Any number of test waveforms can be stored on a PC and sent to the test system controller as required.  These waveforms can be collected by storing the actual current waveforms when arcs are generated with hardware that simulates real-world conditions.  Even though waveforms suitable for breaker testing may not be easy to generate due to experimental difficulties, they only have to be captured once and then subsequent experiments are straightforward and repeatable.

A library of digitized waveforms can be easily recalled for analysis and testing.  These can also be very useful during the early stages of product development for analysis by software that simulates the breaker's operation.  Raw waveform data can be preprocessed by the PC to simulate nonlinearity and bandwidth limitations in the breaker's analog input circuits.  The basic software algorithms can be developed and run on the PC in a high level programming environment that significantly expedites product development.  After the basic principles of the algorithms are determined, they can be converted to code for the breaker's microprocessor and evaluated further using a hardware emulator and the original stored waveforms as test input data.

Other advantages to using stored waveforms are they can be scaled to modify the effective current level or even time scaled so the same signal can be used for 60Hz and 50Hz tests. Waveforms can be combined by the computer to simulate complex real-world conditions like several appliances operating simultaneously on one power line.

Because of the complexity of discriminating hazardous arcs from non-hazardous arcs, the fine tuning process may go on for a long period of time.  As new requirements turn up, they can be evaluated without having to build prototypes, set up lab equipment, and go through a laborious test process again.  New waveforms can be added to the library and a large number of tests can be run quickly to check product revisions.

The test current can be generated in a variety of ways.  Figure 9 shows three different current sources which cover three ranges.  The neutral wire input to the breaker has a resistance of less than 0.1 ohm.  It can be thought of as a summing junction and the total effective current seen by the breaker's internal circuitry is the sum of the individual currents:

• Current Source 1 - This supplies the low frequency part of the waveform.  It can have a total bandwidth on the order of 10KHz to 100KHz using off-the-shelf components.  For many breaker designs this bandwidth may be sufficient for all the arc waveforms.

• Current Source 2 - This supplies accurate current in the milliamp range for testing the GFCI circuit and for sending commands to the breaker to program the test mode.  It might be incorporated in Current Source 1 as a second range, since the two sources are not needed simultaneously.

• Current Source 3 - For breakers that sense high frequencies, another source may be desired.  Operation in the megahertz range can be easily achieved with relatively low power circuitry.

The controller is synchronized with the AC line voltage.  By syncing the externally generated waveforms with the same AC signal that is used to synchronize the breaker's internal circuits, the test results will be repeatable.

In this basic test system we see how the test current actually flows through the neutral conductor of an AFCI/GFCI breaker and since this conductor is grounded (for test purposes), the current source only has to supply enough voltage to accomodate the wire and contact resistance plus a small voltage for the current sensing resistor inside the current source.  In general, the peak signal voltage can be in the 5 to 10 volt range.  By reducing the peak voltage from 170 volts to 10 volts, we've reduced the power level by a factor of 17.  The combination of lower voltage and lower current now require roughly 100 watts peak power, which is not difficult to generate with conventional linear amplifiers.  Since arcing tests only run for a few seconds, the average power dissipated in the current source is only a few watts.

A special circuit is used to determine whether the contacts of the breaker-under-test are open or closed.  When the contacts are closed, the voltage at the neutral wire input junction is always low.  When the contacts open the magnitude of this voltage changes abruptly, depending on whether the current is positive or negative at the time.  (A DC offset of a few milliamps can be injected at the test junction so the current will not be exactly zero when the contacts open.) This change in voltage is detected and a logic signal to the controller stops the test.  In this way, the breaker's trip time can be measured very accurately.

For breakers that need test currents over 30 amperes, it may be an advantage to generate these larger currents with a different test system using more conventional triacs and SCR's.  These high current waveforms are characterized primarily by their large scale parameters, like peak value and conduction angle. The exact shape is not so critical as long as the waveshape is consistently reproduced each time the test is run.  If desired, low-power high-frequency signals can be injected at the same time to get a combination of high power and high frequencies.  Again, this approach takes advantage of the fact that the neutral input acts like a summing junction to combine the outputs of different signal sources.  The voltage at this junction is typically no more than a few volts, even at high currents, so the input power is much less than if the test signals were being injected on the hot (black) wire.

Figure 10 - Test system in the Zlan laboratory. The two cables from the tester are designed to plug into adapters for various models of AFCI/GFCI breakers. This unit programs an equivalent load current up to 75 amperes.

CONCLUSIONS

New developments in electronic circuit breaker technology have reduced the difficulty of manufacturing and testing.  By having a collection of standard test waveforms in a library, the testing process is streamlined and product development time is significantly reduced.  The breaker test system can be physically small (breadbox size), light weight (10-20 lbs) and moderate cost (few $K).  The result is a better product, developed in less time. When AFCI/GFCI breakers are tested with this new approach, their response is exactly the same as it would be if full load power were being applied.  There are no approximations involved.  Consequently, this method of testing is comprehensive and offers the advantages of computer automation, especially repeatability and objectivity. Even after a basic product is completed and approved, there will be a continuing need for testing of upgraded models and new products.  Throughout the manufacturing process, quality assurance testing is needed.  More thorough testing of circuit breakers means a better, more reliable and safer product for consumers and gives the manufacturer greater confidence knowing that the product has been tested with state-of-the-art techniques.