In thinking about an answer to a different question, I came up with a likely-not-ideal solution to the following:

  • Given a square-wave input of 125 Hz whose minimum is 0V and maximum is somewhere between 1-7V
  • Assuming an ideal source with zero output impedance
  • Not having access to a negative rail
  • Needing to support an output load of at most 10 mA that varies at the same frequency but between +Vin and -Vin; i.e. zero-centered and with double the amplitude
  • Wanting as simple a circuit as possible, ideally with only passive, discrete components

The complex and obvious solution is a switching supply to generate the negative rail, and then a comparator. I came up with the following charge pump as a simpler solution, shown at Vin=5V, RL = 5k:


transient simulation

Since it doesn't need to generate a negative rail, it doesn't need an inductor. It works fine, but I don't love it, for the following reasons -

  • I feel like it can be further simplified
  • The output is asymmetric - the positive cycle is stable but the negative cycle follows the capacitor discharging

Is there a simpler solution? The alternate one below does work, courtesy @tobalt.


simulate this circuit – Schematic created using CircuitLab

  • \$\begingroup\$ @tobalt Slightly smaller like 0.5V difference? If so I'd be interested in seeing it \$\endgroup\$
    – Reinderien
    Commented May 22, 2021 at 17:59
  • \$\begingroup\$ @Reinderien I'm kind of flummoxed by the difficulties. But it's probably that I didn't read you as well as I should have. I recently wrote about something like this: here. The inverter part is on the right side of the circuit, there. It's what you drew, isn't it? Seems to work. Can you spot the difference? The key is the load. It totally changes the behavior, depending. So maybe you didn't compute things right? (Read the theory section there.) \$\endgroup\$
    – jonk
    Commented May 22, 2021 at 23:22
  • \$\begingroup\$ @jonk to be clear: there are two circuits above; the first works fine and the second (suggested by tobalt) does not. I'm going to try adding one more FET like he suggested in the comments. \$\endgroup\$
    – Reinderien
    Commented May 23, 2021 at 3:35
  • \$\begingroup\$ @Reinderien I guess I'm just thinking about your comment, "the alternate one below does not seem to work when I simulate it..." But perhaps I didn't read you well, as I mentioned. That's okay. If the link I provided isn't helpful, then it isn't. Only you'd know. \$\endgroup\$
    – jonk
    Commented May 23, 2021 at 4:13
  • \$\begingroup\$ @Reinderien Ive updated my answer \$\endgroup\$
    – tobalt
    Commented May 23, 2021 at 8:47

1 Answer 1


Sorry for my previous answer. It took me a while to understand what you were trying to achieve.

enter image description here

I think this does what you want.

  • \$\begingroup\$ I've tried to simulate it, results in my question, and it's not working - it's inverting to a DC rail instead of a square wave. \$\endgroup\$
    – Reinderien
    Commented May 22, 2021 at 18:08
  • \$\begingroup\$ Could you sketch a diagram of what the FET connection would look like? \$\endgroup\$
    – Reinderien
    Commented May 22, 2021 at 21:04
  • \$\begingroup\$ Brilliant! Thank you. \$\endgroup\$
    – Reinderien
    Commented May 23, 2021 at 15:08
  • 1
    \$\begingroup\$ @Reinderien I have edited it, to use MOSFET instead of BJT. That way you get rid of the two resistors. \$\endgroup\$
    – tobalt
    Commented May 23, 2021 at 15:21

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.