0
\$\begingroup\$

I have been designing a power supply for some time now and I'd like to reiterate. No, this is not an XY problem. I really, am just designing a precision power supply.

At the output I have a semi-elaborate use of op-amps to produce the precision voltage. It works just fine in DC range (simulations show it's accurate to a few microvolts). Yet when I have it produce a low-frequency sine wave, all these wild oscillations keep springing up.

There is a lack of versioning control software on my part, so I've lost the exact schematic "snippet" that produced the oscillations. What I can describe, is that I had a rail that was virtual and I was trying to stabilize it with capacitors. It turns out it was only prompting the op-amps to oscillate wildly.

enter image description here

I've watched a few op-amp tutorial videos about input bias current, input noise, etc., but nothing seems to discuss this issue of oscillations induced by large capacitances.

I take it that this is the op-amp attempting equalize the 2 inputs, in a zigzag probing manner. Whichever case, how do I avoid, or at least mitigate it?

\$\endgroup\$
4
  • 2
    \$\begingroup\$ Op amps can't generally drive large capacitive loads. Put a small resistor in series with the output to stabilize it. \$\endgroup\$
    – Hearth
    Sep 5, 2022 at 3:49
  • 1
    \$\begingroup\$ This is almost certainly a widely understood problem, and well discussed, about compensating opamps -- especially when driving large capacitive loads. And you probably were using a nice higher-bandwidth opamp, too. You can do as Hearth suggests -- stick a series resistor between the opamp output and the capacitance. But it's not really all that good, though as 'cheap-fixes' go, it's often used. I usually go further and add at least a small capacitance between opamp output and the (-) input -- if your topology is right. Or the small cap becomes a small cap and a resistor in some cases. \$\endgroup\$
    – jonk
    Sep 5, 2022 at 3:58
  • 1
    \$\begingroup\$ Give us a behavior circuit that approximates what existed. Doesn't need to be exactly the same as some iteration. Just get the idea across about how your opamp was part of your 'precision' power voltage regulation system. We need to see the general topology that was used. \$\endgroup\$
    – jonk
    Sep 5, 2022 at 4:20
  • \$\begingroup\$ If you dont need a large bandwidth, you can use an opamp with an overcompensating pin where you can connect a capacitor to reduce your bandwidth. \$\endgroup\$
    – Rahmany
    Sep 5, 2022 at 12:22

2 Answers 2

1
\$\begingroup\$

Oscillators occur because of unexpected positive feedback with gain >=1, usually from parasitic capacitance.

possible causes.

Also, adding lower ESR bulk caps used to suppress the output ripple causes the closed loop AC gain to increase by shunting the feedback.

When the noise triggers a rapid change that is limited by Op Amp current, slew rate limiting can occur which reduces the feedback gain and adds more phase shift result and loss in phase margin.

The precision of DC error means nothing until you examine the step load response, which is an indication of the phase margin and gain margin or response to any disturbance.

Adjust the feedback ratio or the reference voltage can incite these healthy oscillations which can be corrected, but design, BOM and layout may be needed to appreciate the subtle flaws.

\$\endgroup\$
1
\$\begingroup\$

That type of oscillation is usually caused by low phase margin where the loop phase is quite close to, but not quite, 360 degrees when the loop gain is one. With this kind of low phase margin situation, limited amplitude oscillation typically starts to appear at the peaks of the sine-wave and increases in amplitude as phase margin is reduced further.

Adding capacitance to ground on the output of an op amp adds a pole in the loop which increases loop phase and reduces phase margin. The added capacitance combines with the output resistance of the op amp.

Three ways to mitigate this situation to better compensate the amplifier and increase phase margin are:

  1. Add a small series resistor after the feedback take-off point at the op amp's output.

  2. Add a small value capacitor across the feedback resistor. This adds a zero to the loop adding loop phase lead.

  3. Place a resistor directly between the op amp's actual inputs. This is referred to as "forcing the noise gain". "Forcing the noise gain" increases the 1/beta value, which reduces the loop gain, but leaves the low frequency closed loop gain essentially unchanged. The drawback of this compensation technique is that closed loop bandwidth will be reduced.

\$\endgroup\$
1

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.