2
\$\begingroup\$

This is ultimately related my previous question here.

And I use the following circuit:

enter image description here

When I try with this MCP opamp I get a DC output which I want since I want the average:

enter image description here

But when I this OPA opamp I get oscillation superimposed on DC as follows:

enter image description here

I wasn't expecting oscillations because both are unity gain stable.

I first soldered and tested with MCP and then plugged off MCP from the IC socket and plugged in the OPA opamp. I noticed OPA has offset pins so pinout is not identical but the rest pins are same for both IC.

enter image description here

OPA has offset pins and they are floating in my case.

I couldn't figure out why OPA is oscillating but not MCP. If not about the offset pins, what specs could explain the root cause of the oscillation when OPA is used?

\$\endgroup\$
4
  • \$\begingroup\$ You have a capacitive load of about 4 nF, which is enough to destabilize many op-amps, especially at low closed loop gain..Try putting a small resistance,.e.g. 100 ohm in series with Cf. \$\endgroup\$ Commented May 23, 2024 at 18:14
  • \$\begingroup\$ I see but why is one oscillating but not the other? Can we say anything by comparing their BW specs or ect? \$\endgroup\$ Commented May 23, 2024 at 18:15
  • \$\begingroup\$ It depends on many things, open loop output impedance, phase and gain vs freq. curves \$\endgroup\$ Commented May 23, 2024 at 18:16
  • \$\begingroup\$ So the only thing I can to is to test on board? Is it super difficult to predict or estimate the necessary feedback capacitor value for stability by using specs? Maybe someone answers about it, let's see. I know stability theory is complicated but thought there are some rule of thumbs. \$\endgroup\$ Commented May 23, 2024 at 18:18

1 Answer 1

Reset to default
2
\$\begingroup\$

Heavy capacitive loading on the op-amp can cause instability. More than 50-100pF is 'heavy' at G=1 and most op-amp datasheets will show you that. They don't tend to explicitly tell you things like open-loop output resistance.

For example, read the capacitive loading on your OPA datasheet:

enter image description here

It looks like you could significantly change the behaviour by forcing it to sink or source some current.

You could add a resistor of perhaps 100-200Ω in series with the 22n but it will change the behaviour somewhat.

One rule of thumb is that if you isolate the capacitance by 100Ω or so (maybe more like 1kΩ for a micropower op-amp) it will not affect the phase margin much.

You can try simulating models to see what they do. If they are good models it may help. Look for underdamped behaviour such as overshoot with a step input rather than limiting yourself to actual oscillation.

\$\endgroup\$
5
  • \$\begingroup\$ I can simulate MCP in LTspice but cannot simulate OPA241 because it doesn't have SPICE model. But can you a bit more explain how that graph is related to G and stability? I didnt get the logical steps there, and probably missing some knowledge. \$\endgroup\$ Commented May 23, 2024 at 19:38
  • \$\begingroup\$ And how about replacing Cf with a large cap like 220nF and overcompensating? Would that fix phase margin issue and oscillation? This would lower the BW but Im interested in DC output. \$\endgroup\$ Commented May 23, 2024 at 19:47
  • \$\begingroup\$ @cm64 The problem is the photodiode capacitance loading the output through the feedback, so increasing the feedback capacitance won't help. \$\endgroup\$ Commented May 23, 2024 at 20:12
  • \$\begingroup\$ @cm64 Something like 3k ohms between the diode and the TIA could help. Your TIA saturates at something like 30 µA, which would result in a forward bias of 90 mV on the diode at saturation, probably not enough to affect it significantly. \$\endgroup\$ Commented May 23, 2024 at 20:27
  • \$\begingroup\$ another fix might be reducing Cf (for more gain). Plus series resistor and some sink current.. There is a number of fixes @cm64. Relying on manufacturer spice models for capacitive load stability is not advisable because it is a very much multidimensional issue that is likely not captured accurately in the models. It's better to e.g. use the level 4 generic op-amp with very conservative specs and then subject it to all sorts of spice torment and adjust until it is profoundly stable. \$\endgroup\$ Commented May 24, 2024 at 5:30

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.