0
\$\begingroup\$

There has been a lot of questions about how to stablise opamp circuits driving capacitive loads. However, I know there are some op amps that are designed to drive capacitive loads but I am struggling to find the search filter criteria to narrow down which opamps would be good at driving capacitive loads. These are the categories found on the analog website. Similarly, TI below. None of the categories specify capacitive driving capabilities as a filter. What are some of the specification of op amps that can help in determining whether or not an opamp will be able to drive a capacitive load?

enter image description here

enter image description here

\$\endgroup\$
3
  • \$\begingroup\$ You can typically drive a capacitive load with most op-amps if you put a small resistor 10~100 ohms in series with the op-amp output. Besides the capacitor, what kind of load are you driving? \$\endgroup\$
    – user4574
    Commented Apr 19, 2022 at 23:11
  • \$\begingroup\$ Thanks for the comment. The opamp I am designing will be used as an excitation source to drive an unknown amount of capacitance but will be large. Cheers. \$\endgroup\$
    – yolt
    Commented Apr 19, 2022 at 23:14
  • \$\begingroup\$ All op-amps can drive capacitive loads if you characterize their output impedance (e.g. using a network analyzer), characterize the load impedance (capacitor and parasitics, often significant), and then design a feedback response that will compensate them adequately. It's a trade-off: op-amps by are limited by their output impedance. To get them to drive capacitors faster, you need lower output impedance - this is often done externally with discrete parts. It's hard to beat a 10GHz ft transistor driven at 50mA collector current. Or ten of them in parallel. \$\endgroup\$ Commented Apr 20, 2022 at 9:42

1 Answer 1

4
\$\begingroup\$

No op-amp is really good at driving capacitive loads. The usual advice, in case a capacitive load cannot be avoided, is to de-couple the capacitive load from the feedback, by connecting a significant resistance in series with the op-amp output. This does affect the design of the feedback network, the frequency response, and the phase response. Remember, for stability, you need to avoid inverting the feedback phase, so for all frequencies where gain is >1V/V, the total amount of phase shift needs to be no more than about 120 degrees (for 60 degree phase margin). A capacitive load chews up 90 degrees of that phase budget, leaving only 30 degrees for the op-amp and the feedback network. This is generally bad practice.

So, an op-amp with poor output characteristics (modeled as a relatively high internal output resistance), would be able to tolerate driving a small capacitive load without the need to connect an additional external series resistor, because in that case the op-amp's internal source resistance is high enough. I don't think anyone deliberately designs an op-amp to have poor output characteristics, especially when it's easy to increase the effective output resistance just by adding a resistor in series.

Putting it in marketing terms rather than engineering terms, what you want is probably something in the "precision op-amp" category. These favor low input offset voltage, as you can see from the fact that they advertise the entire category in terms of Vos and its temperature coefficient. The tradeoff is that gain-bandwidth product and settling time may be worse. You may also need to consider further isolating the capacitive load using a transistor in emitter-follower configuration.

Another marketing term to look into is "line driver"; something that is intended to drive a big load without using any feedback. The application these are designed for is driving long lengths of cable in an industrial setting, so the capacitive load is unavoidable. These sacrifice the ability to precisely regulate the voltage on the load, but if all you want to do is drive some digital on/off keying through a mile of twisted-pair cable, that's what to look for.

\$\endgroup\$

Your Answer

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

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