LM3668: Unexpectedly high ripple voltage during high loads

Bob Wayner

Part Number: LM3668

Support Path: /Product/Development and troubleshooting/

Under non-switching 600mA load we are seeing a ripple voltage of +/-500mV in buck/boost mode and boost mode. Under a non-switching 400mA load we are seeing a +/- 500mV ripple in the boost region.

Increasing output capacitance has not fixed this high ripple voltage,

Our output capacitor is a ceramic 33uF. Chosen inductor is 3.3uF.

It seems the regulator cannot handle these current loads as advertised in the datasheet. Our goal is to have no more than +/- 100mV of ripple votlage. Would TI happen to know known fixes for this issue?

Thank you,

- Bob

over 7 years ago

0 Vijay86929 over 7 years ago

TI__Genius 11490 points

Hi Bob,

Can you share the schematic and waveforms showing ripple?

Thanks

0 Bob Wayner over 7 years ago in reply to Vijay86929

Prodigy 210 points

0 Bob Wayner over 7 years ago in reply to Bob Wayner

Prodigy 210 points

Note: R18 is depopulated.

0 Vijay86929 over 7 years ago in reply to Bob Wayner

TI__Genius 11490 points

The inductor value L1 and the output capacitor C12 are different than what is shown in the datasheet and the user guide. We know that the inductor has effect on loop stability.
Also it will be good to see the conditions under which you are seeing the instability as well as VOUT and IL waveforms. These will give us better insight into what is happening.
Regards

0 Bob Wayner over 7 years ago in reply to Vijay86929

Prodigy 210 points

600mA testing results:

-All regions of testing yielded a DC component voltage of 3.28-3.29V.

-Buck mode: Ripple voltage not much of an issue: 50mV ripple observed.

-Buck/boost mode: Ripple voltages observed as high as 200mV.

-Boost mode: Ripple voltages observed as high as 300mV.

In 417.7mA testing the buck/boost mode ripple is about +/- 100mV and boost yields +/- 450mV ripple.

The load to the regulator was a resistor tied to ground. The regulator has been severed from the rest of the board to accomplish this testing.

0 Vijay86929 over 6 years ago in reply to Bob Wayner

TI__Genius 11490 points

Hi Bob,

Thanks for the information.

I suggest using datasheet recommended L1/COUT. Also we need to look at the waveforms as requested above.

Regards

0 Bob Wayner over 6 years ago in reply to Vijay86929

Prodigy 210 points

Vijay,

Here is a captured waveform at boost mode(2.9V input) and a pure resistive load of 420 mA. I will use TI's recommended values and retest when said components are in my hands.

Thank you,

- Bob

0 Vijay86929 over 6 years ago in reply to Bob Wayner

TI__Genius 11490 points

Hi Bob,

This looks more like switching noise/ripple than any instability. If you measure with short GND probes directly across the COUT it should look better.

Regards

0 Bob Wayner over 6 years ago in reply to Vijay86929

Prodigy 210 points

Vijay, I have replaced the inductor with a 2.2uH value one, as TI recommends. In addition I have changed out the output capacitor to TI value recommended 22uF ceramic. Using active probes with short traces the waveform looks about the same, with a +/- 250mV ripple. Please observe the attached waveforms which were taken with 400mA constant current in buck/boost and boost mode.

Buck/boost:

Boost:

To re-iterate, the regulator now has TI recommended capacitor and inductor values in addition to using active probes to measure the voltage.

Please advise to reduce the ripple to below or equal to 100mV.

Thank you,

- Bob

Power management

Power management forum

LM3668: Unexpectedly high ripple voltage during high loads