Tektronix 11A32 User's Reference Manual page 27

Operating the 11A32 in 11400-series Mainframes
Overdrive occurs when any 11A32 channel is driven out of its linear range of
approximately ±15 divisions.
The 11A32 has extraordinarily good overdrive recovery, and this feature may be
used to greatly extend measurement resolution. For example, suppose a signal
changes from -1.7 V to +0.8 V in 1 ns. The 11A32 could be used to determine if
the signal stabilized immediately at +0.8 V or if perhaps had some small
aberration following the transition. By setting the 11A32 offset to +0.8 V and the
sensitivity to 1 mV/di vision, aberrations of just 0.1% of the original transition
will be 2.5 divisions in amplitude (0.1% of 2.5 V is 2.5 mV or 2.5 divisions at
1 mV/division).
Any amplifier will ultimately reach an equilibrium value after an input step
(although its accuracy will determine how far that equilibrium value is from
the correct value). The HA32/s ability to settle quickly to within a very small
fraction of its equilibrium value is exceptional. The time it takes the 11A32 to
settle to within a stated fraction of the equilibrium value is its overdrive
recovery time.
Measuring the overdrive recovery time of an 11A32 takes some care and can
lead to some surprising results. An interesting experiment is to use a very flat
pulse generator, such as the Tektronix PG 506, to pulse the 11A32. Connect the
generator's fast-rise output to the 11A32 input through a short (one foot or less)
coaxial cable and select 50 Q input impedance. Adjust the pulse amplitude to
1 volt. Set the 11A32 sensitivity to 1 mV/division. Each division now represents
0.1% of the input signal. Trigger the oscilloscope and observe the recovery of the
11A32 using 50 ns/division sweep rate. Now increase the cable length by about
three feet (for example, use a 42-inch length of RG-58 cable, Tektronix part
012-0057-01) and observe the new waveform. See Figure 3-2.
The waveform change is due to skin effect loss in the longer cable. What is
surprising is that the skin effect loss persists for over 200 ns even though the
total cable delay is only 5 ns. At 400 ns the loss is still 0.02%.
This experiment shows the importance of using a short cable to test overdrive
recovery.
A second surprise is that skin effect loss disappears almost completely (after
two .cabledelays) when one end is unterminated.
1
" " —
To observe this, use the short cable again, and select 1 MQ input impedance on
the 11A32. Insert a 2X attenuator between the cable and the pulse generator to
improve the reverse termination and to provide the same amplitude signal as
before. Observe the response and change cables again. Even at 0.1% per division
the skin effect loss is hard to detect without the forward termination. The reason
for this is that the skin effect loss is an increase in the effective series resistance
of the cable. Without current, the cable develops no series voltage drop.
See Figure 3-3.
11A32 User Reference Supplement
3-5