Version: 02.12, 01/20/11 Edits and technical details.
Version: 02.11, 12/17/06 Inrush thermistor testing.
Author: Bob Wilson, 9011 Randall Rd, Huntsville, AL


The computer controlled engine and battery system provides a unique opportunity to make a Prius into an intelligent, gas powered, UPS. Testing revealed this can provide ~1 kW of 110 VAC, enough to power a gas furnace.

Theory of Operation

The following circuit sketch shows how we will get 110 VAC from the Prius:

The main advantage is the engine runs ONLY when the main traction battery voltage falls below a threshold. Then the engine starts, recharges the traction battery and shuts off again. For example, the following chart shows typical charge-discharge cycling of the Prius in operation:

Using the "Toyota Prius Electrical Wiring Diagram 2003 Model," an analysis of the circuit fuses reveals:

BTW, there is a fuse in the inverter:

XPower 1000 Inverter

A review of the XPower 1000 manual reveals their recommended maximum battery fuse is "150 ADC" and a minimum cable of "No. 2 AWG" for less than 5 ft. They also recommended minimizing inductance of the power cable 'loop'. The inverter to frame ground needs to be "No. 8 AWG" but that also means an "earth ground" needs to attach to the car during operation. Finally, they also report it can drive a 1/2 HP motor and our gas furnace uses a 1/3 HP motor.

BTW, there are 1 wK inverters that have built in voltage and current indicators and an optional remote control capability. But our inverter was left over from an earlier emergency power project.

The challenge is to find:


The ideal location for a 12 V. inverter is adjacent to the auxiliary battery. This can be reached on the left hand side of the trunk by removing the felt cover.

This photo shows the power cables for the inverter from an earlier test. One problem is the ground cable was connected to the battery terminal which risked overloading that relatively small wire with inverter power requirements, 90 A. In our implementation, both the battery ground and the inverter ground are terminated in the battery strap bolt. We had to Dremil the washer to make it work but it is a cleaner implementation.

Remote Control

Already having a 1 kW inverter, it lacked a remote ON/OFF switch. This was solved by mounting a 5 VDC, 1 A, double-throw relay and a mini-audio connector.

The circuit uses two of the three mini-plug conductors, not the ground, to operate the 5 VDC relay. Both of the 1 A. relay poles are wired in parallel with the manual switch. Ordinarily the manual switch is left OFF and the inverter is operated via the remote switch which gets power from the cigarette lighter via an under the carpet jumper:
The cigarette lighter circuit is relay controlled by the ignition that insures the inverter cannot be left on without the ignition key on and drain down the 12 V. battery. To operate the inverter on JUST the battery, it has to be turned on in the trunk (NOT RECOMMENDED!)

We are reworking the cabin outlet mount to make a better mount at the base of the armrest. We've made a clay mold of the cover and will reform a utility cover that will be mechanically more secure.

Electrical Connection

A plastic protector covers the battery B+ so a notch has to be nibbled out.

One tricky part was unbolting the 13 mm. bolt that connects the battery clamp to the fuse and fusible link holder. This bolt was attached quite firmly so expect to use both a socket and vice-grip to loosen it.

The inverter vendor recommends minimizing battery cable inductance so the toroidal cores will be removed. Furthermore, the B+ cable and AC power lines will be fitted into flex-duct to avoid chafing. All exposed wires were wrapped in plastic electrical tape to prevent accidental shorts. Other lesson's learned:


The mounting panel is first modeled in cardboard. The cardboard template is taped to the 2 ft. x 2 ft., 1/4 inch board and any rough or interference parts are trimmed out.

The plywood panel is the mount for the inverter.

One valued suggestion from the Yahoo Prius Technical group was to make it a hinged panel.

This really worked out well. BTW, we used an adhesive to hold the inverter to the plywood panel so the four bolts are just clamping the inverter while the adhesive sets.

Once complete, the unit normally is closed and used to provide power for laptops or other low-power applications. But during high-power operation, the panel is lowered to maximize cooling (thanks "Hobbit"):

This is what it looks like buttoned up:

For now, we're using bungee cords to hold it closed.

UPS Testing

There is more work to be done: adding felt, running the remote ON/OFF control line, AC power, and installing the AC outlet. But the real question is, how well does it work?

We setup of a three-level, space heater as a load, used a clamp Hall-effect current sensor, and tested the system. All currents were measured on the battery cable that comes from the engine compartment and voltages at the inverter terminals. The Prius was started with the parking brake set to defeat the day-light running lights. The results were:


|   Start Time   |    End Time    |  Hrs  | Load | US Gal |  Load  |   Gas   | Effic |
|                |                |       |  kW  |        |   kWH  |   kWH   |       |
| 11/19/05 17:00 | 11/20/05 09:30 |  16.5 |   1  |  4.215 |   16.5 |  151.74 | 10.9% |
| 11/25/05 22:00 | 11/26/05 10:40 |  12.7 |   0  |  0.811 |    0.0 |   29.20 |  0.0% |

* 0.25 gal/hour fuel burn with 1 kW load
* 0.06 gal/hour fuel burn w/o load
* includes warm-up fuel burn

Even with the day-light drive lights out, the car continued to operate other systems such as the displays, the trunk light and by accident, the cabin fan on low. To get a handle on the 'overhead', we repeated the test with just the car running, not the inverter. The 'idle only' test revealed a substantially lower fuel burn per hour (temperatures were similar), 20% of the loaded value. In a second test, we verified the MPG display does update the MPG while the car is parked and idling. Using 'stealth mode', we returned home showing 99.9 MPG but the next morning it was down to 1 or 2 MPG.

Relative Efficiency

As pointed out before, the Prius operate the engine long enough to put a charge on battery and then shuts down:

Under light or no loads, the engine charges the main traction battery at 5 A. The normal vehicle overhead discharges the battery at ~1.7 A until the state of charge (SOC) reaches the threshold that requires recharging. Under a full, 1 KW load, the traction battery discharge increases to 5 A so the engine cycles on and off at roughly a 50:50 duty cycle. When the engine is running, the power needed for the inverter comes directly from the generator which is why we only see the traction battery charging and discharging current in the graph.

To compare the Prius to standard, gas-only, powered generators:

Using data from a Great Northern Tools catalog of portable generators listing half-power and full-power fuel consumption rates, we see the Prius inverter data point easily follows the half-power curve of these commercial generators. One reason the Prius is not more efficient is the vehicle overhead that is in the 350-500 W range depending upon outside air temperature and additional loads. But consider this, to generate 1 kW of output power, the Prius overhead of 0.5 kW has to be generated too. If we can increase the Prius AC power output so the overhead is a smaller percentage, the efficiency could easily exceed the power of a standalone generator. Altneratively, there may be efficiencies if we can map the overhead power drains and see if they can be minimized when in Park. For example, the power steering, power brake and air bag systems along with daylight running lights might be powered off when in Park or a 'special' park.

One interesting observation is the calculation of traction battery SOC versus voltage.

Understand that SOC is a synthetic number calculated by the battery controller using unpublished, Toyota algorthms. What is interesting is the battery hysterisis that shows how charging SOC lags the rate of traction battery charge voltage then lags again on discharge at a different rate. Battery charge-discharge is not 100% efficient and the area inside the curve maps out this small but measurable loss.

Furnace Testing

Operating a furnace, especially in anticipation of an 'ice storm' emergency needs to be carefully tested against these risks:

We ran a 7 hour test after instrumenting the gas furnace with a motor case temperature probe. We also measured the VAC and current:

121.4(VAC)  6.6(A)  81(F) - house line power (801 W.) at furnace
103.1(VAC)  6.4(A)  85(F) - Prius inverter power at furnace

5.09 (G) = 191 (M) / 37.5 (MPG) - ending fuel burn at 23:40
3.96 (G) = 191 (M) / 48.2 (MPG) - starting fuel burn at 16:35
1.13 (G) fuel burn / 7.08 hr -> 0.16 gal/hr.

* We drove a mile to refill the tank with 4.7 gal., not the calculated 5.09 gal.
* The lower fuel burn reflected not only the lower load but higher
  efficiencies from not running the inverter and Prius 12 V. system near
  their peak power setting.
Our furnace has a 1/3 Hp. motor rated at just under 600 W. but the measured utility load was 800 W. The higher load may be due to bearing lubrication and back-pressure from an aging gas furnace. Unloaded, the inverter puts out 120 VAC using a modified sine wave and was rated to run a 1/2 Hp. motor. However, we observed a significant voltage drop, ~17 VAC, during the furnace load test.

We brought the inverter power into the house using a 25 ft. contractor grade and a 50 ft. outdoor extension to the furnace. The contractor cord has heavier gage wire and provides three outlets into the primary living room. We measured a 3 VAC voltage drop using utiliity power through both cables to the furnace. This suggests the bulk of the voltage loss was capacitance through the cables.

Another problem, not unexpected, was the digital thermostat ran the gas burner constantly while on modified sine-wave inverter power. Once the furnace went back to utility power, the thermostat returned to normal operation. We will install a mechanical thermostat for power outage use.

Inrush Thermistor Testing

One problem under load is the effects of inrush current spiking the inverter. Modern, switching power supplies in TVs and laptops often have a full-wave, bridge feeding a substantial capacitor. Although the average power usage is small, the initial capacitor charge can overload one or two cycles that an inverter detects and protects itself by shutting down briefly. After the timeout, the inverter comes on again, sees the capacitor inrush and the whole system slowly oscillates . . . with all of the loads.

To mitigate the effects, certainly with the laptop, an NF08AA0330M (33 ohm/2.2W) thermistor was wired in series in a pig-tail outlet. The inrush current sees a room temperature, high resistance thermistor, so it can not overload the inverter. But the current that gets through warms up the thermistor to ~150 F and its resistance goes down allowing full power, current flow. Testing with the laptop revealed an order of magnitude reduction of inrush current at a power loss of 0.42W from the power needed to keep the thermistor warm.

Hot Inrush

By unplugging and plugging in while the thermister is hot, we see a less attenuated, inrush current spike:

Cold Inrush

A cold, higher resistance thermistor extends the initial capacitor charge so several power cycles are needed to fully charge the capacitor. Several samples are displayed to try and capture a worst case event:

The sampling rate was fast enough to capture individual cycles but slow enough that several seconds are captured. The thermistor slows the capacitor charging so it takes more than one cycle to fully charge the capacitor. Then a couple of seconds later, the power supply begins drawing power at much lower currents to operate the 45 W laptop. It is the sudden, unlimited charging that triggers the overload protection built into the inverter.


Our goal was to operate the gas furnace and testing revealed out unit would be safe and stable. We do not anticipate further changes until the gas furnace and air conditioner are replaced with more modern and efficient units. We would also like to thank contributors from the Yahoo Group, "Prius_Tehncial_Stuff" for their valuable suggestions.