I am designing an isolation amplifier that consists of the following blocks:

  • A comparator which produces a PWM signal
  • A coupling capacitor that achieves galvanic isolation
  • A voltage follower that acts like a buffer
  • A 2nd order LPF with cutoff frequency at ~20 kHz

At the output, the output signal can recover the input waveform with a slight phase shift. However the output signal also introduces an unkown DC offset (~1.85V). I have plotted the voltage waveforms at different nodes and found out that there is a DC offset at the buffer input. I think it is this DC offset at the buffer input that causes the DC offset at the output. But I have no idea what circuit component introduces such DC offset in my circuit.

I have attached my schematic as well as my simulation waveforms. I am designing my circuit using LTSpice and AD8513 as my opamps. Greatly appreciated if I can have any advice or help regarding this issue and how to improve my circuit design.

Circuit Schematic Screenshot Circuit Simulation Waveform

Thank you for any advice in advance.

  • 2
    \$\begingroup\$ Tip: ground symbols should point down towards, ehm, the ground, not into the sky. If you are interested, Rules and guidelines for drawing good schematics is well worth a read. \$\endgroup\$
    – Transistor
    Apr 10, 2021 at 14:26
  • \$\begingroup\$ How can you get galvanic isolation if the grounds on both sides of the capacitor are connected together? \$\endgroup\$
    – Ste Kulov
    Apr 10, 2021 at 15:37

1 Answer 1


You have no DC path to ground on the AD8513 input. A resistor to ground or whatever other DC reference you want to use will provide that. The bias current of the 8513 is pA or nA over a reasonable temperature range, but it still needs a path. Also the (likely bigger) factor of the DC leakage through a 220uF capacitor needs to be dealt with. The resistor value needs to be low enough that the sum of the leakage and bias currents does not cause excessive offset.

220uF does not really provide isolation the way most would think of it, but I won't address that issue.

  • \$\begingroup\$ Thank you very much for your help. I attached a pull-up and a pull-down resistors to bother the +/- 5V power supply and it seems the DC offset at the filter output was removed. It also comes with another issue. I am aiming to use such PWM signal to modulate a DC input signal into an "AC" signal so that it can pass through the coupling capacitor. By these pull-up/down resistors. The DC input signal cannot be coupled to the output even with the help of PWM. Is there any reason suggesting the failure of coupling DC signal? Thank you very much \$\endgroup\$ Apr 11, 2021 at 3:01
  • \$\begingroup\$ If you want to continue with this concept, look up DC restoration circuits as used with video and think how to make it precision. \$\endgroup\$ Apr 11, 2021 at 3:06

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.