DigitAl sAmpling oscilloscopes (DSOs) are used to perform picosecond pulse measurements. For these oscilloscopes, three major uncertainties exist: impulse response, vertical noise, and timing jitter. In an application note titled, "Removing Jitter From Picosecond Pulse Measurements," Picosecond Pulse Labs discusses jitter in detail. It also describes a couple of methods that may be used to remove jitter from oscilloscope measurements.

The note begins by detailing the oscilloscope's transient response, which can be characterized as either its impulse response, h(t), or its step response. An oscilloscope manufacturer usually only specifies the step response as the 10-to-90- percent rise-time transition and/or the -3-dB bandwidth. Yet the designer can make more precise measurements if he or she knows the exact nature of the oscilloscope's impulse response. With deconvolution, the scope's impulse-response waveform distorting effects can be removed from a measurement. The result is a more accurate picture of the true pulse waveform.

Because an oscilloscope is essentially a timeresolved voltmeter, its ability to measure voltage is limited by its noise floor. For an instrument like a sampling oscilloscope, it will see more white noise if it has a wider bandwidth. As a result, it will have a higher noise floor. With signal averaging of measured sampled voltages, one can considerably improve the noise floor of an equivalent time, sequential sampling oscilloscope. The key is that such oscilloscopes fundamentally require that the waveform be measured on a recurring, unchanging waveform rather than a single transient. If the engineer is willing to use a longer data-acquisition time, many reoccurrences of the same signal can be measured and averaged.

For its part, jitter deserves special consideration when observing very fast pulse edges. If its width is comparable to the pulse rise or fall time, it becomes quite difficult to make accurate waveform measurements. The signal-averaging feature can provide a step pulse waveform that appears very clean even though it may suffer from rise-time distortion. The author of this application note, Picosecond Pulse Labs Founder and former President James R. Andrews, has written several MatLab programs to generate and analyze pulse waveforms with various controlled amounts of vertical noise and horizontal timing jitter. The note ends by explaining these jitter-deconvolution tools and showing how they can provide nearideal reconstruction when jitter is much less than the fastest transient in the signal.

Picosecond Pulse Labs, P.O. Box 44, Boulder, CO 80306; (303) 443-1249, FAX: (303) 447-2236, internet: