Opamp stability using non-invasive methods Opamps are frequently use in instrumentation systems as unity gain analog buffers, voltage reference buffers and ADC input buffers as well as low gain preamplifiers. The stability of these opamp circuits is critical in order to obtain the necessary performance. The assessment is often difficult as the PCB and interconnected devices can play a significant role in the stability, requiring the measurement to be made in-circuit. The provisions for such in-circuit stability assessment are not often provided and if the provisions do exist the injection circuit will likely alter the measurement results. This article presents a method of assessing the opamp stability from a closed loop, in-circuit measurement using a low cost VNA and a wideband DC block. The opamp used for this evaluation is a TLC071C 10MHz wideband device, shown as part of a Picotest demonstration board. The method described here is applicable to all opamp circuits, limited only by the bandwidth of the VNA and the inclusion of the stability assessment software. The opamp circuit Figure 1 shows a schematic of the opamp circuit, which includes a set of probe pads (H201) for a 1- port probe and a 2 position dipswitch (S201) used to connect additional load capacitors directly to the output of the opamp. Switch position 1 is used to connect an additional 22pF to the opamp output while switch position 2 is used to connect an additional 68pF capacitor to the opamp output. Both switches can be used together to add 90pF of load capacitance to the opamp.
Figure 1 Schematic of the opamp circuit used to demonstrate the stability assessment The 1-port measurement setup A picture of the measurement setup, along with a close-up of the opamp circuit on the demonstration board is shown in Figure 2.
Figure 2 The Bode 100 is connected to a 1-port probe via a Picotest J2130A DC Bias injector, used as a wideband DC block. Baumi, you could put a connection diagram here, but I tried to hold the figure count down for you. The OMICRON Lab Bode 100 OUTPUT is connected to a PICOTEST J2130A DC Bias injector, used as a 100Hz DC block. This DC block is necessary to isolate the 50Ω VNA impedance from the 2.5V output of the opamp. The output of the DC Block is connected to a 1-port probe, which is inserted into the probe pads (H201). In the absence of such pads, the probe can be placed directly on the output capacitors or the output pin of the opamp. The instrument is set to perform a 1-port impedance measurement, providing plots for the impedance magnitude using a log display and the Q determined from group delay. These are both selected from the display format on the right panel of the Bode Analyzer Suite software. The measurement is calibrated using the traditional OPEN, SHORT, LOAD scheme common to all VNA instruments. The detailed instructions for calibrating the OMICRON Lab Bode 100 can be found in section 8.4 of the the users manual (Calibration in the Impedance/Reflection Mode).
For convenience the Picotest demonstration board includes an OPEN, SHORT, LOAD calibrator, though any appropriate calibrator can be used. Once the calibration is complete the impedance measurements can be performed. Making Measurmements The Bode 100 is first formatted to display capacitance directly (format Cs) and the capacitance is measured at H201 with the demonstration board unpowered in all combinations of switch settings. The results are shown in Figure 3. Not that the measurment indicates 30pF with both switches off. This is mostly due to the opamp itself, as removing the opamp resulted in a reading of 8pF. Figure 3 The capacitance measured at the probe pads (H201) with the demonstration board unpowered and all combinations of switch settings. The power is then applied to the demonstration board and the impedance measurement is formatted for magnitude while the Y-scale is set to log display. A second trace records the closed loop Q, derived from the group delay of the impedance measurement. The Bode Analyzer Suite includes Picotest proprietary software that uses these two traces to determine the stability of the opamp circuit using a cursor measurement. The impedance measurements are shown in Figure 4 for three different switch settings. With both load capacitors connected (90pF) in addition to 8pF from the circuit board capacitance the stability is very poor with an equivalent of 2.4 degree phase margin. The equivalent phase margin for the 68pF load capacitor is 7 degrees and for the 22pF capacitor it is 17 degrees.
Figure 4 Impedance and closed loop Q measurements for three switch positions along with the extracted stability information. The results of the assessment show that the stability of the opamp at unity gain is poor with all three load capacitors, with representitive phase margins of 17 deg, 7 deg and 2.4 deg for the 22pF, 68pF and 90pF load capacitors respectively. The poor stability will result in high noise, significant peaking and possible ringing in the vicinity of 12MHz. Improving opamp circuit stability As a rule it is best to keep the phase margin of these sensitive opamp circuits above 60 degrees. There are several ways to improve the stability of these opamp circuits. 1. Use opamps with better inherent unity gain stability, which in general means opamps designed to operate at unity gain and generally for sensitive circuits. 2. Insert a resistor between the opamp output and the load capacitor (often referred to as a null resistor) though this will generally degrade the DC performance and bandwidth of the opamp circuit
3. Add a resistor in series with the load capacitors if possible to improve stability 4. Operate the opamp at a higher gain, where the amplifier is more stable. Conclusion We have demonstrated a method of assessing the stability of opamps from a closed loop impedance measurement using a low cost VNA and a DC block. This method is applicable to many other types of circuits and also higher bandwidth opamps, limited by the bandwidth of the VNA and the inclusion of the proprietary stability assessment software.