As discussed earlier, if you want to minimize the transit time spread as a function of input pulse size, you can add capacitance to the A1 compensation. This will lower the slew rate of A1, and cause the larger pulses to come later, thereby lessening the transit-time spread. We have achieved transit times (50% to 50%) from 8.2 nS for large pulses (>8db above start of logging) to 9.4 nS for pulses at the start of logging. We are still working on optimizing the high-speed performance, and will send out a note as soon as we fully define the optimization procedure. We are currently limited in our test capability by our HP8082A pulse generator. While the output pulse may appear perfect when observed on a scope, we have found that there are small imperfections at the 0.1% level which is unacceptable for accurate characterization of the chip which has a 4 decade range. We are currently working on a high speed Schottky bridge which should provide a settling time to .01%, which is fast enough for us to characterize the chip at the highest frequencies.Temperature compensation: Please note that A1 has a much larger drift than A1 had on the L17-C. However this drift is linear and easily corrected. However, when the drift is large, it is essential to make a first order correction based on the observed drift of A1 using a small temperature range around room temperature. It is sufficient to go between room temp minus 10C to room temp +10C. The reason for this is that when there is a large change in the output voltage, second order effects come in they will make the output vs. temperature curve look non linear. Once the first order correction is in place, the range may be extended. We recommend loading the first order compensation that we give below when you assemble the board. Another feature that may be useful for temperature compensation procedures is the ability to measure the on chip temperature. This can be done by measuring the voltage difference between the auxiliary output and the main output, pins 17 and 18. Pin 17 is the emitter of an emitter follower that drives the base of the output transistor. This means that you are measuring a diode drop. However, the value will only measure the temperature change accurately if there is no change in the output transistor current. When using this method there should be no load on the output. The voltage dependence of this voltage difference is linear over a large temperature range (-55 to +85C).The temperature correction procedure used with the L-17C cannot be used without a DC correction pin on A1. Since the DC correction pin on the L-17D is suspect, another temperature correction method needs to be used, as the DC offset correction is made by offsetting the front end amplifier. With the L-17C we utilized the fact that the temperature drift in a transistor depends on the current flowing through the transistor. We applied a voltage offset to the input pair and corrected the resulting output offset with the DC correction resistor. Since we are using the input offset to remove the output offset we utilize another method of temperature compensation. Some of our users have used this method in the past, as they feel it is simpler. |
|||||
28 February 20, 2008 |
|||||
|
