1
\$\begingroup\$

Lets say I have a circuit which involves an OpAmp and a comparator. They're both single supply running from a noisy USB power supply.

Each component has local bypass capacitors, a big and a small one (10µF and 0.1µF, here simplified as one capacitor).

enter image description here

The comparator drives a load of +20mA and during switching I see the signal feeding back into the OpAmp over the power supply line. This cause my signal to degrade quite a bit and I need more hysteresis that I would like to.

I thought about adding some kind of noise isolation between the two components by placing inductors between them like this:

enter image description here

Inductors are in the range of 10µH and have series resistance around 2Ohm.

Simulation shows that the interaction over the shared power supply is almost non existent afterwards. Also the improved filtering reduces the noise from the USB supply a lot. This is to be expected because it is a CLC or Pi-filter.

The question: Is this a good or bad idea? I'm specifically asking because I've never seen this done in commercial circuits so far.

\$\endgroup\$
  • 1
    \$\begingroup\$ I see lots of input filtering with common mode and differential mode chokes and filtering caps all around, in AC inputs as well as DC inputs, so maybe you just looked at commercial circuits that did not care about it so far? You could also just try and see if a big beef reservoir cap at your power input already helps enough. \$\endgroup\$ – PlasmaHH May 22 '15 at 10:21
  • \$\begingroup\$ As you observe, this is an LC filter, so it's a perfectly legitimate low-dc-impedance way to filter out power supply noise. \$\endgroup\$ – Nick Johnson May 22 '15 at 10:27
  • \$\begingroup\$ How are you so sure it isn't a pcb layout or breadboard problem? \$\endgroup\$ – Andy aka May 22 '15 at 12:24
  • \$\begingroup\$ @PlasmaHH if you turn your comment into an answer I'll happily accept it. :-) \$\endgroup\$ – Nils Pipenbrinck May 26 '15 at 21:15
2
\$\begingroup\$

The obvious solution is to use a linear Low-Drop Out regulator (LDO) to feed both the Vdd opins on the op-amps and comparitors as well as any reference voltages needed.

You will continue to power your load directly from the Vusb supply but the op-amps & comparitors should be nicely insulated from noise on the USB supply.

The total current that the LDO needs to supply will be very low so long as it is not required to power the load.

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

It's good idea with following consideration (if you are trying it on PCB)

  1. Bypass capacitor nearer to VDD/VCC pins.
  2. Ferrite bead (or inductor in your case) nearer to VDD/VCC i.e supply pins.
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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