1
\$\begingroup\$

I have a signal processing circuit that requires symmetrical +/-5V, about 0,5A each with a requirement that the output ripple should be as low as possible (ultra-low ripple.)

I have tested various step-up converters as I wanted to power the device from USB. Every time, the ripple badly distorted my desired signal (it might have been also due to EM interference from coils.)

After some initial search for a solution, I couldn't find anything that would suit my needs: charge pumps, linear regulators or tested converters.

Finally, I tried to connect the converter with a linear regulator in series, so that the converter boosts the voltage for the linear regulator and the linear regulator reduces the ripple from the boost module. Does connecting the boost converter outputs to linear regulators make any technical sense, or there are better solutions to the problem?

\$\endgroup\$
3
  • \$\begingroup\$ Pick your favourite circuit and then add loads of filtering to the output? \$\endgroup\$ Commented Sep 13, 2021 at 9:42
  • 1
    \$\begingroup\$ @user253751 I tried this, but the ripple attenuation was too low and sometimes the converter behaved strangely as if the filters were causing some additional oscillations/excitation. \$\endgroup\$ Commented Sep 13, 2021 at 11:14
  • \$\begingroup\$ +5V 1A total exceeds available power on USB from a PC USB port. With a "phone charger" it will be fine. \$\endgroup\$
    – bobflux
    Commented Sep 13, 2021 at 17:27

2 Answers 2

2
\$\begingroup\$

Does connecting the boost converter outputs to linear regulators make any technical sense, or there are better solutions to the problem?

Yes it makes sense, if you need less noise. But the switching frequency is not that easy to completely remove, you do need a LC lowpass filter before the LDO, at least.

Ultralow Noise Switching Power Supplies

enter image description here

\$\endgroup\$
5
\$\begingroup\$

Does connecting the boost converter outputs to linear regulators make any technical sense, or there are better solutions to the problem?

Yes, but.

Typically linear regulators have good PSRR at low frequency ; if you still want good PSRR upwards of a few kHz you have to do your shopping more carefully, some are really bad and some are quite good.

You will also want to minimize dropout voltage for efficiency, however a LDO's PSRR decreases at low dropout voltage because at low Vce (or Vds) the pass transistor's performance decreases, its capacitance increases, etc. At the "minimum dropout voltage" datasheet spec, PSRR is usually non existent. So make sure you read the fine print and look at PSRR graphs at the dropout voltage you'll actually use. Half a volt of margin makes a lot of difference.

Then all LDO's are feedback systems which means the effectiveness of the regulation falls with increasing frequency. Basically at the rather high frequency of the switching regulator, and especially for its spiky harmonics, the LDO will have poor rejection. So you need a LC filter, preferably with a ferrite bead instead of the inductor. A CLC filter works well. Make sure it is well damped and does not ring at the resonance frequency of the LC circuit.

If you want a DC-DC optimized for low noise, check SilentSwitcher from Analog/LT.

\$\endgroup\$

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.