2
\$\begingroup\$

I have an OPA690 that works fine during normal operation, but weird things happen when it saturates.

After hitting the negative power rail (-5V), the amplifier produces a damped sine response (sine * negative exponential). It is a very small signal (mV) that takes lots of averaging to be able to see. It also takes a very long time to fade (milliseconds). The amplifier is setup to have a bandwidth of a few hundred MHz, so it is surprising to see something last milliseconds.

I can't provide much more information. I am just wondering if anyone has heard of something like this before.

I will delete this question if it is too vague and I do not get any responses.

Thanks.

\$\endgroup\$
8
  • \$\begingroup\$ What frequency and what circuit is it used in? \$\endgroup\$ – Andy aka Dec 22 '16 at 18:39
  • 1
    \$\begingroup\$ If it's truly saturated it's likely not a small-signal oscillation due to the op-amp's feedback loop (since it's no longer closed). Could it be that the load current is sinusoidal, and the r-on of the output stage is causing the voltage to move with load current? \$\endgroup\$ – John D Dec 22 '16 at 18:41
  • \$\begingroup\$ The amplifier is in a complex system with hacks that try to boost the sample rate. The details make the measured frequency of the oscillation not totally useful. I was just wondering if anyone has ever seen any amplifier oscillate only after saturating. \$\endgroup\$ – DavidG25 Dec 22 '16 at 18:42
  • 3
    \$\begingroup\$ What's providing power to the op-amp? Could the power supply voltage be oscillating in response to the step-like change in load current when you drive the op-amp into saturation? \$\endgroup\$ – The Photon Dec 22 '16 at 18:44
  • 1
    \$\begingroup\$ What is your load and circuit? \$\endgroup\$ – Voltage Spike Dec 22 '16 at 19:05
1
\$\begingroup\$

I have seen this in an op-amp designed to go to the rail. It oscillated continuously with a very small amplitude when the output was very near the negative rail (tens of mV). Probably the frequency compensation was ineffective in this range.

Quite a few years ago, so I don't remember the details, but I think it was an LMV324 type.

Ended up not using the part that behaved that way.

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

If an opamp is not designed to go rail-to-rail then its response will become non-linear as the output voltage gets close to a rail. For example: if the output "drops" because it can no longer drive the low side output driver then this effectively increases your loop gain dramatically pushing you closer to instability.

You can add some type of limiting circuit:

  1. Some kind of clamp on the input to limit the maximum input to the circuit.
  2. Adding non-linear feedback that kicks in when output is too large, such as back-to-back zener diodes that bypass the feedback resistor. This reduces the opamp gain to unity when the zener voltage is hit.

Hope that helps, -Vince

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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