Other Parts Discussed in Thread: TINA-TI,
Tool/software: TINA-TI or Spice Models
In trying to verify the accuracy of the TPS50601-SP average model, a simulation was run using the external components shown on the eval board (from the EVM User Guide). The EVM User Guide includes a measured Bode Plot of the loop response which was then compared to the Pspice response. We don't have an EVM to make an real comparison.
The phase response was pretty close, but the gain was about 3-6dB lower which made the crossover freq near 3kHz instead of 6.3Khz shown in the EVM User Guide. If the RI in the model is changed from 0.05 to 0.015, the response gets pretty close to the EVM response, but this seems like a big change for the current sense parameter of the model and is probably not the correct adjustment to make.
Also, a loop response was simulated on Webench with components as close to the EVM as possible. This resulted in a crossover at 3.7kHz, close to the Pspice response. I assume Webench uses the same or similar average model.
Has the model been checked against hardware for accuracy? Are we making a mistake in this comparison approach?
