0
\$\begingroup\$

I've been designing a Simulink/Simscape model of a TI buck converter launchpad (https://www.ti.com/tool/BOOSTXL-BUCKCONV) and I'm trying to have my model fit the behavior of the device.

The board can be programmed to drive the buck converter in both synchronous or asynchronous modes. There is a 7.5 Ω static load, and it is possible to add a parallel 2 Ω active load by driving an additional MOSFET.

When using the static load only, the simulation results fit pretty well to the measurements I took in both synchronous and asynchronous driving, and with switching frequencies between 100-500 kHz.

  • In synchronous mode, the same linear behavior (CCM) for all switching frequencies so that Vout = D*Vin
  • In asynchronous mode, a parabolic behavior (DCM) until a certain duty cycle where a transition from DCM to CCM occurs. The higher the switching frequency, the smaller the transition duty cycle. Their values are the ones I expected from theoretical calculus.

My problem is the following: when I add the parallel 2 Ω resistor, the load current grows and I should not be able to observe DCM in synchronous nor in asynchronous driving. The simulation results I obtain in this case have me confused.

  • Synchronous and asynchronous modes both give the same results (looks like a proof of CCM for both modes)
  • Observing the inductor current also tells me that the conduction is continuous
  • BUT, the behavior is not linear, and it does not fit the measurements I took where Vout grows linearly with duty cycle.

Here is an example with a switching frequency of 200 kHz.

enter image description here

My question is: is it possible that a buck converter works in CCM and yet has a non linear behavior for light current loads? I have looked into my simulation model and I can't find any abnormal reason for this difference.

\$\endgroup\$
3
  • 2
    \$\begingroup\$ Bonjour, the thing is that \$V_{out}=DV_{in}\$ is an overly simplified expression for determining the output voltage of a CCM buck converter. See my article published in How2Power where I did include losses in the final expression. These losses can also involve reverse recovery losses, shoot-trough currents etc. and I doubt they are of linear behavior. \$\endgroup\$
    – Verbal Kint
    Commented Aug 2, 2023 at 14:45
  • \$\begingroup\$ Bonjour ! Thank you for your answer. I understand that losses affect the viability of Vout = Dvin, but the measures I took show me a behavior that is very close to linear (red trace on the plot), unlike my simulations. Maybe my model simulates too much losses due to incorrect parameters like mosfets voltage drop or components internal resistors. \$\endgroup\$
    – Reval
    Commented Aug 3, 2023 at 9:28
  • \$\begingroup\$ I see, I would then start with the simplest model (perfect switches, lossless diodes etc.) and verify linearity from there. Then, enhance the model petit à petit (step by step) until the deviation shows up. \$\endgroup\$
    – Verbal Kint
    Commented Aug 3, 2023 at 11:32

0

Reset to default