[Arm-netbook] [EOMA68] RS232 on SoCs

[luke.leighton]

On Sat, Sep 7, 2013 at 11:34 PM, joem <joem at martindale-electric.co.uk> wrote:
>
> On Fri, Sep 6, 2013 at 10:02 PM, Henrik Nordström
> <henrik at henriknordstrom.net> wrote:
>> tis 2013-09-03 klockan 08:21 +0000 skrev joem:
>>
>>> Use RS232-USB converters that require only 3 pins to operate; and which
>>> operate down to 3.3V for its input.
>>
>> Does not help in this case. I can reproduce AXP chip hang on most of my
>> boards with using only verified (and high quality) 3.3V UART cable if RX
>> pin is connected before applying power to the board.  Some boards are
>> less sensitive than others but not 100%.
>>
>> reproduced on A10, A10s, A13, A20.
>>
>> Only the boards with a diode and internal pullup on the RX pin surives
>> 100%.
>>
>>                     3.3V pullup to board 3.3v power rail (NOT UART cable)
>>                       |
>> UART CABLE TX   --|<--+-- AW SoC RX pin
>>                 diode
>>
>> This very effectively stops any current from the UART cable TX pin to enter the CPU.
>
>  that looks distinctly.... odd!  like it's going to entirely block
> transmit.  but then i realised if Tx goes to 0V that current flows ...
>
>  ok do you mean 3.3V with a resistor?  so the connection goes:
>
>  3.3V - resistor (e.g. 1k) - AW-Rx & also off to diode >| - UART Tx
>
> like that?  that would almost make sense, except that Tx had better
> pull down an extra 0.7v below the logic level expected by the Rx pin
> otherwise not a lot's going to happen.
>
> is that actually the case?  do 3.3V UART cables pull down to 0V (or
> close to it)?
>
> ---
> This entire thread sounds wrong.

 joe: it's not the A20 that's the problem, it's the AXP209 Power
Management Integrated Circuit.

 when the AXP209 Power Management Integrated Circuit is fed 3.3V into
its pins whilst it is powered off, it can get damaged.

 absolutely nothing to do with the CPU whatsoever.

l.