[Arm-netbook] [EOMA68] RS232 on SoCs

[joem]

On Tue, 2013-09-03 at 11:19 +0100, Philip Hands wrote:
> joem <joem at martindale-electric.co.uk> writes:
> 
> > On Tue, 2013-09-03 at 09:10 +0200, Vladimir Pantelic wrote:
> >> On 08/31/2013 04:59 PM, Henrik Nordström wrote:
> >> 
> >> >
> >> > It's not about excessive current. It's about powering the chip via I/O
> >> > pins instead of power pins. Now the Allwinner chips do not really power
> >> > up via the I/O pins, but there is at least sufficient power leakage back
> >> > to the power rails to cause problems for the AXP power management
> >> > controller chip, and maybe some internal I/O functions as well.
> >> 
> >> and a 1k series resistor on the UART line is not enough to stop this?
> >
> > No - as soon as you exceed 3.3V, the IO line's protection diodes start
> > conducting and you will end up powering the CPU through the power line
> > from the UART and the chip can no longer function properly.
> >
> > Use RS232-USB converters that require only 3 pins to operate; and which
> > operate down to 3.3V for its input.
> >
> > http://www.gplsquared.com/eoma_boot/eoma_boot.html - photo of what looks
> > like a Silab CP2102 chip based dirt cheap RS232-USB converter
> >
> > Works very well. And that particular model has LED to tell you if the
> > board is powered up.
> >
> > They used to be common - but I don't see them as often
> > http://www.aliexpress.com/item/Free-shipping-2pcs-New-build-in-CP2102-USB-To-TTL-RS232-COM-Converter-Module-6-pin/581643413.html
> >
> > The latest seems to be PL2303 chip based designs - but I can't vouch for
> > those ones, but someone else who uses it might be able to:
> > http://www.aliexpress.com/item/USB-To-RS232-TTL-PL2303HX-Auto-Converter-Module-Converter-Adapter-5V-3-3V-Output/691986109.html
> 
> I have one of these:
> 
>   http://www.ebay.co.uk/itm/USB-TO-RS232-TTL-PL2303-CONVERTER-FOR-RASPBERRY-Pi-ADAPTER-CONSOLE-CABLE-/121070687261?pt=UK_Computing_Other_Computing_Networking&hash=item1c3060141d
> 
> it worked well on the first A10 board I had, when connected via a
> breadboard.  Now that I've been using it with the MEB it only
> occasionally works, perhaps dependent on the order ////i plug things in,
> but more probably at random.


Just repeat what I say earlier in case it applies.
Avoid connecting micro USB serial cable at any time
and bin that cable if you can (for now).
The effects reported are exactly what I was getting
because that cable was allowing power to reach the EOMA
and MEB and leaving it in an unknown state.
This seemingly trivial problem gets
compounded with having no LEDs to either board
to warn when they are powered up. (I wired LEDs to my boards.)
When working with new hardware, its easy to get
caught out in the excitement and fail to realise
the boards are not power cycling properly if the micro USB
cable is coincidentally connected.

Although device can work
off the USB alone, under testing I find sending
character streams via the RS232 while only on USB power
will crash the A20 probably because it wants to do something
and that something needs more power than 500mA limited
micro USB cable source can provide.