The approximate values needed for RT1 can be calculated by noting that a 1 mV offset of the +
input to A1 (pin 2) will give a shift of 1/5 mV over 65
0
C, referred to the input of A1. As an example,
if the gain of A1 is set at 10, the overall low-power gain will approximate 10 x 4 x 16 x 6.5 = 4160
(for an output slope of 50 mV/dB). An offset of 0.2 mV at pin 2 will therefore produce an output
shift of 0.2/5 mV x 4160 = 166 mV, or about 3.3 dB, as a result of a 65
0
C temperature change.
The offset procedure is the same as that used when compensating A2, i.e. re-zero A1 after adding
RT1.
8.3.5
Non-Linear Temperature-Induced Shifts:
A great deal of effort went into ensuring that
temperature-induced changes in the gain or baseline of the L-17C be linear in nature, so that they
can be easily corrected. Non-linearities which we have observed have been traced to poor
regulators, which drift excessively. We have, however, noticed that the previously-discussed
temperature-induced translations of the transfer functions at high power tend to be slightly
asymmetric, with the -55
0
C curve being displaced upward from the room-temperature curve by
slightly less than the +85
o
C curve is displaced downward. This is rarely a problem, since the
effect is small. However, if the best possible alignment over temperature is desired, there is a
procedure which can be used to correct any such slight asymmetry by adjusting the value of R17,
as discussed below.
We have found by measurement that altering the value of R17 from its normal match to the output
amplifier feedback resistance will shift the +85
0
C and -55
0
C transfer curves asymmetrically with
respect to the room-temperature transfer curve. For example, at a transfer slope of 50 mV/dB, the
output amplifier feedback resistance will be about 2200 ohms. The "normal" value of R17 would
then also be 2200 ohms. We found that changing R17 from 2200 ohms to 1600 ohms displaced
the +85
0
C curve downward by 72 mV and the -55
0
C transfer curve upward by 120 mV. This effect
may be used to correct any observed asymmetry in the hot and cold transfer function shifts prior to
implementing the temperature compensation procedures of sub- sections 8.3.1 and 8.3.2.
The magnitude and direction of this asymmetric shift depends on whether R17 is increased or
decreased, and by how much, and also on the output slope. We have made measurements for
various values of R17 at several output slopes, and this information is discussed in section 8.3.6.
Users who require the best possible transfer matching over temperature may experiment with this
approach. It is possible to make the required measurements with both R17 and the output amp
feedback resistor net located outside the temperature chamber, if a slowdown net is used to kill
any oscillation which may result. This procedure greatly speeds up the measuring process, since
it is then not necessary to open the temperature chamber between measurements.
8.3.6
A Simplified Temperature Compensation Procedure:
This same effect of R17 can be
used as described below to achieve an approximate but quite good level of temperature
compensation. This is done without using the procedures of sub-sections 8.3.1 and 8.3.2 (which
utilize the components RT3, S2, and RT4). The need for these components is thereby eliminated.
-30-
 |
 |
 |
 |
 |