Noise from an OPA552 opamp.

[Hank314]

I have an opamp circuit with the two resistor configuration. I am using the OPA552 with rail voltages of -9V and 41V. The resistors are 30.1K and 10K. See picture. In TINA, I get clean amplification that follows (1 + R1/R2). On the PCB, I get the same amplification with noise. In the attached picture, I am seeing the output of a low frequency sawtooth wave. The "noise" you see in the scope picture is a clean "sine" wave of 1.2 MHz when I stretch it out(See picture). I look at the input (0-8V) and the rail voltages with the scope and they are all clean. I can change the input frequency to DC and the noise is still there. I changed the load resistance by adding 3K ohms in parallel and no change. The PCB is surface mount design and the resistors are close to the opamp. It looks like I need a capacitor across R1 and/or R2. Is that a correct fix or am I missing something ?

 

[Hank314]

I think the problem is that my opa552 does not like low gain.

 

[Harald Kapp]

I am using the OPA552 with rail voltages of -9V and 41V

The datasheet states max. +30 V, not 41 V.

In TINA, I get clean amplification

The simulation very likely does not take into account the overvoltage on the positive supply. Also, simulation assumes ideal components. Real components are not ideal.

The "noise" you see in the scope picture is a clean "sine" wave of 1.2 MHz

The opam,p is obviously self-oscillating. One (or two) obvious missing component is a bypass capacitor from each power supply pin of the opamp to gnd. A typical value is 100 nF (ceramic or foil) from each power supply pin to gnd. This will not change the simulation in any way (other than maybe take a few milliseconds longer), but will greatly reduce noise sensitivity of the real circuit.

It looks like I need a capacitor across R1 and/or R2

Observe chapter 8.2.2.3 in the opamp's datasheet. It shows how to compensate the frequency response of the opamp for low gains (less than 5 as the opamp is designed to be used with gains greater than 5 - your's is only 4).