Hi Garst,
The devil is usually in the details, when it comes to good layout.
Make sure the gound plane really wraps around everything it can near the
MCU. Make it as interconnected as possible, to minimise the effects
breaks in the plane. Keep the 0.1uF *very* close to the relevant pins.
The 10uF would be better being near the MCU, it just isn't as critical
as getting the 0.1uF really close. Putting it on another board doesn't
seem the best choice. Don't be fooled into trying to make an analogue
ground plane and a digital ground plane, with only one point of contact
between them. That very rarely works, because it is rare for there to be
genuinely one point of contact. On those rare ocassions separate ground
planes connected very close to the MCU can be very good. One really
intermeshed plane is normally far better.
Regards,
Steve
Garst R. Reese wrote:
Thanks Rick,
That is exactly the info I needed. Normally I would cut a lot for a
reply, but this is too important.
I have two boards, one with the analog stuff and a digital board with
the msp, an mmc card, an irda chip, and an LCD. Both have ground planes.
They are tied together with a molex .5mm flex cable. I bring out AVss
and DVss and tie them together on the analog board, which supplies AVcc.
I put a .1uF across AVss and AVcc on the digital board and put the 10uF
+ another .1uF cap from AVcc to analog ground on the analog board at the
connector. DVcc is supplied by a TI TPS79730 on the digital board. AVcc
is supplied by a MAX6129EUK30 .04% 3V reference on the analog board, so
they should be pretty close. The peripherals have separate supplies.
Rick Jenkins wrote:
On Sunday 10 October 2004 22:06, Garst R. Reese wrote:
The connection examples for these two supply voltages all show them
externally tied together. Can they actually be separate supplies with
different voltages? I assumed they could be, but got worried when I
could not find an example.
If you are actually using the analog facilities on the chip, performance will
be poor if these are merely tied. My experience is that careful filtering of
AVCC is necessary, both for the comparator and for the ADC12. Poor
decoupling of DVCC is a wonderful way to get exotic, erratic, and
non-repeatable bugs, as a look through the archives of this mailing list
would show. If you don't need the analogue stuff, tie the supplies together
and dcouple them as one.
The AVCC and DVCC voltages should be nominally equal, though a difference of
a few tens of millivolts seems not to matter.
Murata, AVX, and Kemet all show 0805 10uF 6V 10% Ceramic capacitors.
Would these require the extra 0.1uF?
Yes. I've tried it.
The large ceramics have enough intrinsic inductance to resonate at a
surprisingly low frequency, and the 0.1uF takes care of high-frequency
decoupling. No free ride here, I'm afraid. The 0.1uF must be as close to
the chip as you can get it, but a centimeter or so of total lead length on
the 10uF is OK. This can be used to advantage to "isolate" the two
capacitors, so that you don't need separate traces to the chip pins, and it
eases layout too.
