8.3 TEMPERATURE-COMPENSATING THE L-17C
DLVAs utilizing the L-17C may require a pre-set correction in order to maintain a constant output
slope over a wide temperature range. Experience has shown that this correction is the same for
all chips from a given wafer (about 900 ICs). The correction is made by using the net R20, S1,
and R15, in place of a simple feedback resistor, from pin 18 to pin 19. When performing initial
tests, a single resistor is used in place of this net, and its value is set to give the desired slope at
room temperature. R17 is chosen to be equal to this feedback resistor value. (However, see
sections 8.3.5 and 8.3.6)
If a slope change is seen over temperature, the slope will decrease at high temperature and
increase at low temperature. To correct this effect, refer to Fig. 8.3(a) and choose the value of
R20 from the appropriate curve, in accordance with the value of the feedback resistor used. S1 is
a 500-ohm positive tempco (0.007/
0
C) sensistor, available from Midwest Components. To
calculate the required value of R15, we use the formula:
R15 = R
FEEDBACK
- (500 ohms || R20).
Once these values have been determined, in the case of all chips from the same wafer the value
of the parallel pair is determined, and the value of R15 can be selected accordingly.
Before running tests over temperature, place the unit (including the detector!) in a box, as
discussed in point no. 3 of sub-section 8.1. Connect external leads to the outputs of A1, A2, A3,
the main output and the power lines. Make connections through 10-20k resistors to prevent
oscillations. Also, connect a ground lead directly to the board or box. This is necessary since the
ground current varies with temperature, which will lead to inaccurate readings if, for example, the
meters are grounded to the power-supply ground. In principle, it is only necessary to measure the
output when making temperature- compensation measurements, but having the other leads is
extremely useful for diagnostic purposes in the event of problems.
Temperature-compensation of the the L-17C is easier than for some other log amplifiers, in that
the behavior of the transfer curves over temperature is relatively simple. The curves at high or low
temperature don't oscillate about the room temperature curve. Basically, at high powers the
curves are parallel but offset, and there is an additional offset at the start of logging. Put simply,
you should see transfer curves prior to trimming which are similar to those shown in Fig. 8.3(b),
and not like those shown in Fig. 8.3(c).
The most effective approach to merging the curves of Fig. 8.3(b) is to consider the high power and
low power parts of the transfer curves separately, i.e. to first make a correction to merge the
curves at high power, and then to transpose the curves analytically to take this into account, in
order to calculate the corrections needed for the low-power portion of the transfer curves. The
resulting set of transfer curves should then exhibit satisfactory behavior.
-27-
 |
 |
 |
 |
 |