7. RECOVERY TIME
7.1
GENERAL COMMENTS ON RECOVERY TIME
Following long high-power input pulses, the output voltage will, in general, show a long low-level
tail. Such tails appear to be caused by thermal effects which predominantly originate in the
detector. Therefore, once corrective recovery networks are defined for a given detector and
amplifier configuration, the values of the network components for subsequent units of this same
type will be unchanged, as long as the detectors used are reasonably matched. Typically, the tail
will increase in length as the input pulse-length is increased, up to about 10 microseconds, beyond
which the tail retains a constant shape. A second pulse which follows the first will not be
accurately measured if it arrives while there is still a significant tail. We define recovery time as
the elapsed time, following the end of any pulse, after which the next pulse will be measured to
within
x dB of its true value. We normally specify the value of x to be one, and also specify that
the second pulse follows an input pulse of the maximum allowed power. Caution should be
exercised in interpreting advertised DLVA recovery-times, since many manufacturers are vague in
their definition of the critical recovery parameters. A second obvious effect of tails is to cause a
shift of the baseline for high-duty-cycle pulses.
The normal way to remove these tails
(5)
is to deliberately use a net which, were it not for the tail,
would produce an undershoot
(6)
after a large pulse. The method used is similar to that described
for removing overshoots in section 4. Notice, however, that because of the nature of the log
transfer function, the corresponding overshoot at the beginning of the pulse will not be observable,
provided the compensating undershoot is produced in A1 or A2. Put simply, there is not enough
residual gain at the top of a large pulse for the overshoot effect to be seen.
-----------------------------------------------------------------------------------------------------------------------
(5)
This method really only works if the thermal tail is linear with pulse input power. Above
about + 5 dBm, tunnel diodes produce highly non-linear tails which are impossible to remove with
linear circuitry. Putting a 15-ohm resistor in series with the detector helps somewhat at very high
powers, but the problem is still severe. The L-17C does have a feature, discussed in section 7.3.1
below, which aids in the removal of high-power tails without causing an undershoot at lower
power.
(6)
It is essential that any induced undershoot not be too large, or the clean appearance of the
pulse in the tail region will be an illusion. This illusion arises from the fact that the output amplifier
will limit at small negative levels, and therefore a large undershoot will not be directly observable
because of this limiting. However, the output is still affected, and the measured amplitude of a
small pulse following a large one will be much smaller than its proper value if the output is limiting
negative.
-20-
 |
 |
 |
 |
 |