detour

PriUPS Update 26 September 2005

hazard

In my previous update I offered a tentative solution to the inverter problem, that solution being to use a computer UPS, which already has an inverter, and has batteries as well.  Advantages include not even needing the Prius for short outages, and the ever popular "cheap."  This update covers progress since I got the inverter.  Specifically:

When we began this saga, one of the first steps was removing all the carstuff that protected the battery from my prying fingers.  That was done with the help of the step-by-step instructions in the voluminous Prius manuals.  The goal was to connect a power-take-off from the HV battery, which I initially did with a simple outlet strip.  However, the current rating for the outlet strip was only 15A, which is fine for testing under benign conditions, but unsatisfactory for actual use.  Pete (who will reappear moments from now) happened to have a nice mating pair of Anderson high power connectors and offered them up for this project. 

I had never actually seen connectors like this - they're pretty nifty.  The have spring loaded blades and mate with significant pressure, but don't lock in a way that makes removal difficult.  The attached cord is for demating - you yank on it and the pair comes apart without stressing the wires themselves.  For the Prius side of the connector, I simply attached a couple of #8 wires with lugs on the end (and the connector on the other end) to the battery terminals in the Prius.  Now I was all set to add a permanent connection, but unfortunately...

While I drove around in my now high-noise Prius, 220V outlet strip dangling in the rear compartment, my alleged brain methodically forgot where all those upholsteries and doohickii were supposed to go.    The very same Pete of recent connector fame said, in effect, "Get out of the way."  I was pleased to see that all the parts (and all the screws) re-found their homes, and the car was back together after his ministrations.  With one addition:
If you fold down the perfectly-normal-appearing rear seat (left) you can access the full potential of the battery at the connector that's otherwise wedged between the seat side and the side of the car.  This is all soft, insulating material and the connector is very snug.

The next step was to get a cable from the garage to the electrical distribution room, fortunately adjacent.  I simply drilled a hole in the wall between the two and ran a short piece of 3-conductor, 6AWG flexible welding cable through it.

The black "hose" is actually a piece of welding cable brought into the electrical distribution area. (DVM for size comparison only.)  On the right is the garage side of the cable, with a connector attached. 

Finally, I made an "extension cord" out of the same welding cable to connect the Prius (using the blade connector behind the rear seat) to the cable going to the electric room.  Both the extension cord and the cable to the electrical room are made from very heavy cable.  It's overkill for the current actually carried, but because the Prius delivers less voltage than the normal rating of the UPS batteries, I was concerned about even the smallest voltage drop.  Using the cables I did I extrapolated less than 1.5V drop between the car and the diode under peak load, and well under 1V drop under normal operation with all the connectors in-line.

Here's a closeup of the heavy-duty connector I used for the "extension cord."  This is a "blind mating" connector.  It's rated at 175 amps - I think it's normally used for charging fork lifts.  Although I mate and demate these connectors manually, it should be possible to make matching fixtures on the wall and the car so that one can simply drive into the garage and be automatically connected. 

This connector is also by Anderson.  I made sure it was rated for enough connect/disconnect cycles so that even with cycling it each time the car is moved it should last as long as the car does.

Having the three segments of cable detailed above gives extra flexibility.  The "extension cord" can be kept in the garage if that's the only place the car will ever be plugged in, or carried in the car if the Prius will be used to provide power elsewhere, since the fixed portion of the system can easily be recreated at a vacation home, work, or wherever.


Connecting The Inverter

The nominal battery voltage of the UPS is 240VDC, and when fully charged it is higher - typically between 250 and 260V.  The Prius, as demonstrated in my previous testing, typically cycles between 210 and 235V.  Because of the difference in voltages, one cannot simply connect the two battery systems together.  In addition to the voltage problem, if the inverter battery were accidentally connected to the Prius when the "ignition" was off, the car would be receiving high voltage out of sequence, i.e., without the low voltage being turned on first.  This would probably damage it car and would discharge the UPS batteries as well. 

The simplest strategy for solving this problem is to connect a diode between the Prius battery and the inverter battery, and that's what I did.  That way, if the UPS battery voltage is higher than that of the Prius, the Prius remains isolated.  This is the normal case even during a short power failure, as it takes a while for the UPS to discharge. 

Shown at the left is a pair of heavy duty diodes.  The bottom one is being used only as a junction at the moment.  The two black wires going to the left terminal on the bottom diode are the DC negative connection.  The white wire is the + output from the Prius, the black wire to its right (with the red marker ink) is the + battery terminal of the UPS.  When the white wire is positive with respect to the black, the diode conducts and power from the Prius goes to the UPS.

The diodes are mounted on an aluminum plate for heat dissipation.  Even though the voltage drop through the diode is less than a volt, it can get quite warm when there's a lot of current flowing.

Recognize this?  This was the power supply I used to charge the test batteries.  It's now a pair of meters repurposed to measure the voltage of the Prius (left) and UPS (right) batteries.  Although the scales have different legends, they're both about 360V full scale, accomplished by putting appropriate resistors in series.  The meters are connected to the two batteries with the skinny wires shown in the photo above.

It would still be a fine power supply if I hadn't accidentally connected a battery to be charged backwards.  (It rarely pays to fix badly damaged older equipment - you just buy a replacement on eBay.)

Let's Do Some Tests

To summarize the update thus far:

 At this point, if we kill AC on the input to the UPS, the UPS should still supply its load until its batteries discharge.  And, if we connect the Prius as well, as soon as the UPS batteries discharge to the level of the Prius batteries, the diode will conduct, the Prius will supply power to the UPS, and the UPS will continue to supply power to its load until either AC power comes back on or the Prius runs out of gas.  With gas at $3+ per gallon, I'm not inclined to run test for quite that long, but a few hours are OK...

The last time I ran the load test, it was by connecting the "dummy load" of ordinary incandescent lamps directly to the Prius battery.  The lamps ran on DC.  For this test, the lamps are connected to the output of the UPS, i.e., 230VAC.  A slight change to the test setup involved connecting the laptop computer I used earlier to the test instruments via ethernet.  You can see the National Instruments "GPIB-ENET" next to the outlet strip on the bottom right.  The thick orange wire is just an AC extension cord; the thin one is the ethernet cable.

The test program is very similar to the original one.  I modified it to scan the voltage from the Prius and the UPS so the operation of both could be analyzed. 

Here's a complete view of the system.  The dummy load lamps are on the left, the switchbox and voltmeter are at the right, the meter box and a small UPS (for the test equipment only) are behind them.  Under the table repose the UPS and (to its right) the battery cabinet.  The battery cabinet's side is visible in the photo above.

The computer running the test program is in a different room.  It gets very hot and very bright in the test room when the load is powered!

I ran a number of brief tests at differing power levels (up to about 2400W) to confirm everything was working correctly and to confirm that the UPS internal metering was working correctly.  I was surprised (and either pleased or concerned) to see that the UPS only claimed about 10% load at the 2KW level.  Its internal readout has very poor resolution on a character bar graph, so I don't know whether to make anything of this.  The UPS is rated at 12kVA / 8kW.  After confirming that the system was fully functional, I ran a full test by starting the system on AC, disconnecting AC, allowing the UPS to run the load, and then connecting the Prius.  This test ran for a total of over three hours.

Test Analysis

This is the first 2.5 hours of the test run at the 1KW level.  (Click on any test panel to see it at full resolution.)  The test starts out with the UPS connected to AC and its batteries about 2/3 charged.  After about 10 minutes the AC is disconnected and the UPS runs the load on its batteries alone, which immediately start discharging.  After another 15 minutes I plug in the Prius (red curve) whose battery discharges due to its own internal load.  Eventually (another 10 minutes) the Prius engine comes on and the Prius battery starts charging.  As soon as the Prius battery voltage reaches the UPS battery voltage, it starts charging the UPS as well.  The cycle continues to repeat, and would do so until the Prius fuel is exhausted if not stopped sooner. 
Here the same test is continued for another 20 minutes, so that details and timing can be more readily seen.  Note that it took a few cycles of Prius engine on/off before the system equilibrated.  After equilibration, this expanded graph shows that the Prius engine runs for about 2.5 min and then the UPS/Prius discharge takes over for about 9 minutes before the engine restarts.  This is essentially identical to the result obtained in my previous test, but is shown here with better resolution.
Finally, after one additional cycle, I reconnected the UPS to the AC line, at which point the battery voltage lines split while the UPS started to charge its batteries from AC to its normal higher voltage.  At this point Prius was no longer called upon to supply power. 

 

This test scenario is identical to what would be encountered during a 3-hour power failure, assuming no great amount of electricity was being used in the house.  Initially I would ignore the power failure as the UPS took over.  When it became obvious that power wouldn't be restored promptly, I'd connect the Prius "extension cord" and turn the car on, at which point it would supply power.  Finally, after a few hours, AC would be restored, I'd turn off the Prius and disconnect it when reasonably certain that the power emergency was over.

As with the previous tests, the amount of gas used can't be measured well or precisely.  It would require a much longer test, starting with a full tank, and a measurement of how much is required to refill the tank, to get a good reading.

New style of romantic lighting?

On to Update #3