The Rev E hardware switched over to the TI TPS54294 buck converter, which is a dual-output component that can supply both the 3.3V and 5V rails.
In testing, it was found that the power board sometimes cannot supply the inrush current needed to start both sides of the converter at once. When this happens, the power rails either fail to come up or initialize unreliably.
Extensive lab testing has reproduced this issue with a test board and bench instruments providing the supply and load.
To solve this problem, three changes are being proposed:
Sequence the power rails, so that 3.3V comes up before 5V.
Slightly delay the startup of the power rails, so that the AC/DC converter's 12V supply has a little time to stabilize and charge its capacitors
[Optional] Add feedforward capacitors to the buck converter's feedback circuit to help stabilize its response
The first change alone may be sufficient to solve the problem, but doing all three should lead to a generally better and more reliable system.
The current plan is to bodge these change into the Rev E hardware to get development and testing back on track, then to incorporate them directly into the Rev F update.
The Rev E hardware switched over to the TI TPS54294 buck converter, which is a dual-output component that can supply both the 3.3V and 5V rails. In testing, it was found that the power board sometimes cannot supply the inrush current needed to start both sides of the converter at once. When this happens, the power rails either fail to come up or initialize unreliably.
Extensive lab testing has reproduced this issue with a test board and bench instruments providing the supply and load.
To solve this problem, three changes are being proposed:
The first change alone may be sufficient to solve the problem, but doing all three should lead to a generally better and more reliable system.
The current plan is to bodge these change into the Rev E hardware to get development and testing back on track, then to incorporate them directly into the Rev F update.