2
\$\begingroup\$

I am designing and building a power supply with TPS54531 switching regulators from TI. I am double LC filtering the output. The 5 volt PH output pin feeds a 9.2uH inductor, then to a 47uF ceramic capacitor then through a 0.47uH inductor and finally to a 15uF ceramic capacitor at the output. My question is where should the V_sense resistors connect, before or after the secondary LC filter?

\$\endgroup\$
  • \$\begingroup\$ What does the data sheet advise is the correct way and where's the link to the data sheet? \$\endgroup\$ – Andy aka Jan 20 at 19:17
  • \$\begingroup\$ Why are you double-filtering the output? \$\endgroup\$ – Stefan Wyss Jan 20 at 19:36
  • \$\begingroup\$ Ideally after the secondary LC filter since that's the actual output, but I don't know if the regulator can handle it. Might be unstable. It's probably better to run it through a linear regulator instead of a second LC filter. Or use a multi-phase linear regulator if ripple is such a concern. \$\endgroup\$ – DKNguyen Jan 20 at 19:51
  • \$\begingroup\$ Note that TI has an excellent online design tool "WEBENCH Power Designer". You select the chip and required input/output parameters and it gives you a schematics ready to go. Before adding an extra filtering I'd recommend plugging in your ripple requirements and see if it can be achieved with conventional schematics \$\endgroup\$ – Maple Jan 20 at 19:57
  • \$\begingroup\$ The data sheet does not show a second output filter. I understand that the two stage output filter greatly reduces the high frequency switching noise. I suppose much like a follow-on linear regulator might do, but I am way past that point. \$\endgroup\$ – RodB Jan 21 at 21:56
6
\$\begingroup\$

My question is where should the V_sense resistors connect, before or after the secondary LC filter?

It should almost certainly NOT go after the 2nd LC filter because you'll turn the circuit into a power oscillator.

With feedback systems you have to take care of not introducing too much phase shift or delay or negative feedback (desirable) becomes positive feedback (turns into an oscillator).

The regular LC filter in a switcher introduces anything up to 180 degrees of shift and this is compensated by good chip design and compensation component selection. See this: -

9.2.2.6 Compensation Components

and this: -

8.3.7

Slope Compensation In order to prevent the sub-harmonic oscillations when operating the device at duty cycles greater than 50%, the TPS54531 device adds a built-in slope compensation which is a compensating ramp to the switch-current signal.

Note the parts about slope compensation. Adding another LC network within the feedback loop is asking for trouble.

\$\endgroup\$
  • \$\begingroup\$ Thanks a ton, Andy. \$\endgroup\$ – RodB Jan 21 at 22:02

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.