Ahhhh, now everything starts to make sense! I wondered the whole time how the I/O can bring a lot of the benefits, when the large linear supply is kept for the Extreme. Well apparently, it‘s the parts like XDMI that benefit the most from the BPS. Alright, alright, now I get it… well, thanks for explaining and removing that question mark from my forehead.
It’s a good example of how things can change during a design process. It starts with an idea, followed by prototype(s), then concrete designs. During that process you may end up making new discoveries. We went through 4 actual switch designs before we had a final design, for the GaN regulators we’re even at the 8th iteration! For the BPS we “only” have 2 designs, for XDMI we are at the second baseboard design but have designed 2 different digital and 4 different analogue output stages with dozens of smaller changes/adjustments. It always ends up different then you expect somehow.
The downside of sharing information during the design process is that what you share in earlier stages may be obsolete by the time the design is finished. That absolutely applies to both the BPS and XDMI. For the BPS the way its applied has changed, smaller multiple supplies in stead of one large supply. For XDMI we may have been a bit to conservative in predicting its performance level.
