Pulse skipping on Vneg of TPS65130

[thmjpr]

TPS65130  
Split-Rail Converter with Dual, Positive and Negative Outputs (300mA typ)
Datasheet: http://www.ti.com/lit/ds/symlink/tps65130.pdf

5V in and getting +/- 8V out.
Positive supply is working fine, 1.3MHz constant, with minimal ripple:



Negative supply is stuck in pulse skipping mode producing a ton of ripple:



The device has power save mode, enabled by pulling PSN (power save negative) or PSP to Vin. Disabled with Gnd.

Soldering:
- PSN and PSP pin continuity checked to Gnd, looks ok under microscope as well.
- I only had the one chip (new from LCSC), so will update when I receive and test another chip. Going to try a TPS65131 as well (higher current version).

Inductor:
- 4.7uH recommended, 10uH is being used. Tried a 4.7uH and skipping was still present.

Load:
- No load or 20mA load didn't change anything.



Anything I might have missed?

Some other discussions:
https://www.eevblog.com/forum/projects/tps65131-split-rail-power-board/
https://e2e.ti.com/support/power-management/f/196/t/717199?TPS65131-negative-output-voltage-abnormal

[SiliconWizard]

Sorry if I've missed it, but I didn't see any capacitor at +Vo and -Vo.
That's probably going to be a problem.

[thmjpr]

Sorry if I've missed it, but I didn't see any capacitor at +Vo and -Vo.
That's probably going to be a problem.

My fault, its in another area of the circuit. 47uF on each +Vo and -Vo to Gnd. 25V 47uF X5R.
Recommended value is 22uF.

Making it larger might help but I suspect it will still skip.

[T3sl4co1l]

Are you sure it's not compensated incorrectly?

Pulse skipping is, well, skipping pulses, but that's a burst waveform.  Some controllers still do that, but I don't know if that's typical for this one.

Tim

[thmjpr]

Are you sure it's not compensated incorrectly?

Pulse skipping is, well, skipping pulses, but that's a burst waveform.  Some controllers still do that, but I don't know if that's typical for this one.

Tim

You are right the terminology is confusing, its more of burst/sleep mode. TI seems to like to call it that:

The converter operates with a fixed frequency PWM control topology and, if power-save mode is enabled, it uses a pulse-skipping mode at light-load currents.

I've tried with Cn, compensation cap removed, no major difference seen there.
With Cf-, feedforward capacitor removed, duty cycle of the pulses increases to ~%50, ripple reduces significantly, and power consumption goes up a bit. So thats on the right path. But removing it is too risky.

Note here about >33uF not recommended, but I'm getting same issue with lower values: https://e2e.ti.com/support/power-management/f/196/t/799796
Also there, at a point below ~80mA on the Vneg, discontinuous conduction occurs.

When testing today I found dropping Vin from 5V to ~4.2V would kick it out of power-save mode and into normal PWM. It would also get into some wild states above 5.3V. Might need to play with that a bit more..