Modern pulse/pattern generators give users a lot of control over the signals they create. Going beyond basic settings will allow designers and test engineers to better match outputs to application needs and improve test results.
By Todd Stocker
Basic configuration settings of pulse/pattern generators include amplitude, pulse width, rise and fall times, frequency, and delays for the voltage and current outputs used in testing a variety of devices. However, the channel-add and pattern-mode functions in dual-channel instruments allow users to create composite signals that go beyond simple square waves. These multifaceted signals are often required to test complex devices, such as semiconductors, memory cells, and serial data devices.
Impedance effects on pulse output
The foundation for using pulse generators to create complex outputs is an understanding how pulse signals are produced. A fundamental issue is the source and device under test (DUT) impedances and their effects on pulse output. For example, understanding impedance effects allows a pulse generator to be used in applications where the device acknowledge function must be tested.
The voltage output of a pulse generator (PG) depends not only on the output voltage setting, but also on the PG’s internal source impedance, and the impedance of the DUT. Usually, a PG’s default source impedance is 50 ohms, but this is selectable by the user (50 ohms or 1 kilo-ohm). The user can also adjust the PG’s LoadZ setting, i.e., a best estimate of the DUT’s impedance. For the actual output voltage to be the same as the set voltage, the LoadZ setting must be the same as the actual DUT impedance. If these two are different, the actual output voltage will not be the same as the set voltage.
The PG can be modeled as a current source, ISRC, in parallel with a source impedance. If a user sets the output to 10 volts and LoadZ to 50 ohms, then the PG calculates a load line according to Ohm’s Law. For 10 volts across two 50 ohms resistances in parallel, I = 400 milliamps. With the source current set based on the source impedance, LoadZ, and output amplitude settings, the only thing that will affect the actual voltage across the DUT will be its actual impedance. If the actual DUT impedance is different, or changes, the actual voltage output will move along the load line.
Device acknowledge applications
Besides achieving the pulse amplitude one wants across the DUT, impedance effects can be used to check the performance of a device that is designed to receive and respond to a serial data stream. For example, the serial stream might include an address byte, instruction byte, and a data byte. The DUT is designed to acknowledge receipt of each of those byte segments. This acknowledgement is achieved by the device pulling down the line. In actuality, the way a device pulls down a line is to change its impedance state.
To test this performance, a complex external circuit or switching routine is often used. Yet, a pulse generator can be used to simplify this test. The PG, configured with 50 ohms output impedance, could send a set of pulses to the DUT, and the voltage across the device would be monitored with an oscilloscope. If, for instance, the DUT’s normal impedance is 1 kilo-ohm, and it drops to 50 ohms to acknowledge receipt of the pulse train, this will pull down the pulse voltage amplitude. The lower amplitude pulses are a clear indication that the acknowledge event has occurred.
Generating complex outputs
Most dual-channel PGs have channel-add and pattern-mode features that allow the creation of complex waveforms. With the channel-add feature, the PG does a point-by-point addition of the amplitude of the pulse trains on the two channels, and the composite waveform is placed on the Channel 1 output port. Of course, the amplitudes, pulse widths, rise times, and polarities of the two channels could be different, and a delay between the start times of the two pulse streams could be set. Using these features, complex wave shapes can be created by taking advantage of the overlapping outputs.
Testing memory cells is an example of where these features could be used. In a typical write/erase cycle, a positive pulse is used for writing and a negative pulse for erasing. Arranging the shapes of channel 1 and channel 2 would allow a user to change the output wave shape to test other memory cell technologies, such as PRAM.
Pattern mode is normally used for serial data simulation, but it is also useful in creating complex outputs. When pattern-mode is used, the PG creates a series of “0” and “1” bits to generate the waveform. In NRZ (non-return to zero) mode, the pulse does not return to the low level after the pulse width is complete. In this mode the pulse will remain high until the next bit. If this bit is again high, the output will continue to be high. The output will only return to the low level once a zero bit is encountered.
By lining up a series of “1” bits, an increasing pulse width sweep can be created. For even more complex output waveshapes, both the PG’s pattern-mode and channel-add functions can be used to create a wider range of output combinations. Patterns can be overlapped or delayed to achieve the desired output. An example is shown in Figure 6 for a device that is controlled from a data communications line.
Other features to consider
Today’s pulse/pattern generators have a wide range of capabilities for creating and using complex waveforms to test sophisticated electronic devices. In addition to the functions described earlier, there are a number of other features and specifications that should be considered: pulse frequency range; amplitude range and accuracy (flatness); range of pulse widths and rise/fall times; availability of a burst mode; voltage overshoot, undershoot, and droop; data communication interfaces available; speed and ease of set-up; and intuitive user interface. Such features provide the flexibility for a broad array of test functions and applications.
Todd Stocker is a marketing manager at Keithley Instruments Inc. in Cleveland.