avr-gcc-list
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [avr-gcc-list] supply pins on atmega32


From: James Hsu
Subject: Re: [avr-gcc-list] supply pins on atmega32
Date: Wed, 15 Sep 2004 04:37:22 -0700
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040113

Reza Naima wrote:

the "final" version, however.

One other question regarding bypass caps -- So, as there are several VCC pins, should there be multiple bypass caps so that they are all close to the relative pins - or would it be sufficient to put them all next one one another, and just run traces between all the VCC pins? The TQFP IC is only 11mm square, so it seems like a bit of an overkill to put the bypass caps all around (if distance is the important factor).

I'm also expecting some lower frequency EMF noise -- like around 1-10khz, so I'm using a .1 and 1 uF caps. I feel like this is a bit of "black magic" as I've not found any references which correlates the capacitance of a bypass cap to the frequency range in which it is effective. I'm including a link to an image of the PCB w/ the bypass caps and xtal placement. I would love some criticism as how to improve the layout w.r.t. the caps. C1 is 1uF and C11 is 0.1uF. Ignore C5.
http://www.reza.net/misc/caps.jpg

<---------------------------------------------------------------------->


Here's a nice introduction article. If you check other capacitor
manufacturers, you can find more details.

https://www.avxcorp.com/docs/techinfo/mlcbypas.pdf

In general there are a lot of "fudge factors" involved in figuring out
some optimal capacitor strategy. A couple are listed in the PDF above
- page 8 goes into detail about figuring out the frequencies of the
decoupling network.  You should keep the cap next to the pin its
decoupling though. Adding in a nice big loop to a power rail is
generally a bad idea.

Lower frequency noise - KHz level probably needs tantalum caps.
Ceramics are low inductance, but they have other artifacts - like
losing half their capacitance when being used at the rated voltage.
If you have a lot of KHz level power supply ripple, you might need a
big chunk of capacitance to filter that out.

..If it matters. Digital circuits are pretty resilient, so everything
should work, but if you need to think about it, you might as well do
the "proper" layout in the final version. (Analog is an entirely
diffrent matter...)

James


reply via email to

[Prev in Thread] Current Thread [Next in Thread]