Author Topic: SMPS compensation does not match simulation  (Read 371 times)

0 Members and 1 Guest are viewing this topic.

Offline zmasterTopic starter

  • Contributor
  • Posts: 10
  • Country: it
SMPS compensation does not match simulation
« on: August 14, 2023, 07:02:39 pm »
Hello everyone,

I'm a bit lost and I could use some advice.

I've designed a board which is essentially a SMPS that takes the voltage from a 1-cell Li-ion battery and boosts it up to 24V to drive some LED strips.
I figured it wouldn't be too difficult as the LED strips don't need a perfectly regulated output, however I didn't consider that driving those LEDs with a PWM I would be constantly unsettling the regulator due to the load switching, so a good control loop was going to be important.

I started by designing the circuit with TI's Power Designer and I got this schematic:



When I made my board, I used that same design, but I made a couple of mistakes:
  • I used the available ceramic caps from JLC-PCB, assuming they would derate to 50% of nominal capacity due to DC bias like the ones picked by TI's designer. I later discovered they actually derate by 90%.
  • To simplify the BOM, in the compensation network I re-used some existing values for Rcomp and Ccomp, aiming to keep a similar RC constant. I later realised that's not how it works.

The 24V output was fine with a 100% PWM, but it would not regulate well at all with a 50% PWM.
I then addressed the points above by adding a 10uF electrolytic in parallel to Cout and bringing Rcomp to 1.5k and Ccomp to 200n.

Still, the regulation is terrible, drops to 17V and overshoots to 35V:



I've found a tool from TI to calculate the compensation loop and with these values it should be stable:
(cells in yellow are the relevant ones, the ones on the top left come from previous calculations)

1851454-2


I've played a bit with different components and got to something that kind of works (Cout=220uF, Ccomp=100n, Rcomp=22k), but it's still not great and I don't know if it will be stable. I suppose I could try all possible conditions (Vin, DC, etc).


In conclusion, my questions would be:
  • Why does the circuit behave so different than the simulation? What am I missing?
  • If it's normal for the simulation to be so unreliable, how can I tune the compensation and ensure it will be stable under all conditions?


Edit: Sorry, I don't know why pictures are not showing inline...
« Last Edit: August 14, 2023, 07:18:39 pm by zmaster »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf