When using multiple regulators in a chain, is there noise added to the input by the regulator?

If I have a SMPS feeding an LDO, and then another SMPS after the LDO. Is the ripple or noise added to the LDO rail by the downstream regulation?

  • 1
    \$\begingroup\$ It depends on the implementation, implement this poorly then yes, the 2nd SMPS will add ripple to the output voltage of the LDO. Implement it in a "better" way and you will get less ripple (but not zero ripple). Adding decoupling capacitors generally helps. Using a "better" LDO also helps. There is generally no way to add zero ripple to the output of an LDO when connecting an SMPS. The SMPS switches and you will see that on the LDO. What you need to consider is how much ripple is acceptable in your application. \$\endgroup\$ – Bimpelrekkie Jun 11 '19 at 14:28
  • 1
    \$\begingroup\$ Note that I'm not saying anything regarding noise (which is a non-deterministic, random signal). SMPSs add so much ripple that the noise is generally not an issue (the ripple is much worse). If you use an SMPS you should generally be unconcerned with the noise as the ripple will be much, much higher in value. So if noise is an issue to you: don't use an SMPS. Get a low-noise LDO and use a non-rippling input voltage for that LDO (a battery would be ideal for that). \$\endgroup\$ – Bimpelrekkie Jun 11 '19 at 14:33
  • 1
    \$\begingroup\$ Expensive LDOs, consuming high standby currents (Iddq), can regulate very well at DC. Typically at 1MHz, the ability to regulate has gone to ZERO. For low-Iddq LDOs, the ability to regulate typically will be poor and even by 10KHz, the regulation has become ZERO. Its YOUR job to provide filters prior to the LDO, to attenuate the noise and ripple at higher frequencies. \$\endgroup\$ – analogsystemsrf Jun 11 '19 at 14:35
  • \$\begingroup\$ And you must design the Ground paths as well. Copper foil (standard thickness 1.5mils, 35 microns, 1ounce/square foot) has 0.000500 (500 microOhms) per square (any size square). Thus one amp flowing thru 5 squares causes 1A * 5sq * 0.0005/sq = 5 milliVolts upset (noise, trash, ripple). \$\endgroup\$ – analogsystemsrf Jun 11 '19 at 14:57
  • \$\begingroup\$ Why do you want the second SMPS2 after an LDO versus directly after SMPS1? \$\endgroup\$ – sstobbe Jun 11 '19 at 15:14

This is what I get out of the setup you envision:

SMPS -> VReg* -> SMPS

And your asking about the rail with the * on it.

In this case, the last SMPS will affect the regulation of the Vreg because it is switching current and because the Vreg has source impedance, and cannot respond infinitely fast to regulate the voltage.

The question then becomes, how much does it affect regulation? This information can be found in good voltage regulation products as they display a graph that shows ripple rejection vs frequency. The SMPS has a switching frequency, and the voltage regulator will reject the ripple and the amount in dB can be found and the amount of rejection calculated.

In most cases, I would find it confusing to place an SMPS after a regulator because

  • Voltage regulators generally have higher losses
  • SMPS usually have wide voltage inputs

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.