Re: [SI-LIST] : problem with crystal

About this list Date view Thread view Subject view Author view

From: sweir (weirsp@a.crl.com)
Date: Fri Feb 18 2000 - 02:08:52 PST


Yasser,

You do not say what the board stack-up looks like, how your are bypassing
your power, or if you have any fast signals from the MCU running around the
board.

Since, you are seeing the oscillator superimposed on your power network,
then your power network's impedance is too high, or you have made a loop
antenna out of your scope probe, or both. The power supply problem is well
above the frequency where electrolytics or tantalums can help. They are
good for taking up the slack from the voltage regulator at some thousands
or tens of thousands of kiloHertz up to about a megaHertz or so.

The two schools for high frequency bypass are: 1. "carpet bomb" with junky
100 nF caps, or 2. use closely spaced, but different values, preferably
with quality, low ESR caps. Both schools seek to get a flat, low impedance
without antiresonance problems over a wide frequency. The folks at SUN
have had great experience with the latter on bleeding-edge hardware. You
can refer to papers that Larry Miller and Ray Anderson have authored and
are available on WWW.

A typical 100 nF SMT capacitor, depending on how it is mounted, and what
your power and ground planes look like typically resonates between 8 and 12
MHz. You probably need to cover at least up to your fifth harmonic, so
some number of 10 nF, ( 24 - 36 MHz ) and 3.3 nF ( 40 - 60 MHz ) capacitors
are called for. The (2) school will argue that even this separation of a
decade is excessive. However, your board sounds small, so you may get away
with this. The quantity of capacitors depends on the size of your board
and the AC current draw. If you have at-least four layer construction,
then if you place the VCC and GND planes very close together, you will get
high frequency bypass above the point where the ceramic capacitors cease to
be effective.

You may also wish to consider a resistive filter, either true R-C, or a
ferrite and capacitor at the power connection to your A-D. Avoid using an
LC filter, unless you understand how to damp it properly.

There are other possible culprits to your A-D problem. Make sure that you
understand where your signal and return currents are flowing. If you are
using a switching power-supply, make sure that the magnetics are not
radiating into any passives or traces in the A-D section.

Panasonic, and Murata both manufacture resonators up to about
50MHz. Resonators are crystals, but they are cut differently, and are not
as precise. 1% to 3% accuracies are typical, versus 0.01% for a typical
crystal. Resonators are usually used to save money where the absolute
frequency accuracy is not very important. Typically, resonators are found
in consumer and automotive electronics.

Regards,

Steve.
At 09:21 AM 2/18/00 +0200, you wrote:
>Dear Sir :
>
>I have two questions :
>
>1- I use a crystal 12 MHz with my design of MCU and the program is
>internal in my MCU and no external RAM. I see on my supply of the board
>the 12 MHz of superimposed on it and in other times i see the second
>harmonic of the 12 MHz. How i can supress this frequency because i makes a
>problem with the A/D on my board. I need also to tell you that i follow
>the recomendations of connection the crystal (it's as close as possible to
>the MCU, crystal capacitors is near the crystal and connected direct to
>the MCU gnd and i made a ground polygoen around the crystal from the two
>layers, and i avoid to pass a track with high current near the crystal).
>What is the problem.
>
>2- I need to use a cermaic resonator crytal with built in capactance. Can
>you help me to new the difference between it and the crystal oscilator.
>Also i need to know the application for which the two is suitable.
>
>Can any one help me.
>
>with my best regards
>Yasser El-haddad

**** To unsubscribe from si-list or si-list-digest: send e-mail to majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Thu Apr 20 2000 - 11:35:06 PDT