I'm considering adding a critical-loads subpanel near my main load center, in conjunction with a small grid-tie solar system, and with connection for backup power. Please critique my rationale for adding a subpanel, and help me decide on its ampacity.

Currently my main load center is congested, with only five "half spaces" (half of one panel "space", good for one 120 V circuit, or half a 240 V circuit, when using Siemens "space saver" breakers). Furthermore, this is before I connect the photovoltaics (PV) inverter and before I provide means for connecting backup electrical power.

I'd like to achieve:

  1. Connecting my grid-tie solar inverter. This could be using line-side taps (which connect to the service-entry cables between the POCO meter and the main breaker. Or by a dedicated double-pole breaker in the main panel.

  2. Providing a safe and convenient way to patch backup power sources into my house (an EV, a generator, or a battery system tied to the solar, when that becomes affordable). This could be by connecting a generator inlet jack to a double-pole breaker (in the main panel) which is interlocked with the main breaker. Or by installing a subpanel, which includes interlocked double-pole breakers, one connected to a generator inlet and the other connected to a breaker in the main panel.

Secondary goals, in order of importance, are:

  1. Alleviate congestion in my main load center.

  2. Provide hooks for upgrading my PV system to hybrid-mode (with battery backup) whenever that becomes affordable.

  3. Avoid line-side taps for connecting the PV inverter, in favor of a dedicated double-pole breaker in the main panel. The latter has these advantages: (a) can run feed from inverter in romex, instead of THWN in metal conduit, (b) needn't be without power in between finishing installation and passing inspection and re-installing meter, and (c) line-side taps seem sketchy, since they're connected to the very large service-entry cables and have no protection other than what's at the transformer.

What if I don't install a subpanel? To meet #2, I have to use a breaker for the generator inlet which is interlocked with the main breaker; since a space-saver breaker won't work with the interlock, the breaker will use two full spaces, leaving only one half-space free. Thus there's no room for a dedicated breaker for the PV inverter, so in order to meet #1 I must therefore use line-side taps. (Actually, on reflection, I must use line-side taps anyhow; because you can't have a breaker for a generator inlet plus a breaker for a PV inverter, both in the same panel. Otherwise, during grid outages the generator might trigger the inverter to start up, which the generator might not be able to handle if the inverter supplies more power than the panel loads absorb). So I've failed on #3 and #5.

If I do install a sub-panel, there's no reason not to use a dedicated solar breaker in the main panel (instead of line-side taps), because congestion will no longer be an issue (since I'll move a number of circuits to the subpanel). So I'll meet all five of my goals.

But what ampacity shall I make this subpanel? (In other words, the rating of the breaker in the main panel that feeds the subpanel, and the gauge of wire to connect it).

I want the following critical loads to be in the subpanel (with corresponding branch-circuit breaker size):

  1. Well pump (1/2 HP) - 15 amperes double pole.
  2. Range (KitchenAid KFDD500ESS dual-fuel) - 40 amperes double pole.
  3. Minisplit heatpump in computer room (Fujitsu 12RLS2, 5.5 amperes nameplate rating) - 20 amperes double-pole.
  4. Refrigerator (KitchenAid KRFC300ESS) - 15 amperes single pole.
  5. Living room lights - 15 amperes, single pole.
  6. Living room outlets - 20 amperes, single pole.
  7. Computer room outlets - 20 amperes single pole.

This adds up to 110 amperes double-pole. I don't expect my backup power source to handle all these loads: during a power failure, I wouldn't hope to run my main HVAC, my fancy vacuum-tube audio gear, all my computer gear (scanners, printers, etc.), or use the oven to bake bread. But I do hope to run the hyper-efficient minisplit in the study, watch TV, have the modem/router/phone work, and access the gas cooktop (requiring minimal power to the range). Therefore those circuits must be in the critical loads subpanel, and therefore it must support all those loads during normal operations.

We know that a breaker supplying a group of subsidiary breakers (for example, the main breaker in a load center) needn't have an ampacity equal to the sum of its subsidiary breakers; my main load center is 200 amperes, and the various breakers in it sum up to 400-500 amperes, but I've never had the main breaker trip because I had too much stuff on at once. So I'm pretty sure I don't need a 110 A breaker supplying my subpanel, with whatever size wire that requires. But what size should I use; how do I figure that out?

Here is some auxiliary information about the loads in question.

Here is wiring diagram for fridge:

Enter image description here

And here is the wiring diagram for the range. Note that even with the ungrounded conductor L2 disconnected, the control module still sees 120vac between L1 and N(eutral) and so the gas cooktop will be functional:

Enter image description here

Enter image description here


3 Answers 3


Welcome to the land of Article 220

The procedure you're describing for determining the appropriate amperage rating of subpanel and feeder is called a load calculation, and is performed using a set of rules found in Article 220 of the NEC. While most residential Article 220 calculations are performed for services, Article 220 calculation rules apply to feeders and branch circuits, as well; the rules are relatively complicated, but can be simplified for most residential work as follows:

  1. We allocate 3VA/ft2 for all area served by the panel, plus 1500VA for each kitchen or laundry small appliance circuit served by the panel. (By Code, this covers kitchen counter receptacle circuits and laundry receptacle circuits, but it's also possible to use this value as an allowance for a circuit feeding fixed 120V kitchen appliances, as well.)
  2. The above sum is then subject to a 35% demand factor for everything beyond the first 3000VA of load.
  3. We then add allowances for fixed power appliances: 8kVA for a range (this is sufficient for all ranges under 12kVA nameplate rating), 5kVA for a dryer (the Code minimum for such), and 4.5kVA for electric storage (tank) water heaters (with a 125% derate for feeders as a tank-type water heater's considered a continuous load). (Tankless heaters use their nameplate power rating instead)
  4. Finally, we address motor and HVAC loads. Table 430.248 in the NEC is used to convert motor horsepower into Full Load Amperes, while the Minimum Circuit Ampacity from the nameplate's used for HVAC loads. The largest load of these is subject to a 125% derate as well, and these amperage values are converted to VA (W) by multiplying by 230V. (Power factor gets ignored here.)
  5. The resulting sum then gets divided by 240V to determine the ampacity required by the feeder or service.

In your case, we have 700 ft2 of served area between the two rooms (study and living room) whose general loads are served by the panel, as well as a 1500VA allowance for the fridge, an 8kVA allowance for the range (which we deem as a shed load for standby purposes, as I'll discuss later), a 1/2HP@230VAC well pump, and a mini-split heat pump with a 5.5A MCA, also at 230VAC.

Given these loads, we start with 2100VA for the served area at 3VA@ft2 and the 1500VA allowance for the fridge, which gives us a 3600VA unfactored general load. Applying the 35% demand factor gives us 3210VA of factored load; we then add the 8kVA range allowance to give us 11,210VA, so far.

Last but not least, we tackle the motor loads. The well pump is counted as 4.9A@230VAC, while the 5.5A@230VAC heat pump is multiplied by 125%, giving us 1581VA for the heat pump, 1127VA for the well pump, and a total load of 13918VA, or 58A@240VAC.

As a result of this, we can use a 60A feeder and feeder breaker, with a 100A or 125A subpanel with a backfed main breaker that can be interlocked with for generator feed, and plenty of spaces (24-30) for future expansion. However, due to bend radius issues, it may not be possible to get a conduit to route directly between the two panel locations; if you can't get a suitably sized conduit to fit between the two panels, I'd just use 6/3 SER (not NM, as NM is limited to 60°C ampacities but SER can be run at 75°C ampacities these days) for the feeder and call it a day. (Making the feeder bigger doesn't make sense here due to the limitations imposed by the way multimode solar systems do transfer, anyway.)

Tackling that oven....

Now, we get to tackle the elephant in the room; namely, your shiny dual-fuel range and its power-hogging electric oven. We'll be covering two approaches to this problem; either way, we get to replace the 8kVA range allownace with a factorable 1500VA small appliance load. This gives us 6478VA of total standby load, or 27A@240VAC, well within the range of a large multimode system's inverter capacity.

Partial load shedding for fun and profit

The good news is that if you only connect L1, N, and G to your range, breaking L2, the oven elements won't work, but everything else should function normally, as per the posted wiring diagram and your test results. This means that we can use a suitably rated, UL listed contactor to break L2 when the standby system is transferred to generator power. (Note: definite purpose contactors can't be used here, because they're only UL component recognized for going into other pieces of equipment.)

This contactor will need to be controlled by the transfer means, though, and that's where the fun lies. With a manual transfer (interlock) setup, you'll need an auxiliary contact on the generator breaker in order to detect when that breaker is switched ON, and control the contactor accordingly. With a multimode solar system, though, the multimode inverter will need to provide an auxiliary output to turn the contactor on and off according to whether AC mains is available or not.

Physically speaking, the standby subpanel is probably the simplest place for the contactor to live; however, some contactors are too physically large to fit into a residential loadcenter, so you'll have to keep that in mind when contactor shopping. You'll also need to supply a control voltage; the good news is that most industrial and lighting contactors have interchangeable coils, so changing the control voltage down the road when you put in the multimode inverter isn't a complete hassle. This means that you can tap 120VAC from the standby supply to serve as the control voltage for now, and then convert it over to some other setup if need be when the inverters go in.

Ditching gas for an alternate solution

If, for some reason, the contactor solution doesn't work, there is an alternative if you're willing to ditch the idea of cooking on gas while the power's out. Modern portable induction cookers are far better than the dumpy electric hotplates of yesteryear; the best easily rival the performance of built-in induction cooktops, providing a degree of control and finesse unmatched even by gas burners! As a result, having a single standby small appliance branch circuit is a feasible, and rather flexible, option to permit cooking during a power outage, as it can power a variety of options: portable induction cooker, slow cooker or electric pressure cooker, toaster oven, or prep appliances such as a blender or food processor, for that matter. This is just like running any other branch circuit to serve the kitchen countertops, save for the fact that you'll want to have distinctly colored/marked receptacles (red is common for standby power circuits in commercial settings) to denote the fact that these receptacles are available when generator powered. (This is something you might want to do with your standby receptacles, anyway, even.)

P.S. Here is a screenshot of the above calculations in spreadsheet form:

enter image description here


I'm sorry, I must speak up at the idea of using precious auxiliary power to heat food, when a gas stove is at-hand.

Randy and the Eel discovered a wonderful vulnerability/feature to the range. The auxiliary circuits so important to running the gas burners only need one pole of power, whilst all the hungry electric coils need both poles.

I also am concerned the contactor will be daunting being as it is so unusual - and potentially quite expensive and unobtanium. Rather than interrupt it, I'd rather just not send power to it in the first place.

So I would use a separate 3-pole, double-throw switch to switch the 2 poles and neutral to the oven/range.

My first thought was build this up out of a 4-space QO and a QO2DTI interlock, but that won't switch neutral, which could create an undesirable neutral loop.

The upshot is that one side of the switch is fed from the utility-only panel, that has the normal range breaker. The other side of the switch (and only one leg) is fed from the smallest breaker possible, and of course is arranged to supply only the leg that provides auxiliary power for the range.

I argue that we get to size the range based on observable, measured power draw, e.g. from a clamp ammeter around the wire, rather than having to provision 1500W to a gas oven ignitor. (or realistically a 25 watt oven light, since you can't avoid incandescent here). I think this thing would count for 40 watts or so.

  • I only provisioned 1500VA for the SABC so that the calculations worked out the same for both approaches I mentioned; I agree that you could lump the igniter and fridge together and treat it as a single SABC for load calc purposes. I seriously doubt the contactor will be that unobtainium, although it's a bit hard to search for because all the definite purpose contactors get in the way :P Jun 9, 2020 at 18:24
  • I don't see how the contactor is difficult. I'm EE (although you might not know it here), have built DIY water-heater timer from contactor, and gave a link above for one that would work and cost less than $20 (if the UL rating is deemed satisfactory). But I totally agree that it'd be nice to use the existing gas. Jun 9, 2020 at 18:54
  • 1
    @RustyShackleford -- that particular power relay (not a contactor) you linked is not suitable as its UL component recognized (or Rather Useless to us). Something like this contactor would be suitable, though -- it's UL listed (not just component recognized), and adequately rated for the duty it will encounter in this application. (Note that NC power poles are much harder to find than NO ones, though.) Jun 9, 2020 at 19:31
  • Youch, that one's a good bit pricier. Good to know about the "component" rating (I guess that means my water-heater timer isn't legit). For it, I bought the DPDT one, so it's NC or NO depending on which terminals I use (I wanted NC, so I assert 2v from that little timer when I want it off, e.g. on-peak hours). Jun 9, 2020 at 21:24
  • 1
    @RustyShackleford -- yeah, either of the last two you linked will work since they're fully UL listed (to UL508) Jun 10, 2020 at 20:17

Providing an alternate answer to my own question ...

OP states wanting add a critical-loads subpanel, and solicits critique of rationale for doing so. From the facts given, adding panel capacity is clearly a good idea. Otherwise, as OP points out, addressing goals #1 and #2 leads to failing to meet #3 and #5.

But a subpanel is not the only way to add panel capacity. How about, instead, replacing the current main load center with a larger main load center, and not installing a subpanel at all ? The current load center is 20-space/40-circuit/200-ampere, and has 5 unoccupied half-spaces. It could be replaced by a 30-space/60-circuit panel (also 200 amperes).

Two full spaces would be dedicated to a breaker for the generator inlet, interlocked with the main breaker. This would leave 10-1/2 unoccupied spaces. Thus all except #5 of the OP's goals would be met. Furthermore, by leaving all loads in the main panel (and not requiring critical loads to be moved to a subpanel) there is far more flexibility in operating on backup power; breakers for loads can be switched on or off, depending upon household needs and on the amount of backup power available (a big fossil-fuel generator, or just a 1kW inverter connected to an EV).

However, note that this approach has one disadvantage. It is not possible to connect a generator breaker and a solar inverter breaker to the same load center (that is, to the same busbars), failing some sort of improbable interlock mechanism that requires both the main breaker as well as the solar breaker to be off, in order for the generator breaker to be turned on. Otherwise, when power is applied through the generator inlet, the solar inverter might think the grid is active, and potentially start generating power; thus the generator and the PV are shorted together, with lord knows what result.

It's not clear to me how difficult it is to replace the main panel. It is mounted in a wall cavity with no sheetrock on other covering. So it seems a simple matter to disconnect all the romex and clamps and pull the cables out of the panel (labeling them carefully, of course). Install the new load center (which is 6" taller) and reverse.

enter image description here

The main complication is the service-entry conductors. Requires getting POCO to remove the meter. And the new load center has to be "threaded" over those cables (if don't want to re-pull the cables) but this doesn't sound too hard. Then have to finish the job, get inspection, and have meter re-installed; so house will be without power for a couple of days probably. Since the new load center would be 6" taller, if the bottom is aligned with the old one, the service-entry cables probably need to be 12" longer to route properly. Simpler may be to mount it 6" lower (align the top of the new one with the old one). But then the 2" PVC conduit coming in from the bottom need to be cut, which probably means removing and re-installing the cables. And since it comes at an angle, that may be a bit awkward, though the framing allows moving the load center up to 1-1/2" to the left, so that can probably be finessed.

One potential gotcha is AFCI breakers; it appears to me as though a circuit need not be upgraded to AFCI if less than 6ft of wiring, and no additional outlets, are added to it. Apparently this applies even to a panel change, or addition of a subpanel.

There's probably some stuff I haven't thought about though; please enlighten me !

  • How far from the floor is the existing main breaker's handle, and how far from the floor is the bottom of the current breaker box, for that matter? Jun 15, 2020 at 23:33
  • Bottom of current load center (enclosure) is 45" above floor, mian breaker handle is 63" above floor - overall it's 30" high. New one would probably be 36" high. Jun 16, 2020 at 0:45
  • However ... looking at the load-calc for subpanel (above), and thinking about all the crap I've added since house was built (EV charger, replaced gas water-heater with two electric ones, added a shipping container workshop with small subpanel, added study with its own minisplit heatpump) has made me think I should do a new load-calc for entire house. So far it doesn't look like I'll come in below 200 amps, but I shall post a new Q&A about that later. If I end up going to a 225 amp panel (and I realize there's lots of other issues there), it'll probably be 42" high. Jun 16, 2020 at 0:50
  • I'd go to a 225A panel with a 200A main breaker anyway just for busbar ampacity reasons with the solar system (225A busses with a 200A main breaker gives you enough room to to feed 60A+ back into main panel, freeing you from that constraint). Jun 16, 2020 at 1:45
  • Wasn't really constraining solar (was gonna use 20amp breaker, 16amp max output of nverter), but capacity down the road couldn't hurt. Jun 16, 2020 at 4:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.