Re: [SI-LIST] : Excessive clock overshoot]

Dennis Tomlinson (det@tellabs.com)
Thu, 14 May 1998 12:48:21 -0500

This is a multi-part message in MIME format.
--------------34DB5EF93C7F110C4C52B56D
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

--------------34DB5EF93C7F110C4C52B56D
Content-Type: message/rfc822
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

Message-ID: <355B2022.2EED0E7@tellabs.com>
Date: Thu, 14 May 1998 11:47:30 -0500
From: Dennis Tomlinson <det@tellabs.com>
X-Mailer: Mozilla 4.04 [en] (X11; I; SunOS 5.5.1 sun4u)
MIME-Version: 1.0
To: fabrizio=zanella%eng%emchop1@fishbowl02.lss.emc.com
Subject: Re: [SI-LIST] : Excessive clock overshoot
References: <vines.8VJ8+h7iKpA@fishbowl02.emc.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

fabrizio zanella wrote:

> I apologize for not giving enough information on yesterday's email about
> the FCT clock overshoot. The termination at each end of the bus for these
> clock lines is thevenin 100-5V/220-gnd. The loaded round trip delay is
> about 4.4ns, so 2.2ns one way.
> Hope this helps, thank you very much for the comments thus far.
> The comment from Andy Ingraham about the FCT clock resonating at 45MHz does
> make sense because on a receiver card at the clock input we see
> oscillations/non-monotonic signals.
> Would the resonance be caused by the internal clock circuitry or the load
> it's driving?

The resonance is due to transmission line effects - aggrivated/complicated by
all the loads hanging off of it (including the driver). As the clock frequency

increases, it's half period approaches one round trip delay on your clock net.

Since your load is under terminated (about 69 Ohms - causing a load reflection
of
+0.47) and your source is over terminated (roughly 10 Ohms - causing a
source reflection of -0.43), you're experiencing the T-line equivalent of
oscillation known as standing waves. Reflected waves are arriving at the
driver
in phase with the driver output signal swings. Then, about 1/4 cycle later,
they add
in phase at the last receiver (and in part, bounce back toward the driver,
etc.).

The fix is to decrease the reflections. One crude but effective method might
be
to add a clamp diode to the upper rail at the receiver furthest from the
driver.

Another method would be to decrease the resistor values in your Thevenin
termination to something closer to a 25 Ohm equivalent. The strength of
your driver will probably limit how low you can go.

Another method you could try is AC termination. On long clock nets like
yours,
using a bypass capacitor in series with 30 to 40 Ohms might give enough
damping.
The bypass will charge to the average voltage, thus emulating a balanced
Thevenin termination. (Note: there's no power to be saved by using a smaller
capacitor on a net this long at this frequency)

Your 100/220 termination gives 3.44V for V-Thevenin. Was there a reason for
this termination voltage? If R-Thevenin is decreased to 25 Ohms, but
V-Thevenin remains the same, it will take about 130 mA. to drive a logic
'0' into this load.

Also, was there a reason you're using 25 Ohm routes instead of something
higher? Lines with Z0 this low are hard to drive. Switching currents required

of the driver must be over 125 mA. to change the T-line's state.

Best of luck,

Dennis

>
>
> Regards, Fabrizio Zanella.
> EMC Corporation
> fzanella@emc.com
> -------------
> Original Text
> From: "D. C. Sessions" <dc.sessions@vlsi.com>, on 5/13/98 5:23 PM:
> To: smtp@Eng@EMCHOP1["SI-List" <si-list@silab.Eng.Sun.COM>]
> Cc: smtp@Eng@EMCHOP1["SI-List" <si-list@silab.Eng.Sun.COM>]
>
> fabrizio zanella wrote:
> >
> > I have a question regarding an FCT clock (TTL levels) driving a heavily
> > loaded backplane. On the driver pin we see excessive overshoot on the
> L-H
> > transition which increases as we increase the clock frequency. This
> > overshoot goes from 5V at 33MHz to 6.5-7.0V at 45MHz. The stub impedance
> > is 75 ohms, backplane impedance 25 ohms loaded. There is a clamping
> diode
> > on the H-L side but not on the L-H. The H-L side does not have any
> > undershoot.
> > I have asked the manufacturer and they have never seen this phenomena,
> nor
> > do they have an explanation for it.
> > Any ideas on what could be causing this?
>
> 1) Is the line unterminated?
> 2) Is the loaded one-way delay anything near 2.7ns?
>
> --
> D. C. Sessions
> dc.sessions@tempe.vlsi.com

--------------34DB5EF93C7F110C4C52B56D--