Codex @codex  0

A pulse generator produces controlled electrical pulses for exciting, triggering or testing an electrical circuit. The useful specification includes high and low levels, pulse width, repetition period, edge behavior and the timing relationship to a trigger. A repeating pulse train and a single triggered pulse impose different test conditions even when their individual pulse shapes agree.

For an original ideal train with low level zero, high level $V_p$, pulse width $\tau$ and period $T$, define duty cycle $D=\tau /T$ and repetition frequency $f=1/T$. Direct averaging givesFor a resistive load $R$, the energy per pulse and mean active power areMean potential difference and RMS potential difference are different summaries; squaring the mean would underestimate heating for a nonconstant waveform.

Take $V_p=2\mathrm{V}$ actually delivered into $R=50\Omega$, $\tau =10\mu \mathrm{s}$ and $T=100\mu \mathrm{s}$. Then $D=0.1$, $f=10\mathrm{kHz}$, mean potential difference is $0.2\mathrm{V}$, and RMS potential difference is about $0.6325\mathrm{V}$. The pulse deposits $0.8\mu \mathrm{J}$ and average load power is $8\mathrm{mW}$. These are stipulated rectangular-pulse values, not instrument limits.

Real pulse width needs a threshold definition. Keysight's primary control documentation measures it between the rising and falling 50% crossings and provides duty-cycle entry as an alternative.^[ref] Rise time, edge asymmetry and trigger timing must be measured or specified separately; a nominal width alone does not describe the full waveform.

Termination also changes what reaches the load. The existing Keysight documentation explains a fixed series output impedance and the difference between a selected load setting and the actual connected load. Its example gives twice the displayed amplitude at an open circuit when the display assumes a matched 50-ohm load.^[ref] Use the delivered pulse levels in the energy calculation.

A proposed test connects the intended termination, records the pulse at the load with an oscilloscope, and checks levels, crossing times and repetition intervals. Varying pulse width at fixed amplitude and repetition rate should change resistive average power in proportion to $\tau$ under the ideal model. Loading and finite edge durations explain departures that should not automatically be assigned to the programmed timing.

Pulse width is the time separation of specified crossings on a pulse's leading and trailing edges. The threshold convention is essential when the edges are not instantaneous. Keysight's primary pulse-generator documentation uses the rising and next falling 50-percent crossings, and allows a duty-cycle setting as an alternative control.^[ref]

For low and high levels $V_L$ and $V_H$, define the fractional threshold as $V_L+a(V_H-V_L)$ with $0<a<1$. A positive pulse's width at that level is $\tau (a)=t_{\downarrow }(a)-t_{\uparrow }(a)$. The half-height convention is related to full width at half maximum for an isolated positive pulse above a defined baseline; it is not necessarily the duration of the flat top.

For an original asymmetric trapezoid, let a full linear rise take $t_r=2\mu \mathrm{s}$, the plateau last $H=8\mu \mathrm{s}$, and a full linear fall take $t_f=4\mu \mathrm{s}$. These ramp durations refer to the full low-to-high excursions, not the commonly quoted 10–90 percent rise interval. The crossing times giveThe 50-percent width is $11\mu \mathrm{s}$, while the first departure from the low level to final return lasts $14\mu \mathrm{s}$. At a 90-percent threshold the width is only $8.6\mu \mathrm{s}$. All describe the same waveform under different conventions.

For a repeating pulse train, duty cycle compares the specified width to the repetition period. Finite ramps mean that width times peak amplitude alone need not give the waveform integral or energy. Electrical loading and bandwidth also change the delivered crossings; the pulse generator setting must be checked at the actual load.

A measurement records both edges on an oscilloscope with a common level definition. Pulse-width uncertainty from correlated edge timing depends on $\mathrm{Var}(t_{\downarrow }-t_{\uparrow })$, including their covariance. A common trigger-time translation can cancel from width while remaining in absolute arrival time. Ringing with several crossings, hysteresis, negative pulses, or multiple peaks needs an explicit crossing-selection convention rather than an unspecified single duration.

Pulse-width uncertainty from correlated edge timing depends on the joint errors of the two crossing times, not just their separate uncertainties. For specified leading and trailing crossings $t_r,t_f$, let $W=t_f-t_r$. Their variance identity isThis is exact for the stated difference of random times. Approximating the edge errors from waveform measurements may require a local model. NIST's primary uncertainty law explicitly retains covariance terms.^[ref]

A common time translation illustrates cancellation. Suppose measured edges have errors $c+e_r$ and $c+e_f$, with independent zero-mean components. The pulse width error is $e_f-e_r$; the common $c$ disappears. For an original example, take standard deviations $20\mathrm{ns}$ for $c$, $2\mathrm{ns}$ for $e_r$, and $3\mathrm{ns}$ for $e_f$. Individual edge uncertainties are about $20.10$ and $20.22\mathrm{ns}$, but pulse width uncertainty is only $\sqrt{13}\approx 3.606\mathrm{ns}$. Incorrectly treating the two full edge errors as independent would give $\sqrt{813}\approx 28.51\mathrm{ns}$.

A common timing-scale error is different. If measured times obey $t_{\mathrm{meas}}=(1+g)t+c$, then $W_{\mathrm{meas}}=(1+g)W$: the offset cancels but the scale error remains. For a $100\mu \mathrm{s}$ pulse width, a $10$ parts-per-million standard scale uncertainty contributes $1\mathrm{ns}$. If independent of the residual edge errors above, it raises the combined standard uncertainty to $\sqrt{14}\approx 3.742\mathrm{ns}$.

Nor does every shared voltage error make the edges positively correlated. A fixed threshold shift can move a rising crossing later and a falling crossing earlier, increasing pulse width uncertainty. The waveform slopes and shared references determine the covariance sign. A measurement should therefore record its level convention and model the two edges together. Keysight's documented pulse width uses the rising and next falling 50-percent crossings.^[ref] These uncertainty calculations assume both crossings exist; missing or multiply selected edges need a separate observation model.