OPA145: Charge amplifier for piezo sensors - weird output at low gain

Hi Duraj,

Operational amplifiers are sensitive to capacitances placed right at the inverting input terminals of the op-amp device, and sensitive to large capacitive loads as well.  There is a video series Tutorial on TI Precision Labs that discuss op-amp stability, stability analysis and compensation. See below:

https://www.ti.com/video/series/precision-labs/ti-precision-labs-op-amps.html

Under Topics, search browse to "Stability" and there are seven sessions discussing op-amp stability.

To answer your question, on the circuit above, the culprit causing instability is the Cpiezo=40nF capacitance.   The output load capacitance, Cout, is in series with a large Rout 100k resistor, therefore, the output load is not causing stability issues.  The OPA2991 (4.5-MHz) and OPA145 (5.5-MHz) amplifiers have similar bandwidth, but the open-loop output impedance characteristic over frequency is different.

The OPA2991 and OPA145 models both simulate accurately AOL, and open-loop output impedance to allow accurate stability analysis.  On the original circuit, with Cpiezo=40nF, CF=40nF capacitance, actually, both circuits show sensitivity to the Cpiezo = 40nF input capacitance: The OPA2991 circuit is at the edge of becoming un-stable, with very marginal or low phase margin of only 9-degrees, while the OPA145 is completely unstable with no phase margin.   

For example, if we perform open-loop stability analysis of the OPA145 circuit, with CF=40nF, and Cpiezo = 40nF, the open-loop stability analysis shows negative phase margin of -11.8-degrees, meaning the circuit is unstable.  If we take a look at the simulation results, the open-loop output impedance of the OPA145 interacts with the feedback components, and the Cpiezo input capacitance, creating a second pole on the loaded AOL around ~120-kHz, creating the instability.  Due to this second pole on the loaded AOL, the intersection of the loaded AOL and 1/Beta have a -40dB rate of closure, making this circuit unstable. This unstable result correlates with your OPA145 transient simulation results showing oscillation/instability.

TINA simulation: 7026.OPA145_forum_10-17-23_open-loop_stability.TSC

Similarly, when we perform open-loop stability analysis of the OPA2991circuit, with CF=40nF, and Cpiezo = 40nF, shows only marginal 9-degrees phase margin, meaning the circuit design is not robust, the OPA2991 circuit is also marginal or at the edge of instability: 

TINA file: 7026.OPA2991_forum_10-17-23_open-loop_stability.TSC

There may be different ways to compensate the circuit.  However, in this case, since this is an integrator circuit that requires CF=40nF and the piezoelectric capacitance is 40nF, you are correct that adding the 100-ohm series at the op-amp inverting terminal stabilizes the circuit.  The 100-ohm adds a zero on the Loaded AOL frequency response cancelling the second pole that was causing the instability, where the Loop-Gain and AOL Rate-of-Closure where the AOL and1/Beta curves intersect is now around -20-dB/decade, making the circuit stable. The second circuit you have suggested with the 100Ω is stable with 75+ degrees of phase margin; and remains stable as you change the feedback capacitor.  See below:

TINA file: 1641.OPA145_forum_10-17-23_100ohm_open-loop_stability.TSC

Thank you and Regards,

Luis