EV load management: charging without a service upgrade

Load management is the “maybe you do not need the upgrade” answer

The parent page, EV charger installation, owns how a home charging circuit actually gets put in: the 240-volt run, the continuous-load multiplier, hardwired versus plug-in, permits. This page owns a narrower question that comes up the moment the load calculation fails: does the house need a larger service, or does it need the charging load managed so the existing service is never asked to carry everything at once?

Those are different jobs. A service upgrade replaces the service conductors, the meter equipment the utility will accept, and usually the panel. An energy management system (or, in the 2026 model code, a power control system) watches current and holds the EV charger — and sometimes other large loads — under a marked setpoint. The electrician still installs a charger. The utility may still have to be involved. What drops out, when the path is accepted, is the larger service.

The dollar gap that path is trying to close is not theoretical. Pacific Northwest National Laboratory’s December 2024 technical brief (PNNL-31576-1) attributes a New Buildings Institute 2022 figure of $1,500 to $4,000 for a 100-to-200-amp panel upgrade, on top of the charging circuits, and an International Council on Clean Transportation 2019 average of $1,400 to install Level 2 in an existing home across the 100 most populous metro areas. A 2015 NREL analysis cited in the same brief put the average high-cost Level 2 install at about $2,900 and named insufficient panel capacity for a 40-ampere circuit as a leading factor, “more prevalent in less-affluent areas.” Those figures are the attribution chain, dated 2015–2022, cited in 2024. They are not 2026 bids. The EV charger installation cost page is where cost intent lives; this page only needs them to show that the gap an EMS is trying to close is thousands of dollars, not tens.

HyreElectrical analysis: almost nobody leads with this option, because it sells less work. That is exactly why it belongs in print. A site that will tell you when the expensive recommendation is unnecessary is the only site whose expensive recommendation is worth believing. This is not a promise that every 100-amp house can take a 48-amp charger. It is a decision page for the cases in which the calculation failed on coincidence rather than on a panel that is already at its limit, already damaged, or already a known-hazard era.

What 625.42 actually does

Article 625 covers the equipment that connects an electric vehicle to premises wiring. Section 625.42 is the rating rule. In every recent edition it does two things: it treats EV charging as a continuous load, and it lets the counted load on a service or feeder be the maximum the management system will permit rather than the charger’s nameplate. The parent of this page already covers the continuous-load consequence for the branch circuit: a charger that draws 32 A is a 40 A class; a charger that draws 48 A is a 60 A class. This page is about the other half of the sentence — the permission to count a managed maximum instead of that nameplate.

The 2017 text, as quoted by ElectricalLicenseRenewal and independently reprinted in a municipal EV-charging guide from the Village of Tinley Park, already said: “Where an automatic load management system is used, the maximum equipment load on a service and feeder shall be the maximum load permitted by the automatic load management system.” That is the whole mechanism in one sentence. There was no 220.70, no Article 750 pointing, no listed-EMS language, and no adjustable-settings paragraph. The ALMS sentence is what a 2017-NEC jurisdiction actually has. Tennessee, on IAEI’s table checked 5 September 2026, is still on that edition statewide, effective 1 October 2018.

The 2020 edition kept that ALMS sentence, added “Service and feeder shall be sized in accordance with the product ratings,” and then added the first explicit pathway for a charger that is simply turned down. ElectricalLicenseRenewal’s 2020 quotation: “Adjustable settings shall be permitted on fixed-in-place equipment only.” If the adjustment changes the rating label, the manufacturer’s instructions govern, and the adjusted rating has to appear on the label with durability to match the environment. Restricted access to the adjusting means is required, accomplished by at least one of: a cover or door that needs a tool, locked doors for qualified personnel, or password-protected commissioning software for qualified personnel. The 2020 book then permits the EVSE’s ampere rating to equal the adjusted current setting, and permits sizing the service and feeder to match. Solar Tech Collective’s January 2026 625.42 history, which we retrieved the same day, calls this “the first clear pathway to using current values lower than the equipment’s rated maximum.” The ALMS sentence was already there. What 2020 added was the static, locked-down derate — a smaller charger on paper, without watching the rest of the house.

The 2023 edition split the idea into two listed paths, reprinted with NFPA permission in Leviton’s Captain Code 2023 and quoted side-by-side with the 2020 text by ElectricalLicenseRenewal. The parent sentence now reads: service and feeder shall be sized in accordance with the product ratings, “unless the overall rating of the installation can be limited through controls as permitted by 625.42(A) or (B).” 625.42(A) Energy Management System: where an EMS in accordance with 750.30 provides load management of EVSE, the maximum equipment load on a service and feeder is the maximum load permitted by the EMS. The EMS may be integral to one piece of equipment or to a listed system of more than one piece, and it must be marked when the control is integral. 625.42(B) EVSE with adjustable settings: EVSE with restricted access to an ampere adjusting means complying with 750.30(C) is permitted, and the adjusted current may be used as the rating. The “fixed-in-place equipment only” restriction was deleted: ElectricalLicenseRenewal’s 2023 commentary records that cord-connected EVSE with the same restricted-access rules can also be compliant.

Separately, 2023 added 220.70. The 2023 wording, as quoted by ElectricalLicenseRenewal’s 2026-code comparison (updated 15 January 2026), is: if an EMS is used to limit current to a feeder or service in accordance with 750.30, “a single value equal to the maximum ampere setpoint of the EMS shall be permitted to be used in load calculations for the feeder or service,” and that setpoint is treated as a continuous load. That is the whole-service version of the same idea — not just the EVSE line, the service. Eaton’s 2023 energy-management commentary, retrieved 5 September 2026, describes 220.70 as a new section recognising that impact.

The 2026 edition renamed the overload-control equipment a Power Control System, moved the calculation rule from 220.70 to 120.7, moved Energy Management Systems from Chapter 7 Article 750 to Chapter 1 as Article 130, and pointed 625.42(A) at Article 130 Part II. ElectricalLicenseRenewal quotes the 2026 120.7(B) text: the control setting must be determined by qualified persons and set to no greater than 80 percent of the overcurrent device rating on the monitored circuit. That 80-percent cap is a 2026 model-code provision. It is not a 2023 rule, and it is not in force in a 2020-NEC state. Solar Tech Collective’s 14 January 2026 analysis of 625.42 in the 2026 book puts the two options in the same place the 2023 book did: (A) a power control system, or (B) EVSE with adjustable settings. The names changed. The two-class split did not.

625.42(A) versus 625.42(B): dynamic cap versus locked-down derate

The 2023 split is the one to keep in your head, because the 2026 book kept it under new names. Two different designs are being permitted. They are not interchangeable, they do not produce the same number in the calculation, and they do not fail the same way.

(A) is the dynamic path. An EMS in accordance with 750.30 — or, in 2026, a Power Control System in accordance with Article 130 Part II — watches current and holds the EVSE, and sometimes other loads, under a marked maximum. Solar Tech Collective’s 2026 worked pair is the cleanest published illustration we retrieved: EVSE 1 with a PCS is a dynamic load, capable of drawing up to its full rating (they use 9,600 W) when other simultaneous loading allows it, supplied by a 50 A breaker; EVSE 2 is a static load based on an adjustable setting of 5,760 W, though its nameplate maximum is also 9,600 W, supplied by a 30 A breaker. The first design can still deliver peak charging speed on a quiet house. The second design never can. That is the point of (A). It is also why (A) needs hardware that actually measures something — typically current transformers on the service or feeder, or a meter the utility or the manufacturer has listed for the job — and why it needs a fail-safe if that measurement or the controller dies.

(B) is the static path. The EVSE itself has a restricted-access ampere adjustment. Nothing else in the house is monitored. The counted load is simply a smaller charger. 2023 625.42(B) points the access rules at 750.30(C); 2020 wrote those rules into 625.42 itself (tool, lock, or commissioning software). The 2026 book, per Solar Tech Collective, requires the setting to be made at installation, to appear on a field-installed rating label, and to be readjusted only by a qualified person. A homeowner who can bump the charger from 24 A to 48 A from a phone app has not installed (B). They have installed an unmanaged nameplate with a software suggestion.

ElectricalLicenseRenewal’s 2023 commentary also records a branch-circuit use of (A) that is easy to confuse with whole-service management. The 2023 625.40 individual-branch-circuit rule still requires EVSE greater than 16 A or 120 V to sit on its own circuit, with a new exception: a single branch circuit may supply more than one EVSE when the loads are managed by an EMS in accordance with 625.42(A) or when the EVSE has adjustable settings under 625.42(B). Their worked example is three 30 A chargers on a 100 A shared branch, with the EMS holding the group at 80 percent of that 100 A. That is a circuit-sharing design among chargers. It is not a whole-service EMS. EC&M’s 2023 “five ways” piece (retrieved in the same research pass) makes the same 625.40 exception point. A house with one charger does not need that exception. A house with two parking spots and one spare two-pole space might.

Which path a quote is actually proposing is a question, not a vibe. If the design watches the service and throttles the car when the dryer starts, it is (A). If the design is a locked 24 A setting on a 48 A-capable wall unit, it is (B). If the design is a device that lets the dryer and the charger share one 240-volt circuit, it is a third class, treated below. Mixing the three in a sentence is how a homeowner gets sold a $400 gadget as if it were 220.70.

220.70, then 120.7: the whole-service setpoint

625.42 lets the EVSE contribution be the managed maximum. 220.70, new in 2023, lets the feeder or service calculation itself use a single EMS ampere setpoint. Those are nested permissions. A 2020-NEC inspector can accept a managed EVSE line under 625.42 without having a code section that says “calculate the whole service at 80 A.” A 2023-NEC inspector has that section, provided the EMS is in accordance with 750.30. A 2026-NEC inspector has 120.7, which is broader: ElectricalLicenseRenewal’s 15 January 2026 commentary says relocating the rule from Article 220 to Article 120 “signals a broader application,” and the 2026 text explicitly permits PCS use for branch-circuit, feeder, or service load calculations.

The 2023 220.70 text, quoted in that same comparison, is two sentences. If an EMS limits current to a feeder or service in accordance with 750.30, a single value equal to the maximum ampere setpoint may be used in the feeder or service load calculations. That setpoint is considered a continuous load. Eaton’s 2023 commentary restates the same: load calculations are permitted to take advantage of an EMS when sizing feeders, service conductors, and equipment; the current value used must be considered continuous. Mike Holt’s 2023 Changes book, third printing, calls 220.70 new and describes it as allowing an EMS to limit the maximum current used by the electrical system — with a candid “I am sure there will be growing pains with this rule.” Growing pains are not a reason to hide the section. They are a reason to ask the AHJ whether it has an inspection protocol for it.

The 2026 120.7 text, quoted by ElectricalLicenseRenewal, is longer because it had to say what 220.70 left to 750.30. 120.7(A): the PCS shall comply with Article 130 Part II. 120.7(B): the control setting shall be determined by qualified persons and be set to no greater than 80 percent of the rating of the OCPD for the circuit being monitored. 120.7(C): the load on the branch circuit, feeder, or service is the sum of controlled loads and noncontrolled loads. If the PCS monitors only controlled loads, the control setting stands in for those loads. If the PCS monitors both controlled and noncontrolled loads, the minimum operating current of the controlled loads is used instead, and noncontrolled loads are calculated the ordinary Article 120 way. An informational note, quoted in the same comparison, says minimum operating current is “a value greater than or equal to zero representing the minimum current of the controlled loads.”

Brian Mehalic’s 12 January 2026 piece in pv magazine USA, writing as a Solar Tech Collective consultant, is the most careful published walk-through of that (C)(1) split we retrieved. His example: an EV charger that can only ramp down to 10 A has a 10 A minimum operating current; that is the number that goes into the calc for the controlled portion when the PCS also sees the rest of the house. He then flags a safety point that is easy to miss: in his worked feeder, a 37.5 A load-calc value does not let you downsize the feeder conductor below the 60 A OCPD that supplies it. The calculation and the conductor are different jobs. He also records two 2026 ambiguities the industry will have to live with: “minimum operating current” is not defined in the NEC beyond that informational note, so the number has to come from the manufacturer; and “qualified person” is defined, but what it means in practice for who sets a residential PCS control setting is an AHJ conversation. Those are limitations, not loopholes.

The 80-percent cap in 120.7(B) is new in 2026. ElectricalLicenseRenewal says it “aligns with existing design examples shown in NEC Annex D” and is a conservative design approach so the OCPD is not asked to carry a continuous load at its full rating. On a 100 A main, that is an 80 A control setting. On a 200 A main, 160 A. It is not a 2023 rule. Quoting it in a 2020-NEC permit package is quoting the wrong book.

750.30 is the machinery 625.42(A) points at

In the 2023 model code, 625.42(A) does not describe how an EMS has to behave. It points at 750.30. Article 750 is Energy Management Systems. Eaton’s 2023 commentary records the listing options at 750.6: listed as a complete energy management system; listed as a kit for field installation in switch or overcurrent-device enclosures; or listed individual components assembled as a system. A pile of current transformers, a Raspberry Pi, and a contractor’s confidence is none of those three.

750.30 itself, as summarised from NFPA first-draft material for the following cycle (which restates the 2023 structure) and from Schneider Electric’s 2023 NEC eGuide and Eaton’s commentary, has three jobs. 750.30(A) load-shedding controls: an EMS shall not override the load-shedding controls that guarantee minimum capacity for fire pumps, emergency systems, legally required standby systems, and critical operations power systems. 750.30(B) disconnection of power: an EMS shall not cause disconnection of power to listed life-safety loads — Eaton and Schneider both name fire pumps (Article 695), healthcare (517), emergency systems (700), legally required standby (701), and critical operations power systems (708); the first-draft restatement also names elevators, escalators, moving walks, stairway lift chairs, and ventilation that exhausts hazardous gas or reclassifies an area. On a dwelling these lists are mostly empty. They are not empty on a mixed-use building, a townhouse with an elevator, or a house that already has a legally required standby generator. An EMS that sheds the wrong thing is not a clever EV solution.

750.30(C) is the overload-control block 625.42(A) actually needs. Schneider’s 2023 eGuide lists the requirements in one place: the EMS must have a maximum current setpoint; monitoring and controls are required; system malfunction automatically ceases current flow; settings must be restricted to qualified personnel; and the power equipment that supplies the circuit must be field marked. Electrical Contractor Magazine’s December 2023 “Keeping Cool” piece quotes the 750.30(C)(4) marking list in full: maximum current setting; date of calculation and setting; identification of loads and sources associated with the current-limiting feature; and the following or equivalent wording: “The setting for the EMS current limiting feature shall not be bypassed.” The markings have to meet 110.21(B) and be located so they are clearly visible to qualified persons before examination, adjustment, servicing, or maintenance. Eaton’s 2023 commentary records the same four-item list. If nobody can say what will be on that label, the managed-setpoint path has not actually been designed.

The malfunction rule is the residential version of fail-safe. Schneider: “System malfunction automatically ceases current flow.” A 2023 EMS that fails to full nameplate has failed the section it is relying on. A 2026 PCS, per Mehalic, must “transition to a controlled state that prevents overload.” PG&E’s ChargeBoost FAQ, retrieved 5 September 2026, describes a utility-meter version of the same idea: if the home Wi-Fi that ties the meter to the charger disconnects, “your charger will default to Level 1 until it is reconnected.” That is one utility’s product behaviour, not a national fail-safe rule. It is evidence that at least one large programme treats loss of communication as a throttle-down, not a throttle-up.

In 2026 this machinery moves. Solar Tech Collective: Energy Management Systems moved from Chapter 7 to Chapter 1 as Article 130; throughout the NEC, EMS was replaced with PCS “in instances where equipment used for overload control is permitted”; the common requirements sit in Article 130 Part II. UL Solutions’ UL 3141 service page, retrieved the same day, is the listing that change is written around. A PCS “monitors the output of power sources and regulates or limit current or power within predefined limits.” Unlike an EMS, “which focuses on optimizing energy usage and cutting costs, the PCS addresses the essential aspects of load control and safety.” EV charging management is named as a load-control application. Panel or utility-upgrade avoidance is named as a consumer benefit. UL 3141 is an Outline of Investigation, not a full standard; Solar Tech Collective dates its publication to 2024 and notes that PCS products were previously certified via a UL 1741 Certification Requirement Decision beginning in 2019. A 2020-NEC inspector may still be looking at “automatic load management system” language and a UL 916 Energy Management Equipment listing. That is an AHJ question, not a shopping question.

What each class actually does on a house

Whole-service monitoring is the class the code sections above are written around, and it is the class that can replace a service upgrade when the constraint is amps rather than slots. Current is measured on the service or feeder. When the sum of house load plus charger approaches the marked setpoint, the charger is told to drop. The control-pilot protocol that Level 2 EVSE already uses to talk to the car — IEC 61851-1, Annex A — is how that drop is communicated: the EVSE advertises available current by pulse-width modulation on the control-pilot line. IEC 61851-1:2010 Table A.6, which we retrieved from a published extract, maps an 8-to-10 percent duty cycle to 6 A, and duty cycles from 10 to 85 percent to available current equal to the duty cycle times 0.6 A. Six amperes is therefore a protocol floor for PWM current advertisement, not a HyreElectrical invention and not a DIY setting. A system that “pauses” charging entirely is a product behaviour on top of that protocol. A system that can only drop to 6 A still occupies 6 A of the service while it waits. 2026’s “minimum operating current” is the code’s name for that floor, and Mehalic’s point stands: you get the number from the manufacturer.

Landing current transformers on service conductors, opening a panel to place them, and commissioning the controller are licensed electrical work. This page does not describe how to place a CT, which conductor to land it on, or how to set a DIP switch. A whole-service design that is actually going to be used in a 220.70 or 120.7 calculation also has to be listed for overload control, field marked, and fail-safe. A current-transformer kit sold as an energy monitor for a phone app is not, by itself, 625.42(A).

Circuit sharing is a different constraint. LBNL estimated that 22 percent of all panels, including 20 percent of 200 A panels, have zero breaker spaces remaining. Two empty adjacent slots for a two-pole breaker is a circuit. Zero slots is not. A listed sharing device on an existing 240-volt dryer circuit does not create a slot in the panel; it reuses one. It also does not watch the range, the HVAC, or the water heater. If the service calculation already fails before the charger is added, sharing the dryer circuit does not repair the service calculation. If the calculation fails because there is no place to land a new two-pole breaker, sharing might. PG&E’s rebate page describes the class in those terms: it “works with existing 240-volt outlets in your home” and “allows two appliances to be plugged into one outlet without overloading the circuit.” That is a utility naming the class as an upgrade-avoidance path. It is not a national listing, and it is not a reason to share a circuit with an unlisted adapter, a dryer-outlet splitter from a hardware aisle, or anything that is not the listed device the electrician is installing.

Charger-native derating is the cheapest of the three in hardware and the most honest about what you give up. A 48 A-capable wall unit locked to 24 A is a 30 A-class circuit and a 5.76 kW charge rate. The parent page’s 125-percent table still applies: 24 A times 1.25 is 30 A. Overnight, for a driver whose daily miles fit in that window, that is often enough. It is not enough for a driver who needs a full pack between a 6 p.m. return and a 7 a.m. departure, and it does nothing for a house whose other loads already overshoot the main. 2023 220.57, new in that edition, also puts a floor under the EVSE line in a standard calculation: 7,200 volt-amperes or the nameplate, whichever is larger. 7,200 VA is 30 A at 240 V, so a 16 A or 24 A unit still counts as 7,200 VA in a 2023-edition standard calculation unless a 625.42 path is actually being used. Jurisdictions on the 2020 or 2017 NEC do not have that floor. Optional calculations under 220.82 are a different path, and some inspectors want the EVSE held at 100 percent rather than riding the 40 percent remainder. That disagreement is why a site visit beats a web form, and why a locked-down (B) setting has to be the number the electrician puts in the calc, not a number the homeowner picked in an app.

PG&E’s Residential EV Charging rebate eligible-equipment list, retrieved 5 September 2026, is the cleanest published grouping of the three classes we found from a named utility. One column is chargers with adjustable amperage “to work with existing service panel capacity.” One column is a circuit-sharing device on existing 240-volt outlets. One column is “EV Energy Management System (EMS) that automatically balances the load of an EV charger and home energy usage to avoid a panel upgrade.” The utility’s own note on that page: “PG&E does not offer recommendations for any product on this eligibility list.” Neither does this page. The list is evidence that a large investor-owned utility treats the three classes as real upgrade-avoidance paths, and that it will rebate them. It is not a shopping guide, and it is not in force outside PG&E territory.

What the mechanism looks like when it is measured

A national laboratory ran the mechanism rather than asserting it. NREL/TP-5500-95492 (August 2025), a CRADA final report with Redwood Energy and NeoCharge, tested a whole-home energy management system in NREL’s Systems Performance Laboratory. To represent older housing, “the electric panel was effectively downsized to 70 A in software (the actual panel limit was 150A).”

The system “successfully throttled the Wallbox EV charger load from 28 A to 6 A,” “reducing the peak energy load from 64 A to 42 A (35% reduction).” Load-management actions started at 49 A — 70 percent of the simulated capacity — to hold the panel under 56 A, 80 percent. When a dryer came on at 6:39 p.m. and again at 6:56 p.m. while the car was at 28 A, the charger dropped to 16 A; after the dryer finished, it returned to 28 A “after approximately one minute.” A Cleveland-winter run throttled the charger from 21 A to 6 A when the panel approached 80 percent. Detection-to-reduction latency was 15–23 seconds.

The report’s own limit on what that proves is the one to keep. The vehicle on the test could accept only 28 A; the authors note that a typical Level 2 charger draws up to 40 A (9.6 kW). The work is a collaborative report on one vendor’s system. It validates the mechanism under those test conditions. It is not a finding that every device in the three classes above will produce a 35 percent peak cut. The 6 A floor in that test is also the IEC 61851-1 PWM floor, not a coincidence. A house whose coincident evening load already sits above the setpoint minus 6 A will see the car trickle, or pause, every evening the dryer and the range run. That is the design working. It is also the design announcing its limit.

Amps and slots fail independently — LBNL’s numbers, with the paper’s own hedges

Lawrence Berkeley National Laboratory and the National Renewable Energy Laboratory estimated, in a February 2026 Journal of Building Engineering paper (Murphy et al., vol. 120, 115576), that 58 percent of U.S. single-family homes have panels of 200 A or more (55 percent at 200 A, 3 percent above) and that 100 A panels are 29 percent, with another 2 percent below 100 A — 31 percent at 100 A or smaller. They also estimated that 22 percent of all panels, and 20 percent of 200 A panels, have zero breaker spaces remaining. Headroom in amps and headroom in slots are different shortages. An EMS addresses the first. It does not create the second.

The vintage and geography in that paper are the parts a national average hides. In ResStock’s 2010s vintage they estimate 77 percent of homes have 200 A panels and 13 percent have 100 A. In homes built before 1940 the figures invert: 56 percent at 100 A, 31 percent at 200 A. Southern-census-region states show the highest share of 200 A-and-above panels (70 percent) and the lowest share of 100 A-and-below (21 percent), which the authors attribute in part to higher prevalence of electric space heating and water heating. California and states in the Great Lakes and Northeast show the greatest prevalence of 100 A-or-smaller panels, where space and water heating are more often fossil-fired. Panel capacity also rises with percent of area median income: in the 0–30 percent AMI bin they estimate 48 percent of homes at 200 A and 33 percent at 100 A; in the 150 percent-and-above bin, 60 percent and 25 percent. The panel-capacity model did not include income as an input; the authors treat that gradient as a correlate of housing stock and of who can pay for replacements.

The paper’s method limitation is printed in it, and it belongs here. More than 97 percent of the field observations used to train the models are from homes in California and Minnesota, two states with very high shares of natural gas space heating. The authors applied those models to ResStock, a nationwide representation. That is a modelled estimate, not a census of panels. EPRI’s household survey, which the paper compares against, found 27 percent of single-family detached homes at 100 A or less and 56 percent at 151–200 A after the authors recalculated EPRI’s bins without null responses. The two studies agree on the shape. Neither one tells you what is in your garage.

On headroom, the paper simulated two NEC methods. The Load Summing method follows 220.83(a): first 8 kVA of existing tabulated load at 100 percent, remainder at 40 percent. The Maximum Demand method follows 220.87: maximum metered demand from the preceding year, times 1.25. Existing loads estimated by Maximum Demand were 40 percent lower than Load Summing for the median single-family home in ResStock, with an interquartile range of 22 to 52 percent. Under Maximum Demand, load in the median 200 A home was 55 A and in the median 100 A home 38 A; median additional headroom ranged from 27 A to 30 A across panel-size bins. Calculated loads exceeded predicted panel capacity for 2.7 percent of homes under Load Summing and 1.5 percent under Maximum Demand. Those are modelled headroom distributions, not a finding that your 100 A panel has 27 A to spare. They are why 220.87 exists as a tool, and why an EMS setpoint is a different tool: both exist because the summing method is conservative on purpose.

Hazard-era, full, or already-maxed: the three stops an EMS does not talk past

The checklist above is the decision. The paragraphs below are why each stop is a stop, not a preference.

On a Federal Pacific Stab-Lok interior, the overcurrent device is the open question. CPSC press release 83-008, 3 March 1983, closed a two-year investigation without a recall because, in the Commission’s words, the data then available did not establish that the breakers posed a serious risk of injury to consumers. The 18 February 2011 staff note on that same release is the sentence that matters for this page: the Commission closed the matter “without making a determination as to the safety of FPE circuit breakers or the accuracy of the manufacturer’s position on the matter.” UL 489 requires a residential moulded-case breaker to trip at or below 135 percent of rated current. Independent testing compiled on the Federal Pacific replacement page documents high rates of failure to trip under that calibration test, including CPSC-contracted Wright-Malta work. An energy management system that throttles a charger so the main never sees 80 A does not make a branch breaker that may not open into a breaker that will. The next job is a listed load-centre replacement, which may or may not change the service size. Zinsco / GTE-Sylvania interiors are a different maker and a different failure mode; they are not Stab-Lok, and they are also not a reason to add a continuous 240-volt load to a panel whose bus and breakers are the problem. Do not remove the dead-front to “check.” Photograph the door label and the breaker faces, and have a licensed electrician confirm.

On a full panel, the constraint is geometry. A dedicated 240-volt breaker usually wants two adjacent spaces listed for a two-pole breaker. Tandem breakers already in the panel are not free slots. Two conductors under a lug listed for one is a defect, not a spare. LBNL’s 22 percent / 20 percent figures are the national modelled rates; your panel is a count, not a percentage. Circuit sharing, as a class, is the one EMS-adjacent path that can reuse an existing 240-volt circuit without a new slot. Whole-service monitoring and charger-native derating both still need a place to land the EVSE unless the design is a listed smart-panel extension that brings its own spaces — PG&E’s PanelBoost FAQ, retrieved 5 September 2026, describes a waitlisted utility programme that supplies a panel extension including 16 breaker spaces, for 150 A-or-smaller single-family services, with PG&E covering the hardware. That is one utility’s programme, currently waitlisted, not a product you can assume exists in your territory. A full 200 A panel with a currently listed interior is a panel upgrade or a subpanel conversation, not an EMS conversation.

On a house that already uses the service, the constraint is physics. 220.87 lets an electrician, where the AHJ accepts it, calculate existing load from a year of maximum demand times 125 percent. If that number plus a 40 A-class charger already sits on the main, a setpoint of 80 A on a 100 A service will spend the evening choosing between the car and the range. The NREL laboratory house recovered in a minute after a dryer cycle because the dryer was the coincident spike and the rest of the house was not already at the rail. A heat pump, an induction range, a tankless water heater and a 48 A charger landing in the same season is a stacked-load problem. The electrification readiness tool is where those stacks get a first look. An EMS that has to starve the charger every evening has not avoided an upgrade. It has deferred one, at the cost of a car that is not full in the morning.

Utilities are not always the party pushing the upgrade

Some utilities now rebate the cheaper path, or install the monitoring themselves. That is worth knowing because homeowners often assume the utility is the reason they were told to upsize. The opposite is also true: some utilities will not energise a service the existing equipment cannot support, EMS or no EMS. The only way to know which sentence applies is to ask the serving utility which constraint is binding.

PG&E’s Residential EV Charging rebate, checked 5 September 2026, is explicit: the programme “minimizes costly electrical service upgrades.” Eligible equipment is grouped into the three classes already named: chargers with adjustable amperage, a circuit-sharing device, and “EV Energy Management System (EMS) that automatically balances the load of an EV charger and home energy usage to avoid a panel upgrade.” Standard rebate is up to 50 percent of the purchase price of one listed option. A licensed California electrician must do hard-wired installation; the FAQ is blunt: “Can I install a 240-volt outlet myself instead of using an electrician? No.” Rebate Plus, for income-qualified customers (under 80 percent of area median income, or enrolment in a named public-assistance programme), goes up to $2,000 for an eligible charger or $5,000 for a panel upgrade plus charger — which is the utility naming both paths, not hiding the expensive one. The programme is funded through California’s Low Carbon Fuel Standard, not customer dollars, and is limited to one rebate per household Electric Service Agreement. Those are PG&E’s rules on the day we retrieved them. Programme rules change.

The same utility’s ChargeBoost programme swaps the meter at no cost and lets a compatible Level 2 charger slow down when the house is busy, so the main panel is not upgraded. PG&E’s own how-it-works list: apply; if approved, PG&E swaps the meter; the customer buys a compatible charger and has a qualified electrician install it; the new meter is connected to the home’s Wi-Fi and then to the charger. Compatible equipment on the page we retrieved was the Emporia Classic Level 2, with a note that additional options may be added. Installation costs PG&E tells customers to expect: Level 2 chargers $400–$700, electrician $200–$800 “depending on the complexity.” If Wi-Fi drops, the charger defaults to Level 1. ChargeBoost is currently described as available for single-family homeowners with no income restriction. It is a meter-side EMS: the utility owns the sensor. That is a different product class from a contractor-installed CT kit, even though the physics is the same.

PanelBoost is a waitlisted smart-panel extension for 150-amp-or-smaller single-family services. PG&E’s own FAQ puts a typical overhead panel-and-service upgrade at $3,000 to $8,000, and $20,000 or more with underground service. The programme covers the PanelBoost hardware, including 16 breaker spaces; the customer pays the electrician’s fees for installation, permits, and appliance connections. During rare high-demand periods the system may briefly pause a lower-priority connected appliance; the customer sets priorities in an app. PG&E’s own “why would I still upgrade” answer names an ADU, more load than PanelBoost can support, or an existing panel that is outdated or unsafe. Those are PG&E’s figures and hedges for its territory, not a national quote and not a finding that a waitlisted programme is available on your street.

Named tariffs, not a national interconnection rule

HyreElectrical does not invent a national rule about how a utility will interconnect a managed charger. This page names two PG&E rate schedules retrieved 5 September 2026, because they are the interconnection-adjacent facts a PG&E customer will actually be offered, and because one of them requires a second meter.

Schedule EV2-A (Home Charging) combines the house and the vehicle on one meter. It is a time-of-use rate. PG&E’s page: costs are lowest from midnight to 3 p.m. every day, including weekends and holidays; peak is 4–9 p.m.; partial-peak is 3–4 p.m. and 9 p.m. to midnight. Customers over 800 percent of baseline allowance over the last 12 months are not eligible. EV2-A customers may enrol in CARE, FERA, SmartRate, and Medical Baseline. The full schedule is published as PG&E’s EV2-A tariff PDF. A managed charger on EV2-A is still one service, one meter, one load calculation. The rate does not change 625.42. It changes when it is cheapest to let the charger run, which is exactly when a whole-service EMS has the most room: overnight, other loads off.

Schedule EV-B separates the vehicle’s electricity from the house and requires the installation of a second meter. PG&E: “Do you want to install a second meter dedicated to your electric vehicle (EV)? You are only eligible for EV-B on that meter.” The house then sits on another rate, not EV2-A. EV-B is not eligible for CARE, FERA, Medical Baseline, or SmartRate. Off-peak on EV-B is 11 p.m. to 7 a.m.; peak is 2–9 p.m. A second meter is a utility interconnection. It is a different service arrangement, not an EMS. It can make the house-side calculation easier by taking the charger off the house meter; it does not, by itself, create spare slots in the house panel, and it does not repair a hazard-era interior. PG&E’s own page tells customers they can change rate schedule twice in the first 12 months and must then remain on the new rate for 12 months.

ChargeBoost’s meter swap, described above, is a third PG&E path: not a new rate schedule, a new meter that talks to a compatible charger. PanelBoost adds a meter upgrade “to support intelligent energy coordination between your home and the grid.” Those are programme descriptions on PG&E’s pages, not Rule 21 interconnection studies, and not a finding about any other utility. If your serving utility is not PG&E, none of these named schedules apply. Ask that utility whether it has a managed-charging programme, a second-meter EV rate, or a service standard that already requires a larger service regardless of 625.42.

Three calculation tools that are not EMS — and why they get confused with it

Load management is one way through a failed calculation. It is not the only one, and mixing the tools is how a quote becomes unreadable.

220.57, new in 2023, is a floor, not a management system. EC&M’s 2023 electrified-home piece and Stallcup’s 2023 change video both state it the same way: the EVSE load is calculated at 7,200 volt-amperes or the nameplate, whichever is larger. 7,200 VA is 30 A at 240 V. A 16 A or 24 A charger still counts as 7,200 VA in a 2023 standard calculation. Code-making panel 2, in Stallcup’s telling, based the floor on a 30 A 240 V single-phase circuit and declined to add demand factors. 220.57 is why a small derated charger does not automatically shrink the service calc in a 2023-NEC jurisdiction unless 625.42 is also being used. A 2017- or 2020-NEC state does not have this section.

220.82, the optional dwelling calculation, is the neighbourhood shape this site’s load calculator follows: first 10 kVA of general load at 100 percent, remainder at 40 percent, HVAC at 100 percent, with EVSE treated as a continuous load at 125 percent on this site’s tool. Some inspectors want the EVSE held at 100 percent of nameplate in the optional method rather than riding the 40 percent remainder, because a charger can run for more than three hours. That disagreement is local. The optional method is a calculation path. It is not an EMS.

220.83 is the existing-dwelling adding-load path LBNL simulated as “Load Summing”: 220.83(a) for existing load (first 8 kVA at 100 percent, remainder at 40 percent); 220.83(b) when the retrofit includes air conditioning or electric heat. 220.87 is the metered-demand path: maximum demand data for a one-year period, that demand calculated at 125 percent, and the new load not exceeding feeder ampacity or service rating. Electrical Contractor Magazine’s December 2022 EV-charging piece states those 220.87 conditions in those words. 220.87 can make a 100 A service pass a charger that 220.83 would fail, if a year of meter data exists and the AHJ accepts it. It still does not create a breaker space, and it does not bind the utility.

220.60, noncoincident loads, is the older, cruder cousin of an EMS: if two loads will not run at once, count the larger. A dryer interlock that physically prevents the charger and the dryer from being on together is a 220.60 design, not a 220.70 design. Eaton’s AbleEdge white paper, retrieved in this research pass, describes interlocking two smart breakers so only one 50 A load in the group runs at a time, as an alternative to upsizing a 125 A load centre for a 50 A charger. That is a listed interlock, installed by an electrician. It is not a phone-app schedule. It is also not whole-service current monitoring. Asking which of 220.57, 220.60, 220.82, 220.83, 220.87, 220.70 / 120.7, and 625.42(A) or (B) the quote is actually using is how you find out whether anyone has done a calculation.