Passive House Electrical System Design (BETA)

I’ve talked through some of the preliminary research and outputs for the electrical system in our passive house, but I haven’t mentioned the requirements that drove the decision making or explained the proposed design in detail. We’re going to fix that.

Some of that is because the list/discussion bits weren’t in a state that I was ready to share. Some of it was for editorial reasons :). The design below is what I have so far. We’re still far enough out from actually needing to commit to anything on there that this is still a “living” design. The most fluid aspect are the photo-voltaic (PV) solar panels, but everything marked with a * is still unset. The PV is also likely to be the last thing we need to nail down, and potentially an area where choosing later is better. Cool things are happening in the PV space.

I completely understand that this configuration is well beyond the complexity and needs of most people. Doing something like this is not required to build a passive house. It’s complicated because our resilience requirements are complicated. You don’t need solar panels (although, why would you not?) or batteries (same question) to build an energy efficient home or a passive house. But if you are building a house, it makes sense to include both aspects in the design. Even if you just want to add it later.

There are many paths here as well. It isn’t necessary to go the way we are, but much of the product selection is driven by value-for-money. Which is why I started the discussion over the last couple weeks there (i.e. the editorial reasons). So even though it’s atypical to do what we’re doing – we should get more capability for less money than if we opted for a more traditional approach using a vendor like Enphase, or Tesla.

The Solution (so far):

  • 2x Sol-Ark 15K hybrid inverters
  • 4x EG4 PowerPro 14.3kWh LFP batteries
  • 2x 200A Siemens service panels*
  • 2x Sol-Ark SmartLoads 14*
  • 64x REC 470W Alpha solar panels*
  • 64x DC Optimizers, providing rapid shutdown (RSD)
  • 2x AC combiners
  • 1x AC bypass

High Level Requirements

  • ~60kWh of battery
  • The ability to easily and cheaply add battery capacity, preferably independent of inverter capacity.
  • Reduced reliance on a single OEM for battery capacity.
  • ~30kW of PV
  • Enough inverter capacity to cover “everything” during a grid down event (see load shedding).
  • Enough inverter capacity to charge the EVs from the battery if the batteries have sufficient SOC to facilitate this goal.
  • The ability to easily and cheaply add inverter capacity, preferably independent of battery capacity.
  • Grid pass-through if loads exceed inverter or battery capacity.
  • System level monitoring
  • Circuit level monitoring
  • PV monitoring
  • Load shedding
  • Off-grid resilience – I don’t expect that we will take the system intentionally off-grid, but I want to build a system designed to be used that way.
  • Bypass the inverter / batteries if necessary. This is mostly for safety and resiliency; things fail. Keeping the house online if there is a system failure is critical. Worst case we run everything on-grid.
  • Reduced reliance on a single OEM (e.g. Savant and Leviton Smart panels use proprietary breakers).
  • GEN input.
  • External access to GEN input.
  • 80A circuit in the garage for two power sharing EVSE and one NEMA 14-50R. This isn’t shown on the layout above because it is mostly informational.


  • Why so much energy storage, isn’t that expensive? – The goal is to use this as much as possible to reduce grid consumption when the sun isn’t shining and provide backup when the grid is offline. Decoupling the inverter and batteries and opting to use the approach we are, makes this very cost effective. I discussed the numbers before, so we’re not going to rehash it. 20kWh is a “large” number to see in many on-grid setups, most of the time it’s much smaller. Quite possibly driven by cost. 20kWh is large enough that you can do some use-shifting, but not a lot if you also want to mitigate against grid events. We want to do both. With 60kWh we have 40kWh to play with and 20kWh of reserve to mitigate grid downtime. That wouldn’t make sense if we were paying Tesla/Enphase money for the inverters and batteries. It’s also important to remember that if you’re doing this in the US, you get a 30% discount on batteries and inverters.
  • Why so much PV? – 30Kw is a guess. We have two EV, I have servers and a lot of POE networking kit. Everything in the house will be electric: HVAC, cooking, water heater, laundry, etc. There won’t be a natural gas hookup. I would like to get as close to 110% as we can. This is no longer the limit for net metering (NM) where the house will be located, but it’s still a reasonable working target. I’m not sure what happens to NM credits if you don’t use them, they might just go away. I don’t expect that to happen, worst case we invite some people over to charge their cars :P. I don’t know if 30kW is even achievable given the size of the roof (it’s probably not). As we get closer in, some load calculations will need to happen. I am using this as the working target for now. I need a number to run the numbers. I like numbers :).
  • Why max the PV now? – That’s a great question. It is likely that PV will keep improving, and I will be able to get better panels in 5-10 years than I can get today. The main reason we’re doing it in the build is incentives. In the US there is a 30% tax credit for everything associated with PV and batteries. In Illinois we have the Solar REC program which is roughly another 30%. Shaving 60% off the cost of the panels and 30% off the batteries, inverters, etc. makes it easier to do it all now. I’m not confident that these incentives will be here in 5-10 years.
  • Why 470W panels? – I was modeling PV in the first few revisions around the LG 435W NeonR, mostly because it’s what the PV installer is using right now. But then I started looking into higher output panels, and flipped to the REC ones because 30kW of PV equals 70x 435W or 64x 470W. I hope that there are even higher output residential panels available when we need to actually buy them. I would love to get a quality 500W+ panel on the roof. That will make it easier to hit our numbers.
    • I don’t have solid pricing data for the LG or REC panels, so I am guessing that fewer, higher potential, panels are likely to be a better option from a cost perspective. I can be confident that they will be when considering extensibility.
    • There are also likely to be packaging and labor cost advantages to this approach. For e.g., 70x LG 435W NeonR would require 139.32 sq ft compared to 135.05 sq ft for 64x REC 470 Alpha. Which is the difference of roughly two additional REC 470W panels. Lowering panel count also lowers RSD count.
  • Why use RSD? – AFAIK, we will need to use some kind of per-panel RSD device in the US to be roof-legal. 64 < 70 there too. Using micro-inverters is a popular way to provide RSD in the US. The PV installer likes Enphase, so I ran some numbers using IQ8 micros.
    • There are a lot of different IQ8 versions that range in price from $189-$242. So between $12,096 and $15,488 for the inverters vs $4,608 for Sol-Ark DC Optimizers ($72/per) or $3,008 for Tigo DC Optimizers ($47/per). That’s a huge saving. I can buy a lot of panels or electricity for the $7-$12k I’m not handing to Enphase.
    • There are valid reasons to opt for micros. We have them on our current house, but the data indicates that they are unlikely to add any real value given the roof design of the new house. Maximizing PV was a design goal for lot selection and roof design. If I could skip RSD, I probably would run straight strings. The two Sol-Ark 15K will support six strings with the built-in MPPT, and is able to couple 39kW of DC PV. They also support 19.2kW per 15K of AC coupled, but that uses the generator (GEN) input. One of these needs to be open, and it would reduce the overall efficiency of the setup. So I don’t see that as viable at least until I max out the DC coupled PV.
  • What’s up with the external GEN connection? – I hope we never use this 🤞.  If there is an extended grid outage that the PV can’t manage, I want to be able to charge the batteries using a portable generator, or maybe even off one of the EV.
  • If modularity is the goal, why use a hybrid inverter, why not completely separate everything? – In theory separating everything is better from support, flexibility, and extensibility perspectives. But there are efficiency, switching speed, and cost advantages to using a hybrid inverter instead of separates. There’s room for a diversity of opinion here, but IMO the hard costs of completely separating everything isn’t worth the negligible benefits.
  • With that much inverter and battery, why add load shedding? – load shedding will be used to mitigate grid down events when the battery SOC is low or inverters are maxed.
    • Can be SPAN, although SPAN doesn’t currently support Sol-Ark hybrid inverters. SPAN claims mid-2024 support, so not off the table, but I’m not completely sold on SPAN for other reasons. I will probably write more about this later.
    • Can also be Sol-Ark SmartLoads-14, availability claimed EOY. Will need to see how this works when real people can play with it.
    • Maybe something else. What other options are there?
  • What are the options for PV, system, and circuit monitoring?
    • SolArk can provide both system PV and loads.
    • SPAN can do discrete and system loads.
    • Emporia can do AC PV, system, and discrete loads.
    • Maybe there are better options for this?
  • Sol-Ark 15K vs EG4 18K? – This was a really hard choice. There’s even some evidence that I was leaning towards the EG4 18K a couple months ago.
    • Sol-Ark has been doing grid legal hybrid inverters for a long time (~10 years). They have more brand recognition, which makes other OEMs more likely to play ball with them (e.g. SPAN). This is the main reason. I want to put the system in the best position to integrate with the other products we’re using. It seems like more vendors have Sol-Ark on the roadmap.
    • Functionally, these units are almost exactly the same. There is a difference at the company level: Sol-Ark is a US company, supported from the US, available from multiple resellers. There’s an official training/certification program and process around the installer network. EG4 is a great option, but it feels more targeted towards the DIY market than Sol-Ark. Having a larger, established, presence in the market makes me feel like if there is a problem I’m less likely to get yelled at by my wife :D.
    • Cost difference is ~$3,000 for the two hybrid inverters. If I wasn’t worried about integration options, that’s enough of a difference that I would probably choose the EG4 18K. It wins on cost, by a mile.
    • Fuzzy (or FUDdy), but it feels like a more defensible position if the inspector challenges the install because Sol-Ark is a known quantity, US company, etc.
  • Why two EVSE and a 14-50R on one 80A circuit? – I would like to have the hardwired EVSE to share the 80% of 80A automagically via power sharing. We don’t usually charge both cars at the same time, but it happens. The 14-50R is there so I can test EVSE. It will not be used under normal conditions. I don’t want to add a dedicated 50A circuit for that purpose. Current front runner is the Tesla Universal Wall Connector (UWC). It’s likely that there will be other EVSE in market with the same capability and flexibility that the UWC provides when the time comes to make that call. So it’s mostly a placeholder that demonstrates a modelable capability.

There’s a lot of information here. I tried to make it as digestible as possible. I’m not confident I succeeded :D. Please ask any questions below. Provide feedback. Tell me how I’m wrong. Etc.

Notify of
Inline Feedbacks
View all comments