Other Parts Discussed in Thread: TINA-TI, , TPS62801
I have spent quite some time in trying to stabilize the outputs of switching regulators provided by TI. I would like to specify two devices here, the TPS63050 and the TPS62801, which I have been working reently. Developing power supplies around these devices is pretty straight forward, thanks to the WEBENCH Power Designer. The simulations look pretty stable. The designs can also be exported from WEBENCH Power Designer into TINA-TI. Here as well the simulations run as smoothly as on the WEBENCH Power Designer.
However, the WEBENCH Power Designer creates power supplies with generic passive components, and it embedds these generic components into the exported designs as well. As soon as one replaces the generic passive component models with the ones provided by the respective manufacturers, the simulation starts running errand, throwing out spikes several times of any voltage level present in the design.
For example, in the TPS63050 design, the problem starts with replacing the generic output capacitor model with manufacturer's, no matter which manufacturer I choose. Changing the output inductor does not exhibit this problem, irrespective of the manufacturer.
However, in the TPS62801 design, the problem starts with replacing the generic output inductor model with manufacturer's, no matter which manufacturer I choose. Changing the output capacitor does not exhibit this problem, irrespective of the manufacturer.
In designing complex systems, it is imperative to have a stable, smooth power sypply. It is only then that the stability and performance of other modules of the systems could be evaluated.
For more details about the problem, please refer to this thread.
I even contacted the Designsoft guys to confirm this is not due to any bug in TINA, and they confirmed that the culiprits are the faulty models. Please find the snapshot of the communications with designsoft guys attached herewith. designsoft.pdf
I tried running the simulations in LTSPICE several time, and it simply crashed the application even before reaching the point of stability.
There are some serious convergance problems with these models. Requesting the TI tech guys to look into the affair and suggest measures on eliminating these spikes and glitches from the output, thereby enabling stable output simulations even with manufacturer provided SPICE models of the passive components. Your customers spend a significant amount of time battling with such problems.
Looking forward to hearing back soon.