From: Greim, Michael ([email protected])
Date: Mon Nov 29 1999 - 05:09:22 PST
As you have seen from some of the posts, there are a number
of good ways to skin this cat. Either a single or multiple
series terminations can be employed to yield good results.
What will work best will depend on loading and trace routing
on each of the legs. Remember, series and thevenin termination
together is generally not a good idea. I have personally seen
better results using one series terminator per leg when the legs
have dis-similar routing and/or loading. However, your mileage
I can't stress enough the value of a good simulator on less
traditional net topologies such as stars, Hs, other (please
take the term traditional loosely). In many cases you may
be surprised with the effectiveness of the results.
You should also be careful with driver loading as multiple
legs on 50 ohm layers will load things down quite quickly if
you aren't watching for it. Many a time I have seen someone
substitute in a high drive driver as a place holder for a model
they were waiting on (say a 4 to 8 mA FPGA/ASIC output).
receipt of another model. Needless to say, you will see
radically different performance when the correct driver is
substituted in. Unfortunately, I have run into a couple of
folks who discovered this difference when they got in the
lab. (Why the would do this is a story for another time).
The bounds of Time, Space or Mechanics should never stand
in the way of a perfectly good idea.......
The time is gone, The email's over, thought I'd
something more to say.........
Michael C. Greim Consulting Engineer
Mercury Computer Systems, Inc email: [email protected]
199 Riverneck Road V: 978-256-0052/x1607
Chelmsford, MA 01824-2820 F: 978-256-4778
> -----Original Message-----
> From: Abe Riazi [SMTP:[email protected]]
> Sent: Saturday, November 27, 1999 1:37 AM
> To: '[email protected]'
> Subject: RE: [SI-LIST] : Star connection termination
> Hi Massimo:
> Series termination is an excellent choice for the star topolgy.
> When the branches are closely matched in delay and impedance, a single
> resistor can be used near the driver. The value of this resistor Rt is
> related to the trace impedance Zo, number of branches N, and the driver
> impedance Ro via:
> Rt = (Zo/N) - Ro
> If the branches are not well matched then one resistor should be
> employed for each branch, becasue using one resistor will be
> ineffective. The value of each branch terminator Rn can be calculated
> from the branch trace impedance Zn and the driver impedance Ro by means
> of the equation:
> Rn = Zn - Ro
> The following publication includes a nice discussion of this topic:
> S. Kuffer, and K. Crisafulli, "Terminating Traces on High Speed
> PCBs", Printed Circuit Design, March 1988, pp. 24 -27.
> I hope you find this information useful.
> Abe Riazi
> [email protected]
> [email protected][SMTP:[email protected]]
> >Sent: Friday, November 26, 1999 7:56 AM
> >To: [email protected]
> >Subject: [SI-LIST] : Star connection termination
> >Does anybody out of there have any suggestion about the following basic,
> >really important and frequent, problem: how a star-point (a point from
> >many traces going to other ICs start) must be series-terminated?
> >Do one have to put a resistance equal to the characteristic impedance of
> >trace before the point from which the traces start? Or is it better to
> >insert a
> >resistance on each line coming up from the star-point? In the latter
> >is the best value for the resistances?
> >Thanks for any help
> >ESAOTE S.p.A. Massimo Polignano
> >Research & Product Development Design Quality Control Mngr
> >Via di Caciolle,15 tel:+39.055.4229402
> >I- 50127 Florence fax:+39.055.4223305
> > e-mail: [email protected]
> >**** To unsubscribe from si-list: send e-mail to
> [email protected]
> >In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP.
> >si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
> **** To unsubscribe from si-list: send e-mail to
> [email protected] In the BODY of message put: UNSUBSCRIBE
> si-list, for more help, put HELP.
> si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
**** To unsubscribe from si-list: send e-mail to [email protected] In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
This archive was generated by hypermail 2b29 : Tue Feb 29 2000 - 11:39:03 PST