Re: [SI-LIST] : PECL oscillator termination

About this list Date view Thread view Subject view Author view

From: Michael Nudelman (mnudelman@tellium.com)
Date: Thu Nov 09 2000 - 19:27:48 PST


Tae,

If they are talking of series termination, than:
I favour parallel termination for very-hi-speed lines for the reason that for the
series (source) one you have to know exactly what your driver's output impedance is:
the resistor should be the difference between it and the trace's impedance.
Otherwise you are up for overshoots or vice versa - lost amplitude.
Which in addition is complicated by the fact that the output impedance of, say,
emitter follower changes dynamically.

Now, some people advocate double termination - source and end. You will have nice
clean signal, but will loose amplitude; it is used for receivers with low working
input thershold.

For a PECL clock, especially point-to-point (is it? what frequency?) I never saw
source termination, though I cannot claim I saw everything.
I use what I described earlier, very typical "cut'n'paste" circuit, that allows
connection to any ECL receiver.
Resistor 330 Ohm to GND (current sink), that a cap, and than diffmode termination
(100 Ohm for 100 Ohm diffpair between 50-Ohm traces), since I don't have common mode
problems.

SInce you use two PECLs and if you can use DC connection (can you?) this is the
recommended termination (Dr. Johnson advocates this one too):
Assume diff. imp. 100Ohm.

Straight connection driver-to-receiver. In the end: each line is terminated by 50Ohm
resistor; ends of the resistors come together to one capacitor (for clock, since it
is balanced signal, it can be large; otherwise - choose one you like for AC
termination for your kind'a signal); the capacitor is then grounded. Also, in
parallel to that capacitor, the resistor equal 330 Ohm minus 50 Ohm (is it 280? I
need a calculator for this) is connected, thus providing needed DC-bias.

Mike.

Tae-Kwang Jeon wrote:

> Hi Mike,
>
> >>
> >> #3 In the case of #2, the resistors are supposed to be grounded, then what
> >> if I make them in series?
> >>
> >
> >In series with what? WIth the praces of the diffpair?
> >The resistors are there to provide bias. Make them in series with the pair -
> and
> >you divide your signal between your resistor and the trx-line. The series
> >termination does just that, but then, if you need AC-coupling, wher's your
> bias?
> >YOu will open can of worms, when everything is simple - connect 330 Ohm to GND,
> >run your traces, put the caps in series (where you like; some people have
> >different preferences - beginning, end) and terminate - either for diff. mode
> >(diff. impedance resistor between traces in the end) of for common mode (one
> >trace impedance resistor to GND from each trace in the end).
> >
> >Mike.
> >
>
> Yes, I agree with you on the DC-biasing resistor.
> The reason for asking #3 is that I read an article the other day from Web -
> I don't remember what the site was right now - saying resistor connection in
> series with the cap could improve the PECL termination. So I was wondering
> if it really works because I'm working on PECL-to-PECL interface for my
> system.
>
> Regards,
> TK
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> TaeKwang Jeon, LSI Logic Corp.
> 1525 McCarthy Blvd Phone: (408) 433-6818
> MS G-715 Fax : (408) 433-2840
> Milpitas, CA 95035 Email: tkjeon@lsil.com
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
> **** 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
> ****

**** 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 : Tue May 08 2001 - 14:30:02 PDT