Once you have a hypothesis, it's important to test it. Stated with as little babble as possible, my hypothesis was "you can get useful - very useful - amounts of electricity from the Prius." Looking at the ads, its components, its "theory of operation," it seemed like a "no brainer." As charming as that notion was, however, I could see a number of ways in which a cortical deficiency might be manifested as reluctant electrons. I had to devise a test that would confirm (or, perhaps, disprove) that my scheme would work. I was handicapped to some extent by a realization that if I were clumsy, stupid, or just plain wrong I would have to endure ridicule (much as I do routinely), have a very expensive lump of metal (or worse, slag) on my hands, and, worst, have to walk home. The latter would be a problem here in suburbia. So, the test plan was designed to:
It was immediately clear that the chief impediment, if impediment there were, would be the manufacturer, Toyota. They could make it impossible for customers to obtain service documentation. They could have anticipated this potential application and thwarted it in software. The second chief impediment would be my own, car-ignorant, self. I'm an electrical guy, not a car guy. I think nothing of buying a $50,000 piece of test equipment on eBay (final price: $50+shipping) and ripping into it. But for cars, I'm clue free and even had to borrow a socket wrench set from a guy keeping his 300Kmile Volvo running single handedly. I want to thank Pete (thanks, Pete) for helping me remove all the bits to get at the HV section, for his car knowledge and advice, and for the fact that when there was a Metric Nut Emergency his bike was functional while the Prius was in pieces.
The plan that didn't work
My original plan was a preliminary and essentially risk-free one. I assembled a computer, HP electronic load, and a cable with a "cigarette lighter plug" at the end. (I think cigarette lighter plugs now connect to "accessory sockets" since cars don't have cigarette lighters any more.) I wrote a program to take progressively greater currents from the socket, while graphing the voltage. In effect, it was exactly the same data I ended up taking from the HV battery connection. What went wrong? The Prius was too smart for me. It turns out that the low voltage auxiliary battery is charged from a DC-DC converter that gets power from the HV battery, and functions indetectably. As I increased the load current, all I could see was a slight, fixed, decrease in voltage due to cable loss. No charging, no discharging, no indication of when or how much the gasoline engine was running to charge the battery.
Even so, this test was useful. It established that for low power applications, the Prius is a good source of 13.8VDC. It provides a highly regulated voltage pretty much indefinitely. The user manual cautions you to take only 120W from this source. It's fused for 25A, which would correspond to about 300W. I didn't investigate this discrepancy and would recommend following the manual's advice since the charging converter may be thermally limited despite the higher peak current available. (New information Jan 2006: Bob Wilson has done some testing of the low voltage system and found that 1kW is the practical limit if you connect directly to the battery.)
The plan that did work
Unlike the LV auxiliary battery that is designed to be used by the vehicle owner, the HV battery is protected by layers of car stuff and interminable cautionary statements in the manual. Even so, it was clear that it would have to be tapped since that is the mother lode of electricity in the vehicle. Unfortunately, testing the HV isn't as simple as the LV. For one thing, I didn't and don't have an HP electronic load capable of absorbing kilowatts at 200V, and such a load would be necessary. I finally decided to make one myself, and visitors to my office for the next week or so felt they were in the middle of a high school science project. This device, built on a plywood plank using common household incandescent lamps, was designed to provide a load varying from 50W (a hopefully unnoticeable 250mA at 200V) up to as much as 5.4KW, all under computer control. The computer would be able to:
After assembling the hardware and connecting it to the car, the plan was to simply make a series of measurements at varying load currents and evaluate the results after each test to determine if they made sense and if it was safe to continue. The first test was at zero external load (measuring voltage only) and the final test was 2.4KW due to the limited rating of some of the interconnecting cables. I didn't feel it necessary to go to higher loads based on my conclusions about the information adduced at 2 and 2.4KW.
Other than a couple of computer crashes during the load tests (the results were preserved by photograph) the system worked as intended, and the data were usable. All of the main and subsidiary goals were accomplished. In particular, I did not suffer death from electrocution, and the Prius remains serviceable, although it now is that ugly champagne color instead of the lovely blue that I orignally ordered.
Updated 07 January 2006