1
\$\begingroup\$

I "designed" a simple buck converter, which is basically a clone of the 5V reference design in the TPS6293x datasheet. The converter is outputting the correct voltage on average, but has quite a lot of ripple. The ripple frequency and voltage depends on the input voltage to the buck converter, but seems to vary between 30-120Hz and 100-300mV. This is with the buck converter unloaded. If I add a load, the ripple voltage stays about the same, but the frequency shoots up into the dozens of kHz. There is no detectable ripple at the input to the buck converter.

Below is the schematic, board layout, and o-scope readings at various input voltages. Note that the board has a bodge due to the schematic initially being wrong.

schematic

board layout

scope 1

scope 2

scope 3

\$\endgroup\$
1
  • \$\begingroup\$ Try a 100pF cap on parallel with r12. Also it seems like a no load behaviour. It shoots up quickly, then shuts down on overvoltage. Try higher load and larger output capacitors. \$\endgroup\$
    – TQQQ
    Nov 22, 2023 at 16:42

1 Answer 1

4
\$\begingroup\$

This is "PFM" mode, or pulse frequency modulation. At light loads, instead of keeping the frequency constant, the converter enters a hysteretic mode where it switches only when the output voltage falls below some threshold. Note fig 10-13 in the datasheet:

enter image description here

If the load is very light it can take a long time for the output voltage to fall back below the PFM threshold and therefore the switching frequency can get very low. The frequency and amplitude of the output ripple are dependent on the load, inductor, and output capacitance.

The output ripple is typically higher in this mode as well. The upside is that the efficiency is greatly improved at light loads. The downsides you have noticed- Variable switching frequency that can sometimes cause audible noise, and higher output ripple. As @TQQQ mentioned a feedforward cap in parallel with R12 may help reduce the ripple, as will adding some minimum load.

If this operation mode is a problem for you, you can replace your controller with the TPS62933F which always operates in forced continuous conduction fixed-frequency mode.

\$\endgroup\$
1
  • \$\begingroup\$ I think you're right. I did the other suggestions but it had little effect, if any, on ripple. Thanks! \$\endgroup\$
    – Stephen D.
    Nov 23, 2023 at 0:32

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.