Sunday, December 20, 2015

Home-brew J1772 EVSEs

I've spent some time this morning and looked at some of the videos on YouTube of people showing off their home-brew J1772 charging stations. I have some concerns about a lot of them.

Now, my criticism doesn't apply to all of them. Quite a few of them are based on OpenEVSE, which does include all of the checks. In fact, an OpenEVSE based design recently obtained full UL approval. That said, there are some things I would change about OpenEVSE's design, and those things wound up being the design for OpenEVSE II.

The most frequent deficiency I see is a lack of a ground-fault interruptor (GFI). A GFI is particularly important in EVSEs for two reasons. One, the business end of the EVSE is intended to be used in all weather conditions. Two, the car itself sits on four rubber tires, which prevent the frame from being grounded other than via the ground connection of the J1772 cable. A ground fault in the vehicle potentially could energize the entire chassis.

The second deficiency I see often is a lack of ground continuity monitoring (GCM). In the above case, where a ground fault energizes the chassis of the vehicle, a proper ground connection would overload the circuit breaker feeding the EVSE if the GFI didn't stop the current flow first. If the ground fault in the vehicle existed and the ground connection was poor (and the GCM didn't detect that), then there would be no ground fault current for the GFI to detect until a circuit to ground was established - likely through the next human to touch the vehicle. Now, the GFI will trip, but that takes some time, so the human will still get a brief shock, which could have been prevented.

Lastly, the specification requires a stuck-relay test. This test insures that when the relay or contactor is switched off that there is no voltage present on the hot lines of the J1772 cable. If the relay fails in a manner where it's stuck closed, then all of the other safety systems are completely moot. A proper EVSE should at least produce a warning so that power can be removed and the unit repaired.

There is also the pilot diode test. The diode test allows the EVSE to tell the difference between a vehicle and a bucket of salt water. Obviously, turning the power on when the handle is "plugged in" to one of the latter would be a bad thing. This is accomplished by testing for both the maximum and minimum voltage seen on the pilot line while the pilot is oscillating. The maximum voltage is used to detect J1772 state transitions, but the minimum voltage should be (close to) -12 volts.

So, yes, you can just charge a car by making a 1 kHz square wave with variable duty cycle and turning a contactor on and off, but there's more to it than that. You don't have to go get your home-brew J1772 EVSE UL approved, but you need to at least be aware of what UL wants to see and why before you play with 7.2 kW or more of power.