What is the best way to configure a isolating 2:1 stepdown transformer, so that an EV can be efficiently charged (using a mode-2 EVSE which has a built-in RCD/GFI) without drawing more than 1 kW from an 230V AC household supply?

The reason for the 1 kW limit is to avoid over-running the power of my small PV array -- which reliably produces 1 kW of power between 10am and 3pm except during winter, and which rarely produces more than 1.5 kW of power except during mid-day in summer months.

Please note that the efficiency of the transformer is important in this application. At a rough estimate: my current configuration (2:1 isolating transformer without any power-factor correction) delivers about 4.6 kWh of charging energy during a 5-hour charging session at 115V AC * 8A = 920 VA; but it consumes about 5.4 kWh of photovoltaic energy. I'd greatly prefer this 0.8 kWh of energy to be heating the water in my electric hot-water cylinder rather than to be heating the transformer in my garage! (BTW I have a solar-controller on my HWC so that, unless it is already "hot", it will consume any "excess" PV energy my household creates.)

After a thorough exploration of the design space (there being more than a dozen ways the four windings of a Magnetek Jefferson 211-0091-055 stepdown transformer could reasonably be interconnected), and after swotting-up on the basics of power engineering, I have identified three plausible options for improving the efficiency of my low-power EV-charging system:

  1. Add a power-correction circuit to my existing 2:1 isolating transformer, tie its neutral output to its ground-supply outlet (to avoid tripping the RCD on my new EVSE), and use the 8A setting of my EVSE.
  2. Rewire the transformer so that it will be a 3:2 autotransformer, and use the 6A setting on my EVSE. Note that (230V)(2/3)(6A) = 920 VA, so this is nominally the same power as option 1.
  3. Rewire the transformer so that it will be a 2:1 autotransformer, and continue to use the 8A setting on my EVSE.

[2 kVA 2:1 stepdown transformer]

I need some guidance in order to complete my design, because -- even after my swotting-up -- I lack even a basic level of competency in power engineering. Accordingly, I'd appreciate your comments and suggestions, in particular: 1) is there a fourth major design option I should consider? 2) am I making any naive mistakes in my analyses and experiments to date as described below? 3) am I missing any important safety considerations? 4) given my measurements to date, could someone who is competent in power engineering identify the "best choice" for a fully-refined design? 5) are there additional measurements I might make (with my very-limited collection of test equipment) which would help to refine the design?

Regarding safety: I'm located in New Zealand. Our household electric supply is 230V (+/- 6%) AC, with MEN earthing. Accordingly: I'm cautious about design option 1 above, because each NZ household has just one tiepoint (called its MEN-tie, and located in its main switchbox) between line-neutral and its earthing rod. Having a floating-ground on the supply to my EVSE would make me a bit nervous when I'm charging my e-NV200 on a rainy day, because it is too tall to fit in my garage where my transformer is located. And I think it may be even more hazardous (as well as an additional expense) for me to install an earthing rod; and furthermore I'm planning to sell this house soon, so I would have to remove that earthing rod before vacating the property to avoid future hazards and confusions.

Regarding my test equipment: I have an uncalibrated consumer-grade mains power meter, an old-fashioned but still quite reliable Fluke 8020B multimeter, a household power meter which reports import/export wattage on its front panel, an EVSE which reports its input voltage and AC input amperage on its front panel, and a Konnwei 902 which plugs into my e-NV200's OBD2 port and communicates with the LeafSpy Pro app on my mobile.

When the output of the 3:2 autotransformer of Circuit #3 below is unloaded, the mains power meter on its input has (in one experimental run) reported 17.3W, 239.0V, 0.59A, 0.12 PF. Its unloaded output voltage (as reported by my EVSE) was variable: ranging between 150.6V to 151.1V. After starting a charge on the 6A setting of my EVSE, its front-panel reported a significant drop in voltage, to about 141V AC (+/- 0.5V), and a variable amperage of 5.8A to 6.5A. LeafSpy instrumentation on the charging VA rate of the battery shows even more variability -- this is the green line on the plot in my experimental notes on Circuit #3.

Circuit #3 below is my current-best design, because it is quite efficient -- indeed I doubt I can do better, unless there are safety hazards I have overlooked which are significant enough to push me back onto a 2:1 isolating transformer (with the additional expense and trouble of installing a power-factor correcting circuit). When wired as a 3:2 autotransformer, it stays cold (even after a 3-hour charging session), and consumes only 920W -- so is very efficient. However, the efficiency of the EVSE and of the charging circuitry in my e-NV200 may be degraded by whatever phenomenon is causing the evident instability of the VA rate of the charging power reaching the battery terminals.

I do not, as yet, know how to estimate any of the relevant parameters of a transformer, such as its input impedance, given the results of an unloaded test. After I learn how to do this (and your suggestions are welcome!) I will adjust my circuit diagrams to show the (estimated) impedance and resistance of each winding of my transformer.

I understand that a short-circuit test of a transformer can be used to estimate some of its parameters, however I do not have a variac.

Circuits I have considered to date, with some test results:

Circuit 1: starting from the isolating 240/120VAC wiring of the table, tie one side (I chose X1-X3) of the secondary to one side of the primary (I chose H2-H4); connect 230VAC mains neutral to this tie-point, and supply 115VAC stepdown neutral from this tie-point. Result: 115VAC at PF approximately 0.67 (according to a cheap-as-chips AliExpress inline multimeter which has no specs -- but which reports voltages and currents that correspond well with what I can measure in different ways). This power was accepted without ground-fault error by my 6A EVSE; the charge rate to my e-NV200's battery, estimated through the LeafSpy app and an aftermarket OBD2 bluetooth dongle, is 430W.


simulate this circuit – Schematic created using CircuitLab

Circuit 2: same as circuit 1, but with the switch closed to put a 40 uF cap across the primary. Experimental findings: PF unchanged, transformer perhaps throwing off a bit more heat, I'd guesstimate 150W rather than the 100W of Circuit 1 (but my measurement system consists of resting my hand on the steel casing of the transformer ;-). LeafSpy/OBD2 reports that %SOC of my e-NV200 is rising by 0.1% every 3 minutes, which translates into (roughly) a 300W charge rate. The mains power to my transformer is unchanged from Circuit 1: 730W.

Circuit 3: 3:2 autotransformer, phased with H1-H3 tied to X2-X4, with a switch-controlled capacitor.

Circuit 3 schematic

simulate this circuit

Experimental findings on Circuit 3 with the capacitor switch open: unloaded, this circuit produces 150V AC when supplied with mains power at 235V AC. When loaded with my EVSE on its "6A" setting, the voltage sags to 140V AC and the current (as reported by my EVSE) varies between 5.8A and 6.5A. The ammeter inline with the mains supply to Circuit 3 reports a power factor which varies between 0.90 and 1.00, and a current varying between 3.8A and 4.3A. My LeafSpy/OBD2 instrumentation on my e-NV200 reports that the EVSE's supply rate to the vehicle (plotted in blue on the screenshot below) varies between 750W and 850W; and that the VA charge rate of the battery (plotted in green on the screenshot) is highly variable but its median value (by my eyeball's assessment) is about 650W. After allowing 7.5 minutes for the stabilisation of battery-charging status, then waiting for the next 0.1% uptick in SOC, I observed a rise of 0.6% SOC over 10.17 minutes; and using my best-fit linear estimator (from other experiments) for the usable capacity (i.e. 0% to 100% SOC) of my 24 kWh 2014 e-NV200 as being (22 kWh)(83% SOH) = 18 kWh, I estimate the charging rate of this session at (0.6%)/(10.16 minutes)(60 minutes/hour)(18 kWh) = 640W. My observations of my household's power meter showed a steady 0.29 kW background-consumption rate, which rises to 1.19 kW (+/- 0.01 kW) when my EVSE is charging my e-NV200 through Circuit 3. This power-meter-reported draw of 1.19-0.29 kW = 900W for Circuit #3 is at the low end of the 900W to 1000W power draw as reported by my mains power meter, and the PF (as reported by my mains power meter) has a median value of 0.98 (by my eyeball's assessment of its variation), and my previous experimentation has revealed that my power meter does not correct for household loads with low power factors; so my best estimate of the actual draw of Circuit #3 is 920W. I tentatively conclude that I have achieved my objective of finding an efficient low-cost "solar-charging" system for my EVs; but I'm nervous about the highly-variable VA charge rate as reported by LeafSpy, because this may be causing some inefficiency in the battery-charging process which would only become evident when I discharge the battery.

Experimental results for a 6A charge using Circuit 3

Experimental findings on Circuit 3 with the capacitor switch closed. This was a 5.2 hour charging session which took my Leaf from a LeafSpy-reported 78.9% SOC to 97.2% SOC. My EVSE reports the ending charge as "Full 100%"; and the GOM ("guessometer") display on the dashboard of my Leaf also reports 100%. My ammeter reports this charge-session as consuming 4.86 kWh from my household's 230V AC supply. My EVSE reports having delivered 4.73 kWh over a session lasting 5h 11min. The difference between these two energy readings is 0.13 kWh, suggesting that my 3:2 transformer was consuming about (0.13 kWh)/(311 minutes)*(60 minutes/hour) = 25W. This corresponds nicely with the slightly-warm feeling of the transformer at the end of this session. I observed significant heat-gains (perhaps 30W?) in the EVSE over this session: its front-panel temperature readout rose from 20.3 degrees C to 30.7 degrees in the first 25 minutes of charging. There are significant inefficiencies in the Leaf's AC-charging circuitry which I have estimated at 300W in other experimental runs. In this run, the Leaf's AC-charging inefficiencies are swamped by (what I believe to be) the temperature-correction terms in the Leaf's computation of an state of charge (SOC) value from its time-averaged measurements of charging voltage and charging current. As indicated in the LeafSpy plot below, the SOC of the battery (magenta) dropped during the first 0.2h of charging. At the start of the charging session, the three battery temperature sensors were reporting 19.0, 19.2, and 18.4 degrees Celsius; at the end of the session they were on 20.9, 21.3, and 20.7. Using my usual method \$(SOCend - SOCstart)(SOH)(22 kWh)\$ for computing the charge energy delivered to the terminals of the battery, I estimate that 2.55 kWh was delivered to the battery, and 4.73 - 2.55 = 2.18 kWh was dissipated as heat either in the Leaf or in the EVSE. This is a believable reading, because I have seen 400W of inefficiency in other low-amperage AC charge sessions, and 2.18 kWh / 5h11m = 420W of combined heat dissipation from the Leaf and the EVSE (whose temperature rose during the charge). Confusingly, LeafSpy reports that this charging session started with the battery having 8.6 kWh of remaining capacity, and ended with the battery having 14.0 kWh of remaining capacity, implying that the battery somehow gained 5.4 kWh from a charge session which consumed 4.86 kWh of mains power. I think some of this disparity may be explained by a temperature-correction factor in the battery-energy estimator; and some (perhaps most) by the sort of user-interface "adjustment" (commonly found on the "fuel tank" gauge of motor vehicles) which transforms some of their dashboard displays into "guessometers" whose readouts -- over time -- a driver may, or may not, be able to learn how to correct. But I digress... the primary experimental result here is that the 25W inefficiency of the 3:2 autotransformer Circuit #3 is a very small fraction of the 420W inefficiency of the charging circuit downstream of Circuit #3.

5h11m 6A charging session of my Leaf from the 3:2 autotransformer of Circuit #3 with the 40 uF shunt capacitor

Circuit 4: 3:2 autotransformer, phased with H1-H3 tied to X1-X3, with a switched capacitor. Untested.

Circuit 5: alternate phasing of neutral-tied 2:1 stepdown transformer, with X2-X4 tied to H2-H4. No capacitor.


simulate this circuit

Experimental results on Circuit 5: Estimated charge rate is 760W. The power-meter on my house during this experiment was ticking over at a steady 970W; and the baseline household draw immediately before and after the experiment was 240W; and this 970-240 = 730W estimate for the mains power into my transformer corresponds to what my cheap-as-chips uncalibrated ammeter says. So... it seems the terrible power-factor in this circuit is "working in my favour" -- when it is powered from mains. But ... I want an efficient circuit for using PV power from my roof!

Circuit 6: as in circuit 5 in the schematic above, but with the switch closed so that there is a 40 uF cap across the 230VAC supply. The estimated charge rate of my e-NV200's battery (as measured over a 15-minute charging session using LeafSpy and my Konnwei KW902 OBD2 dongle) is 900W. Wow, and if I run it from mains power (as in my experiment) I'd pay my electricity retailer only for 730W! But... I really don't care to find out what would happen if I tried this experiment when my PV panels are producing power. The 3 kVA inverter on my home's solar system can tolerate some range of power factors on its load, I'm sure... but if this circuit really has the 0.67 PF reported by my AliExpress multimeter then I definitely want to know how to get a better power factor!

Circuit 7: a 2:1 autotransformer using the H1-H2 and H3-H4 windings, i.e. with H2 tied to H3, 230 VAC mains power across H1 and H4 (with H4 "neutral"), and 115 VAC supplied from H2-H3 as "hot" and H4 as "neutral". Untested. I'm concerned about the kVA rating of this circuit. I have only a very-rudimentary knowledge of how a power-rating for a transformer can be calculated; but if this wiring cuts the rated capacity roughly in half (to 1 kVA) then the detailed calculations will matter. My mains supply is 230 VAC +/- 6%, see sec 28(1) of https://legislation.govt.nz/regulation/public/2010/0036/latest/whole.html#DLM2763653, so it may go as high as 248.4 VAC -- and for all I know, my poorly-specified EVSE on its 8A setting might, at some ambient temperatures, draw more than 1 kVA through this autotransformer.

Circuit 8: alternate phasing of the 2:1 autotransformer of Circuit 7, with H2 tied to H4. Untested.

Circuit 9: a 3:1 autotransformer, formed by tying H1-H3 to 230 VAC "hot" mains, H2-H4-X1-X3 to "hot" step-down supply, and X2-X4 to "neutral" on both mains and step-down supply. Testing results: this circuit dimly lights my 35W Christmas bulbs at (roughly) 80 VAC at a time when my mains voltage was (roughly) 240 VAC. This circuit is unsuitable for charging either my e-NV200 or my 2013 24 kWh Leaf, using the 6A setting on my EVSE, because the minimum charge-rate for a Leaf is 600W (this being 6A at the Japanese 100 VAC mains supply). It might work for trickle-charging my EVs if I used the 8A setting of my EVSE but ... in my vague understanding of autotransformers, I'd get better electrical efficiency with a 2:1 or a 3:2 autotransformer.

Circuit 10: a 3:1 autotransformer, formed by tying H1-H3 to 230 VAC "hot" mains, H2-H4-X2-X4 to "hot" step-down supply, and X1-X3 to "neutral" on both mains and step-down supply. Somewhat surprisingly, despite the phase-change from Circuit 9, this circuit also dimly lights my 35W Christmas bulbs at (roughly) 80 VAC at a time when my mains voltage was (roughly) 240VAC. Perhaps: half of the turns on each winding of this transformer are clockwise around the core, and half are counter-clockwise?

Circuit 11: A 2:1 autotransformer using all available windings. This would be preferable to the 2:1 autotransformers of Circuits 6, 7 and 8 (which use only the H windings) if it would have a higher kVA rating; but I don't know how to calculate a conservative kVA rating for a 2 kVA isolating 2:1 stepdown transformer which has been rewired into a 2:1 stepdown autotransformer. Do you?


simulate this circuit

  • 2
    \$\begingroup\$ This is a question and answer site - you need to ask a question and not propose a lab exercise. \$\endgroup\$
    – Andy aka
    Commented Feb 11, 2022 at 9:36
  • 2
    \$\begingroup\$ In some parts of the world mains plugs are non-polar, so there is a 50% chance the live and neutral are swapped by the user plugging it it. There is no way the charger can work if it requires a low impedance path between neutral and earth/ground. \$\endgroup\$
    – Justme
    Commented Feb 11, 2022 at 9:50
  • 4
    \$\begingroup\$ Please, schematics, not descriptions. Schematics are the language of EEs. \$\endgroup\$
    – winny
    Commented Feb 11, 2022 at 11:30
  • 3
    \$\begingroup\$ Welcome to EE.SE. As said, a schematic is better than words. You can add one in using the CircuitLab button on the editor toolbar. Double-click a component to edit its properties. 'R' = rotate, 'H' = horizontal flip. 'V' = vertical flip. Note that when you use the CircuitLab button on the editor toolbar and Save and Insert on the editor an editable schematic is saved in your post. That makes it easy for us to copy and edit in our answers. You don't need a CircuitLab account, no screengrabs, no image uploads, no background grid. \$\endgroup\$
    – Transistor
    Commented Feb 11, 2022 at 19:24
  • 1
    \$\begingroup\$ tl;dr Can you cut down your question to a couple of pages, with paragraphs of no more than a few sentences each? \$\endgroup\$ Commented Feb 12, 2022 at 20:14

1 Answer 1


Circuit #3 is an excellent match to my requirements. It is efficient, dissipating only about 25W. It meets my 920W charging goal when my EVSE is set for a 6A charge, and (unlike a 2:1 autotransformer) Circuit #3 also offers somewhat higher charging rates on the 8A setting of my EVSE.

When compared with a 2:1 isolating transformer, the 3:2 autotransformer of Circuit #3 has a significant advantage in that it doesn't need a power-factor correcting circuit for the (highly variable!) loads my EVSE draws, and I don't have to navigate the regulatory and safety issues of installing a second earthing stake on my property (so that my isolating transformer would be supplying into a TT circuit. See https://electrical.theiet.org/wiring-matters/years/2020/80-may-2020/the-iet-code-of-practice-for-electric-vehicle-charging-equipment-installation-4th-edition/ for a discussion of the safety issues.

I do not see any point in designing and testing any other circuits, although I can see a few other vaguely-promising ones, notably the 3:2 autotransformers which use only two of the four windings on my transformer (with two phasings). There are also many alternative phasings of the four windings in the 3:2 autotransformer of Circuit #3! However, EE.SE readers who live in areas with 230V AC supply on earthing systems other than MEN may be able to gain the safety advantages of an isolating transformer by bonding its ground input to its ground output (and to the chassis of the transformer), and they might be able to find a way to correct its power factor to the point that it is not dissipating the (roughly) 150W that made it run so "hot" when I was using it to power my old-fashioned EVSE (without a built-in GFI/RCD -- so it was ok with having a ground supply that was electrically isolated from its neutral supply).

Within a few years, I'd hope to find it possible to abandon my low-power AC EV-charging setup in favour of a mode-4 (DC) EVSE that is designed to meet the needs of households with a small PV array, an electric hot water cylinder, and an EV. The "entry level" model of such an EVSE might supply a maximum of 3.5 kVA, so that its power electronics and charging cable -- and its installation costs -- could be significantly less expensive than a 7.0 kVA unit. I'm almost certain that a DC charge of an EVSE can proceed at very low charging rates (perhaps as low as 0.5A at 480V) and still be reasonably efficient; whereas an AC charge of a Leaf or an e-NV200 has a rather lossy AC-DC inverter which (as far as I can determine) draws a minimum of 300W at low charging rates, rising to about 400W when the charge is at the maximum AC rate of 3.1 kW.

At the distinct risk of boring EE.SE readers with more marginally-relevant experimental findings... Below is a plot of the PV output from my household's 230V AC inverter on the day that I ran the 5h11m charging session described in my experimental notes on Circuit #3. I started that session at 10:34am, and it ended at 3:45pm; with the last half-hour being an intermittent trickle-charge which slowly brought the Leaf's batteries all the way up to their maximum-allowable charge level. During almost all of this period, my PV array was producing the 920W required by a 6A EVSE charge through Circuit #3; and the solar-controller on my HWC cylinder was offering any PV power that wasn't used by Circuit #3 to the HWC (which became fully "hot" during the day). Of the 7.8 kWh produced by my solar system: 4.73 kWh was consumed by my EV charging system (with 2.55 kWh of this 4.73 kWh reaching the terminals of the Leaf's battery); 0.52 kWh was "charging" my household's hot-water "storage battery"; about 0.6 kWh was consumed by other devices in my household; and the remaining 7.8 - 4.73 - 0.52 - 0.6 = 2.5 kWh was exported to the grid. My household "PV efficiency" today was thus 2.5 (to the Leaf battery) + 1.1 (to other household uses) / 7.8 = 46%. I can't see how I could ever expect to do much better than this, unless I have guests who are drawing a lot of hot water (or I'm baking many pies in the oven, for some reason!) on a sunny summer day... until such time (if ever!) that I have a low-power solar-controlled mode 4 (DC) EVSE.

Power delivered by the 230V AC inverter of my solar array on 16 Feb


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.