The AVS power solution on the EVM board doubles as a characterization circuit, and the board is large, so I don't think it's a bad choice, but I was looking at optimization for a product. I searched the TI sight for "AVS regulator", and was delighted to see some National parts pot up. Congrats on the logistics challenge there.
Anyway, The EVM is using GPIOs, and most of the AVS regulator silicon is using a two-wire interface. I read up on AVS on the Wiki, and I note that the AVS driver, as documented at http://processors.wiki.ti.com/index.php/DM816x_C6A816x_AM389x_AVS_Driver_Guide, has selections for the TPS65023, LP3972, TPS6507X, Intersil ISL6271A, etc.
Are these working in the AVS driver in the Netra EZSDK?
It's scary to change the regulator, since you need to develop the software on the board powered by the regulator. Mistakes could be expensive. I'm thinking of putting my production choice on the eval board I'm designing with a dummy load for experimentation, along with the EVM solution powering the core.
I think a solution more compact than the EVM's, documented with schematics and driver, based on TI/National silicon would benefit most designers in my situation.
