Thoughts on EV Charging Interfaces

Phil Karn
revised 18 Sep 1998

I've been driving an EV1 for about 4 months now, plenty long enough to have been exposed to the Great Inductive (GM) Vs Conductive (Honda/Ford) Charging debate.

Everyone agrees that the success of electric vehicles will depend in large part on the widespread deployment of a standard charging infrastructure. It goes without saying that this infrastructure should be based on the best technology available, and that a costly and confusing VHS-vs-Beta standards battle should be avoided.

This essay is my attempt to analyze these relative strengths and weaknesses of the various approaches to EV charging in the hope of informing the debate. I think I can maintain a fairly neutral position, as I personally feel that the jury is still out on the issue.

I am specifically excluding non-technical issues such as the current popularity of each type in public installations and the amount of marketing muscle being exerted by each system's champions, as this discussion is focused on the underlying technical merits of each approach.

Charging an EV

Though they do present special challenges of size, power and safety, an EV charger is fundamentally no different from any other kind of battery charger.

The most obvious way to charge a big battery from the AC power line is with a switching power supply controlled by battery voltage, current, temperature and time. Switching power supplies are far lighter, cheaper and more efficient than conventional "linear" power supplies. A switching power supply can also operate at full power to rapid charge a battery until it is nearly full and then taper off to prevent excessive heating and gassing.

Separable Issues

Many of the arguments frequently advanced for either inductive or conductive charging have more to do with a specific implementation than with some inherent property of either one. This section attempts to identify those issues.

Leaving It Behind Vs Carrying It Around

An argument frequently made for the conductive-style interface used with the Honda EV+ is the simplicity of its conductive "wall charger" (a simple 240V outlet) as compared to the EV1's big and heavy inductive Magnecharger. But this misses the point. Both systems almost certainly have similar charging circuitry. The real difference is where this circuitry is placed: the Honda EV+ carries everything on board while GM puts most of it in the offboard charger.

The offboard approach has the definite advantage of offloading much of the weight and volume that would otherwise have to be carried around in the car. It also gives the charger designer a somewhat more relaxed weight and volume budget. Among other things, this lets him do a better job in suppressing radio frequency interference, as good high power RFI line filters tend to be physically large. (As evidence of this, the 6.6 kW Magnecharger seems rather quiet, radio-wise, while the portable 1.2 kW charger is quite noisy on the lower HF bands up to about 5.5 MHz.)

Both the GM EV1 and the Honda EV+ appear to use switching power supplies in their chargers. The GM approach splits the switching supply right down the middle of the high frequency transformer. Everything up to and including its primary is in the charger, and the secondary and beyond is in the car. The Honda EV+ (and presumably the Ford Ranger) have onboard chargers, so the external conductive interface simply carries high current 240V AC.

An offboard charger that produces high frequency AC is fundamental to the inductive approach; an inductive coupler rated for, say, 6.6 kW at 60 Hz would be impractically massive. But an offboard charger could also be built with a conductive interface, with the same reductions in onboard size and weight.

For example, you replace the existing paddle and charge coupler in the EV1 with a conventional high frequency transformer in the car whose primary is hardwired to the charger. You'd also replace the existing 915 MHz RF car/charger control link with an optoisolator or signal transformer. Then you'd "cut the cord" on the existing Magnecharger and install conductive connectors. Instead of carrying the control link across auxiliary conductive contacts, you could make this an optical link.

This approach would still allow a common conductive charger to be used with a variety of EV battery voltages. You'd establish a standard voltage on the conductive connector, and the onboard transformer would have the appropriate windings to step this up or down to the desired pack voltage.

In short, both the inductive and conductive approaches lend themselves equally well to a stationary charger that can be left behind while driving.

Noise

A very common complaint about the GM Magnecharger is acoustic noise. The Magnecharger's blower is loud, and the wall mounting also conducts vibration. The Honda EV+ seems much quieter while charging.

But this too is not a fundamental difference between inductive and conductive charging. The Honda's onboard charger is only a 4.4 kW unit, so it doesn't have quite as much heat to dissipate; one would expect it to be at least a little quieter. I don't think there's any fundamental reason why an offboard inductive charger couldn't be made as quiet as an offboard conductive charger of the same power rating.

It's possible (and this is speculation) that it's inherently easier to quietly cool an onboard charger like that in the Honda EV+ by using the vehicle's existing liquid cooling system. But this has more to do with the onboard/offboard issue than with the choice of inductive or conductive coupling, and we've already established that an offboard charger is preferable for reasons of weight and volume.

Safety

GM makes a big deal about the safety of the inductive coupler in their marketing literature. It's true; the inductive coupler is inherently quite safe. But in so doing they insinuate, rather unfairly in my opinion, that there's something inherently unsafe about conductive charging. This is just not true. While safety is clearly an important consideration, it is not impossible to make a safe conductive interface. It just takes a little more work. Ground fault interrupters have been around for three decades, and they work very well. It is also not rocket science to design a connector with very effective physical and electrical interlocks.

I've seen several public charging stations with nearby disconnect and breaker panels that can be easily opened to expose bare 240V and even 480V conductors. Obviously these are far more of a potential hazard to, say, a young child than a well-designed conductive charging connector.

Reliability

The inductive Magnechargers do seem to fail distressingly often, based on reports to the EV1 mailing list and also from my personal experience (two failures observed in San Diego County in the past few months). Some of these are really the installer's fault, e.g., not installing concrete-filled steel pipes as barriers to protect them from being hit by vehicles. But even the failures of the chargers themselves probably have more to do with being on the top of the learning curve than with any inherent failing of the inductive approach.

Again, the seemingly greater complexity of the inductive charger is really due to it containing circuitry that would otherwise be carried on the vehicle even if it had a conductive connector. And if this circuitry is to fail, it's better to have it happen in a stationary charger than inside your car. If a charger fails, you can always go to another one. If circuitry inside your car fails, you're stuck.

Conductive interfaces have also been criticized as unreliable. And perhaps they are inherently somewhat more so than an inductive coupler with no exposed metal surfaces to corrode. But an inductive coupler is not immune either to mechanical damage, e.g., to sensor microswitches, control antennas, door covers, etc.

In sum, I suspect that the learning curves will eventually produce both inductive and conductive couplers that are quite reliable.

The Real Issues

Now we come to what I believe are the intrinsic differences between the two approaches.

User Simplicity and Convenience

Perhaps the biggest advantage of the GM paddle-and-slot system is its extreme ease of use. Like most EV1 drivers, I attract a curious crowd of onlookers almost everywhere I go. Before I was even able to get out of my car at the Laguna Niguel Costco, one lady asked me how I charged it. I challenged her to try figuring it out on her own. With no prompting (and presumably no prior experience), she opened the door of the Magnecharger, pulled out the paddle and stuck it into the charge coupler. It took her perhaps 5 seconds to figure it all out. She did hesitate for a moment wondering which way it should go before she realized it didn't matter.

While we engineers tend to dismiss such issues, the fact is that ease-of-use is very important to the mainstream adoption of any new technology. And it seems unlikely that any high-power, high voltage conductive connector can ever be quite as easy to use as the inductive paddle.

Efficiency

The inductive coupler is a high frequency transformer that is designed to be repeatedly put together and taken apart. While it is surprisingly efficient, it cannot possibly be more efficient than a conventional transformer that is assembled only once. Magnetic fields don't like having to cross air gaps, even the very small ones in the GM inductive coupler. A conventional transformer has no gaps, so the efficiency seems inherently higher.

The efficiency issue also creates a cooling problem. The EV1 already runs coolant to the coupler, though it's not clear if this is really needed for 6.6 kW charging or is there in anticipation of future Level 3 (high power - 50 kW) charging. It is already known that the Level 3 paddles will have to be liquid cooled through lines down their cables.

In contrast, it seems doubtful that any well-designed conductive connector would dissipate so much power as to require liquid cooling, even at the 50 kW power level.

Direct AC Charging

At the beginning I assumed that any electric vehicle would use a switching power supply as its charger. This is not necessarily the case. It's an interesting coincidence that the EV1's battery pack voltage, 312 volts, is not much below that produced by direct rectification of the 240 VAC power line (340 volts). Given that the EV1's inverter (the electronic circuit that converts DC into AC for the induction motor) looks very much like a DC switching power supply in reverse, it's possible to feed 240V AC line current directly back into the inverter. This would use the inverter as if it were doing regenerative braking -- with the AC power line taking the place of the motor/generator.

This approach, called reductive charging by Alan Cocconi (one of the original EV1 designers) has the advantage of requiring neither an offboard charger like the Magnecharger or the nearly equivalent onboard charger. The inverter would double as the charger, and with its existing liquid cooling system it could probably handle some fairly high charging power levels -- at least as high as those encountered in highway cruising.

This approach would dictate a conductive interface.

For safety's sake, this approach would require the complete isolation of the propulsion battery and all related circuitry from the chassis of the car. Both the negative and positive lines from the propulsion battery would be "hot" with AC with respect to ground while charging. The EV1 is already designed this way. But Cocconi reports that even with this isolation, small leakage currents can exist that will trip a GFCI in the supply circuit and careful design of the drive motor is required to eliminate them.

As an alternative, a stationary AC isolation transformer could be provided on the outlet used for charging. But because this would have to be a 60 Hz transformer rated for full charging power, it would be physically large and heavy. It could easily be heavier and possibly even more expensive than the existing Magnecharger.

The existing switch-mode chargers, for all their faults, do provide AC ground isolation "for free" through their high-frequency transformers -- either the one built into the car (conductive approach) or the one formed by the paddle and coupler in the inductive approach. And they do adapt readily to changes in AC line voltages. Both problems would have to be solved in any direct AC charging scheme.

Discussion

The reductive charging scheme seems attractive because it requires no special charging hardware, either on the car or at the charging site. But it turns out that the charger is only a fraction of the cost of many charging sites. Simply bringing power to the parking spot turns out to be much more costly in most cases. Even 120V convenience outlets, sufficient only for very slow EV charging, are rare in parking lots. Higher power feeds have to be installed specifically for EV charging. This often requires trenching under asphalt in a parking lot or running hundreds of feet of conduit in a parking garage. Sometimes the runs are long enough to require a 480V feed and a step-down transformer near the charger, further increasing the cost.

It's also reasonable to expect that with volume production, Magnechargers will come down in price faster than the cost of installing them.

Comments on this essay are invited.