Content originally posted in LPCWare by bavarian on Tue Aug 18 05:18:00 MST 2015
For more than one reason the protection with the voltage divider is not a good solution. As you have already experienced, the VBUS pin changes its characteristics depending on settings made by the DFU bootcode. This, in addition to other effects like static power consumption, makes this voltage divider risky to use.
Applying a voltage more than 3.6V to VBUS for a short time will not destroy the pin structure, it only has a long term effect. Over time it impacts the diffusion of the silicon. So don't worry about short periods with more than 3.6V.
A much better solution is the switch in front of the VBUS pin (discrete FET or a small analog switch like 74LVC1G66).
The comparator levels behind the VBUS pin make things also more complicated. For USB OTG there are different levels defined, the lowest level is normally enough to recognize the connection of a cable which provides power (i.e. the 5V). In a self powered device this event is enough, because the processor is in operation and can work on this event.
In a nutshell: if you want to avoid problems with one or the other application use case, take the switch solution instead of the voltage divider. I will discuss with relevant people to get the voltage divider proposal out of the documentation, or at least restrict it to a very specific use case where it is doing a reasonable job.
Note: the problem with the higher power consumption during DFU bootmode has been solved in the bootcode of the LPC4357A (and the other flash devices). This version is in production since beginning of 2015 and replaced the earlier LPC4357 '-' version.
Regards,
NXP Support Team