[Arm-netbook] KiCAD based A10 break out / prototyping PCB finished
jm
joem at martindale-electric.co.uk
Wed Feb 6 09:41:30 GMT 2013
On Wed, 2013-02-06 at 12:32 +0530, Ajith Kumar wrote:
>
>
> On Tue, Feb 5, 2013 at 10:19 PM, jm <joem at martindale-electric.co.uk>
> wrote:
> Hi,
>
> KiCAD based A10 break out / prototyping PCB is finished.
>
> This is a 6 layer A10 prototyping / break out board with all
> the pins
> except ground lines brought out to pads from where it can be
> wired to
> make any kind of device.
>
> Great work. Would it be better to have the power supply controller,
> like AXP209, also onboard. In fact I do not whether we can power A10
> without something like that.
>
> ajith
Firstly Ajith, anyone can download KiCAD
http://www.kicad-pcb.org/display/KICAD/KiCad+EDA+Software+Suite
learn it, and do the modifications and upload it, make it, sell it etc.
Its all GPL'd - so no need to wait around for approval
from anybody. :-)
There is another variant being planned with LCD + micro SD + 200 pin
SO-DIMM + DRAM + USB + RS232 + decoupling capacitors.
The power chip as far as I know is not 100% essential, because again as
far as I know, it then requires the software to be working to control
it. (Anyone who knows better can correct me on this.)
Still having said that, it is probably better to have it on the board so
that it can be wired in as needed. Hardly any product will ship without
that chip. The CN new year holidays are still on going, so I don't
plan to make the board for 2 to 3 weeks yet, and in that time, I might
be able to add in the power chip - unless someone beats me to it and
uploads something we can all use. :-)
The break out board will require a lot of decoupling capacitors
to make the board work well.
Having looked at various boards and the frequency with which they
crash, my suspicion is that the decoupling capacitors used are under
rated. So it is an opportunity to fit tantalums everywhere
with 1uF to 10uF rating and clock it to higher frequencies to
see if crashing can be limited or eliminated.
We know the A10 chip is under NDA which means the engineers who work
with these chips can call out for help on issues like this because they
then have to describe the chip and break the NDA.
Doh!
So the designs they churn out are sterile without all the proper in
depth testing needed to make robust PCBs. The break out PCB is a chance
for electronics engineers to go do some extreme testing with it.
If you have long wires, you should twist them with a ground wire
for better high frequency operation. Also twisting them makes the wires
equilength (*cough*)!!
Sorry, I always poke fun at the equilength bus camp because I don't feel
as though they understand anything about electronics when someone says a
bus has to be equilength.
It is most likely that bus wires and/or power wires have to be twisted
to prevent surge propagation. Equilength as described in the data sheets
as far as my engineering understanding goes, makes it worse because all
the energy will get radiated away on parallel wires, while if each bus
wire is twisted with a ground wire, it will reduce that likely hood.
Also twisting one wire with another makes it automatically equilength
even on PCB tracks. So a pair of USB data wires should twist around each
other on PCB tracks which means they are close together as they should
be to reduce emissions, and it makes them nearly equilength by default!
Simple rules like that all get forgotten / lost and wasted when NDAs
prevent engineers from discussing their designs. But KiCAD is here, and
we can share designs and make best designs.
Many of the Wits-Tech and Cubie designs are power wasting designs that
may have just worked for somebody at some time. Again because of NDAs,
the reasons for failure don't get discussed because discussing it in
open forums require the NDA to be broken, and so something half baked
gets made and propagated.
Hopefully we can address all those issues step by step and make best
EOMA / SoM / Internet of Things boards with KiCAD.
>
>
>
> The Gerbers are included, but you can modify and generate your
> own from
> KiCAD. If you build and try to use it, the ground at the back
> needs a
> direct connection with a thick wire as there is not enough
> ways to get
> fat tracks into that region with this arrangement of wiring.
>
> The aim of the board is to allow a slow speed operation of the
> A10 as a
> standard microprocessor with some 400 GPIO pins to hand. Many
> items such
> as SDCard, real time clock, RS232, and LCD should work
> at low speeds if you were to add it on by connecting it from a
> breadboard to this board.
> Linux won't run without a lot of RAM, so you may be able to
> wire
> something up and run
> it at low speeds if you plan running Linux from say the
> SDCard.
> (A future derivative will have the RAM chips to eliminate the
> wiring.)
>
> An ideal situation is to get something to boot from SDCard and
> flash a
> simple LED. Providing that part is fully documented, the
> software
> together with this board becomes a useful stepping stone to
> learn to build more complex derivative projects.
>
> 05-February 2013
> http://www.gplsquared.com/SoM2/SoM2_2013_02_05.tar.bz2
> (finished)
> http://www.gplsquared.com/SoM2/SoM2.html (web page)
>
>
>
>
> _______________________________________________
> arm-netbook mailing list arm-netbook at lists.phcomp.co.uk
> http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook
> Send large attachments to arm-netbook at files.phcomp.co.uk
>
>
>
> --
> Dr. Ajith Kumar B.P.
> Scientist SH
> Inter-University Accelerator Centre
> Aruna Asaf Ali Marg
> New Delhi 110067
> www.iuac.res.in
> Ph: (off) 91 11 26893955 (Ext.230)
> (res)91 11 26897867
> (mob) 91 9868150852
> _______________________________________________
> arm-netbook mailing list arm-netbook at lists.phcomp.co.uk
> http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook
> Send large attachments to arm-netbook at files.phcomp.co.uk
More information about the arm-netbook
mailing list