[SI-LIST] Re: On-chip Terminations
- From: "ruston, matt" <ruston_matt@xxxxxxx>
- To: "'si-list@xxxxxxxxxxxxx'" <si-list@xxxxxxxxxxxxx>
- Date: Fri, 13 Jul 2001 14:58:15 -0400
Mark:
Hi. I have to disagree. There still has to be a spec on how well the chip
can match the reference. If I have 50 ohm +/- 1% references, and Xcite
matches that to within +/- 20 ohms, I'm not going to be very happy with
Xcite, no matter if the matching is feedback, adaptive, digital control, or
anything else that you care to call it. Adaptive does not make it accurate.
You say that since I can use the resistors in a series or up/down or pullup
type configuration, etc., that it "... makes characterization from a percent
tolerance standpoint a metric that's not very useful." Why can't you specify
the overall system tolerance (i.e. if used as a series output term, then the
impedance of the driver AND sterm is within +/- 5% of the reference). In
parallel mode, +/- X%, etc. I would like to see some type of spec (tested or
guaranteed by design) or design goals at least. A thorough application note
would be helpful too (explain the process/algorithms of matching, how many
LSBs can the error be, how to take package and on-chip interconnect
resistance into account, etc.).
Other specs would be helpful (power ratings, voltage linearity, etc.).
Finally, I believe the Ibis model inputs are only for matching into a 50
ohm system. What happens if my trace impedances are different; will there be
Ibis support for these? Also, Hyperlinx and spice sims have to use an
external resistor for these input sims. Will this be fixed?
I think that on-chip termination is a beautiful thing, but I get nervous
when everywhere I look, I can't find specs or schematics; all I find are
Macromedia Shockwave presentations that espouse the same benefits over and
over. More meat please.
Regards,
Matt
-----Original Message-----
From: Mark Alexander [mailto:Mark.Alexander@xxxxxxxxxx]
Sent: Friday, July 13, 2001 1:54 PM
To: si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: On-chip Terminations
Matt,
You're right -- the Xcite resistors have a feedback mechanism which matches
them
to external reference resistors. Therefore they're adaptive over
temperature
and voltage, and maintain a good impedance match over all corners (process,
temperature, voltage).
We do not give a tolerance value for these terminations because it doesn't
make
sense to. Since there are a number of IO standards that have on-chip
termination in each IO block, the circuitry is used in a number of different
configurations. This combined with the active nature of the implementation
(FETs instead of fixed resistors), makes characterization from a percent
tolerance standpoint a metric that's not very useful.
Instead, an accurate picture of the behavior of these on-chip resistors can
best
be obtained from simulation. The IBIS models contain the information for
simulation of the Xcite IO. If anyone would like example simulations of
this,
drop me an email.
There's one other point I'd like to make in reference to the tolerance
issue.
When using even a precision external termination resistor, say 1%, changes
in
the impedance of the driver with temperature and voltage can cause much more
than 1% of impedance mismatch in the channel, leading to a wide range of
rise/fall times and possible SI problems. Since Xcite compensates the
driver as
well as the termination, edges are very consistent and well-matched.
If anyone is interested in additional information on Xcite, there is a paper
available that I delivered at PCB West. It's on the Proceedings CD, or drop
me
an email and I can send it to you.
-mark
"ruston, matt" wrote:
> Patrick:
>
> Hi. I've been looking into the Xcite terminations too and cannot find a
> tolerance. The "resistors" are really multiple parallel FET transistors
with
> different impedances that can get turned on or off to change the overall
> resistance. It should be much better than an on-chip implanted,
> poly-silicon, or pinch resistor (15%+). Xcite uses external reference
> resistors to tweak the on-chip FETs. It is done at startup to dial into
the
> proper range, then it is done in use to make minor adjustments due to
> voltage and temperature changes. You already know this, just filling in
> details for the others.
>
> Pat Z. was correct to say that a major win with these on-chip terms is
stub
> elimination. It also frees up a lot of PCB real estate and eliminates tons
> of vias. They are not just for OC48 (don't know why that was thrown in).
> They can be used on any I/O (in Virtex-II line) and come in several
> topologies/configs.
>
> Still, I'd like to know tolerance, power handling, etc. And no, I'm not a
> Xilinx salesman.
>
> Regards,
>
> Matt
>
> -----Original Message-----
> From: Chuck Hill [mailto:chuckh@xxxxxxxxxxx]
> Sent: Friday, July 13, 2001 11:34 AM
> To: si-list@xxxxxxxxxxxxx; 'si-list@xxxxxxxxxxxxx'
> Subject: [SI-LIST] Re: On-chip Terminations
>
> Good question.
>
> Also, what is the temperature coefficient of these resistors? I'll bet
its
> not the 50ppm/C that the discrete resistors have.
>
> Chuck
>
> At 09:13 AM 7/13/01, Patrick Francq wrote:
> >Greetings,
> > I recently read a brief paper on Xilinx's XCITE
technology.
> >
> >The only problem is I couldn't find any tolerance values for these
on-chip
> >resistors.
> >
> >Discrete resistors have a 1% tolerance.
> >Buried resistors have a 10-15% tolerance.
> >On-chip resistors have a ??? tolerance.
> >
> >Does anybody have a "ball-park" number on this?
> >
> >Thanks all,
> >Patrick
> >
> >
> >
> >
> >Patrick Francq
> >Hardware Designer / SI Specialist <http://intranet/home.asp>
> >
> ><?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office"
> />
> ><mailto:pfrancq@xxxxxxxxxxxxx> pfrancq@xxxxxxxxxxxxx <
> ><mailto:pfrancq@xxxxxxxxxxxxx> mailto:pfrancq@xxxxxxxxxxxxx>
> >
> > <http://intranet/home.asp>
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >------------------------------------------------------------------
> >To unsubscribe from si-list:
> >si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
> >For help:
> >si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
> >
> >List archives are viewable at:
> > //www.freelists.org/archives/si-list
> >or at our remote archives:
> > http://groups.yahoo.com/group/si-list/messages
> >Old (prior to June 6, 2001) list archives are viewable at:
> > http://www.qsl.net/wb6tpu
> >
>
> ------------------------------------------------------------------
> To unsubscribe from si-list:
> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
> For help:
> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>
> List archives are viewable at:
> //www.freelists.org/archives/si-list
> or at our remote archives:
> http://groups.yahoo.com/group/si-list/messages
> Old (prior to June 6, 2001) list archives are viewable at:
> http://www.qsl.net/wb6tpu
>
> ------------------------------------------------------------------
> To unsubscribe from si-list:
> si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
> For help:
> si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>
> List archives are viewable at:
> //www.freelists.org/archives/si-list
> or at our remote archives:
> http://groups.yahoo.com/group/si-list/messages
> Old (prior to June 6, 2001) list archives are viewable at:
> http://www.qsl.net/wb6tpu
>
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
