3.3v, 3.3v_a

I see there is need for (what I believe it to be) an “analog” 3.3v supply, in addition to just 3.3v. In Rings, Streams, I imagine all the older MI modules, it seems to be achieved through L3, a ferrite bead I think. In Plaits, Marbles, and I assume MI modules to come, it’s done through a linear voltage regulator, while the rest is powered by a switching regulator.
I know it’s necessary as I originally noticed it in the datasheet for the PCM5100, that’s what started this investigation!
Can someone please take the time to enlighten me? Why can’t these 3.3v regulators hack it on their own? Why does one need both?

Fast and power-hungry processors are more likely to have rapid and pronounced changes in their current needs that a voltage regulator can’t immediately react too. Also to consider, the resistance of the copper track between the regulator and the MCU, think of it as trying to guide something with a slightly flexible and wobbly stick. That’s why the 3.3V line powering the MCU is always a little bit wiggly and noisy, which is not good if it’s also powering ADCs and DACs (more particularly if this same 3.3V is used as a reference for the ADC).

The few modules in which a common linear 3.3V regulator serves both the digital and analog side (with some filtering provided by a ferrite) are either heavily space or cost-constrained, or can tolerate a lot of filtering on their CV inputs (Rings, Warps, Peaks, Streams). Big downside: a lot of heat will have to be dissipated by a single regulator! On Rings and Warps we’re very close to the limit in terms of heat dissipation, the board is hot!

In the case of Stages, Marbles, Plaits, Tides: I wanted to reduce current consumption, hence a switching regulator. Their output is not very clean (oh the 50 mVpp trash at 800kHz…), so I used a separate linear regulator for the analog section.

1 Like