3
\$\begingroup\$

enter image description hereI am using a TSZ121 op-amp with a unity-gain configuration driving 100 pF via 20 Ω at the output. Power supply is 5 V. My input is a slow moving signal between 0.4 and 1 V from a temperature sensor.

I noticed a continuous 20 kHz oscillation at the output with a 200 mV amplitude. According to the datasheet, my circuit should be stable. Once I replaced the 20 Ω resistor with 1 kΩ the oscillation disappeared. I suppose this is related to instability. I always thought that oscillations should be rail to rail when not stable. Can it be just 200 mV?

\$\endgroup\$
4
  • 1
    \$\begingroup\$ If you could add a photograph of your circuit to the question that might help forestall a large number of questions people may ask about layout, test setup, etc. A schematic would also be appreciated. \$\endgroup\$
    – user57037
    Oct 7, 2022 at 16:58
  • 3
    \$\begingroup\$ Is a bypass capacitor used directly across the TSZ121's power pins? Try bending and touching a 0.1µF polypropylene through-hole capacitor to it's power pins. \$\endgroup\$
    – rdtsc
    Oct 7, 2022 at 19:06
  • \$\begingroup\$ It is generally good practice to balance the impedances of the inverting and NI inputs of an op-amp, so perhaps 300k in series with pin 4 may help. You might also try connecting the negative feedback directly to the capacitor, although that might make it worse. Proper bypass capacitors, of course, should always be used, and might solve the problem. \$\endgroup\$
    – PStechPaul
    Oct 7, 2022 at 20:52
  • \$\begingroup\$ Is it a circuit board or did you implement this with some type of non-solder bread board? Sometimes there can be parasitic capacitance depending on how you implement the circuit. If there is parasitic capacitance between R2 and the op-amp that could maybe be part of the problem. \$\endgroup\$
    – user57037
    Oct 8, 2022 at 0:22

3 Answers 3

Reset to default
5
\$\begingroup\$

See note 5.9 on page 17 of the data sheet. Because this device is a chopper based amplifier there is a high frequency switching circuit internal to the device. The 20khz may be a sub-harmonic or beat frequency from the chopper system. Because of this the note stresses the importance of placing bypass capacitors as close as possible to the supply pins. Per the note: "Due to the modulation of the chopper, the decoupling capacitance also helps to reject the small ripple that may appear on the output."

Alternately if your system does not required an ultra low noise output you might try using a device without chopper stabilization.

\$\endgroup\$
2
  • \$\begingroup\$ Thanks for your reply. But when I increase R2 to 1k, the op amp output (pin 1) becomes much cleaner. Is it still possibly due to the chopper architecture? Thanks again. \$\endgroup\$
    – jsmith0910
    Oct 9, 2022 at 17:48
  • \$\begingroup\$ Adding or removing any decoupling caps on the supply pins may help determine what the true cause of the noise is, (chopper noise or instability). \$\endgroup\$
    – Nedd
    Oct 10, 2022 at 7:01
1
\$\begingroup\$

A small (limited amplitude) high frequency oscillation perhaps riding on a lower frequency waveform can be caused by too low (but still positive) a phase margin. This would be described as oscillatory rather than unstable. Instability results in an ever increasing output and is a result of zero or negative phase margin. Obviously this "ever increasing output" would, in reality, be limited by the power rails.

The load capacitance at the output can act with the amplifier's output resistance to put an extra pole in the loop and reduce phase margin. By adding a correctly sized isolation resistor outside the loop at the output it's possible to add some loop phase lead which will increase phase margin and put the amplifier further from instability.

You could also improve phase margin by increasing the closed loop gain to a higher value (less feedback). Unity non-inverting gain is the least stable gain configuration.

20 kHz seems quite a low frequency oscillation to be related to insufficient phase margin.

\$\endgroup\$
0
\$\begingroup\$

There is a section on capacitative loading in the linked data sheet section 5.7.

Since the frequency is fairly high relative to the gain bandwidth of the amplifier the amplitude of oscillation cannot be rail to rail although it does seem to be les than expected by the available slew rate.

Rail to rail output opamps are always going to be rather sensitive to capacitance loading as the output stage is configured as common source/emitter devices giving a high-impedance output. The more conventional source/emitter follower has a much low output impedance before feedback is applied.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.