely you will need a new DC circuit in conduit or MC to each array, leading down to the string inverter. And then within the array use PV cable and RSD devices for each module to be converted.

If you want to bench test microinverters, you should buy a solar panel to plug into it, install a 240V receptacle to plug into the microinverter, and use an energy monitor like Emporia or IotaWatt to capture the data.

 

[zanydroid]

I think your model for how long microinverters will last is off. I’m pretty sure Enphase is not this bad, otherwise the forums would be up in arms given their market penetration in the US. The most recent university study I saw mailed around, claimed 10-15 years lifespan for microinverters, using data from installations sampled by that institute.

Your electricity from the utility is cheap and you get 75% of retail. That is a great subsidy. It will be hard to break even just on the levelized cost for server rack LFP. Here I can at most make $0.10/KWh from shifting my usage / self consuming, and that is not enough to justify it for the trouble. My grid is pretty solid, if it was flaky then that would tip the scale.

 

[diWhy]

@TorC, thanks for all the information. I'll spend some time digesting that as soon as I can.

@zanydroid,
>Are you planning to do DC to code? EG comply with rapid shutdown, etc.

Well, that's partly why I'd like to do something where it's not grid tied, so I don't have to think about or deal with anything more to do with code.
I was thinking that my DC system would not be grid tied at all, it would be something separate like my generator is. When I run my generator, a transfer switch instantly disconnects from mains so I end up with a fully isolated AC source.

The conduit is already in place and I'd like to use that if possible.
I am thinking I would use weatherprrof combiners on the canopy to get the strings wired up then use the conduit piping to run the wiring back to the house with the AC cables. I have a junction box mounted on the house on one corner and the rest of the panels/meter etc are on the opposite end.

My thinking was to use that first junction box as my entry point into the house for the DC so everything would be nice and clean.

In terms of my inverters, I'm not making that up, many of them have died and the rest were disabled by the installer. I'm not sure if you read the entire thread but I have Pantheon II inverters that started dying only three years into owning the system. Then they became very hard to find and installers started converting people over to Enphase which sound much more reliable.

In my investigation, meaning, having to hack into the manager, I found that all of the inverters were output disabled so more than likely, most of the inverters are in fact working but are software locked to not output anymore.

 

[diWhy]

BTW, I also posted many times hoping to find someone who knew the manager software and more specifically the Pantheon inverters. I think their problem wasn't hardware but software related. I hoped to find someone that could help me but even reaching out to old employees, no one would talk about it.

 

[zanydroid]

Well, that's partly why I'd like to do something where it's not grid tied, so I don't have to think about or deal with anything more to do with code.

Your AHJ may disagree with this. Unless your state has an exclusion for non-grid-tied electrical work. It's permanent wiring on your house. The main difference is that POCO will not let you connect until they see your evidence for passing final inspection, and then they also impose some extra hoops on top of the process.

The generator is covered by a different section of NEC and building code than PV systems. There's like 1-2 chapters of excruciating detail on PV systems.

DC systems have much more complexity with which to hang yourself, safety wise and blowing up your equipment wise. Microinverters are significantly simpler in that regard. Personally I'm OK with smaller, lower current/voltage DC arrays, or arbitrary size microinverter arrays, in terms of doing something safe and reliable with a reasonable amount of education and no apprenticeship-style supervision (IE 20 hours).

The RSD protection required on DC strings is ostensibly to protect first responders, so in theory they would not be happy with a non-code-compliant installation. In my city one of the AHJ is the fire marshal, and they check all the plans for compliance to their access standards, along with the correct positioning / signage around shutdown devices. Fortunately since you are not moving/adding panels, your roof access should be compliant / grandfathered in.

IMO if you are not an electrician / solar installer the AHJ's extra hoops (code check, inspection, etc.) is the only check against your own skill or lack thereof. That is how I've done most of my recent electrical / solar projects. Get plans drawn up (either by myself or outsourced), go back and forth until approved, etc. If you try to get a forum to check for you, you're going to get a couple dozen cooks in the kitchen like is happening in this thread. It is somewhat harder in your case b/c it is an odd retrofit, which to me means you cannot as readily outsource doing the plans to a service.

In terms of my inverters, I'm not making that up, many of them have died and the rest were disabled by the installer.

Oh I believe that. So that is solid evidence you should NOT use that brand of microinverter. But I don't think you can infer from this that DC is the right answer.

Enphase is pretty open with their documentation, DIY friendly with their training and providing documentation, and has US-based support and corporate so they're here for the long haul. However they also cost quite a bit. I decided to go with a cheaper microinverter, but that comes at a risk of long term support. For 50% cheaper I'm sort of OK with that. If it was only 10% cheaper, maybe not.

 

[Hedges]

Was RSD the law at the time your microinverters were installed?
I would think inverter replacement might not require RSD installation, if not required back then. (of course, microinverter complies as well. But they existed before RSD, or at least before module-level RSD was required. That law probably came about as a result of microinverter companies trying to legislate away competition.) Maybe you would need an array disconnect switch, nothing more. Depends on date of permit.

As far as PG&E is concerned, you can replace older inverters with other that comply with requirements in effect as of original permission to connect. So if UL-1741 back then, you can install another UL-1741, don't need UL-1741-SA.

I think you would be allowed by code to replace existing equipment with equipment meeting same requirements, such as ground-fault, arc-fault, rapid-shutdown, that were in effect at the time.

Oh I believe that. So that is solid evidence you should NOT use that brand of microinverter. But I don't think you can infer from this that DC is the right answer.

Quality string inverters are ... quality.

 

[zanydroid]

I think the difficulty here is that swapping microinverters to string inverters and all the rewiring that entails may not feel like a repair, which is my rule of thumb for continuing on grandfathered code.

And note there are situations where repairs are NOT entitled to carry forward grandfathered code (EG exposing knob and tube, lead, or asbestos during a remodel). Probably doesn't apply to solar unless there is some ancient code that admitted comparably dangerous situations.

In California (my town and San Francisco) I think a lot of places provide a list of construction activities that are explicitly allowed to be done without a permit. Which implies the rest are not (de jure at least)

Quality string inverters are ... quality.

What do you mean?

 

[Think About It]

Insurance companies many not cover a claim for for fire or liability if a permit was not pulled and inspections were not performed.

Rapid Shutdown was introduced in 2014 NEC. I believe he said the system was installed 3 years ago.

If he's going to use batteries DC to DC would be more efficient.

My advice to the OP is to hire an electrician that is well versed in solar.

 

[zanydroid]

Rapid Shutdown was introduced in 2014 NEC. I believe he said the system was installed 3 years ago.

That's array-level RSD. Module level RSD is NEC2017 but I believe it had some incremental time-based triggering of when different clauses came into effect.

 

[Think About It]

That's array-level RSD. Module level RSD is NEC2017 but I believe it had some incremental time-based triggering of when different clauses came into effect.

Right, and each state adopts the latest version in their own sweet time, sometimes many years later.

If I were contracted to do his work I would pull a permit and ask my local electrical inspector.

 

[Hedges]

I think the difficulty here is that swapping microinverters to string inverters and all the rewiring that entails may not feel like a repair, which is my rule of thumb for continuing on grandfathered code.

If I unplug MC cable between PV panel & microinverter, daisy chain to next microinverter, slap a string inverter in place ... feels like an equipment replacement to me! So long as RSD wasn't in effect at the time of initial install. If it is, slap in RSD box in place of microinverter, but make sure MC4 are compatible. And select a string inverter that supports RSD.

My planned install on a new place, TriPower doesn't send keep-alive so I bought separate Tigo transmitter. It does discharge input capacitors. More complex, expensive setup; look for an inverter that does everything you need, including DC disconnect.

In California (my town and San Francisco) I think a lot of places provide a list of construction activities that are explicitly allowed to be done without a permit. Which implies the rest are not (de jure at least)

Mine (San Jose) says "replacement of equipment".

What do you mean?

I mean an SMA Sunny Boy or Fronius is likely to run without problems for 20 years. I had several, couple failures, but 34 year MTBF at the time I took them down for an upgrade.

In a few years, today's models are available used or new-old-stock. I've picked up spares for 10% of MSRP.
String inverter is very fast replacement, nothing like microinverter in the middle of an array.

 

[zanydroid]

If I unplug MC cable between PV panel & microinverter, daisy chain to next microinverter, slap a string inverter in place ... feels like an equipment replacement to me! So long as RSD wasn't in effect at the time of initial install. If it is, slap in RSD box in place of microinverter, but make sure MC4 are compatible. And select a string inverter that supports RSD.

Hmm but now you have one or more high voltage DC circuit in the house that didn't previously exist. There's a good chance you would need to add a new branch circuit for the string inverter.

Where I am on the peninsula you need to submit drawings showing location of outlets and panel for new AC circuits, even ones I would think are trivial...

You need a new disconnect switch and DC conduit / circuit labeling + other signage (since the labeling and signage is different from micro inverter)

String inverter is very fast replacement, nothing like microinverter in the middle of an array.

Are module-level RSD supposed to fail in a manner that tries to maintain functionality? If they fail open and interrupt the string, that is pain and you have to climb up there and do the same work as a microinverter replacement. If they just bypass the panel on module failure, then the availability should be about the same as before module level RSD. I'm actually very curious about this.

Since microinverters are paralleled into the trunk, as long as they fail stop, instead of doing something like a fail close into a short, the system will continue to operate. As long as the failures across microinverters in my array are de-correlated, I might even just leave it not working for a while.

Fast replacement does not address my concerns about the wiring standards and wiring topology being more complex than microinverters

 

[Hedges]

Maybe string inverter could mount right on the roof, connected directly to MC cables, no conduit. But shade preferred, for reliability.

String inverter would tie into AC disconnect same as a bunch of microinverters. Same wattage too, except if the micros were clipping and you sized string not to.

When micros fail, they just put zero power onto the AC wire. They don't short it out. So the rest continue to operate.

Module-level RSD might have a bypass diode so if the rest fails, string keeps working. Don't know for sure. But you would lose that panel, unlike before RSD.

Problem is if they fail with fire. Many fires due to incompatible "MC4" connectors. Now, only single brand/model, or mated pair that was tested and UL listed.

 

[zanydroid]

When micros fail, they just put zero power onto the AC wire. They don't short it out. So the rest continue to operate.

Certainly I've seen situations where the micros weren't connected properly to the trunk or panel, and the rest of the trunk was fine. If you do that with optimizers in the middle of a string, it just wouldn't work. But then the shitty installer wouldn't be able to leave until they found the bad connection, while a shitty installer that messes up some micros can get away with it.

This is my standard mental model, but I realized while writing my post that I don't have info to back up this up with mechanism of action. Not an EE, but quick Google says a lot of semiconductor devices fail-short. And this random block diagram I found shows that there is no isolation. If the right things in the DC-DC or PWM inverter fail-short, and the disconnect semiconductor switch next to the coupling inductor fail short too, you could have a single module failure take out the whole branch.

One would hope that semiconductor switch is engineered to add an extra layer of safety against this.