2
\$\begingroup\$

Problem, parameters

I have an 5V output power supply, with 2S LiPo input (6-8.4V), implemented with AP5002 (PWM control 2A step-down switching regulator, docsheet), and for the life of me, I can't make it stable. The PSU is chirping, dropping voltage irregularly, and in general shows a chaotic output voltage.

unstable output voltage

This is at around 300mA load. The full load my circuit draws is 900mA (mostly LEDs).

The initial circuit is as follows, I'm using ceramic capacitors both for input and output caps. initial schematic Note: C40 is far from PSU, close to the battery. initial layout Ground plane with green: initial layout ground plane

My attempts

I started calculating the feedback loop, but I neither have enough experience in this, and also several parameters of the AP5002 are missing (eg. the internal amplifier circuit gain, transconductance, etc...). I calculated however the zeros and poles of the output filter network:

output filter network

$$ W(s) = \frac{U_{out}}{U_{sw}} = \frac { R_e C s + 1 } { L C s^2 + R C s +1 } \\ p_{1,2} = -109 +-4.545 \cdot 10^4 j \\ z = -9.47 \cdot 10^6 \\ k = 218 $$

I started to implement a compensation network based on this docsheet of a similar regulator by ST. The most stable configuration I was able to come up with is this: compensation network

The values again (close to the theoretical values calculated based on the ST doc):
Rs=160R
Cs=2.2nF
Rf=6.8k
Cf=2.2nF

This was better, but became unstable at around 400mA load. I tried countless variations of values, even soldering in 47k trimpots for Rs and Rf to be able to sweep values from 0-47k. No success. I also tried a 22uF tantal cap for output filter, hoping the larger ESR will make a diff, but nope; the loop behavior hasn't changed.

Any comment or advice is appreciated, thanks!

\$\endgroup\$
2
  • \$\begingroup\$ How does your ground trace get from the input side (C47/C48) to the output side (D16/C46)? \$\endgroup\$ Feb 14, 2015 at 17:40
  • \$\begingroup\$ @BruceAbbott With a detour... I wanted to prevent the switching node noise to couple into the ground plane. I edited the post and added an image of the layout with the ground plane. \$\endgroup\$
    – okaresz
    Feb 14, 2015 at 18:56

1 Answer 1

1
\$\begingroup\$

Wow, thanks @Bruce for reminding me of the ground loop! I looked at the ground plane layout again, and was not pleased... I guess after clearing the ground plane under the switching node, I forgot to check the important should-be-low-impedance loops around the PSU.

Indeed, the input and output side grounds are far-far away from each other, and might be stitched with too few vias as well. I soldered a beefy cable between the input filter caps GND and the D16 diode anode:

layout with ground patch

And without further changes, the PSU is stable. Output voltage @ around 800mA:

output voltage waveform

With an additional 100nF output filter cap, the peaks can be dampened a bit:

output voltage waveform with 100nf filter

Lesson well learned, keep your loops tight when designing a switcher regulator!

\$\endgroup\$
1
  • \$\begingroup\$ Additionally, here is a good article by ROHM with a little explanation and further tips on layout design. \$\endgroup\$
    – okaresz
    Feb 15, 2015 at 8:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.