A1:    (pins 1, 2, 3, 4, 6, 34, 35)

These instructions assume that the amplifier is being used in the inverting mode. It can be run as a non-inverting amplifier, but we have chosen to illustrate procedures with the inverting mode, as that is the more commonly used configuration.

Select the A1 feedback resistor (R17) to give the required gain.

Connect pin 6, or pin 34, to pin 32. (A1 to A2) This is already connected on the PC17D-2 test board. For AC coupling, connect through an appropriate capacitor. (See above) It will also be necessary to provide base drive for the positive input to A2 when capacitive coupling is used. When connecting capacitively, connect a 200Ω resistor from pin 32 to ground.

Put a resistor (R1) of the same value as the output impedance of your source from pin 2 (the positive input) to ground, and put a 1 nF cap (C1) in parallel, to cut the noise.

Ignore pins 3 and 4. They are unnecessary for normal operation.

A1 is very stable. With a 390-ohm feedback resistor, it requires no compensation on the PC17D-2 board. If it is oscillating with no compensation or cap across the feedback, proceed as follows.  Add 1 or 2 pF from the compensation pin (pin 1) to ground (C4). To remove overshoots, do not increase the compensation capacitance or you will slow down the slew rate unnecessarily. Instead, use a cap across the feedback capacitance (C2). The value you will need depends on the gain you set on A1 and the input capacitance of your setup.  With the test board, we used a 1 pF capacitor and the A1 output pulse was perfect with a 3 nS rise time. Another feedback combination that works well is to use a 690Ω resistor, R3 in series with a 3.3pf capacitor, C3. The comp cap was not needed.  For a 20 MHz setup, you may want to slow A1 down a little by using 3pF or 4 pF to compensate. For applications where it is desirable to minimize the transit time difference between small and large input pulses, it may prove useful to add 4 or 5pf to the compensation as A1 will retain most of its small signal bandwidth, but large pulses will have a longer transit time because of the decreased slew rate. This compensates for the fact that the higher voltage parts of the larger pulses do not have to go through A2 and A3, and thus minimizes the transit time differences. The above discussion applies to the 50% input time to 50% output transit time.

Run a large resistor (R4) from the A1 positive input to VCC. We start with the following values, and put it in before switching on +/-6V: 82K to VCC, +/-8V: 135 K to VCC. The board is set up to make this correction on the positive input. It is also a good idea to put in a zero order temperature compensation when assembling the unit. See the section on temperature compensation below.

The pin 6 output can drive a 50-ohm load, but the pin 34 output current should be limited to 15 ma, as this output trace is not sized for heavy currents. It is convenient to use for the feedback, as it cuts the length of the traces. In general the input and feedback leads for A1 should be kept as short as possible, or they should be shielded by running them between power, or ground, planes. The input transistors to A1 are extremely fast and if these lines are exposed and there is any very high frequency oscillation on the board (>1 Ghz), you will not see it affect pulse shapes, but is may be rectified by the input transistors acting as square law detectors, and this can cause DC drifts. For this reason, we strongly recommend using pin 34 for the feedback output. The Anadyne board is set up to use pin 34.