Op-amp capacitive load stability is not guaranteed by an amplifier being stable on its own. A capacitor on the output, whether it is a deliberate bypass, a length of coaxial cable, or the input of something downstream, combines with the amplifier’s open-loop output impedance to add a pole inside the feedback loop. Enough added phase lag and the circuit stops amplifying and starts oscillating.
Why a capacitor on the output matters at all
An operational amplifier has a real open-loop output impedance, largely resistive over a wide frequency band. Put a capacitor from the output to ground and you have created a low-pass filter between the amplifier’s internal output and the node the feedback network senses. That filter contributes phase lag, and because it sits inside the feedback loop, the lag adds directly to whatever the amplifier’s own compensation already contributes. Stability depends on the total phase shift around the loop at the frequency where loop gain passes through unity. An amplifier designed with comfortable phase margin can have that margin eroded by the added pole until the loop is marginally stable, producing ringing and overshoot, or genuinely unstable, producing sustained oscillation that may be well above the frequencies the circuit was designed to handle.
What the failure looks like in practice
Mild cases show as overshoot and ringing on a step response, an output that looks fine on a slow signal and untidy on a fast one, or peaking in the frequency response near the top of the band. Severe cases oscillate continuously, often at a frequency far above the signal band, which means a slow meter shows nothing wrong while the amplifier runs hot and the supply current is elevated. Other symptoms are indirect: distortion that increases with cable length, a circuit that misbehaves only when a particular probe is attached, or an output that starts oscillating when a load is disconnected. Because oscilloscope probes add capacitance themselves, a low-capacitance probe or a properly compensated ten-to-one probe is essential, as a cheap probe can create the very problem you are hunting.
Where the capacitance comes from
Deliberate output capacitors are the obvious source, often added with good intentions to filter noise. Cables are the common hidden source, since coaxial and multi-conductor cable presents capacitance proportional to length and even a modest run can be significant. Long PCB traces contribute. The input of a following stage contributes, especially sampling converters that switch a capacitor onto their input. Piezoelectric transducers, some sensors, and certain actuators are fundamentally capacitive. Test equipment at the end of a bench lead adds more. The important habit is to total all of it rather than considering only the part you drew on the schematic, and to remember that the worst case may be a service condition, such as a long cable fitted in the field, rather than the configuration on your bench.
Fixes that work and fixes that do not
The most reliable remedy is an isolation resistor in series with the output, placed inside the loop from the feedback network’s point of view so that the feedback is taken from the amplifier side of the resistor. That resistor separates the amplifier’s output impedance from the load capacitance and restores phase margin, at the cost of output impedance and a voltage drop under load. A refinement takes DC feedback from the load side and AC feedback from the amplifier side, preserving DC accuracy while keeping the loop stable. Choosing an amplifier explicitly specified to drive a stated capacitive load is another sound approach, since some parts include internal compensation for exactly this. What does not work is adding more capacitance, increasing supply decoupling in the hope it helps, or reducing gain without checking whether the noise gain actually changed.
Verifying stability rather than hoping for it
Test with a small square wave superimposed on the operating point and observe the output edges under the worst-case load, including the longest cable the product will ever see and the coldest and hottest temperature. Overshoot and ringing tell you how much phase margin remains; a clean, well-damped edge indicates comfortable margin. Watch the supply current too, since a sharp rise usually means something is oscillating out of the band you are observing. Check both with the load connected and disconnected, because some configurations are least stable at light load. Finally, consult the amplifier datasheet for a graph of overshoot against capacitive load, which many manufacturers provide, and treat that graph rather than a rule of thumb as the authority for the part you are using.
What to check before you act
- Total every source of load capacitance, including cables and probes.
- Test the step response with the longest cable the product will ever use.
- Add a series isolation resistor before adding more output capacitance.
- Watch supply current for signs of oscillation outside your scope’s view.
Common questions
Does a unity-gain buffer handle capacitive loads better?
No, usually worse. A follower has the highest noise gain bandwidth and therefore the least phase margin to spare, which is why buffers driving cables are the classic failure case.
How large should the isolation resistor be?
Large enough to damp the response and small enough that the voltage drop and output impedance remain acceptable. Determine it experimentally from the step response with the real load rather than from a generic figure.
The practical takeaway
Count every capacitance hanging on an amplifier output, cables included, and assume it is inside the feedback loop until proven otherwise. Verify the step response with the worst-case load, reach for a series isolation resistor or a part specified for capacitive drive, and confirm the fix at both temperature extremes.
References and further reading
- FCC: electromagnetic compatibility and radio frequency safety questions answered
- Department of Energy: how electronics losses turn into wasted energy and heat
- CPSC: safety guidance for electronics and electrical products indoors
Numerical scenarios are illustrative unless identified otherwise. Follow the exact product instructions; component ratings and local installation requirements can differ.
